Results publication

workflowr

• Summary
• Checks
• Past versions

Last updated: 2023-01-17

Checks: 7 0

Knit directory: emlr_obs_analysis/analysis/

This reproducible R Markdown analysis was created with workflowr (version 1.7.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

---

R Markdown file: up-to-date

Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(20210412)

The command set.seed(20210412) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Cache: none

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

File paths: relative

Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.

Repository version: 30b15de

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.

The results in this page were generated with repository version 30b15de. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:

Ignored files:
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    data/
    Ignored:    output/other/
    Ignored:    output/presentation/
    Ignored:    output/publication/

Untracked files:
    Untracked:  code/RScript_Jens_globalMap.R
    Untracked:  code/choky-barb_reprex.R
    Untracked:  code/choky-barb_reprex.md
    Untracked:  code/choky-barb_reprex_files/
    Untracked:  code/raster_dataframe.R
    Untracked:  code/results_publication_backup_incl_ensemble_uncertainty_20221111.Rmd
    Untracked:  code/tidyterra_spatRaster.R
    Untracked:  code/write_ocean_raster_netcdf.R
    Untracked:  ok-coqui_reprex_files/

Unstaged changes:
    Modified:   analysis/G19_column_inventories.Rmd
    Deleted:    analysis/MLR_target_budgets.Rmd
    Deleted:    analysis/MLR_target_column_inventories.Rmd
    Deleted:    analysis/MLR_target_zonal_sections.Rmd
    Modified:   analysis/_site.yml
    Modified:   analysis/basics.Rmd
    Modified:   code/Workflowr_project_managment.R

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

---

These are the previous versions of the repository in which changes were made to the R Markdown (analysis/results_publication.Rmd) and HTML (docs/results_publication.html) files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view the files as they were in that past version.

File	Version	Author	Date	Message
Rmd	30b15de	jens-daniel-mueller	2023-01-17	changed label
html	fdf5f22	jens-daniel-mueller	2023-01-11	Build site.
Rmd	657f205	jens-daniel-mueller	2023-01-11	revised plots for talk
html	2cca4eb	jens-daniel-mueller	2023-01-10	Build site.
Rmd	532c24a	jens-daniel-mueller	2023-01-10	revised plots
html	8ded995	jens-daniel-mueller	2023-01-05	Build site.
Rmd	80fac85	jens-daniel-mueller	2023-01-05	converted all maps to robinson projection
html	b9d5a73	jens-daniel-mueller	2022-12-15	Build site.
Rmd	fe0def4	jens-daniel-mueller	2022-12-15	included robinson raster maps
html	0f2c9d5	jens-daniel-mueller	2022-12-08	Build site.
Rmd	ee5cdab	jens-daniel-mueller	2022-12-08	minor figure updates
html	445761e	jens-daniel-mueller	2022-12-02	Build site.
Rmd	ab1269e	jens-daniel-mueller	2022-12-02	GCB budget tables compiled
html	ef50aa6	jens-daniel-mueller	2022-11-30	Build site.
Rmd	104a756	jens-daniel-mueller	2022-11-29	refined BIM color scale
html	439cb85	jens-daniel-mueller	2022-11-29	Build site.
Rmd	c01f61e	jens-daniel-mueller	2022-11-29	refined BIM estimates
html	9048dc4	jens-daniel-mueller	2022-11-29	Build site.
Rmd	533e469	jens-daniel-mueller	2022-11-29	included BIM estimates
html	3a7bab0	jens-daniel-mueller	2022-11-28	Build site.
Rmd	5f54639	jens-daniel-mueller	2022-11-28	minor figure revisions
html	322b1c1	jens-daniel-mueller	2022-11-25	Build site.
Rmd	0447054	jens-daniel-mueller	2022-11-25	1 and 2 sigma uncertainty
html	491f8b1	jens-daniel-mueller	2022-11-25	Build site.
Rmd	6a7dfb4	jens-daniel-mueller	2022-11-25	2 sigma uncertainty
html	4da4d66	jens-daniel-mueller	2022-11-25	Build site.
Rmd	18f0c96	jens-daniel-mueller	2022-11-25	1 sigma uncertainty
html	534ee74	jens-daniel-mueller	2022-11-25	Build site.
Rmd	d682344	jens-daniel-mueller	2022-11-24	tested OceanSODA surface dcant based on regression trend
Rmd	9e119d5	jens-daniel-mueller	2022-11-23	revised penetration depth estimates, adapted new uncertainty estimates throughout
html	57d1a9b	jens-daniel-mueller	2022-11-17	Build site.
Rmd	a56474c	jens-daniel-mueller	2022-11-17	new emissions uncertainty
html	f39bfd4	jens-daniel-mueller	2022-11-17	Build site.
Rmd	71e9efe	jens-daniel-mueller	2022-11-17	converted scaling uncertainty contribution to 1 sigma
html	e924c80	jens-daniel-mueller	2022-11-17	Build site.
Rmd	cc118f6	jens-daniel-mueller	2022-11-17	revised additive uncertainty calculation
html	8e1702d	jens-daniel-mueller	2022-11-16	Build site.
Rmd	5cf1489	jens-daniel-mueller	2022-11-16	revised inventory uncertainty calculation
html	ea34027	jens-daniel-mueller	2022-11-15	Build site.
Rmd	be95f92	jens-daniel-mueller	2022-11-15	exported residual correlation plot
html	8ac0da0	jens-daniel-mueller	2022-11-15	Build site.
Rmd	63fb2fb	jens-daniel-mueller	2022-11-15	more units expressed per decade
html	5520de8	jens-daniel-mueller	2022-11-15	Build site.
Rmd	751a519	jens-daniel-mueller	2022-11-15	units expressed per decade
html	1dad51a	jens-daniel-mueller	2022-11-15	Build site.
Rmd	ab89db3	jens-daniel-mueller	2022-11-15	surface emlr extrapolated without surface obs
html	375300e	jens-daniel-mueller	2022-11-14	Build site.
Rmd	40f357d	jens-daniel-mueller	2022-11-14	included C* and DIC target cases
html	ebc85e2	jens-daniel-mueller	2022-11-14	Build site.
Rmd	beb709c	jens-daniel-mueller	2022-11-14	included C* and DIC target cases
html	cc337dd	jens-daniel-mueller	2022-11-11	Build site.
Rmd	c53fea0	jens-daniel-mueller	2022-11-11	rebuild website with woa18 clim
html	6edad9b	jens-daniel-mueller	2022-11-11	Build site.
Rmd	9937e55	jens-daniel-mueller	2022-11-11	with additive uncertainty estimates
html	16beb51	jens-daniel-mueller	2022-11-08	Build site.
Rmd	6276ead	jens-daniel-mueller	2022-11-08	run with all contribution, but exclude AT and reoccupation in inventories
html	b92657d	jens-daniel-mueller	2022-11-08	Build site.
Rmd	3f9fad2	jens-daniel-mueller	2022-11-08	run without C*(P only) and SD factor 2
html	0b939ca	jens-daniel-mueller	2022-11-08	Build site.
Rmd	5c705d7	jens-daniel-mueller	2022-11-08	run with C*(P only)
html	ec60f68	jens-daniel-mueller	2022-11-07	Build site.
html	7275091	jens-daniel-mueller	2022-11-05	Build site.
Rmd	ba55b15	jens-daniel-mueller	2022-11-05	include run with C*(P only)
html	99610ed	jens-daniel-mueller	2022-11-01	Build site.
Rmd	c972beb	jens-daniel-mueller	2022-11-01	updated reoccupation run
html	7060801	jens-daniel-mueller	2022-10-31	Build site.
html	ecb84c1	jens-daniel-mueller	2022-10-30	Build site.
html	2345935	jens-daniel-mueller	2022-10-29	Build site.
html	0b40366	jens-daniel-mueller	2022-10-26	Build site.
Rmd	0465788	jens-daniel-mueller	2022-10-26	implemented additive uncertainty assesment
html	91052ae	jens-daniel-mueller	2022-10-20	Build site.
Rmd	33486a1	jens-daniel-mueller	2022-10-20	updated plots
html	084a41c	jens-daniel-mueller	2022-10-13	Build site.
Rmd	ffef558	jens-daniel-mueller	2022-10-13	land sink assesment
html	46c163a	jens-daniel-mueller	2022-10-10	Build site.
Rmd	78161f9	jens-daniel-mueller	2022-10-10	revised figures
html	8105380	jens-daniel-mueller	2022-10-07	Build site.
Rmd	d663118	jens-daniel-mueller	2022-10-07	penetration depth analysis
html	e93fdfd	jens-daniel-mueller	2022-09-23	Build site.
Rmd	f01db23	jens-daniel-mueller	2022-09-23	added uncertainty to dcant maps and section
html	8dc5a3b	jens-daniel-mueller	2022-09-09	Build site.
Rmd	25bff3d	jens-daniel-mueller	2022-09-09	added global section location to maps
html	77e6a80	jens-daniel-mueller	2022-09-08	Build site.
Rmd	188dc63	jens-daniel-mueller	2022-09-08	added uncertainty to global section
html	7cf954e	jens-daniel-mueller	2022-09-08	Build site.
Rmd	6f53c28	jens-daniel-mueller	2022-09-08	column inventory uncertainty stippling added
html	051b911	jens-daniel-mueller	2022-09-07	Build site.
Rmd	c87415f	jens-daniel-mueller	2022-09-07	testrun integration 3000
html	75bd58c	jens-daniel-mueller	2022-09-07	Build site.
Rmd	1024099	jens-daniel-mueller	2022-09-07	testrun integration 1000
html	1c6de0d	jens-daniel-mueller	2022-09-07	Build site.
Rmd	4106aa4	jens-daniel-mueller	2022-09-07	testrun integration 10000
html	5ca8778	jens-daniel-mueller	2022-09-07	Build site.
Rmd	784849d	jens-daniel-mueller	2022-09-07	included uncertainty assesment for column inventories
html	4eb9ed2	jens-daniel-mueller	2022-08-29	Build site.
html	392f3d7	jens-daniel-mueller	2022-08-29	Build site.
Rmd	282421f	jens-daniel-mueller	2022-08-29	added contour line to global section
html	c975141	jens-daniel-mueller	2022-08-29	Build site.
Rmd	8f5c3c2	jens-daniel-mueller	2022-08-29	finalized global section
html	cf69673	jens-daniel-mueller	2022-08-29	Build site.
html	4810db6	jens-daniel-mueller	2022-08-26	Build site.
Rmd	88ce2d9	jens-daniel-mueller	2022-08-26	added global section
html	74dca9d	jens-daniel-mueller	2022-08-17	Build site.
Rmd	37ab61f	jens-daniel-mueller	2022-08-17	added adjustment delta map
html	ec4f1d0	jens-daniel-mueller	2022-08-11	Build site.
Rmd	d1cd39a	jens-daniel-mueller	2022-08-11	revised figure aspect ratio
html	78dbb27	jens-daniel-mueller	2022-08-11	Build site.
Rmd	2155209	jens-daniel-mueller	2022-08-11	rebuild with 3 basin separation as standard case
html	302d289	jens-daniel-mueller	2022-08-11	Build site.
Rmd	77719db	jens-daniel-mueller	2022-08-11	rebuild with 5 basin separation as standard case
html	318fefe	jens-daniel-mueller	2022-08-11	Build site.
Rmd	60dae4b	jens-daniel-mueller	2022-08-11	delta dcant budget analysis
html	5b77c4f	jens-daniel-mueller	2022-08-10	Build site.
Rmd	c25333a	jens-daniel-mueller	2022-08-10	bias sections and maps added
html	9c3be27	jens-daniel-mueller	2022-08-09	Build site.
Rmd	a940880	jens-daniel-mueller	2022-08-09	improve delta dcant bias assesment
html	a691b29	jens-daniel-mueller	2022-08-09	Build site.
Rmd	d136a86	jens-daniel-mueller	2022-08-09	include delta dcant bias assesment
html	910f96e	jens-daniel-mueller	2022-08-08	Build site.
Rmd	e064958	jens-daniel-mueller	2022-08-08	include bias assesment
html	e99640e	jens-daniel-mueller	2022-07-29	Build site.
html	f68fcfa	jens-daniel-mueller	2022-07-26	Build site.
Rmd	abec0de	jens-daniel-mueller	2022-07-26	changed color scale back
html	f2615fb	jens-daniel-mueller	2022-07-26	Build site.
Rmd	8da318e	jens-daniel-mueller	2022-07-26	changed color scale
html	99a82db	jens-daniel-mueller	2022-07-25	Build site.
Rmd	1d6985d	jens-daniel-mueller	2022-07-25	stat analysis of layer budgets and CIs
html	4af29f9	jens-daniel-mueller	2022-07-24	Build site.
Rmd	b2c4ff6	jens-daniel-mueller	2022-07-24	revised GCB plot
html	e2adac6	jens-daniel-mueller	2022-07-22	Build site.
Rmd	a9c8af4	jens-daniel-mueller	2022-07-22	revised section plots and tables
html	9bbc6a4	jens-daniel-mueller	2022-07-20	Build site.
Rmd	f937de2	jens-daniel-mueller	2022-07-20	revised plots
html	0c6db30	jens-daniel-mueller	2022-07-20	Build site.
Rmd	a42e52d	jens-daniel-mueller	2022-07-20	revised plots
html	b18b250	jens-daniel-mueller	2022-07-20	Build site.
Rmd	36691bf	jens-daniel-mueller	2022-07-20	revised plots
html	a80b59b	jens-daniel-mueller	2022-07-20	Build site.
Rmd	f9c0545	jens-daniel-mueller	2022-07-20	revised budget stats
html	fea41c1	jens-daniel-mueller	2022-07-20	Build site.
Rmd	b27e42f	jens-daniel-mueller	2022-07-20	rerun with corrected Canyon-B talk gap-filling in standard case
html	9ce772d	jens-daniel-mueller	2022-07-19	Build site.
Rmd	aaf37cf	jens-daniel-mueller	2022-07-19	coverage maps with gap filling
html	d803308	jens-daniel-mueller	2022-07-19	Build site.
Rmd	933285c	jens-daniel-mueller	2022-07-19	revised plots
html	b1f7ab3	jens-daniel-mueller	2022-07-18	Build site.
Rmd	ab5220b	jens-daniel-mueller	2022-07-18	revised budget over atm pCO2 plot
html	d2ae54c	jens-daniel-mueller	2022-07-18	Build site.
Rmd	85f482c	jens-daniel-mueller	2022-07-18	include scaled Sabine data for budgets
html	2695085	jens-daniel-mueller	2022-07-17	Build site.
Rmd	8624c93	jens-daniel-mueller	2022-07-17	use global output from MLR basins
html	535196a	jens-daniel-mueller	2022-07-17	Build site.
html	d20faeb	jens-daniel-mueller	2022-07-17	Build site.
Rmd	b6ef86b	jens-daniel-mueller	2022-07-17	revised budget plots
html	0160c40	jens-daniel-mueller	2022-07-16	Build site.
Rmd	d2b1090	jens-daniel-mueller	2022-07-16	cleaned code
html	efa414b	jens-daniel-mueller	2022-07-16	Build site.
Rmd	e84b169	jens-daniel-mueller	2022-07-16	added bias global analysis
html	7267b08	jens-daniel-mueller	2022-07-16	Build site.
Rmd	b91f619	jens-daniel-mueller	2022-07-16	analysed dcant budget stats
html	08c00b4	jens-daniel-mueller	2022-07-16	Build site.
html	692c937	jens-daniel-mueller	2022-07-16	Build site.
html	afb27ad	jens-daniel-mueller	2022-07-15	Build site.
Rmd	53e63d5	jens-daniel-mueller	2022-07-15	plot no surface data equi
html	b492b46	jens-daniel-mueller	2022-07-15	Build site.
Rmd	31f29fe	jens-daniel-mueller	2022-07-15	plot decadal sink trends as boxplots
html	bd24a0f	jens-daniel-mueller	2022-07-15	Build site.
Rmd	efbb1dc	jens-daniel-mueller	2022-07-15	include no surface equi cases
html	022fd60	jens-daniel-mueller	2022-07-14	Build site.
Rmd	5a9e419	jens-daniel-mueller	2022-07-14	implemented beta budget analysis
Rmd	d2092c1	jens-daniel-mueller	2022-07-14	implemented beta budget analysis
html	f1d7f80	jens-daniel-mueller	2022-07-13	Build site.
Rmd	6811ae4	jens-daniel-mueller	2022-07-13	refined profil plots
html	ea46812	jens-daniel-mueller	2022-07-13	Build site.
Rmd	defdcfe	jens-daniel-mueller	2022-07-13	additional analyis
html	17cd1d1	jens-daniel-mueller	2022-07-13	Build site.
Rmd	1bf1708	jens-daniel-mueller	2022-07-13	rerun reoccupation
html	26e9496	jens-daniel-mueller	2022-07-12	Build site.
Rmd	5d121f0	jens-daniel-mueller	2022-07-12	applied dcant scaling and improved ensemble member analysis
html	8fb595c	jens-daniel-mueller	2022-07-12	Build site.
Rmd	e2d97ef	jens-daniel-mueller	2022-07-12	excluded no adjustment from ensemble
html	003b161	jens-daniel-mueller	2022-07-12	Build site.
Rmd	54a0d14	jens-daniel-mueller	2022-07-12	added cruise based adjustment
html	b44c72a	jens-daniel-mueller	2022-07-03	Build site.
html	37f56b3	jens-daniel-mueller	2022-07-01	Build site.
Rmd	8c4a9f8	jens-daniel-mueller	2022-07-01	added basin separation analysis
html	232909e	jens-daniel-mueller	2022-07-01	Build site.
Rmd	7f85db3	jens-daniel-mueller	2022-07-01	added basin separation analysis
html	df21d31	jens-daniel-mueller	2022-07-01	Build site.
Rmd	2bbfba0	jens-daniel-mueller	2022-07-01	rebuild
html	6be73e0	jens-daniel-mueller	2022-06-30	Build site.
Rmd	6e173bf	jens-daniel-mueller	2022-06-30	updated regional budget plots
html	6e173bf	jens-daniel-mueller	2022-06-30	updated regional budget plots
html	8ab4a87	jens-daniel-mueller	2022-06-29	Build site.
Rmd	d8c7cb9	jens-daniel-mueller	2022-06-29	ensemble with unadjusted data
html	7629c78	jens-daniel-mueller	2022-06-29	Build site.
Rmd	4758911	jens-daniel-mueller	2022-06-29	ensemble without unadjusted data
html	f6786c8	jens-daniel-mueller	2022-06-29	Build site.
Rmd	6f694a1	jens-daniel-mueller	2022-06-29	included Cstar N in ensemble
html	f09080e	jens-daniel-mueller	2022-06-28	Build site.
Rmd	d457610	jens-daniel-mueller	2022-06-28	save figures for publication
html	ee30748	jens-daniel-mueller	2022-06-28	Build site.
Rmd	fe42644	jens-daniel-mueller	2022-06-28	GCB emissions ratio included
html	9393c07	jens-daniel-mueller	2022-06-28	Build site.
Rmd	2a3cf97	jens-daniel-mueller	2022-06-28	included basin-hemisphere bias, and bias contributions
html	0825298	jens-daniel-mueller	2022-06-28	Build site.
Rmd	d35ddb1	jens-daniel-mueller	2022-06-28	included GCB ocean sink data as boxplot
html	a13a7cf	jens-daniel-mueller	2022-06-28	Build site.
html	fb59a6f	jens-daniel-mueller	2022-06-27	Build site.
Rmd	7ff568c	jens-daniel-mueller	2022-06-27	included GCB ocean sink data
html	a26a21d	jens-daniel-mueller	2022-06-27	Build site.
Rmd	9425709	jens-daniel-mueller	2022-06-27	scaled sabine 2004 to full area
html	748aa43	jens-daniel-mueller	2022-06-27	Build site.
Rmd	82d5793	jens-daniel-mueller	2022-06-27	ensemble scatter plot
html	457e640	jens-daniel-mueller	2022-06-27	Build site.
Rmd	426c5bf	jens-daniel-mueller	2022-06-27	cleaned read-in section
html	16dc3af	jens-daniel-mueller	2022-06-27	Build site.
Rmd	dd2063a	jens-daniel-mueller	2022-06-27	ensemble member analysis added
html	87e9eb8	jens-daniel-mueller	2022-06-27	Build site.
Rmd	09a3348	jens-daniel-mueller	2022-06-27	1 as standard case and new ensemble members
html	b52b159	jens-daniel-mueller	2022-06-27	Build site.
html	09b0780	jens-daniel-mueller	2022-05-24	Build site.
html	25da2fb	jens-daniel-mueller	2022-05-24	Build site.
html	1d73ec9	jens-daniel-mueller	2022-05-16	Build site.
Rmd	e117b7b	jens-daniel-mueller	2022-05-16	rerun w/o data adjustments
html	2ffbdda	jens-daniel-mueller	2022-05-16	Build site.
Rmd	6fc8438	jens-daniel-mueller	2022-05-16	plot individual zonal sections
html	c71227f	jens-daniel-mueller	2022-05-16	Build site.
Rmd	55e9ac6	jens-daniel-mueller	2022-05-16	plot individual zonal sections
html	3c1100c	jens-daniel-mueller	2022-05-16	Build site.
Rmd	fdb111d	jens-daniel-mueller	2022-05-16	plot individual zonal sections
html	2ca0109	jens-daniel-mueller	2022-05-02	Build site.
Rmd	63d51bd	jens-daniel-mueller	2022-05-02	rerun with adjusted data
html	dcf2eaf	jens-daniel-mueller	2022-05-02	Build site.
Rmd	a3a2cf9	jens-daniel-mueller	2022-05-02	IO sections seperate
html	eff4fd7	jens-daniel-mueller	2022-05-02	Build site.
Rmd	39e920a	jens-daniel-mueller	2022-05-02	IO sections seperate
html	08607eb	jens-daniel-mueller	2022-05-02	Build site.
Rmd	0791c1b	jens-daniel-mueller	2022-05-02	modified plots
html	b018a9a	jens-daniel-mueller	2022-04-29	Build site.
Rmd	02ede93	jens-daniel-mueller	2022-04-29	standard case uncorrected data
html	e09320d	jens-daniel-mueller	2022-04-12	Build site.
Rmd	0dec180	jens-daniel-mueller	2022-04-12	3 data adjustment procedures implemented
html	8dca96a	jens-daniel-mueller	2022-04-12	Build site.
Rmd	e5e9288	jens-daniel-mueller	2022-04-12	3 data adjustment procedures implemented
html	2f20ea6	jens-daniel-mueller	2022-04-11	Build site.
html	209c9b6	jens-daniel-mueller	2022-04-10	Build site.
Rmd	537aff7	jens-daniel-mueller	2022-04-10	no data adjustment implemented
html	acad2e2	jens-daniel-mueller	2022-04-09	Build site.
html	3d81135	jens-daniel-mueller	2022-04-07	Build site.
html	0f5d372	jens-daniel-mueller	2022-04-04	Build site.
Rmd	ac5121f	jens-daniel-mueller	2022-04-04	added zonal mean beta distribution new figure
html	1b5a309	jens-daniel-mueller	2022-04-04	Build site.
Rmd	8b96a9d	jens-daniel-mueller	2022-04-04	added zonal mean beta distribution new figure
html	a74e341	jens-daniel-mueller	2022-04-04	Build site.
Rmd	c0432be	jens-daniel-mueller	2022-04-04	added zonal mean beta distribution ensemble sd
html	b599680	jens-daniel-mueller	2022-04-04	Build site.
Rmd	e406c39	jens-daniel-mueller	2022-04-04	added zonal mean beta distribution
html	ca7e590	jens-daniel-mueller	2022-03-22	Build site.
Rmd	8cb4b78	jens-daniel-mueller	2022-03-22	use 1800 as tref for sabine estimates
html	5a6be34	jens-daniel-mueller	2022-03-22	Build site.
Rmd	cee203c	jens-daniel-mueller	2022-03-22	rerun with NP 2021 talk correction
html	bd9e11d	jens-daniel-mueller	2022-03-22	Build site.
html	2501978	jens-daniel-mueller	2022-03-21	Build site.
Rmd	9e12898	jens-daniel-mueller	2022-03-21	use 1800 as tref for sabine estimates
html	c3a6238	jens-daniel-mueller	2022-03-08	Build site.
Rmd	775eb4f	jens-daniel-mueller	2022-03-08	moving eras analysis implemented
html	094bfa0	jens-daniel-mueller	2022-02-18	Build site.
Rmd	fa258cc	jens-daniel-mueller	2022-02-18	updated plots
html	ba2d62e	jens-daniel-mueller	2022-02-17	Build site.
Rmd	5420131	jens-daniel-mueller	2022-02-17	added Seaflux data
html	192504c	jens-daniel-mueller	2022-02-17	Build site.
Rmd	2394302	jens-daniel-mueller	2022-02-17	added Seaflux data
html	251c7cf	jens-daniel-mueller	2022-02-17	Build site.
Rmd	9ea0d8c	jens-daniel-mueller	2022-02-17	adapted beta factor
html	565224d	jens-daniel-mueller	2022-02-17	Build site.
Rmd	33422e3	jens-daniel-mueller	2022-02-17	scaled budgets to global coverage
html	2116dd3	jens-daniel-mueller	2022-02-09	Build site.
Rmd	e281bf3	jens-daniel-mueller	2022-02-09	updated plots
html	6fe70a1	jens-daniel-mueller	2022-02-05	Build site.
Rmd	8955c85	jens-daniel-mueller	2022-02-05	cleaned plots
html	a6b33aa	jens-daniel-mueller	2022-02-04	Build site.
Rmd	73efef8	jens-daniel-mueller	2022-02-04	atm co2 time series plotted, uncertainties revised
html	4b48475	jens-daniel-mueller	2022-02-04	Build site.
Rmd	6afd7b1	jens-daniel-mueller	2022-02-04	calculated uncertainty of basin budget changes
html	fec5a1e	jens-daniel-mueller	2022-02-04	Build site.
Rmd	325da3a	jens-daniel-mueller	2022-02-04	calculated uncertainty of basin budget changes
html	d2191ad	jens-daniel-mueller	2022-02-04	Build site.
Rmd	79a54fe	jens-daniel-mueller	2022-02-04	new coaverage map
html	4c5b079	jens-daniel-mueller	2022-02-03	Build site.
Rmd	66d89a9	jens-daniel-mueller	2022-02-03	added surface flux model products
html	4077397	jens-daniel-mueller	2022-02-03	Build site.
Rmd	c0c3be1	jens-daniel-mueller	2022-02-03	added surface flux products
html	0d0f790	jens-daniel-mueller	2022-02-02	Build site.
Rmd	049346f	jens-daniel-mueller	2022-02-02	shifted year labels
html	4673df5	jens-daniel-mueller	2022-02-02	Build site.
Rmd	1185397	jens-daniel-mueller	2022-02-02	rearranged plots
html	60727e6	jens-daniel-mueller	2022-02-02	Build site.
Rmd	db5308b	jens-daniel-mueller	2022-02-02	rearranged plots
html	c7b4984	jens-daniel-mueller	2022-02-02	Build site.
Rmd	a3d8469	jens-daniel-mueller	2022-02-02	ensemble uncertainties in global buget
html	7fb28a2	jens-daniel-mueller	2022-02-02	Build site.
Rmd	a332fe6	jens-daniel-mueller	2022-02-02	ensemble uncertainties in time series
html	49097e8	jens-daniel-mueller	2022-02-02	Build site.
Rmd	f98c33b	jens-daniel-mueller	2022-02-02	incl ensemble uncertainties in plot
html	fe11bfd	jens-daniel-mueller	2022-02-02	Build site.
Rmd	d3f7f05	jens-daniel-mueller	2022-02-02	incl ensemble uncertainties
html	fa46251	jens-daniel-mueller	2022-02-02	Build site.
Rmd	33feacf	jens-daniel-mueller	2022-02-02	incl sabine budgets
html	7655085	jens-daniel-mueller	2022-02-02	Build site.
Rmd	09dd555	jens-daniel-mueller	2022-02-02	incl sabine budgets
html	226d67d	jens-daniel-mueller	2022-02-02	Build site.
Rmd	ed056e2	jens-daniel-mueller	2022-02-02	incl sabine budgets
html	ed903f7	jens-daniel-mueller	2022-02-02	Build site.
Rmd	9d11695	jens-daniel-mueller	2022-02-02	scaled budget to atm pco2 increase
html	32e9682	jens-daniel-mueller	2022-02-02	Build site.
Rmd	5cc5572	jens-daniel-mueller	2022-02-02	included Sabine column inventory as reference
html	913e42f	jens-daniel-mueller	2022-02-01	Build site.
Rmd	c4b2de9	jens-daniel-mueller	2022-02-01	updated profile plots
html	189de95	jens-daniel-mueller	2022-02-01	Build site.
Rmd	b0c17bc	jens-daniel-mueller	2022-02-01	updated profile plots
html	ab001eb	jens-daniel-mueller	2022-01-31	Build site.
Rmd	ccf6723	jens-daniel-mueller	2022-01-31	filled step plot for layer budgets
html	d2ae5fe	jens-daniel-mueller	2022-01-31	Build site.
Rmd	278523c	jens-daniel-mueller	2022-01-31	step plot for layer budgets
html	b62308d	jens-daniel-mueller	2022-01-31	Build site.
Rmd	714c5cc	jens-daniel-mueller	2022-01-31	step plot for layer budgets
html	ec7fe7e	jens-daniel-mueller	2022-01-31	Build site.
Rmd	5c948ae	jens-daniel-mueller	2022-01-31	added time series vs atm pco2
html	de557de	jens-daniel-mueller	2022-01-28	Build site.
html	5f2aed0	jens-daniel-mueller	2022-01-27	Build site.
Rmd	54c9e26	jens-daniel-mueller	2022-01-27	added layer budget profiles
html	eccd82b	jens-daniel-mueller	2022-01-26	Build site.
Rmd	c5577d3	jens-daniel-mueller	2022-01-26	added meand sd to offset mean concentrations profiles
html	c6fe495	jens-daniel-mueller	2022-01-26	Build site.
Rmd	e0e7974	jens-daniel-mueller	2022-01-26	added offset mean concentrations profiles
html	9753eb8	jens-daniel-mueller	2022-01-26	Build site.
html	b1d7720	jens-daniel-mueller	2022-01-21	Build site.
Rmd	0210ed5	jens-daniel-mueller	2022-01-21	added mean concentrations profiles per 5 basins
html	d6b399a	jens-daniel-mueller	2022-01-21	Build site.
Rmd	da17a07	jens-daniel-mueller	2022-01-21	added mean concentrations profiles
html	c499be8	jens-daniel-mueller	2022-01-21	Build site.
Rmd	d941871	jens-daniel-mueller	2022-01-21	run color map test
html	e572075	jens-daniel-mueller	2022-01-21	Build site.
Rmd	99b6c92	jens-daniel-mueller	2022-01-21	run color map test
html	4fe7150	jens-daniel-mueller	2022-01-21	Build site.
Rmd	0379e99	jens-daniel-mueller	2022-01-21	script cleaning
html	49b41cf	jens-daniel-mueller	2022-01-21	Build site.
Rmd	2c82651	jens-daniel-mueller	2022-01-21	added map of scaled absolute change
html	c0807e8	jens-daniel-mueller	2022-01-21	Build site.
Rmd	5dd3d7a	jens-daniel-mueller	2022-01-21	added map of scaled relative change
html	22b421f	jens-daniel-mueller	2022-01-21	Build site.
Rmd	2c3fa75	jens-daniel-mueller	2022-01-21	cleaned alluvial plots
html	1a35f1f	jens-daniel-mueller	2022-01-20	Build site.
Rmd	e58f510	jens-daniel-mueller	2022-01-20	added relative changes to alluvial plots
html	b503ae1	jens-daniel-mueller	2022-01-20	Build site.
Rmd	2eb2567	jens-daniel-mueller	2022-01-20	added relative changes to alluvial plots
html	cc31f4b	jens-daniel-mueller	2022-01-20	Build site.
Rmd	416e107	jens-daniel-mueller	2022-01-20	added delta dcant map
html	11a800b	jens-daniel-mueller	2022-01-20	Build site.
Rmd	81a40d5	jens-daniel-mueller	2022-01-20	updated alluvial plots
html	3087804	jens-daniel-mueller	2022-01-20	Build site.
Rmd	2ae5966	jens-daniel-mueller	2022-01-20	updated alluvial plots
html	6d566d5	jens-daniel-mueller	2022-01-20	Build site.
Rmd	4901b0f	jens-daniel-mueller	2022-01-20	updated alluvial plots
html	44796b1	jens-daniel-mueller	2022-01-20	Build site.
Rmd	cdbd92c	jens-daniel-mueller	2022-01-20	created alluvial plots
html	48ec4c6	jens-daniel-mueller	2022-01-19	Build site.
Rmd	0fb2ae5	jens-daniel-mueller	2022-01-19	printed column inv from AIP standard runs
html	f347cd7	jens-daniel-mueller	2022-01-18	Build site.
Rmd	86b711c	jens-daniel-mueller	2022-01-18	plot hemisphere budgets and publication results

---

center <- -160
boundary <- center + 180
target_crs <- paste0("+proj=robin +over +lon_0=", center)
# target_crs <- paste0("+proj=eqearth +over +lon_0=", center)
# target_crs <- paste0("+proj=eqearth +lon_0=", center)
# target_crs <- paste0("+proj=igh_o +lon_0=", center)

worldmap <- ne_countries(scale = 'small',
                         type = 'map_units',
                         returnclass = 'sf')

worldmap <- worldmap %>% st_break_antimeridian(lon_0 = center)
worldmap_trans <- st_transform(worldmap, crs = target_crs)

# ggplot() +
#   geom_sf(data = worldmap_trans)

coastline <- ne_coastline(scale = 'small', returnclass = "sf")
coastline <- st_break_antimeridian(coastline, lon_0 = 200)
coastline_trans <- st_transform(coastline, crs = target_crs)

# ggplot() +
#   geom_sf(data = worldmap_trans, fill = "grey", col="grey") +
#   geom_sf(data = coastline_trans)


bbox <- st_bbox(c(xmin = -180, xmax = 180, ymax = 65, ymin = -78), crs = st_crs(4326))
bbox <- st_as_sfc(bbox)
bbox_trans <- st_break_antimeridian(bbox, lon_0 = center)

bbox_graticules <- st_graticule(
  x = bbox_trans,
  crs = st_crs(bbox_trans),
  datum = st_crs(bbox_trans),
  lon = c(20, 20.001),
  lat = c(-78,65),
  ndiscr = 1e3,
  margin = 0.001
)

bbox_graticules_trans <- st_transform(bbox_graticules, crs = target_crs)
rm(worldmap, coastline, bbox, bbox_trans, bbox_graticules)

# ggplot() +
#   geom_sf(data = worldmap_trans, fill = "grey", col="grey") +
#   geom_sf(data = coastline_trans) +
#   geom_sf(data = bbox_graticules_trans)

lat_lim <- ext(bbox_graticules_trans)[c(3,4)]*1.002
lon_lim <- ext(bbox_graticules_trans)[c(1,2)]*1.005

# ggplot() +
#   geom_sf(data = worldmap_trans, fill = "grey90", col = "grey90") +
#   geom_sf(data = coastline_trans) +
#   geom_sf(data = bbox_graticules_trans, linewidth = 1) +
#   coord_sf(crs = target_crs,
#            ylim = lat_lim,
#            xlim = lon_lim,
#            expand = FALSE) +
#   theme(
#     panel.border = element_blank(),
#     axis.text = element_blank(),
#     axis.ticks = element_blank()
#   )

1 Libraries

2 Read files

2.1 Integration depth

params_global$inventory_depth_standard <- 1000

2.2 Paths and Versions

### cases included in uncertainty budget

# standard case

version_id_pattern <- "1"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_s <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# cruise-by-cruise adjustments

version_id_pattern <- "c"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_c <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# gap-filling uncertainty

version_id_pattern <- "g"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_g <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# C* from N and TA

version_id_pattern <- "n"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_n <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# Surface with eMLR(C*)

version_id_pattern <- "h"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_e <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# BGC predictors from WOA18

version_id_pattern <- "w"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_w <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)


### cases considered as sensitivity tests but not in uncertainty budget

# no data adjustments

version_id_pattern <- "d"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_d <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# reoccupation filter

version_id_pattern <- "o"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_o <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# C* from TA

version_id_pattern <- "a"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_a <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# C*

version_id_pattern <- "x"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_x <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)

# TCO2 target

version_id_pattern <- "t"

Version_IDs_1 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_1", version_id_pattern))

Version_IDs_2 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_2", version_id_pattern))

Version_IDs_3 <- list.files(path = "/nfs/kryo/work/jenmueller/emlr_cant/observations",
                            pattern = paste0("v_3", version_id_pattern))

Version_IDs_t <- c(Version_IDs_1, Version_IDs_2, Version_IDs_3)
rm(Version_IDs_1, Version_IDs_2, Version_IDs_3)




Version_IDs_ensemble <- c(
  Version_IDs_s, Version_IDs_c, Version_IDs_g, Version_IDs_n, Version_IDs_e, Version_IDs_w,
  Version_IDs_d, Version_IDs_o, Version_IDs_a, Version_IDs_x, Version_IDs_t
  )

Version_IDs_ensemble_uncertainty <- c(
  Version_IDs_s, Version_IDs_c, Version_IDs_g, Version_IDs_n, Version_IDs_e, Version_IDs_w
  )

Version_IDs_ensemble_sensitivity <- c(
  Version_IDs_s, Version_IDs_d, Version_IDs_o, Version_IDs_a, Version_IDs_x, Version_IDs_t
  )


rm(
  Version_IDs_s,
  Version_IDs_c,
  Version_IDs_g,
  Version_IDs_n,
  Version_IDs_e,
  Version_IDs_w,
  Version_IDs_d,
  Version_IDs_o,
  Version_IDs_a,
  Version_IDs_x,
  Version_IDs_t
)

# subset Standard case
Version_IDs <- Version_IDs_ensemble[str_detect(Version_IDs_ensemble, "103")]

2.3 Parameters

for (i_Version_IDs in Version_IDs_ensemble) {
  
  path_version_data     <-
    paste(path_observations,
          i_Version_IDs,
          "/data/",
          sep = "")
  
  params_local <-
    read_rds(paste(path_version_data,
                   "params_local.rds",
                   sep = ""))
  
  params_local <- bind_cols(
    Version_ID = i_Version_IDs,
    tref1 = params_local$tref1,
    tref2 = params_local$tref2,
    MLR_basins = params_local$MLR_basins
  )
  
  tref <- read_csv(paste(path_version_data,
                         "tref.csv",
                         sep = ""))
  
  params_local <- params_local %>%
    mutate(
      median_year_1 = sort(tref$median_year)[1],
      median_year_2 = sort(tref$median_year)[2],
      duration = median_year_2 - median_year_1,
      period = paste(median_year_1, "-", median_year_2)
    )
  
  if (exists("params_local_all_ensemble")) {
    params_local_all_ensemble <- bind_rows(params_local_all_ensemble, params_local)
  }
  
  if (!exists("params_local_all_ensemble")) {
    params_local_all_ensemble <- params_local
  }
  
  
}

rm(params_local,
   tref)


params_local_all_ensemble <- params_local_all_ensemble %>%
  select(Version_ID, period, MLR_basins, tref1, tref2)

params_local_all_ensemble <-
  params_local_all_ensemble %>%
  mutate(
    Version_ID_group = str_sub(Version_ID, 4, 4),
    Version_ID_group = case_when(
      Version_ID_group == "1" ~ "Standard case",
      Version_ID_group == "c" ~ "Cruise adjustment",
      Version_ID_group == "g" ~ "Gap filling",
      Version_ID_group == "n" ~ "C* with NO3,TA",
      Version_ID_group == "h" ~ "OceanSODA DIC",
      Version_ID_group == "w" ~ "WOA18 predictors",
      Version_ID_group == "d" ~ "No data adjustments",
      Version_ID_group == "o" ~ "Reoccupation filter",
      Version_ID_group == "a" ~ "C* with PO4 only",
      Version_ID_group == "t" ~ "DIC target",
      Version_ID_group == "x" ~ "No tref adjustment",
      TRUE ~ Version_ID_group
    )
  )

params_local_all_ensemble <-
  params_local_all_ensemble %>%
  mutate(
    MLR_basins = case_when(
      MLR_basins == "AIP" ~ "3",
      MLR_basins == "SO_AIP" ~ "3+SO",
      MLR_basins == "SO_5" ~ "5+SO",
      TRUE ~ MLR_basins
    )
  )

params_local_all <- params_local_all_ensemble %>% 
  filter(Version_ID %in% Version_IDs)

path_version_data     <-
  paste(path_observations,
        Version_IDs[1],
        "/data/",
        sep = "")

params_local <-
  read_rds(paste(path_version_data,
                 "params_local.rds",
                 sep = ""))

2.4 eMLR stats

for (i_Version_IDs in Version_IDs) {
  # i_Version_IDs <- Version_IDs[1]
  
  path_version_data     <-
    paste(path_observations,
          i_Version_IDs,
          "/data/",
          sep = "")
  
  # load and join data files
  
  # Model performance: RMSE, n_predictors, VIF
  GLODAP_glanced <-
    read_csv(paste(path_version_data,
                   "lm_model_metrics.csv",
                   sep = ""))
  
  GLODAP_glanced <- GLODAP_glanced %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("GLODAP_glanced_all")) {
    GLODAP_glanced_all <-
      bind_rows(GLODAP_glanced_all, GLODAP_glanced)
  }
  
  if (!exists("GLODAP_glanced_all")) {
    GLODAP_glanced_all <- GLODAP_glanced
  }
  
  # The chosen models
  lm_best_target <-
    read_csv(paste(path_version_data,
                   "lm_best_target.csv",
                   sep = ""))
  
  lm_best_target <- lm_best_target %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("lm_best_target_all")) {
    lm_best_target_all <-
      bind_rows(lm_best_target_all, lm_best_target)
  }
  
  if (!exists("lm_best_target_all")) {
    lm_best_target_all <- lm_best_target
  }
  
  lm_best_predictor_counts <-
    read_csv(paste(path_version_data,
                   "lm_best_predictor_counts.csv",
                   sep = ""))
  
  lm_best_predictor_counts <- lm_best_predictor_counts %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("lm_best_predictor_counts_all")) {
    lm_best_predictor_counts_all <-
      bind_rows(lm_best_predictor_counts_all, lm_best_predictor_counts)
  }
  
  if (!exists("lm_best_predictor_counts_all")) {
    lm_best_predictor_counts_all <- lm_best_predictor_counts
  }
  
  lat_residual <-
    read_csv(paste(path_version_data,
                   "lm_lat_residual.csv",
                   sep = ""))
  
  lat_residual <- lat_residual %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("lat_residual_all")) {
    lat_residual_all <-
      bind_rows(lat_residual_all, lat_residual)
  }
  
  if (!exists("lat_residual_all")) {
    lat_residual_all <- lat_residual
  }
  
  lat_residual_offset <-
    read_csv(paste(path_version_data,
                   "lm_lat_residual_offset.csv",
                   sep = ""))
  
  lat_residual_offset <- lat_residual_offset %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("lat_residual_offset_all")) {
    lat_residual_offset_all <-
      bind_rows(lat_residual_offset_all, lat_residual_offset)
  }
  
  if (!exists("lat_residual_offset_all")) {
    lat_residual_offset_all <- lat_residual_offset
  }
  
  
  spatial_residual <-
    read_csv(paste(path_version_data,
                   "lm_spatial_residual.csv",
                   sep = ""))
  
  spatial_residual <- spatial_residual %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("spatial_residual_all")) {
    spatial_residual_all <-
      bind_rows(spatial_residual_all, spatial_residual)
  }
  
  if (!exists("spatial_residual_all")) {
    spatial_residual_all <- spatial_residual
  }
  
  spatial_residual_offset <-
    read_csv(paste(path_version_data,
                   "lm_spatial_residual_offset.csv",
                   sep = ""))
  
  spatial_residual_offset <- spatial_residual_offset %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("spatial_residual_offset_all")) {
    spatial_residual_offset_all <-
      bind_rows(spatial_residual_offset_all, spatial_residual_offset)
  }
  
  if (!exists("spatial_residual_offset_all")) {
    spatial_residual_offset_all <- spatial_residual_offset
  }
  
  

  
}

rm(GLODAP_glanced,
   lat_residual,
   lat_residual_offset,
   spatial_residual,
   spatial_residual_offset,
   lm_best_predictor_counts,
   lm_best_target)

2.5 Inventories

for (i_Version_IDs in Version_IDs_ensemble) {
  # i_Version_IDs <- Version_IDs[1]

  path_version_data     <-
    paste(path_observations,
          i_Version_IDs,
          "/data/",
          sep = "")
  
  # load and join data files
  
  dcant_budget_basin_MLR <-
    read_csv(paste(path_version_data,
                   "dcant_budget_basin_MLR.csv",
                   sep = ""))
  
  dcant_budget_basin_MLR_mod_truth <-
    read_csv(paste(
      path_version_data,
      "dcant_budget_basin_MLR_mod_truth.csv",
      sep = ""
    ))
  
    
  dcant_budget_basin_MLR <- bind_rows(dcant_budget_basin_MLR,
                                      dcant_budget_basin_MLR_mod_truth)
  
  dcant_budget_basin_MLR <- dcant_budget_basin_MLR %>%
    mutate(Version_ID = i_Version_IDs)
  

  if (exists("dcant_budget_basin_MLR_all")) {
    dcant_budget_basin_MLR_all <-
      bind_rows(dcant_budget_basin_MLR_all, dcant_budget_basin_MLR)
  }
  
  if (!exists("dcant_budget_basin_MLR_all")) {
    dcant_budget_basin_MLR_all <- dcant_budget_basin_MLR
  }
  
  
  dcant_budget_basin_MLR_bias_decomposition <-
    read_csv(
      paste(
        path_version_data,
        "dcant_budget_basin_MLR_bias_decomposition.csv",
        sep = ""
      )
    )
  
  
  dcant_budget_basin_MLR_bias_decomposition <-
    dcant_budget_basin_MLR_bias_decomposition %>%
    mutate(Version_ID = i_Version_IDs)
  
  
  if (exists("dcant_budget_basin_MLR_bias_all_decomposition")) {
    dcant_budget_basin_MLR_bias_all_decomposition <-
      bind_rows(
        dcant_budget_basin_MLR_bias_all_decomposition,
        dcant_budget_basin_MLR_bias_decomposition
      )
  }
  
  if (!exists("dcant_budget_basin_MLR_bias_all_decomposition")) {
    dcant_budget_basin_MLR_bias_all_decomposition <-
      dcant_budget_basin_MLR_bias_decomposition
  }

}

rm(
  dcant_budget_basin_MLR,
  dcant_budget_basin_MLR_mod_truth,
  dcant_budget_basin_MLR_bias_decomposition
)

dcant_budget_basin_MLR_all <- dcant_budget_basin_MLR_all %>%
  filter(estimate == "dcant", 
         method == "total") %>% 
  select(-c(estimate, method)) %>% 
  rename(dcant = value)

dcant_budget_basin_MLR_all_10000 <- dcant_budget_basin_MLR_all %>%
  filter(inv_depth == 10000)

dcant_budget_basin_MLR_all_1000 <- dcant_budget_basin_MLR_all %>%
  filter(inv_depth == 1000)

dcant_budget_basin_MLR_all <- dcant_budget_basin_MLR_all %>%
  filter(inv_depth == params_global$inventory_depth_standard)

dcant_budget_basin_MLR_bias_all_decomposition <- 
  dcant_budget_basin_MLR_bias_all_decomposition %>%
  filter(inv_depth == params_global$inventory_depth_standard)

2.6 Column inventories

for (i_Version_IDs in Version_IDs_ensemble) {
  # i_Version_IDs <- Version_IDs_ensemble[19]
  
  path_version_data     <-
    paste(path_observations,
          i_Version_IDs,
          "/data/",
          sep = "")
  
  # load and join data files
  
  dcant_inv <-
    read_csv(paste(path_version_data,
                   "dcant_inv.csv",
                   sep = ""))
  
  p_map_cant_inv(dcant_inv %>% filter(data_source == "obs"))
  
  
  dcant_inv_mod_truth <-
    read_csv(paste(path_version_data,
                   "dcant_inv_mod_truth.csv",
                   sep = "")) %>%
    filter(method == "total") %>%
    select(-method)
  
  dcant_inv_bias <-
    read_csv(paste(path_version_data,
                   "dcant_inv_bias.csv",
                   sep = "")) %>%
    mutate(Version_ID = i_Version_IDs)
  
  dcant_inv <- bind_rows(dcant_inv,
                         dcant_inv_mod_truth) %>%
    mutate(Version_ID = i_Version_IDs)
  
  dcant_budget_lat_grid <-
    read_csv(paste(path_version_data,
                   "dcant_budget_lat_grid.csv",
                   sep = "")) %>%
    mutate(Version_ID = i_Version_IDs)
  
  dcant_budget_lon_grid <-
    read_csv(paste(path_version_data,
                   "dcant_budget_lon_grid.csv",
                   sep = "")) %>%
    mutate(Version_ID = i_Version_IDs)
  if (exists("dcant_inv_all")) {
    dcant_inv_all <- bind_rows(dcant_inv_all, dcant_inv)
  }
  
  if (!exists("dcant_inv_all")) {
    dcant_inv_all <- dcant_inv
  }
  
  if (exists("dcant_inv_bias_all")) {
    dcant_inv_bias_all <- bind_rows(dcant_inv_bias_all, dcant_inv_bias)
  }
  
  if (!exists("dcant_inv_bias_all")) {
    dcant_inv_bias_all <- dcant_inv_bias
  }
  
  if (exists("dcant_budget_lat_grid_all")) {
    dcant_budget_lat_grid_all <- bind_rows(dcant_budget_lat_grid_all, dcant_budget_lat_grid)
  }
  
  if (!exists("dcant_budget_lat_grid_all")) {
    dcant_budget_lat_grid_all <- dcant_budget_lat_grid
  }
  
  if (exists("dcant_budget_lon_grid_all")) {
    dcant_budget_lon_grid_all <- bind_rows(dcant_budget_lon_grid_all, dcant_budget_lon_grid)
  }
  
  if (!exists("dcant_budget_lon_grid_all")) {
    dcant_budget_lon_grid_all <- dcant_budget_lon_grid
  }
}

rm(dcant_inv,
   dcant_inv_bias,
   dcant_inv_mod_truth,
   dcant_budget_lat_grid,
   dcant_budget_lon_grid)

dcant_inv_all_10000 <- dcant_inv_all %>%
  filter(inv_depth == 10000)

dcant_inv_all <- dcant_inv_all %>%
  filter(inv_depth == params_global$inventory_depth_standard)

dcant_budget_lat_grid_all <- dcant_budget_lat_grid_all %>% 
  filter(inv_depth == params_global$inventory_depth_standard)

dcant_budget_lon_grid_all <- dcant_budget_lon_grid_all %>% 
  filter(inv_depth == params_global$inventory_depth_standard)

dcant_budget_lat_grid_all <- dcant_budget_lat_grid_all %>%
  pivot_wider(names_from = estimate,
              values_from = value) %>%
  filter(period != "1994 - 2014",
         method == "total")

dcant_budget_lon_grid_all <- dcant_budget_lon_grid_all %>%
  pivot_wider(names_from = estimate,
              values_from = value) %>%
  filter(period != "1994 - 2014",
         method == "total")

dcant_inv_all %>% 
  filter(Version_ID == "v_1103",
         data_source == "obs") %>% 
  select(lon, lat, dcant) %>% 
  write_csv(here::here("output/dcant_column_inventor_example.csv"))

2.7 Sections / profiles

for (i_Version_IDs in Version_IDs_ensemble) {

  path_version_data     <-
  paste(path_observations,
        i_Version_IDs,
        "/data/",
        sep = "")
  
  # load and join data files
  
  dcant_zonal <-
    read_csv(paste(path_version_data,
                   "dcant_zonal.csv",
                   sep = ""))
  
  dcant_zonal_mod_truth <-
    read_csv(paste(path_version_data,
                   "dcant_zonal_mod_truth.csv",
                   sep = ""))
  
  dcant_zonal <- bind_rows(dcant_zonal,
                           dcant_zonal_mod_truth)
  
  dcant_global_section <-
    read_csv(paste(path_version_data,
                   "dcant_global_section.csv",
                   sep = ""))
  
  # dcant_profile <-
  #   read_csv(paste(path_version_data,
  #                  "dcant_profile.csv",
  #                  sep = ""))
  # 
  # dcant_profile_mod_truth <-
  #   read_csv(paste(path_version_data,
  #                  "dcant_profile_mod_truth.csv",
  #                  sep = ""))
  # 
  # dcant_profile <- bind_rows(dcant_profile,
  #                            dcant_profile_mod_truth)
  
  
  dcant_profile_basin_MLR <-
    read_csv(paste(path_version_data,
                   "dcant_profile_basin_MLR.csv",
                   sep = ""))
  
  
  dcant_profile_basin_MLR_mod_truth <-
    read_csv(paste(
      path_version_data,
      "dcant_profile_basin_MLR_mod_truth.csv",
      sep = ""
    ))

  dcant_profile_basin_MLR <- bind_rows(dcant_profile_basin_MLR,
                                       dcant_profile_basin_MLR_mod_truth)
  
  
  # dcant_budget_basin_AIP_layer <-
  #   read_csv(paste(path_version_data,
  #                  "dcant_budget_basin_AIP_layer.csv",
  #                  sep = ""))

  dcant_budget_basin_MLR_layer <-
    read_csv(paste(path_version_data,
                   "dcant_budget_basin_MLR_layer.csv",
                   
                   sep = ""))
  
  dcant_budget_basin_MLR_layer_mod_truth <-
    read_csv(paste(
      path_version_data,
      "dcant_budget_basin_MLR_layer_mod_truth.csv",
      sep = ""
    ))
  
  dcant_budget_basin_MLR_layer <-
    bind_rows(dcant_budget_basin_MLR_layer,
              dcant_budget_basin_MLR_layer_mod_truth)
  
  dcant_zonal_bias <-
    read_csv(paste(path_version_data,
                   "dcant_zonal_bias.csv",
                   sep = ""))
  
  dcant_penetration_depth_all_lat_mean <-
    read_csv(paste0(path_version_data,
                   "dcant_penetration_depth_all_lat_mean.csv"))
  

  dcant_zonal <- dcant_zonal %>% 
    mutate(Version_ID = i_Version_IDs)

  dcant_global_section <- dcant_global_section %>% 
    mutate(Version_ID = i_Version_IDs)
  
  # dcant_profile <- dcant_profile %>% 
  #   mutate(Version_ID = i_Version_IDs)

  dcant_profile_basin_MLR <- dcant_profile_basin_MLR %>% 
    mutate(Version_ID = i_Version_IDs)
  
  # dcant_budget_basin_AIP_layer <- dcant_budget_basin_AIP_layer %>% 
  #   mutate(Version_ID = i_Version_IDs)
  
  dcant_budget_basin_MLR_layer <- dcant_budget_basin_MLR_layer %>% 
    mutate(Version_ID = i_Version_IDs)
  
  dcant_zonal_bias <- dcant_zonal_bias %>% 
    mutate(Version_ID = i_Version_IDs)
  
  dcant_penetration_depth_all_lat_mean <-
    dcant_penetration_depth_all_lat_mean %>%
    mutate(Version_ID = i_Version_IDs)
  
  
  if (exists("dcant_zonal_all")) {
    dcant_zonal_all <- bind_rows(dcant_zonal_all, dcant_zonal)
  }
  
  if (!exists("dcant_zonal_all")) {
    dcant_zonal_all <- dcant_zonal
  }

  
  if (exists("dcant_global_section_all")) {
    dcant_global_section_all <- bind_rows(dcant_global_section_all, dcant_global_section)
  }
  
  if (!exists("dcant_global_section_all")) {
    dcant_global_section_all <- dcant_global_section
  }

  # if (exists("dcant_profile_all")) {
  #   dcant_profile_all <- bind_rows(dcant_profile_all, dcant_profile)
  # }
  # 
  # if (!exists("dcant_profile_all")) {
  #   dcant_profile_all <- dcant_profile
  # }

  if (exists("dcant_profile_basin_MLR_all")) {
    dcant_profile_basin_MLR_all <- bind_rows(dcant_profile_basin_MLR_all, dcant_profile_basin_MLR)
  }
  
  if (!exists("dcant_profile_basin_MLR_all")) {
    dcant_profile_basin_MLR_all <- dcant_profile_basin_MLR
  }

  # if (exists("dcant_budget_basin_AIP_layer_all")) {
  #   dcant_budget_basin_AIP_layer_all <-
  #     bind_rows(dcant_budget_basin_AIP_layer_all,
  #               dcant_budget_basin_AIP_layer)
  # }
  # 
  # if (!exists("dcant_budget_basin_AIP_layer_all")) {
  #   dcant_budget_basin_AIP_layer_all <- dcant_budget_basin_AIP_layer
  # }

  if (exists("dcant_budget_basin_MLR_layer_all")) {
    dcant_budget_basin_MLR_layer_all <-
      bind_rows(dcant_budget_basin_MLR_layer_all,
                dcant_budget_basin_MLR_layer)
  }
  
  if (!exists("dcant_budget_basin_MLR_layer_all")) {
    dcant_budget_basin_MLR_layer_all <- dcant_budget_basin_MLR_layer
  }

  if (exists("dcant_zonal_bias_all")) {
    dcant_zonal_bias_all <- bind_rows(dcant_zonal_bias_all, dcant_zonal_bias)
  }
  
  if (!exists("dcant_zonal_bias_all")) {
    dcant_zonal_bias_all <- dcant_zonal_bias
  }

  if (exists("dcant_penetration_depth_all_lat_mean_all")) {
    dcant_penetration_depth_all_lat_mean_all <-
      bind_rows(
        dcant_penetration_depth_all_lat_mean_all,
        dcant_penetration_depth_all_lat_mean
      )
  }
  
  if (!exists("dcant_penetration_depth_all_lat_mean_all")) {
    dcant_penetration_depth_all_lat_mean_all <-
      dcant_penetration_depth_all_lat_mean
  }

}

rm(dcant_zonal, dcant_global_section, dcant_zonal_bias, dcant_zonal_mod_truth,
   dcant_budget_basin_MLR_layer,
   dcant_profile_basin_MLR, dcant_penetration_depth_all_lat_mean)

2.8 Observations coverage

for (i_Version_IDs in Version_IDs) {
  path_version_data     <-
    paste(path_observations,
          i_Version_IDs,
          "/data/",
          sep = "")
  
  # load and join data files
  
  GLODAP_grid_era  <-
    read_csv(paste(
      path_version_data,
      "GLODAPv2.2020_clean_obs_grid_era.csv",
      sep = ""
    ))

  
  GLODAP_grid_era <- GLODAP_grid_era %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("GLODAP_grid_era_all")) {
    GLODAP_grid_era_all <-
      bind_rows(GLODAP_grid_era_all, GLODAP_grid_era)
  }
  
  if (!exists("GLODAP_grid_era_all")) {
    GLODAP_grid_era_all <- GLODAP_grid_era
  }
  
  
  
  GLODAP <- read_csv(paste(
    path_version_data,
    "GLODAPv2.2020_clean.csv",
    sep = ""
  ))
  
  GLODAP <- GLODAP %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("GLODAP_all")) {
    GLODAP_all <-
      bind_rows(GLODAP_all, GLODAP)
  }
  
  if (!exists("GLODAP_all")) {
    GLODAP_all <- GLODAP
  }
  
}

rm(GLODAP_grid_era, GLODAP)


GLODAP_grid_era_all <- full_join(GLODAP_grid_era_all,
                                 params_local_all)

GLODAP_all <- full_join(GLODAP_all,
                        params_local_all)


GLODAP_expocodes <-
  read_tsv(
    paste(
      "/nfs/kryo/work/updata/glodapv2_2021/",
      "EXPOCODES.txt",
      sep = ""
    ),
    col_names = c("cruise", "cruise_expocode")
  )

GLODAP_all <-
  left_join(GLODAP_all, GLODAP_expocodes)

2.9 Atm CO2

co2_atm <-
  read_csv(paste(path_preprocessing,
                 "co2_atm.csv",
                 sep = ""))

co2_atm_reccap2 <-
  read_csv(paste(path_preprocessing,
                  "co2_atm_reccap2.csv",
                  sep = ""))

2.10 Sabine total Cant

tcant_inv <-
  read_csv(paste(path_preprocessing,
                  "S04_tcant_inv.csv", sep = ""))

tcant_inv <- tcant_inv %>% 
  filter(inv_depth == 3000) %>% 
  rename(dcant = tcant,
         dcant_pos = tcant_pos) %>% 
  mutate(tref1 = 1800,
         tref2 = 1994)

2.11 GCB carbon sink

Ocean_Sink <-
  read_csv(paste(path_preprocessing,
                  "Ocean_Sink.csv",
                  sep = ""))

Global_Carbon_Budget <-
  read_csv(paste(path_preprocessing,
                  "GCB_Global_Carbon_Budget.csv",
                  sep = ""))

2.12 Periods

two_decades <- unique(params_local_all_ensemble$period)[1:2]

3 Define labels

dcant_pgc_label <- expression(Delta * C["ant"]~inventory~(Pg~C~dec^{-1}))
dcant_bias_pgc_label <- expression(Delta * C["ant"]~inventory~ bias ~(Pg~C~dec^{-1}))
delta_dcant_bias_pgc_label <- expression(Delta * Delta * C["ant"]~bias~(Pg~C~dec^{-1}))
beta_inv_label <- expression(Mean~beta~column~inventory~(mol~m^{-2}~µatm^{-1}))

dcant_umol_label <- expression(Delta * C[ant]~(mu * mol ~ kg ^ {-1}~dec^{-1}))
ddcant_umol_label <- expression(Delta * Delta * C[ant]~(mu * mol ~ kg ^ {-1}~dec^{-1}))

dcant_layer_budget_label <- 
  expression(Delta ~ C[ant] ~ "inventory per 500m depth layer"~(Pg~C~dec^{-1}))

dcant_CI_label <- expression(Delta * C[ant]~(mol ~ m ^ {-2}~dec^{-1}))
delta_dcant_CI_label <- expression(Delta * Delta * C[ant]~(mol ~ m ^ {-2}~dec^{-1}))
beta_CI_label <- expression(atop(beta,
                                 (mol ~ m ^ {-2} ~ µatm ^ {-1})))

4 Additive uncertainty

MLR_basins_reference <- unique(params_local_all$MLR_basins)

MLR_basins_sensitivity <- unique(params_local_all_ensemble$MLR_basins)
MLR_basins_sensitivity <- MLR_basins_sensitivity[!MLR_basins_sensitivity %in% MLR_basins_reference]

additive_uncertainty <- function(df){

df_regional <-
  df %>% 
  pivot_wider(names_from = MLR_basins,
              values_from = dcant) %>% 
  pivot_longer(all_of(MLR_basins_sensitivity),
               names_to = "MLR_basins",
               values_to = "dcant") %>% 
  mutate(delta_dcant = dcant - !!sym(MLR_basins_reference)) %>% 
  select(-c(all_of(MLR_basins_reference), MLR_basins, dcant))


df_regional_stat <-
  df_regional %>% 
  group_by(across(-delta_dcant)) %>%
  summarise(sd = mean(abs(delta_dcant))) %>%
  ungroup() %>%
  filter(Version_ID_group == unique(params_local_all$Version_ID_group)) %>%
  select(-Version_ID_group)

Version_ID_group_uncertainty <-
  df %>% distinct(Version_ID_group) %>% pull
Version_ID_group_uncertainty <-
  Version_ID_group_uncertainty[Version_ID_group_uncertainty != unique(params_local_all$Version_ID_group)]

df_configuration <-
  df %>% 
  pivot_wider(names_from = Version_ID_group,
              values_from = dcant) %>%
  pivot_longer(
    all_of(Version_ID_group_uncertainty),
    names_to = "Version_ID_group",
    values_to = "dcant"
  ) %>% 
  mutate(delta_dcant = dcant - !!sym(unique(params_local_all$Version_ID_group)))


df_configuration <-
  df_configuration %>%
  filter(MLR_basins == unique(params_local_all$MLR_basins)) %>%
  rename(sd = delta_dcant) %>% 
  select(-c(MLR_basins, dcant, unique(params_local_all$Version_ID_group)))

df_factorial <-
  bind_rows(
    df_regional_stat %>%
      mutate(Version_ID_group = "Regional clustering"),
    df_configuration
  )

df_factorial_stat <-
  df_factorial %>% 
  select(-Version_ID_group) %>% 
  group_by(across(-sd)) %>% 
  summarise(sd = sqrt(sum(sd^2))) %>% 
  ungroup()

df_stat <-
  full_join(
    df_factorial_stat,
    df %>%
      filter(
        Version_ID_group %in% unique(params_local_all$Version_ID_group),
        MLR_basins == MLR_basins_reference
      ) %>%
      rename(std_case = dcant) %>%
      select(-c(MLR_basins, Version_ID_group))
  )

df_stat <- df_stat %>%
  mutate(signif_single = if_else(abs(std_case) >= sd,
                          1,
                          0),
         signif_double = if_else(abs(std_case) >= sd*2,
                          1,
                          0))

df_stat_delta <-
  df_stat %>%
  select(-starts_with("signif_")) %>%
  arrange(period) %>%
  group_by(across(-c(period, sd, std_case))) %>%
  mutate(
    delta_dcant = std_case - lag(std_case),
    RSS = sqrt(sd ^ 2 + lag(sd ^ 2)),
    signif_single = if_else(abs(delta_dcant) >= RSS,
                            1,
                            0),
    signif_double = if_else(abs(delta_dcant) >= RSS*2,
                            1,
                            0)
  ) %>%
  ungroup() %>%
  drop_na()


df_out <- list(df_stat, df_factorial, df_stat_delta)

return(df_out)

}

4.1 Uncertainty factor

sd_factor <- 2

5 Coverage maps

cruises_phosphate_gap_fill <-
  c("33MW19930704",
    "33RO20030604",
    "33RO20050111",
    "33RO19980123")

cruises_talk_gap_fill <-
  c("06AQ19980328")

cruises_talk_calc <-
  c("06MT19900123",
    "316N19920502",
    "316N19921006")

GLODAP_all_coverage <- GLODAP_all %>% 
  distinct(lat, lon, era, cruise_expocode) %>% 
  mutate(gap_filling = case_when(
    cruise_expocode %in% cruises_phosphate_gap_fill ~ "Phosphate from CANYON-B",
    cruise_expocode %in% cruises_talk_gap_fill ~ "TA from CANYON-B",
    cruise_expocode %in% cruises_talk_calc ~ "TA calculated from other variables",
    cruise_expocode %in% "31DS19940126" ~ "Nitrate qc-flag not met",
    TRUE ~ "All quality criteria fulfilled"
  )) %>% 
  distinct(lat, lon, era, gap_filling)

colour("bright")(7)

     blue       red     green    yellow      cyan    purple      grey 
"#4477AA" "#EE6677" "#228833" "#CCBB44" "#66CCEE" "#AA3377" "#BBBBBB" 
attr(,"missing")
[1] NA

coverage_map <- map +
  geom_tile(data = GLODAP_all_coverage,
            aes(lon, lat)) +
  facet_wrap( ~ era, ncol = 1) +
  scale_color_okabeito() +
  scale_fill_okabeito() +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        legend.position = c(0.7,0.3),
        legend.title = element_blank())

coverage_map

Past versions of coverage_maps-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
91052ae	jens-daniel-mueller	2022-10-20
fea41c1	jens-daniel-mueller	2022-07-20
9ce772d	jens-daniel-mueller	2022-07-19
16dc3af	jens-daniel-mueller	2022-06-27
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04

ggsave(plot = coverage_map,
       path = here::here("output/presentation"),
       filename = "Fig_observations_coverage_map.png",
       height = 6,
       width = 4)


GLODAP_all_coverage <- GLODAP_all %>% 
  filter(between(depth, 250, 750)) %>% 
  group_by(lat, lon, era) %>%
  summarise(tco2 = mean(tco2, na.rm = TRUE)) %>% 
  ungroup()

map +
  geom_tile(data = GLODAP_all_coverage,
            aes(lon, lat, fill = tco2, col = tco2,
                height = 2, width = 2)) +
  facet_wrap( ~ era, ncol = 1) +
  scale_fill_viridis_c(name = expression(
    atop(DIC~at~"500m",(µmol~kg^{-1})))) +
  scale_color_viridis_c(guide = "none") +
  labs(title = "Observations",
       subtitle = "Decadal coverage") +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        legend.position = "bottom")

Past versions of coverage_maps-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
0b40366	jens-daniel-mueller	2022-10-26
91052ae	jens-daniel-mueller	2022-10-20

GLODAP_all_coverage_raster <- rast(GLODAP_all_coverage %>% 
                                     relocate(lon, lat) %>% 
                                     pivot_wider(names_from = era,
                                                 values_from = tco2),
                                   crs = "+proj=longlat")


GLODAP_all_coverage_raster <- project(GLODAP_all_coverage_raster, target_crs,
                                      method = "near")

GLODAP_all_coverage_tibble <- GLODAP_all_coverage_raster %>%
  as.data.frame(xy = TRUE, na.rm = FALSE) %>%
  as_tibble() %>%
  rename(lon = x, lat = y) %>%
  pivot_longer(cols = -c("lon", "lat"),
               names_to = "era",
               values_to = "tco2") %>% 
  drop_na()

coverage_map <- 
  ggplot() +
  geom_tile(data = GLODAP_all_coverage_tibble,
            aes(
              x = lon,
              y = lat,
              height = 3e5,
              width = 3e5,
              fill = tco2
            )) +
  geom_sf(data = worldmap_trans, fill = "grey90", col = "grey90") +
  geom_sf(data = coastline_trans, linewidth = 0.3) +
  geom_sf(data = bbox_graticules_trans, linewidth = 0.5) +
  coord_sf(
    crs = target_crs,
    ylim = lat_lim,
    xlim = lon_lim,
    expand = FALSE
  ) +
  scale_fill_viridis_c( expression(
    atop(DIC~at~"500m",(µmol~kg^{-1})))) +
  facet_wrap( ~ era, ncol = 1) +
  theme(
    axis.title = element_blank(),
    axis.text = element_blank(),
    axis.ticks = element_blank(),
    panel.border = element_rect(colour = "transparent"),
    strip.background = element_blank(),
    legend.position = "bottom"
  )



coverage_map

Past versions of coverage_maps-3.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

ggsave(plot = coverage_map,
       path = here::here("output/presentation"),
       filename = "Fig_observations_coverage_map_dic.png",
       height = 6,
       width = 4,
       dpi = 600)

rm(coverage_map, 
   GLODAP_all, GLODAP_all_coverage, GLODAP_expocodes,
   GLODAP_grid_era_all)

rm(cruises_phosphate_gap_fill,
   cruises_talk_gap_fill,
   cruises_talk_calc)

6 Atm CO2

co2_atm_roll <- co2_atm %>% 
  mutate(pCO2_roll = zoo::rollmean(pCO2, 10, fill = NA))

p_pco2_atm <- co2_atm_roll %>% 
  ggplot() +
  geom_point(aes(year, pCO2, col="annual")) +
  geom_path(aes(year, pCO2, col="annual")) +
  geom_point(aes(year, pCO2_roll, col="10yr roll")) +
  geom_path(aes(year, pCO2_roll, col="10yr roll"))

co2_atm_roll <- co2_atm_roll %>% 
  mutate(delta_pCO2_roll = pCO2_roll - lag(pCO2_roll,10),
         delta_pCO2 = pCO2 - lag(pCO2,10))

p_delta_pCO2_atm <- co2_atm_roll %>% 
  ggplot() +
  geom_point(aes(year, delta_pCO2, col="annual")) +
  geom_path(aes(year, delta_pCO2, col="annual")) +
  geom_point(aes(year, delta_pCO2_roll, col="10yr roll")) +
  geom_path(aes(year, delta_pCO2_roll, col="10yr roll"))

p_pco2_atm / p_delta_pCO2_atm

Past versions of atm_CO2-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
0160c40	jens-daniel-mueller	2022-07-16
022fd60	jens-daniel-mueller	2022-07-14

rm(p_pco2_atm, p_delta_pCO2_atm)

delta_pCO2_atm <-
  left_join(
    params_local_all %>%
      distinct(period, tref1, tref2),
    co2_atm_roll %>%
      select(
        tref1 = year,
        pCO2_1 = pCO2,
        pCO2_roll_1 = pCO2_roll
      )
  )

delta_pCO2_atm <-
  left_join(
    delta_pCO2_atm,
    co2_atm_roll %>%
      select(
        tref2 = year,
        pCO2_2 = pCO2,
        pCO2_roll_2 = pCO2_roll
      )
  )

delta_pCO2_atm <- delta_pCO2_atm %>% 
  mutate(delta_pCO2 = pCO2_2 - pCO2_1,
         delta_pCO2_roll = pCO2_roll_2 - pCO2_roll_1) %>% 
  select(period, delta_pCO2, delta_pCO2_roll)

delta_pCO2_atm <- delta_pCO2_atm %>%
  select(period, delta_pCO2)

delta_pCO2_atm <- 
  bind_rows(
    delta_pCO2_atm,
    tibble(period = "1800 - 1994",
           delta_pCO2 = co2_atm %>% filter(year==1994) %>% pull(pCO2) - 280)
  )

rm(co2_atm_roll)

7 Globale scaling

• storage beneath 3000m: +2%
• unmapped regions south of 65N: +1.5%
• Nordic Seas: +1%
• Arctic Ocean: +2%

dcant_scaling <- 1.02*(1+0.02+0.015+0.01+0.003)
dcant_scaling_uncertainty <- 0.5

8 eMLR stats

8.1 Predictor selection

lm_best_predictor_counts_all %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  pivot_longer(aou:temp,
               names_to = "predictor",
               values_to = "count") %>% 
  # mutate(count = as.factor(count)) %>% 
  group_by(predictor, count, period, Version_ID_group) %>% 
  summarise(n = n()) %>% 
  ungroup() %>% 
  ggplot(aes(x = count, y = n)) +
  scale_color_manual(values = c("#EE7733", "#009988"), name = "Decade") +
  scale_fill_manual(values = c("#EE7733", "#009988"), name = "Decade") +
  geom_path(aes(col = period)) +
    geom_point(aes(fill = period), shape = 21) +
  coord_flip() +
  scale_x_continuous(breaks = seq(0,10,2),
                     limits = c(0,10),
                     name = "Presence among 10 best models") +
  labs(y = "Counts across density slabs, regional clusters and basins") +
  facet_wrap(~predictor)

Past versions of predictors-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

8.2 Residual distribution

spatial_residual_all %>%
  filter(gamma_slab %in% params_local$plot_slabs[2],
         data_source == "obs",
         period %in% two_decades) %>%
  p_map_dcant_slab(
    var = ".resid_mean",
    col = "divergent",
    title_text = "Residual distribution"
    ) +
  facet_grid(era ~ period) +
  theme_dark()

Past versions of residual_latitude-1.png

Version	Author	Date
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

spatial_residual_offset_all %>%
  filter(
    gamma_slab %in% params_local$plot_slabs[1],
    data_source == "obs",
    period %in% two_decades
  ) %>%
  p_map_dcant_slab(var = ".resid_mean_offset",
                   col = "divergent",
                   title_text = "Residual offset distribution (era 2 - era 1)") +
  facet_grid(. ~ period) +
  theme_dark()

Past versions of residual_latitude-2.png

Version	Author	Date
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

lat_residual_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>%
  ggplot(aes(
    .resid_mean,
    lat_grid,
    col = era,
    group = interaction(basin, era, gamma_slab, Version_ID)
  )) +
  geom_vline(xintercept = 0) +
  # geom_point() +
  geom_path(alpha = 0.5) +
  facet_grid(. ~ period)

Past versions of residual_latitude-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

lat_residual_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>%
  select(-c(Version_ID, tref1, tref2)) %>% 
  mutate(sampling_period = if_else(period == "2004 - 2014" & 
                                     era == "2000-2009",
                                   "first",
                                   "second"),
         sampling_period = if_else(period == "1994 - 2004" & 
                                     era == "1989-1999",
                                   "first",
                                   sampling_period)) %>% 
  select(-era) %>% 
  pivot_wider(names_from = sampling_period,
              values_from = .resid_mean) %>% 
  ggplot(aes(first, second)) +
  geom_vline(xintercept = 0) +
  geom_hline(yintercept = 0) +
  geom_bin2d(binwidth = 0.5) +
  geom_abline(slope = 1, col = "red") +
  # geom_smooth(method = "lm") +
  scale_fill_viridis_c(trans = "log10", name = "n") +
  coord_fixed(ylim = c(-10, 10),
              xlim = c(-10, 10)) +
  labs(
    y = expression(atop("First sampling period", "mean MLR residual in 5° latitude bin (µmol k"*g^{-1}*")")),
    x = expression(atop("Second sampling period", "mean MLR residual in 5° latitude bin (µmol k"*g^{-1}*")"))
    )

Past versions of residual_latitude-4.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

  # facet_wrap(~period)

ggsave(path = here::here("output/publication"),
       filename = "FigS_MLR_residual_correlation_latitude_bin.png",
       height = 5,
       width = 5,
       dpi = 600)


lat_residual_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>%
  select(-c(Version_ID, tref1, tref2)) %>% 
  mutate(sampling_period = if_else(period == "2004 - 2014" & 
                                     era == "2000-2009",
                                   "first",
                                   "second"),
         sampling_period = if_else(period == "1994 - 2004" & 
                                     era == "1989-1999",
                                   "first",
                                   sampling_period)) %>% 
  select(-era) %>% 
  pivot_wider(names_from = sampling_period,
              values_from = .resid_mean) %>% 
  nest_by(period) %>%
  mutate(mod = list(lm(second ~ first, data = data))) %>%
  summarize(broom::tidy(mod))

# A tibble: 4 × 6
# Groups:   period [2]
  period      term        estimate std.error statistic  p.value
  <chr>       <chr>          <dbl>     <dbl>     <dbl>    <dbl>
1 1994 - 2004 (Intercept)   -0.140    0.0842     -1.66 9.75e- 2
2 1994 - 2004 first          0.647    0.0465     13.9  2.23e-36
3 2004 - 2014 (Intercept)    0.142    0.0899      1.58 1.15e- 1
4 2004 - 2014 first          0.604    0.0415     14.6  3.76e-39

lat_residual_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>%
  select(-c(Version_ID, tref1, tref2)) %>% 
  mutate(sampling_period = if_else(period == "2004 - 2014" & 
                                     era == "2000-2009",
                                   "first",
                                   "second"),
         sampling_period = if_else(period == "1994 - 2004" & 
                                     era == "1989-1999",
                                   "first",
                                   sampling_period)) %>% 
  select(-era) %>% 
  pivot_wider(names_from = sampling_period,
              values_from = .resid_mean) %>% 
  drop_na() %>% 
  group_by(period) %>%
  summarize(correlation = cor(first, second))

# A tibble: 2 × 2
  period      correlation
  <chr>             <dbl>
1 1994 - 2004       0.566
2 2004 - 2014       0.580

lat_residual_all_mean <- lat_residual_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>% 
  group_by(period, era, lat_grid) %>% 
  summarise(residual_mean = mean(abs(.resid_mean)),
            residual_sd = sd(.resid_mean),
            residual_iqr = IQR(.resid_mean)) %>% 
  ungroup()

lat_residual_offset_all_mean <- lat_residual_offset_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>% 
  group_by(period, lat_grid) %>% 
  summarise(residual_mean = mean(abs(.resid_mean_offset)),
            residual_sd = sd(.resid_mean_offset),
            residual_iqr = IQR(.resid_mean)) %>% 
  ungroup()

p_residual_lat <- 
ggplot() +
  geom_vline(xintercept = 0) +
  geom_ribbon(
    data = lat_residual_all_mean,
    aes(
      xmin = residual_mean - residual_iqr,
      xmax = residual_mean + residual_iqr,
      y = lat_grid,
      fill = era
    ),
    alpha = 0.3
  ) +
  geom_path(data = lat_residual_all_mean,
            aes(residual_mean,
                lat_grid,
                col = era)) +
  scale_fill_highcontrast() +
  scale_color_highcontrast() +
  coord_cartesian(xlim = c(-3,7)) +
  facet_grid(. ~ period)


p_residual_lat_offset <- 
ggplot() +
  geom_vline(xintercept = 0) +
  geom_ribbon(
    data = lat_residual_offset_all_mean,
    aes(
      xmin = residual_mean - residual_iqr,
      xmax = residual_mean + residual_iqr,
      y = lat_grid
    ),
    alpha = 0.3
  ) +
  geom_path(data = lat_residual_offset_all_mean,
            aes(residual_mean,
                lat_grid)) +
  coord_cartesian(xlim = c(-3,7)) +
  facet_grid(. ~ period)

p_residual_lat / p_residual_lat_offset

Past versions of residual_latitude-5.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

rm(p_residual_lat, p_residual_lat_offset)


p_residual_lat <-
ggplot() +
  geom_vline(xintercept = 0) +
  geom_path(data = lat_residual_all %>% 
              filter(data_source == "obs",
                     period %in% two_decades,
                     gamma_slab %in% params_local$plot_slabs[2]),
            aes(.resid_mean,
                lat_grid,
                col = era)) +
  scale_fill_highcontrast() +
  scale_color_highcontrast() +
  coord_cartesian(xlim = c(-3,7)) +
  facet_grid(basin ~ period)

p_residual_lat_offset <-
ggplot() +
  geom_vline(xintercept = 0) +
  geom_path(data = lat_residual_offset_all %>% 
              filter(data_source == "obs",
                     period %in% two_decades,
                     gamma_slab %in% params_local$plot_slabs[2]),
            aes(.resid_mean_offset,
                lat_grid)) +
  scale_fill_highcontrast() +
  scale_color_highcontrast() +
  coord_cartesian(xlim = c(-3,7)) +
  facet_grid(basin ~ period)


p_residual_lat / p_residual_lat_offset

Past versions of residual_latitude-6.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01

rm(p_residual_lat, p_residual_lat_offset)

lat_residual_offset_all %>% 
  ggplot(aes(.resid_mean, .resid_mean_offset)) +
  geom_bin_2d() +
  scale_fill_viridis_c(trans = "log10")

Past versions of residual_latitude-7.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
7275091	jens-daniel-mueller	2022-11-05

lat_residual_offset_all %>%
  pivot_longer(starts_with(".resid"),
               names_to = "type",
               values_to = "residual") %>%
  ggplot(aes(residual, fill = type, col = type)) +
  coord_cartesian(xlim = c(-5,5)) +
  geom_density(alpha = 0.5)

Past versions of residual_latitude-8.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
7275091	jens-daniel-mueller	2022-11-05

left_join(lm_best_target_all,
          GLODAP_glanced_all) %>% 
  summarise(mean_rmse = mean(rmse, na.rm = TRUE),
            sd_rmse = sd(rmse, na.rm = TRUE))

# A tibble: 1 × 2
  mean_rmse sd_rmse
      <dbl>   <dbl>
1      3.10    1.63

9 Column inventories

dcant_inv_all <- bind_rows(
  dcant_inv_all,
  tcant_inv %>% mutate(period = paste(tref1, tref2, sep = " - "),
                       data_source = "obs")
)


dcant_inv_all <- left_join(dcant_inv_all,
                           delta_pCO2_atm)


dcant_inv_all <- dcant_inv_all %>% 
  mutate(beta = dcant / delta_pCO2,
         beta_pos = dcant_pos / delta_pCO2) %>% 
  select(-starts_with("pCO2"))

dcant_inv_all_ensemble <- dcant_inv_all %>% 
  filter(Version_ID %in% Version_IDs_ensemble_uncertainty)

dcant_inv_all_sensitivity <- dcant_inv_all %>% 
  filter(Version_ID %in% Version_IDs_ensemble_sensitivity)

dcant_inv_all <- dcant_inv_all %>% 
  filter(Version_ID %in% Version_IDs | is.na(Version_ID))

9.1 Sensitivity cases

dcant_inv_all_ensemble_sensitivity <-
  dcant_inv_all_sensitivity %>%
  filter(period %in% two_decades,
         data_source != "mod_truth") %>%
  select(lon, lat, period, data_source, MLR_basins, Version_ID_group, dcant)

dcant_inv_sensitivity <- additive_uncertainty(dcant_inv_all_ensemble_sensitivity)

delta_dcant_inv_sensitivity <- dcant_inv_sensitivity[[3]]
dcant_inv_sensitivity_contributions <- dcant_inv_sensitivity[[2]]
dcant_inv_sensitivity <- dcant_inv_sensitivity[[1]]


dcant_inv_sensitivity %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Synthetic data",
      data_source == "obs" ~ "Observations"
    )
  ) %>%
  p_map_cant_inv(var = "sd",
                 breaks = c(seq(0, 16, 2), Inf),
                 title_text = NULL) +
  facet_grid(period ~ data_source) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_sensitivity_cases-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
5520de8	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
cc337dd	jens-daniel-mueller	2022-11-11

dcant_inv_sensitivity_contributions %>%
  filter(data_source == "obs",
         Version_ID_group != "Regional clustering") %>% 
  rename(dcant_diff = sd) %>% 
  p_map_cant_inv(var = "dcant_diff",
                 col = "bias",
                 breaks = c(-Inf, seq(-2,2,1), Inf),
                 title_text = NULL) +
  facet_grid(Version_ID_group ~ period) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_sensitivity_cases-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
e924c80	jens-daniel-mueller	2022-11-17
5520de8	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
cc337dd	jens-daniel-mueller	2022-11-11

legend_title <- expression(atop(Delta * C["ant"] ~ offset,
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_map_mdim_robinson(
  df = dcant_inv_sensitivity_contributions %>%
    filter(data_source == "obs",
           Version_ID_group != "Regional clustering") %>% 
    select(-data_source),
  dim_col = "period",
  dim_row = "Version_ID_group",
  var = "sd",
  breaks = c(-Inf, seq(-8, 8, 2), Inf),
  legend_title = legend_title,
  col = "divergent"
)

Past versions of column_inventory_additive_uncertainty_sensitivity_cases-3.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_column_inventory_uncertainty_sensitivity_cases.png",
       height = 8,
       width = 8,
       dpi = 600)

9.2 Uncertainty cases

dcant_inv_all_ensemble_uncertainty <-
  dcant_inv_all_ensemble %>%
  filter(period %in% two_decades,
         data_source != "mod_truth") %>%
  select(lon, lat, period, data_source, MLR_basins, Version_ID_group, dcant)

dcant_inv_uncertainty <- additive_uncertainty(dcant_inv_all_ensemble_uncertainty)

delta_dcant_inv_uncertainty <- dcant_inv_uncertainty[[3]]
dcant_inv_uncertainty_contributions <- dcant_inv_uncertainty[[2]]
dcant_inv_uncertainty <- dcant_inv_uncertainty[[1]]


dcant_inv_uncertainty %>%
  filter(data_source == "obs") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Synthetic data",
      data_source == "obs" ~ "Observations"
    )
  ) %>%
  p_map_cant_inv(var = "sd",
                 breaks = c(seq(0, 8, 1), Inf),
                 title_text = NULL) +
  facet_grid(. ~ period) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_fine_grid-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
57d1a9b	jens-daniel-mueller	2022-11-17
e924c80	jens-daniel-mueller	2022-11-17
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
cc337dd	jens-daniel-mueller	2022-11-11

legend_title <- expression(atop(Delta * C["ant"],
                                atop("1" * sigma ~ -uncertainty,
                                     (mol ~ m ^ {
                                       -2
                                     } ~ dec ^ {
                                       -1
                                     }))))

p_map_mdim_robinson(
  df = dcant_inv_uncertainty %>%
    filter(data_source == "obs") %>%
    select(lon, lat, period, sd),
  dim_col = "period",
  var = "sd",
  breaks = c(seq(0, 8, 1), Inf),
  legend_title = legend_title
)

Past versions of column_inventory_additive_uncertainty_fine_grid-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
cc337dd	jens-daniel-mueller	2022-11-11

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_column_inventory_uncertainty.png",
       height = 3.5,
       width = 8,
       dpi = 600)

dcant_inv_uncertainty_contributions %>%
  filter(data_source == "obs") %>% 
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Synthetic data",
      data_source == "obs" ~ "Observations"
    )
  ) %>%
  rename(dcant_diff_uncertainty = sd) %>% 
  p_map_cant_inv(var = "dcant_diff_uncertainty",
                 col = "bias",
                 breaks = c(-Inf, seq(-2,2,1), Inf),
                 title_text = NULL) +
  facet_grid(Version_ID_group ~ period) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_fine_grid-3.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

legend_title <- expression(atop(Delta * C["ant"] ~ offset,
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_map_mdim_robinson(
  df = dcant_inv_uncertainty_contributions %>%
    filter(data_source == "obs") %>%
    select(-data_source),
  dim_col = "period",
  dim_row = "Version_ID_group",
  var = "sd",
  breaks = c(-Inf, seq(-4, 4, 1), Inf),
  legend_title = legend_title,
  col = "divergent"
)

Past versions of column_inventory_additive_uncertainty_fine_grid-4.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_column_inventory_uncertainty_contributions.png",
       height = 9,
       width = 8,
       dpi = 600)

dcant_inv_all_ensemble_uncertainty <-
  dcant_inv_all_ensemble %>%
  filter(period %in% two_decades,
         data_source != "mod_truth") %>%
  select(lon, lat, period, data_source, MLR_basins, Version_ID_group, dcant)

dcant_inv_all_ensemble_uncertainty <- 
  m_grid_horizontal_coarse(dcant_inv_all_ensemble_uncertainty)

dcant_inv_all_ensemble_uncertainty <-
  dcant_inv_all_ensemble_uncertainty %>% 
  group_by(lat_grid, lon_grid, data_source, period, MLR_basins, Version_ID_group) %>% 
  summarise(dcant = mean(dcant),
            n = n()) %>% 
  ungroup() %>% 
  filter(n == 25) %>% 
  rename(lon = lon_grid,
         lat = lat_grid)

dcant_inv_uncertainty <- additive_uncertainty(dcant_inv_all_ensemble_uncertainty)

delta_dcant_inv_uncertainty <- dcant_inv_uncertainty[[3]]
dcant_inv_uncertainty_contributions <- dcant_inv_uncertainty[[2]]
dcant_inv_uncertainty <- dcant_inv_uncertainty[[1]]


dcant_inv_uncertainty %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Synthetic data",
      data_source == "obs" ~ "Observations"
    )
  ) %>%
  p_map_cant_inv(var = "sd",
                 breaks = c(seq(0, 8, 1), Inf),
                 title_text = NULL) +
  facet_grid(period ~ data_source) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_coarse_grid-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
cc337dd	jens-daniel-mueller	2022-11-11

delta_dcant_inv_uncertainty %>% 
  p_map_cant_inv(var = "delta_dcant",
                 col = "bias",
                 title_text = NULL) +
  facet_grid(data_source ~ .) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_coarse_grid-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
cc337dd	jens-daniel-mueller	2022-11-11

delta_dcant_inv_uncertainty %>% 
  p_map_cant_inv(var = "RSS",
                 breaks = c(seq(0, 8, 1), Inf),
                 title_text = NULL) +
  facet_grid(data_source ~ .) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_coarse_grid-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
cc337dd	jens-daniel-mueller	2022-11-11

delta_dcant_inv_uncertainty %>% 
  filter(signif_single == 1) %>% 
  p_map_cant_inv(var = "delta_dcant",
                 col = "bias",
                 title_text = NULL) +
  facet_grid(data_source ~ .) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_coarse_grid-4.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
cc337dd	jens-daniel-mueller	2022-11-11

delta_dcant_inv_uncertainty %>% 
  filter(signif_double == 1) %>% 
  p_map_cant_inv(var = "delta_dcant",
                 col = "bias",
                 title_text = NULL) +
  facet_grid(data_source ~ .) +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of column_inventory_additive_uncertainty_coarse_grid-5.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25

9.3 dCant - absolute

p_CI_absolute <- dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source %in% c("obs")) %>%
  p_map_cant_inv(var = "dcant",
                 title_text = NULL) +
  facet_wrap(~ period, ncol = 1) +
  theme(axis.text.x = element_blank()) +
  geom_point(
    data = dcant_inv_uncertainty %>%
      filter(signif_double == 0,
             data_source == "obs"),
    aes(lon, lat),
    shape = 20,
    size = 0.1,
    col = "grey"
  ) +
    geom_point(
    data = dcant_inv_uncertainty %>%
      filter(signif_single == 0,
             data_source == "obs"),
    aes(lon, lat),
    shape = 20,
    size = 1,
    col = "grey"
  )

p_CI_absolute +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of inventory_maps_absolute_obs-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

dcant_inv_uncertainty %>%
  select(data_source, lon, lat, period, starts_with("signif")) %>%
  mutate(area = earth_surf(lat,lon)) %>% 
  group_by(data_source, period, signif_single) %>% 
  summarize(area_signif = sum(area)) %>% 
  ungroup() %>% 
  group_by(data_source, period) %>% 
  mutate(area_total = sum(area_signif)) %>% 
  ungroup() %>% 
  mutate(area_fraction = 100 * area_signif / area_total)

# A tibble: 12 × 6
   data_source period      signif_single area_signif area_total area_fraction
   <chr>       <chr>               <dbl>       <dbl>      <dbl>         <dbl>
 1 mod         1994 - 2004             0     2.67e11    1.11e13        2.40  
 2 mod         1994 - 2004             1     1.09e13    1.11e13       97.5   
 3 mod         1994 - 2004            NA     1.65e10    1.11e13        0.148 
 4 mod         2004 - 2014             0     3.57e10    1.11e13        0.320 
 5 mod         2004 - 2014             1     1.11e13    1.11e13       99.5   
 6 mod         2004 - 2014            NA     1.65e10    1.11e13        0.148 
 7 obs         1994 - 2004             0     8.74e11    1.13e13        7.76  
 8 obs         1994 - 2004             1     1.04e13    1.13e13       92.2   
 9 obs         1994 - 2004            NA     9.81e 9    1.13e13        0.0872
10 obs         2004 - 2014             0     1.07e12    1.13e13        9.48  
11 obs         2004 - 2014             1     1.02e13    1.13e13       90.4   
12 obs         2004 - 2014            NA     9.81e 9    1.13e13        0.0872

9.4 Inventory stats

9.4.1 Basin

dcant_inv_stat_mean <-
  full_join(basinmask,
            dcant_inv_all) %>%
  mutate(surf_area = earth_surf(lat, lon)) %>%
  group_by(basin, period, data_source) %>%
  summarise(dcant = weighted.mean(dcant, surf_area)) %>%
  ungroup() %>%
  filter(!is.na(dcant))

dcant_inv_stat_ensemble <-
  full_join(basinmask,
            dcant_inv_all_ensemble) %>%
  mutate(surf_area = earth_surf(lat, lon)) %>%
  group_by(basin, data_source, period, MLR_basins, Version_ID_group) %>%
  summarise(dcant = weighted.mean(dcant, surf_area)) %>%
  ungroup() %>%
  filter(!is.na(dcant)) %>%
  group_by(basin, data_source, period) %>%
  summarise(dcant_sd = sd(dcant) * 2) %>%
  ungroup()

dcant_inv_stat <-
  full_join(dcant_inv_stat_mean,
            dcant_inv_stat_ensemble)

rm(dcant_inv_stat_mean,
   dcant_inv_stat_ensemble)

dcant_inv_stat %>%
  filter(period %in% two_decades) %>%
  group_by(data_source, basin) %>%
  mutate(delta_dcant = dcant - lag(dcant),
         RSS = sqrt(dcant_sd ^ 2 + lag(dcant_sd ^ 2))) %>%
  ungroup() %>%
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

basin	period	data_source	dcant	dcant_sd	delta_dcant	RSS
N. Pacific	1994 - 2004	mod	3.363703	0.3712442	NA	NA
N. Pacific	1994 - 2004	mod_truth	2.915800	0.0000000	NA	NA
N. Pacific	1994 - 2004	obs	3.076873	1.8983189	NA	NA
N. Pacific	2004 - 2014	mod	3.614817	0.4403483	0.2511143	0.5759591
N. Pacific	2004 - 2014	mod_truth	3.352012	0.0000000	0.4362120	0.0000000
N. Pacific	2004 - 2014	obs	3.415398	1.5195985	0.3385254	2.4316238
S. Pacific	1994 - 2004	mod	4.826130	1.2656840	NA	NA
S. Pacific	1994 - 2004	mod_truth	4.931744	0.0000000	NA	NA
S. Pacific	1994 - 2004	obs	7.482479	1.3015988	NA	NA
S. Pacific	2004 - 2014	mod	6.476528	0.8515919	1.6503976	1.5255048
S. Pacific	2004 - 2014	mod_truth	6.060215	0.0000000	1.1284704	0.0000000
S. Pacific	2004 - 2014	obs	6.402648	1.5413944	-1.0798310	2.0174380
N. Atlantic	1994 - 2004	mod	4.616514	0.9132060	NA	NA
N. Atlantic	1994 - 2004	mod_truth	4.057536	0.0000000	NA	NA
N. Atlantic	1994 - 2004	obs	9.867769	0.6490727	NA	NA
N. Atlantic	2004 - 2014	mod	5.075378	0.6845292	0.4588643	1.1412823
N. Atlantic	2004 - 2014	mod_truth	4.878549	0.0000000	0.8210126	0.0000000
N. Atlantic	2004 - 2014	obs	8.025765	1.5821054	-1.8420048	1.7100740
S. Atlantic	1994 - 2004	mod	5.064488	0.9895347	NA	NA
S. Atlantic	1994 - 2004	mod_truth	4.706655	0.0000000	NA	NA
S. Atlantic	1994 - 2004	obs	7.252470	1.6530468	NA	NA
S. Atlantic	2004 - 2014	mod	6.393315	0.7347411	1.3288264	1.2324867
S. Atlantic	2004 - 2014	mod_truth	5.251166	0.0000000	0.5445113	0.0000000
S. Atlantic	2004 - 2014	obs	9.960798	2.0284147	2.7083280	2.6166830
Indian	1994 - 2004	mod	5.899824	2.1898564	NA	NA
Indian	1994 - 2004	mod_truth	5.570926	0.0000000	NA	NA
Indian	1994 - 2004	obs	8.154922	1.5925219	NA	NA
Indian	2004 - 2014	mod	7.583920	1.2120903	1.6840960	2.5029251
Indian	2004 - 2014	mod_truth	6.501337	0.0000000	0.9304108	0.0000000
Indian	2004 - 2014	obs	6.519972	0.7627870	-1.6349503	1.7657775

9.4.2 Gyres

dcant_inv_stat_ensemble <-
  full_join(basinmask,
            dcant_inv_all_ensemble) %>%
  mutate(surf_area = earth_surf(lat, lon),
         lat_grid = cut(lat, c(-50, -20, 20, 50))) %>%
  group_by(basin,
           data_source,
           period,
           MLR_basins,
           Version_ID_group,
           lat_grid) %>%
  summarise(dcant = weighted.mean(dcant, surf_area)) %>%
  ungroup() %>%
  filter(!is.na(dcant),
         !is.na(lat_grid),
         lat_grid != "(-20,20]")

dcant_inv_stat_uncertainty <- additive_uncertainty(dcant_inv_stat_ensemble %>%
                                                     filter(period %in% two_decades))

delta_dcant_inv_stat_uncertainty <- dcant_inv_stat_uncertainty[[3]]
dcant_inv_stat_uncertainty <- dcant_inv_stat_uncertainty[[1]]


dcant_inv_stat_uncertainty %>%
  filter(data_source == "obs") %>%
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

basin	data_source	period	lat_grid	sd	std_case	signif_single	signif_double
N. Pacific	obs	1994 - 2004	(20,50]	0.9900593	3.9829040	1	1
N. Pacific	obs	2004 - 2014	(20,50]	2.3041469	3.3833004	1	0
S. Pacific	obs	1994 - 2004	(-50,-20]	1.4359046	9.8048932	1	1
S. Pacific	obs	2004 - 2014	(-50,-20]	1.1921319	8.5667582	1	1
N. Atlantic	obs	1994 - 2004	(20,50]	0.5388493	12.1361324	1	1
N. Atlantic	obs	2004 - 2014	(20,50]	0.6289353	8.4461849	1	1
S. Atlantic	obs	1994 - 2004	(-50,-20]	0.9183569	8.4912458	1	1
S. Atlantic	obs	2004 - 2014	(-50,-20]	1.9497286	13.1557314	1	1
Indian	obs	1994 - 2004	(-50,-20]	0.7599534	10.7970711	1	1
Indian	obs	1994 - 2004	(20,50]	4.2501720	-0.3174586	0	0
Indian	obs	2004 - 2014	(-50,-20]	1.0554965	8.2468486	1	1
Indian	obs	2004 - 2014	(20,50]	2.1877358	7.0876343	1	1

delta_dcant_inv_stat_uncertainty %>%
  filter(data_source == "obs") %>%
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

basin	data_source	period	lat_grid	sd	std_case	delta_dcant	RSS	signif_single	signif_double
N. Pacific	obs	2004 - 2014	(20,50]	2.3041469	3.383300	-0.5996036	2.5078498	0	0
S. Pacific	obs	2004 - 2014	(-50,-20]	1.1921319	8.566758	-1.2381350	1.8662799	0	0
N. Atlantic	obs	2004 - 2014	(20,50]	0.6289353	8.446185	-3.6899474	0.8282018	1	1
S. Atlantic	obs	2004 - 2014	(-50,-20]	1.9497286	13.155731	4.6644856	2.1551847	1	1
Indian	obs	2004 - 2014	(-50,-20]	1.0554965	8.246849	-2.5502226	1.3006160	1	0
Indian	obs	2004 - 2014	(20,50]	2.1877358	7.087634	7.4050929	4.7801830	1	0

rm(dcant_inv_stat_ensemble, 
   dcant_inv_stat_uncertainty, 
   delta_dcant_inv_stat_uncertainty)

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source != "obs") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Reconstruction",
      data_source == "mod_truth" ~ "Model truth"
    )
  ) %>%
  p_map_cant_inv(var = "dcant",
                 title_text = NULL) +
  facet_grid(period ~ data_source) +
  theme(axis.text.x = element_blank())

Past versions of inventory_maps_delta_dcant_mod_bias-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

legend_title <- expression(atop(Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_CI_absolute_mod_delta <- p_map_mdim_robinson(
  df = dcant_inv_all %>%
    filter(period %in% two_decades,
           data_source != "obs") %>%
    mutate(
      data_source = case_when(
        data_source == "mod" ~ "Reconstruction",
        data_source == "mod_truth" ~ "Model truth"
      )
    ) %>% 
    select(lon, lat, period, data_source, dcant),
  dim_row = "period",
  dim_col = "data_source",
  var = "dcant",
  legend_title = legend_title
)

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source != "obs") %>%
  select(data_source, lon, lat, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
         period = "Decadal offset") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Reconstruction",
      data_source == "mod_truth" ~ "Model truth"
    )
  ) %>%
  p_map_cant_inv(
    var = "dcant_offset",
    title_text = NULL,
    subtitle_text = NULL,
    col = "bias"
  ) +
  facet_grid(. ~ data_source)

Past versions of inventory_maps_delta_dcant_mod_bias-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
ebc85e2	jens-daniel-mueller	2022-11-14
7275091	jens-daniel-mueller	2022-11-05

        legend_title <- expression(atop(Delta * Delta * C["ant"],
                                        (mol ~ m ^ {-2}~dec^{-1})))

p_CI_bias_delta <- p_map_mdim_robinson(
  df = dcant_inv_all %>%
    filter(period %in% two_decades,
           data_source != "obs") %>%
    select(data_source, lon, lat, basin_AIP, period, dcant) %>%
    pivot_wider(names_from = period,
                values_from = dcant) %>%
    mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
           period = "Decadal offset") %>%
    mutate(
      data_source = case_when(
        data_source == "mod" ~ "Reconstruction",
        data_source == "mod_truth" ~ "Model truth"
      )
    ) %>% 
    select(lon, lat, data_source, dcant_offset),
  dim_col = "data_source",
  var = "dcant_offset",
  breaks = c(-Inf, seq(-8, 8, 2), Inf),
  legend_title = legend_title,
  col = "divergent"
)

p_CI_absolute_mod_delta / p_CI_bias_delta + 
  plot_annotation(tag_levels = 'A')

Past versions of inventory_maps_delta_dcant_mod_bias-3.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_delta_dcant_column_inventory_synthetic_data.png",
       height = 7,
       width = 8,
       dpi = 600)

rm(p_CI_absolute_mod_delta, p_CI_bias_delta)

9.5 dCant - delta

delta_dcant_inv_all <-
  dcant_inv_all %>%
  filter(data_source %in% c("obs", "mod"),
         period %in% two_decades) %>%
  select(data_source, lon, lat, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
         period = "(2004 - 2014) - (1994 - 2004)")

p_CI_delta <-
  delta_dcant_inv_all %>%
  filter(data_source == "obs") %>%
  p_map_cant_inv(
    var = "dcant_offset",
    title_text = NULL,
    subtitle_text = NULL,
    col = "bias"
  ) +
  geom_point(
    data = delta_dcant_inv_uncertainty %>%
      filter(signif_single == 0,
             data_source == "obs") %>% 
      mutate(period = "(2004 - 2014) - (1994 - 2004)"),
    aes(lon, lat),
    shape = 20,
    size = 1,
    col = "black"
  ) +
  geom_point(
    data = delta_dcant_inv_uncertainty %>%
      filter(signif_double == 0,
             data_source == "obs") %>% 
      mutate(period = "(2004 - 2014) - (1994 - 2004)"),
    aes(lon, lat),
    shape = 20,
    size = 0.1,
    col = "black"
  ) +
  facet_wrap(~ period)

p_CI_absolute / p_CI_delta + 
  plot_annotation(tag_levels = 'A')

Past versions of inventory_maps_absolute_delta-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
e2adac6	jens-daniel-mueller	2022-07-22
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
ea46812	jens-daniel-mueller	2022-07-13
f09080e	jens-daniel-mueller	2022-06-28
16dc3af	jens-daniel-mueller	2022-06-27
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
094bfa0	jens-daniel-mueller	2022-02-18
d2191ad	jens-daniel-mueller	2022-02-04
5f2aed0	jens-daniel-mueller	2022-01-27
9753eb8	jens-daniel-mueller	2022-01-26
b1d7720	jens-daniel-mueller	2022-01-21
d6b399a	jens-daniel-mueller	2022-01-21

p_CI_delta +
  theme(axis.text = element_blank(),
        axis.ticks = element_blank())

Past versions of inventory_maps_absolute_delta-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
e2adac6	jens-daniel-mueller	2022-07-22
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
ea46812	jens-daniel-mueller	2022-07-13
f09080e	jens-daniel-mueller	2022-06-28
16dc3af	jens-daniel-mueller	2022-06-27
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
094bfa0	jens-daniel-mueller	2022-02-18
d2191ad	jens-daniel-mueller	2022-02-04
5f2aed0	jens-daniel-mueller	2022-01-27
9753eb8	jens-daniel-mueller	2022-01-26
b1d7720	jens-daniel-mueller	2022-01-21
d6b399a	jens-daniel-mueller	2022-01-21

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_column_inventory_delta.png",
       height = 3.2,
       width = 6)

rm(p_CI_absolute, p_CI_delta)

dcant_inv_dec1 <- dcant_inv_all %>%
  filter(data_source == "obs",
         period == "1994 - 2004") %>%
  select(lon, lat, dcant)

dcant_inv_dec2 <- dcant_inv_all %>%
  filter(data_source == "obs",
         period == "2004 - 2014") %>%
  select(lon, lat, dcant)

delta_dcant_inv <- delta_dcant_inv_all %>%
  filter(data_source == "obs") %>%
  select(lon, lat, dcant_offset)

dcant_inv_uncert_dec1 <- dcant_inv_uncertainty %>%
  filter(data_source == "obs",
         period == "1994 - 2004") %>%
  select(lon, lat, signif_single, signif_double) %>%
  st_as_sf(coords = c("lon", "lat"), crs = "+proj=longlat")

dcant_inv_uncert_dec2 <- dcant_inv_uncertainty %>%
  filter(data_source == "obs",
         period == "2004 - 2014") %>%
  select(lon, lat, signif_single, signif_double) %>%
  st_as_sf(coords = c("lon", "lat"), crs = "+proj=longlat")

dcant_inv_uncert_delta <- delta_dcant_inv_uncertainty %>%
  filter(data_source == "obs") %>%
  select(lon, lat, signif_single, signif_double) %>%
  st_as_sf(coords = c("lon", "lat"), crs = "+proj=longlat")


legend_title <- expression(atop(Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))


p_dcant_inv_dec1 <-
  p_map_mdim_robinson(df = dcant_inv_dec1,
                      df_uncertainty = dcant_inv_uncert_dec1,
                      var = "dcant",
                      legend_title = legend_title) +
  geom_label(aes(
    label = "1994 - 2004",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_dec1

Past versions of dcant_column_inventory_maps_robinson-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

p_dcant_inv_dec2 <-
  p_map_mdim_robinson(df = dcant_inv_dec2,
                      df_uncertainty = dcant_inv_uncert_dec2,
                      var = "dcant",
                      legend_title = legend_title) +
  geom_label(aes(
    label = "2004 - 2014",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_dec2

Past versions of dcant_column_inventory_maps_robinson-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

legend_title <- expression(atop(Delta * Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_dcant_inv_delta <-
  p_map_mdim_robinson(df = delta_dcant_inv,
                      df_uncertainty = dcant_inv_uncert_delta,
                      var = "dcant_offset",
                      breaks = c(-Inf, seq(-8, 8, 2), Inf),
                      legend_title = legend_title,
                      col = "divergent") +
  geom_label(aes(
    label = "Decadal difference",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_delta

Past versions of dcant_column_inventory_maps_robinson-3.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

(p_dcant_inv_dec1 / p_dcant_inv_dec2 + plot_layout(guides = "collect")) -
  p_dcant_inv_delta +
  plot_layout(ncol = 1) +
  plot_annotation(tag_levels = 'A')

Past versions of dcant_column_inventory_maps_robinson-4.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "Fig_dcant_column_inventory.png",
       height = 10,
       width = 8,
       dpi = 600)

legend_title <- expression(atop(Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))


p_dcant_inv_dec1 <-
  p_map_mdim_robinson(df = dcant_inv_dec1,
                      var = "dcant",
                      legend_title = legend_title) +
  geom_label(aes(
    label = "1994 - 2004",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_dec2 <-
  p_map_mdim_robinson(df = dcant_inv_dec2,
                      var = "dcant",
                      legend_title = legend_title) +
  geom_label(aes(
    label = "2004 - 2014",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_dec1 / p_dcant_inv_dec2 + 
  plot_layout(guides = "collect") +
  plot_layout(ncol = 1)

Past versions of dcant_column_inventory_maps_robinson_presentation-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_column_inventory.png",
       height = 7,
       width = 8,
       dpi = 600)

p_dcant_inv_dec1 <-
  p_map_mdim_robinson(df = dcant_inv_dec1,
                      df_uncertainty = dcant_inv_uncert_dec1,
                      var = "dcant",
                      legend_title = legend_title) +
  geom_label(aes(
    label = "1994 - 2004",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_dec1

Past versions of dcant_column_inventory_maps_robinson_presentation-2.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

p_dcant_inv_dec2 <-
  p_map_mdim_robinson(df = dcant_inv_dec2,
                      df_uncertainty = dcant_inv_uncert_dec2,
                      var = "dcant",
                      legend_title = legend_title) +
  geom_label(aes(
    label = "2004 - 2014",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_dec2

Past versions of dcant_column_inventory_maps_robinson_presentation-3.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

p_dcant_inv_dec1 / p_dcant_inv_dec2 + 
  plot_layout(guides = "collect") +
  plot_layout(ncol = 1)

Past versions of dcant_column_inventory_maps_robinson_presentation-4.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_column_inventory_uncertainty.png",
       height = 7,
       width = 8,
       dpi = 600)




legend_title <- expression(atop(Delta * Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_dcant_inv_delta <-
  p_map_mdim_robinson(df = delta_dcant_inv,
                      var = "dcant_offset",
                      breaks = c(-Inf, seq(-8, 8, 2), Inf),
                      legend_title = legend_title,
                      col = "divergent") +
  geom_label(aes(
    label = "Decadal difference",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_delta

Past versions of dcant_column_inventory_maps_robinson_presentation-5.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

ggsave(path = here::here("output/presentation"),
       filename = "Fig_delta_dcant_column_inventory.png",
       height = 4,
       width = 8,
       dpi = 600)

p_dcant_inv_delta <-
  p_map_mdim_robinson(df = delta_dcant_inv,
                      df_uncertainty = dcant_inv_uncert_delta,
                      var = "dcant_offset",
                      breaks = c(-Inf, seq(-8, 8, 2), Inf),
                      legend_title = legend_title,
                      col = "divergent") +
  geom_label(aes(
    label = "Decadal difference",
    x = lon_lim[1] * 0.55,
    y = lat_lim[2] * 0.85
  ))


p_dcant_inv_delta

Past versions of dcant_column_inventory_maps_robinson_presentation-6.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

ggsave(path = here::here("output/presentation"),
       filename = "Fig_delta_dcant_column_inventory_uncertainty.png",
       height = 4,
       width = 8,
       dpi = 600)

9.6 Bias

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source != "obs") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Reconstruction",
      data_source == "mod_truth" ~ "Model truth"
    )
  ) %>%
  p_map_cant_inv(var = "dcant",
                 title_text = NULL) +
  facet_grid(data_source ~ period) +
  theme(axis.text.x = element_blank())

Past versions of inventory_maps_absolute_mod-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

legend_title <- expression(atop(Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

 p_CI_absolute_mod <- p_map_mdim_robinson(
  df = dcant_inv_all %>%
    filter(period %in% two_decades,
           data_source != "obs") %>%
    mutate(
      data_source = case_when(
        data_source == "mod" ~ "Reconstruction",
        data_source == "mod_truth" ~ "Model truth"
      )
    ) %>% 
    select(lon, lat, period, data_source, dcant),
  dim_col = "period",
  dim_row = "data_source",
  var = "dcant",
  legend_title = legend_title
)


p_CI_absolute_mod

Past versions of inventory_maps_absolute_mod-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source != "obs") %>%
  select(data_source, lon, lat, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = data_source,
              values_from = dcant) %>%
  mutate(dcant_bias = mod - mod_truth,
         data_source = "Reconstruction bias") %>%
  p_map_cant_inv(
    var = "dcant_bias",
    title_text = NULL,
    subtitle_text = NULL,
    col = "bias"
  ) +
  facet_grid(data_source ~ period) +
  theme(strip.text.x = element_blank(),
        strip.background.x = element_blank())

Past versions of invetory_maps_absolute_mod_bias-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
5b77c4f	jens-daniel-mueller	2022-08-10

legend_title <- expression(atop(Delta * C["ant"] ~ bias,
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_CI_bias <- p_map_mdim_robinson(
  df = dcant_inv_all %>%
    filter(period %in% two_decades,
           data_source != "obs") %>%
    select(data_source, lon, lat, basin_AIP, period, dcant) %>%
    pivot_wider(names_from = data_source,
                values_from = dcant) %>%
    mutate(dcant_bias = mod - mod_truth,
           data_source = "Reconstruction bias") %>%
    select(lon, lat, data_source, period, dcant_bias),
  dim_row = "data_source",
  dim_col = "period",
  var = "dcant_bias",
  breaks = c(-Inf, seq(-8, 8, 2), Inf),
  legend_title = legend_title,
  col = "divergent"
)

p_CI_absolute_mod / p_CI_bias + 
  plot_annotation(tag_levels = 'A')

Past versions of invetory_maps_absolute_mod_bias-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_column_inventory_synthetic_data.png",
       height = 7,
       width = 8,
       dpi = 600)

rm(p_CI_absolute_mod, p_CI_bias)

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source != "obs") %>%
  select(data_source, lon, lat, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = data_source,
              values_from = dcant) %>%
  mutate(dcant_bias = mod - mod_truth,
         data_source = "Reconstruction bias") %>% 
  mutate(area = earth_surf(lat,lon),
         dcant_bias_int = cut(abs(dcant_bias), c(-Inf,2,4,Inf))) %>% 
  group_by(dcant_bias_int) %>% 
  summarise(area = sum(area)) %>% 
  ungroup() %>% 
  mutate(area_fraction = 100 * area / sum(area))

# A tibble: 3 × 3
  dcant_bias_int    area area_fraction
  <fct>            <dbl>         <dbl>
1 (-Inf,2]       5.74e14        87.1  
2 (2,4]          8.23e13        12.5  
3 (4, Inf]       2.54e12         0.385

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source == "mod_truth") %>%
  mutate(area = earth_surf(lat,lon),
         dcant_int = cut(abs(dcant), c(-Inf,2,4,Inf))) %>% 
  group_by(dcant_int) %>% 
  summarise(area = sum(area)) %>% 
  ungroup() %>% 
  mutate(area_fraction = 100 * area / sum(area))

# A tibble: 3 × 3
  dcant_int    area area_fraction
  <fct>       <dbl>         <dbl>
1 (-Inf,2]  6.22e13          9.43
2 (2,4]     2.33e14         35.3 
3 (4, Inf]  3.64e14         55.3 

dcant_inv_all %>%
  filter(period %in% two_decades,
         data_source != "obs") %>%
  select(data_source, lon, lat, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = data_source,
              values_from = dcant) %>%
  mutate(dcant_bias = mod - mod_truth,
         data_source = "Reconstruction bias") %>%
  arrange(period) %>% 
  group_by(lon, lat) %>% 
  summarise(delta_dcant_bias = dcant_bias - lag(dcant_bias)) %>% 
  ungroup() %>% 
  drop_na() %>% 
  p_map_cant_inv(
    var = "delta_dcant_bias",
    title_text = NULL,
    subtitle_text = NULL,
    col = "bias"
  ) +
  theme(strip.text.x = element_blank(),
        strip.background.x = element_blank())

Past versions of inventory_maps_absolute_mod_bias_decade-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

legend_title <- expression(atop(Delta * Delta * C["ant"] ~ bias,
                                (mol ~ m ^ {
                                  -2
                                } ~ dec ^ {
                                  -1
                                })))

p_CI_bias_decade <-
  p_map_mdim_robinson(
    df = dcant_inv_all %>%
      filter(period %in% two_decades,
             data_source != "obs") %>%
      select(data_source, lon, lat, basin_AIP, period, dcant) %>%
      pivot_wider(names_from = data_source,
                  values_from = dcant) %>%
      mutate(dcant_bias = mod - mod_truth,
             data_source = "Reconstruction bias") %>%
      arrange(period) %>%
      group_by(lon, lat) %>%
      summarise(delta_dcant_bias = dcant_bias - lag(dcant_bias)) %>%
      ungroup() %>%
      drop_na(),
    var = "delta_dcant_bias",
    breaks = c(-Inf, seq(-8, 8, 2), Inf),
    legend_title = legend_title,
    col = "divergent"
  )



p_CI_bias_decade

Past versions of inventory_maps_absolute_mod_bias_decade-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_delta_dcant_column_inventory_synthetic_data_bias.png",
       height = 3.5,
       width = 5,
       dpi = 600)

rm(p_CI_bias_decade)

9.7 Density distributions

bind_rows(
  dcant_inv_all %>%
    filter(period %in% two_decades) %>%
    mutate(
      data_source = case_when(
        data_source == "mod" ~ "Reconstruction",
        data_source == "obs" ~ "Observations",
        data_source == "mod_truth" ~ "Model truth"
      )
    ) %>%
    select(data_source, dcant, period),
  dcant_inv_bias_all %>%
    filter(period %in% two_decades) %>%
    mutate(data_source = "Reconstruction bias") %>%
    select(data_source, dcant = dcant_pos_bias, period)
) %>%
  ggplot(aes(dcant, col = data_source)) +
  scale_color_brewer(palette = "Dark2") +
  scale_fill_brewer(palette = "Dark2") +
  geom_vline(xintercept = 0) +
  geom_density(alpha = 0.3) +
  coord_cartesian(xlim = c(-2, 18)) +
  facet_grid(period ~ .) +
  labs(x = dcant_CI_label) +
  theme(legend.title = element_blank())

Past versions of dcant_column_inventory_density_distribution-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
5ca8778	jens-daniel-mueller	2022-09-07

# ggsave(path = here::here("output/publication"),
#        filename = "FigS_dcant_column_inventory_density_distribution.png",
#        height = 4,
#        width = 7)

bind_rows(
  dcant_inv_all %>%
    filter(period %in% two_decades) %>%
    select(data_source, lon, lat, period, dcant) %>%
    pivot_wider(names_from = period,
                values_from = dcant) %>%
    mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
           period = "(2004 - 2014) - (1994 - 2004)")  %>%
    mutate(
      data_source = case_when(
        data_source == "mod" ~ "Reconstruction",
        data_source == "obs" ~ "Observations",
        data_source == "mod_truth" ~ "Model truth"
      )
    ) %>%
    select(data_source, dcant_offset, period),
  dcant_inv_bias_all %>%
    filter(period %in% two_decades) %>%
    select(lon, lat, period, dcant_bias) %>%
    pivot_wider(names_from = period,
                values_from = dcant_bias) %>%
    mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
           period = "(2004 - 2014) - (1994 - 2004)")  %>%
    mutate(data_source = "Reconstruction bias") %>%
    select(data_source, dcant_offset, period)
) %>%
  ggplot(aes(dcant_offset, col = data_source)) +
  scale_color_brewer(palette = "Dark2") +
  scale_fill_brewer(palette = "Dark2") +
  geom_vline(xintercept = 0) +
  geom_density(alpha = 0.3) +
  coord_cartesian(xlim = c(-10, 10)) +
  labs(x = delta_dcant_CI_label) +
  theme(legend.title = element_blank())

Past versions of delta_dcant_column_inventory_density_distribution-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
5ca8778	jens-daniel-mueller	2022-09-07

# ggsave(
#   path = here::here("output/publication"),
#   filename = "FigS_delta_dcant_column_inventory_density_distribution.png",
#   height = 3,
#   width = 7
# )

9.8 Beta

dcant_inv_all %>%
  filter(data_source %in% c("obs")) %>% 
  p_map_cant_inv(var = "beta",
                 breaks = c(-Inf, seq(0, 1, 0.1), Inf),
                 title_text = NULL,
                 legend_title = beta_CI_label) +
  facet_wrap(~ period, ncol = 2)

Past versions of beta_CI_maps-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
e924c80	jens-daniel-mueller	2022-11-17
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26
fea41c1	jens-daniel-mueller	2022-07-20
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
efa414b	jens-daniel-mueller	2022-07-16

p_map_mdim_robinson(
  df = dcant_inv_all %>%
    filter(data_source %in% c("obs")) %>%
    select(lon, lat, period, beta),
  legend_title = beta_CI_label,
  var = "beta",
  breaks = c(-Inf, seq(0, 1, 0.1), Inf),
  dim_row = "period"
)

Past versions of beta_CI_maps-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_beta_column_inventory.png",
       height = 10,
       width = 8,
       dpi = 600)

9.9 Zonal means

dcant_inv_all <- m_grid_horizontal_coarse(dcant_inv_all)

dcant_inv_all_lat_mean <- dcant_inv_all %>% 
  group_by(basin_AIP, period, lat_grid, data_source) %>% 
  summarise(beta = mean(beta, na.rm = TRUE),
            dcant = mean(dcant, na.rm = TRUE)) %>% 
  ungroup()

dcant_inv_all_ensemble <- m_grid_horizontal_coarse(dcant_inv_all_ensemble)

dcant_inv_all_ensemble_lat_mean <- dcant_inv_all_ensemble %>% 
  group_by(basin_AIP, period, lat_grid, Version_ID, data_source) %>% 
  summarise(beta_mean = mean(beta, na.rm = TRUE),
            dcant_mean = mean(dcant, na.rm = TRUE)) %>% 
  ungroup()

dcant_inv_all_ensemble_lat_mean_average <- dcant_inv_all_ensemble_lat_mean %>% 
  group_by(basin_AIP, period, lat_grid, data_source) %>% 
  summarise(beta_sd = sd(beta_mean, na.rm = TRUE),
            beta_mean = mean(beta_mean, na.rm = TRUE),
            dcant_sd = sd(dcant_mean, na.rm = TRUE),
            dcant_mean = mean(dcant_mean, na.rm = TRUE)) %>% 
  ungroup()

dcant_inv_all_lat_mean <- full_join(dcant_inv_all_lat_mean,
                                    dcant_inv_all_ensemble_lat_mean_average)

9.9.1 dcant

dcant_inv_all_lat_mean %>%
  filter(period != "1800 - 1994") %>% 
  ggplot(aes(lat_grid, dcant, fill = period)) +
  geom_hline(yintercept = c(0,10)) +
  geom_vline(xintercept = c(-10,10)) +
  geom_vline(xintercept = c(-40,-30,30,40)) +
  geom_ribbon(aes(ymin = dcant - dcant_sd,
                  ymax = dcant + dcant_sd),
              alpha = 0.3) +
  geom_path(aes(col = period)) +
  geom_point(shape = 21) +
  scale_color_brewer(palette = "Dark2", name = "Mean ± SD") +
  scale_fill_brewer(palette = "Dark2", name = "Mean ± SD") +
  coord_flip() +
  facet_grid(data_source ~ basin_AIP) +
  labs(x = "Latitude (°N)",
       y = dcant_CI_label) +
  scale_x_continuous(breaks = seq(-75,75, 15))

Past versions of zonal_mean_storage_dcant-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
5b77c4f	jens-daniel-mueller	2022-08-10
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
ea46812	jens-daniel-mueller	2022-07-13

dcant_inv_all_lat_mean %>%
  filter(period != "1800 - 1994") %>% 
  mutate(lat_grid = cut(lat_grid, c(-40,-30,-10,10,30,40))) %>% 
  group_by(lat_grid, data_source) %>% 
  summarise(dcant_mean = mean(dcant_mean, na_rm = TRUE)) %>% 
  ungroup()

# A tibble: 18 × 3
   lat_grid  data_source dcant_mean
   <fct>     <chr>            <dbl>
 1 (-40,-30] mod              11.7 
 2 (-40,-30] mod_truth        10.9 
 3 (-40,-30] obs              13.9 
 4 (-30,-10] mod               8.65
 5 (-30,-10] mod_truth         7.73
 6 (-30,-10] obs              11.1 
 7 (-10,10]  mod               4.60
 8 (-10,10]  mod_truth         4.09
 9 (-10,10]  obs               6.87
10 (10,30]   mod               5.31
11 (10,30]   mod_truth         5.03
12 (10,30]   obs               6.78
13 (30,40]   mod               6.23
14 (30,40]   mod_truth         6.46
15 (30,40]   obs              10.6 
16 <NA>      mod               5.61
17 <NA>      mod_truth         5.59
18 <NA>      obs               7.08

dcant_inv_all_lat_mean %>%
  filter(period != "1800 - 1994") %>% 
  mutate(lat_grid = cut(lat_grid, c(-40,-30,-10,10,30,40))) %>% 
  group_by(lat_grid, basin_AIP, data_source) %>% 
  summarise(dcant_mean = mean(dcant_mean, na_rm = TRUE)) %>% 
  ungroup()

# A tibble: 51 × 4
   lat_grid  basin_AIP data_source dcant_mean
   <fct>     <chr>     <chr>            <dbl>
 1 (-40,-30] Atlantic  mod              11.3 
 2 (-40,-30] Atlantic  mod_truth        10.1 
 3 (-40,-30] Atlantic  obs              15.1 
 4 (-40,-30] Indian    mod              13.3 
 5 (-40,-30] Indian    mod_truth        12.6 
 6 (-40,-30] Indian    obs              13.8 
 7 (-40,-30] Pacific   mod              10.5 
 8 (-40,-30] Pacific   mod_truth         9.97
 9 (-40,-30] Pacific   obs              12.8 
10 (-30,-10] Atlantic  mod               7.55
# … with 41 more rows

dcant_inv_all_lat_mean %>%
  filter(period != "1800 - 1994") %>% 
  mutate(lat_grid = cut(lat_grid, c(-90,0,90))) %>% 
  group_by(lat_grid, basin_AIP, period, data_source) %>% 
  summarise(dcant_mean = mean(dcant_mean, na_rm = TRUE)) %>% 
  ungroup()

# A tibble: 54 × 5
   lat_grid basin_AIP period      data_source dcant_mean
   <fct>    <chr>     <chr>       <chr>            <dbl>
 1 (-90,0]  Atlantic  1994 - 2004 mod               4.19
 2 (-90,0]  Atlantic  1994 - 2004 mod_truth         4.06
 3 (-90,0]  Atlantic  1994 - 2004 obs               6.68
 4 (-90,0]  Atlantic  1994 - 2014 mod               9.50
 5 (-90,0]  Atlantic  1994 - 2014 mod_truth         8.72
 6 (-90,0]  Atlantic  1994 - 2014 obs              14.4 
 7 (-90,0]  Atlantic  2004 - 2014 mod               5.35
 8 (-90,0]  Atlantic  2004 - 2014 mod_truth         4.66
 9 (-90,0]  Atlantic  2004 - 2014 obs               7.77
10 (-90,0]  Indian    1994 - 2004 mod               5.77
# … with 44 more rows

dcant_inv_all_lat_mean %>%
  filter(period %in%two_decades,
         data_source == "obs") %>% 
  mutate(lat_grid = cut(lat_grid, c(-50,-20,20,50))) %>% 
  group_by(lat_grid, basin_AIP, period, data_source) %>% 
  summarise(dcant_mean = mean(dcant_mean, na_rm = TRUE)) %>% 
  ungroup() %>% 
  filter(data_source == "obs") %>% 
  arrange(basin_AIP)

# A tibble: 24 × 5
   lat_grid  basin_AIP period      data_source dcant_mean
   <fct>     <chr>     <chr>       <chr>            <dbl>
 1 (-50,-20] Atlantic  1994 - 2004 obs               8.55
 2 (-50,-20] Atlantic  2004 - 2014 obs              11.4 
 3 (-20,20]  Atlantic  1994 - 2004 obs               5.23
 4 (-20,20]  Atlantic  2004 - 2014 obs               8.50
 5 (20,50]   Atlantic  1994 - 2004 obs              12.0 
 6 (20,50]   Atlantic  2004 - 2014 obs               8.07
 7 <NA>      Atlantic  1994 - 2004 obs               7.39
 8 <NA>      Atlantic  2004 - 2014 obs               4.07
 9 (-50,-20] Indian    1994 - 2004 obs              10.8 
10 (-50,-20] Indian    2004 - 2014 obs               8.43
# … with 14 more rows

dcant_inv_all_ensemble_lat_mean %>% 
  ggplot(aes(lat_grid, dcant_mean, fill=period, group=Version_ID)) +
  geom_hline(yintercept = 0) +
  geom_path(aes(col=period), alpha = 0.3) +
  scale_color_brewer(palette = "Dark2") +
  scale_fill_brewer(palette = "Dark2") +
  coord_flip() +
  facet_grid(data_source ~ basin_AIP) +
  labs(x = "Latitude (°N)",
       y = dcant_CI_label)

Past versions of zonal_mean_storage_dcant-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
5b77c4f	jens-daniel-mueller	2022-08-10
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
0160c40	jens-daniel-mueller	2022-07-16
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
ea46812	jens-daniel-mueller	2022-07-13

9.9.2 beta

dcant_inv_all_lat_mean %>%
  ggplot(aes(lat_grid, beta, fill = period)) +
  geom_hline(yintercept = 0) +
  geom_ribbon(aes(ymin = beta - beta_sd,
                  ymax = beta + beta_sd),
              alpha = 0.3) +
  geom_path(aes(col = period)) +
  geom_point(shape = 21) +
  scale_color_brewer(palette = "Dark2", name = "Mean ± SD") +
  scale_fill_brewer(palette = "Dark2", name = "Mean ± SD") +
  coord_flip() +
  facet_grid(data_source ~ basin_AIP) +
  labs(x = "Latitude (°N)",
       y = beta_CI_label) +
  scale_x_continuous(breaks = seq(-75,75, 15))

Past versions of zonal_mean_storage_beta-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
5b77c4f	jens-daniel-mueller	2022-08-10
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
ea46812	jens-daniel-mueller	2022-07-13

dcant_inv_all_ensemble_lat_mean %>% 
  ggplot(aes(lat_grid, beta_mean, fill=period, group=Version_ID)) +
  geom_hline(yintercept = 0) +
  geom_path(aes(col=period), alpha = 0.3) +
  scale_color_brewer(palette = "Dark2") +
  scale_fill_brewer(palette = "Dark2") +
  coord_flip() +
  facet_grid(data_source ~ basin_AIP) +
  labs(x = "Latitude (°N)",
       y = beta_CI_label)

Past versions of zonal_mean_storage_beta-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
5b77c4f	jens-daniel-mueller	2022-08-10
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
0160c40	jens-daniel-mueller	2022-07-16
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
ea46812	jens-daniel-mueller	2022-07-13

9.10 Write files

dcant_inv_out <- dcant_inv_all_ensemble %>%
  filter(
    period %in% two_decades &
      data_source == "obs" &
      (MLR_basins == "3" |
         Version_ID_group == "Standard case")
  )

Version_IDs_out <- dcant_inv_out %>% 
  distinct(Version_ID, period, Version_ID_group, MLR_basins)

dcant_inv_out <- dcant_inv_out %>%
  select(lon,
         lat,
         period,
         MLR_basins,
         Version_ID_group,
         dcant)

# convert from tibble to stars object
dcant_inv_out_stars <- st_as_stars(dcant_inv_out,
                                   dims = c("lon", "lat", "period", "MLR_basins", "Version_ID_group"))

# assign coordinate system to horizontal (lon, lat) coordinates
# st_crs(dcant_inv_out_stars) <- "+proj=longlat"
st_crs(dcant_inv_out_stars) <- st_crs(4326)

# plot variable per depth level
ggplot() +
  geom_stars(data = dcant_inv_out_stars %>% 
               filter(period == "1994 - 2004")) +
  facet_grid(MLR_basins~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")

file_out <- paste0(path_out, "inv_dcant_emlr_cstar_mueller_94-04-14.nc")

# write stars object to .nc file
dcant_inv_out_stars %>%
  write_mdim(file_out)



# reopen .nc file with all available stars functions
dcant_inv_read_ncdf <-
  read_ncdf(file_out)
dcant_inv_read_mdim <-
  read_mdim(file_out)

# plot variable in object reopened

ggplot() +
  geom_stars(data = dcant_inv_read_ncdf %>% 
               filter(period == "1994 - 2004")) +
  facet_grid(MLR_basins~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")

ggplot() +
  geom_stars(data = dcant_inv_read_mdim %>% 
               filter(period == "1994 - 2004")) +
  facet_grid(MLR_basins~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")

ggplot() +
  geom_stars(data = dcant_inv_read_ncdf %>% 
               filter(MLR_basins == "3")) +
  facet_grid(period~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent") +
  coord_quickmap()


dcant_inv_read_tidync <-
  tidync::tidync(file_out)

dcant_inv_read_tidync %>%
  tidync::hyper_tibble() %>%
  filter(period == "1994 - 2004") %>%
  ggplot(aes(lon, lat, fill = dcant)) +
  geom_raster() +
  facet_grid(MLR_basins ~ Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")

# library(ncdf4)
# 
# nc <- nc_open(file_out)
# print(nc)
# 
# # get longitude and latitude
# lon <- ncvar_get(nc,"period")
# nlon <- dim(lon)
# 
# 
# ncatt_get(nc, "period")
# ncvar_get(nc)
# 
# nt <- dim(time)
# 
# nc_close(nc)

delta_dcant_inv_uncertainty_out <- delta_dcant_inv_uncertainty %>%
  filter(data_source == "obs") %>%
  mutate(period = "decadal difference") %>% 
  select(lon, lat, period, dcant = delta_dcant, uncertainty = RSS)

dcant_inv_uncertainty_out <- dcant_inv_uncertainty %>% 
  filter(data_source == "obs") %>% 
  select(lon, lat, period, dcant = std_case, uncertainty = sd)

dcant_inv_uncertainty_out <- bind_rows(dcant_inv_uncertainty_out,
                                       delta_dcant_inv_uncertainty_out)


# convert from tibble to stars object
dcant_inv_uncertainty_out_stars <- st_as_stars(
  dcant_inv_uncertainty_out,
  dims = c("lon", "lat", "period"))

# assign coordinate system to horizontal (lon, lat) coordinates
# st_crs(dcant_inv_out_stars) <- "+proj=longlat"
st_crs(dcant_inv_uncertainty_out_stars) <- st_crs(4326)

# plot variable per depth level
ggplot() +
  geom_stars(data = dcant_inv_uncertainty_out_stars) +
  facet_grid(period ~ .) +
  scale_fill_binned_divergingx(
    breaks = c(-Inf, seq(-16, 16, 4), Inf),
    palette = "RdBu",
    rev = TRUE,
    na.value = "transparent"
  )

ggplot() +
  geom_stars(data = dcant_inv_uncertainty_out_stars,
             aes(fill = abs(dcant) - uncertainty)) +
  facet_grid(period ~ .) +
  scale_fill_binned_divergingx(
    breaks = c(-Inf, 0 , Inf),
    palette = "RdBu",
    rev = TRUE,
    na.value = "transparent"
  )


file_out <- paste0(path_out,"inv_dcant_emlr_cstar_mueller_94-04-14_uncertainty.nc")

# write stars object to .nc file
dcant_inv_uncertainty_out_stars %>%
  write_mdim(file_out)

# reopen .nc file with all available stars functions
dcant_inv_uncertainty_read_ncdf <-
  read_ncdf(file_out)
dcant_inv_uncertainty_read_mdim <-
  read_mdim(file_out)

# plot variable in object reopened

ggplot() +
  geom_stars(data = dcant_inv_uncertainty_read_ncdf,
             aes(fill = abs(dcant) - uncertainty)) +
  facet_grid(period ~ .) +
  scale_fill_binned_divergingx(
    breaks = c(-Inf, 0 , Inf),
    palette = "RdBu",
    rev = TRUE,
    na.value = "transparent"
  )

ggplot() +
  geom_stars(data = dcant_inv_uncertainty_read_mdim,
             aes(fill = abs(dcant) - uncertainty)) +
  facet_grid(period ~ .) +
  scale_fill_binned_divergingx(
    breaks = c(-Inf, 0 , Inf),
    palette = "RdBu",
    rev = TRUE,
    na.value = "transparent"
  )

dcant_inv_uncertainty_read_tidync <-
  tidync::tidync(file_out)

dcant_inv_uncertainty_read_tidync %>%
  tidync::hyper_tibble() %>%
  ggplot(aes(lon, lat)) +
  geom_raster(aes(fill = abs(dcant) - uncertainty)) +
  facet_grid(period ~ .) +
  scale_fill_binned_divergingx(
    breaks = c(-Inf, 0 , Inf),
    palette = "RdBu",
    rev = TRUE,
    na.value = "transparent"
  )

# library(ncdf4)
# nc_open(file_out)
# nc_close(file_out)

10 Global section

10.1 Additive uncertainty

delta_dcant_global_section_all <-
  dcant_global_section_all %>%
  filter(data_source %in% c("obs", "mod"),
         period %in% two_decades) %>%
  select(data_source, dist, depth, period, dcant) %>%
  drop_na() %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
         period = "(2004 - 2014) - (1994 - 2004)")

dcant_global_section_all_uncertainty <-
  dcant_global_section_all_ensemble %>%
  filter(data_source %in% c("obs", "mod"),
         period %in% two_decades) %>%
  select(dist,
         depth,
         period,
         data_source,
         MLR_basins,
         Version_ID_group,
         dcant) %>%
  mutate(dist_grid = as.numeric(as.character(cut(
    dist, seq(0, 50, 1), seq(0.5, 49.5, 1)
  ))),
  # depth_grid = depth
  depth_grid = as.numeric(as.character(cut(
    depth,
    c(seq(0, 500, 100), seq(1000, 10000, 500)),
    c(seq(50, 450, 100), seq(750, 10000, 500)),
    right = FALSE
  )))
  ) %>%
  group_by(dist_grid,
           depth_grid,
           data_source,
           period,
           MLR_basins,
           Version_ID_group) %>%
  summarise(dcant = mean(dcant)) %>%
  ungroup() %>%
  rename(dist = dist_grid,
         depth = depth_grid)

dcant_global_section_all_uncertainty <-
  additive_uncertainty(dcant_global_section_all_uncertainty)


delta_dcant_global_section_all_uncertainty <- dcant_global_section_all_uncertainty[[3]]
dcant_global_section_all_uncertainty_contributions <- dcant_global_section_all_uncertainty[[2]]
dcant_global_section_all_uncertainty <- dcant_global_section_all_uncertainty[[1]]

dcant_global_section_all_uncertainty <-
  dcant_global_section_all_uncertainty %>%
  mutate(band = "dummy")


dcant_global_section_all_uncertainty  %>% 
  filter(data_source == "obs",
         period == "1994 - 2004") %>% 
  p_section_global(
    var = "std_case",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  )

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
6edad9b	jens-daniel-mueller	2022-11-11

dcant_global_section_all_uncertainty  %>% 
  filter(data_source == "obs",
         period == "1994 - 2004") %>% 
  p_section_global(
    var = "sd",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  )

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11

ggplot() +
  geom_point(
    data = dcant_global_section_all_uncertainty %>%
      filter(signif_single == 0),
    aes(dist, depth),
    shape = 4
  ) +
  facet_grid(data_source~period) +
  scale_y_reverse()

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11

ggplot() +
  geom_point(
    data = dcant_global_section_all_uncertainty %>%
      filter(signif_double == 0),
    aes(dist, depth),
    shape = 4
  ) +
  facet_grid(data_source~period) +
  scale_y_reverse()

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-4.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11

delta_dcant_global_section_all_uncertainty <-
  delta_dcant_global_section_all_uncertainty %>%
  mutate(band = "dummy")

delta_dcant_global_section_all_uncertainty  %>% 
  filter(data_source == "obs") %>% 
  p_section_global(
    var = "delta_dcant",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  )

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-5.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11

delta_dcant_global_section_all_uncertainty  %>% 
  filter(data_source == "obs") %>% 
  select(-delta_dcant) %>% 
  rename(delta_dcant = RSS) %>% 
  p_section_global(
    var = "delta_dcant",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  )

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-6.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11

ggplot() +
  geom_point(
    data = delta_dcant_global_section_all_uncertainty %>%
      filter(signif_single == 0),
    aes(dist, depth),
    shape = 4
  ) +
  scale_y_reverse() +
  facet_grid(data_source~.)

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-7.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25

ggplot() +
  geom_point(
    data = delta_dcant_global_section_all_uncertainty %>%
      filter(signif_double == 0),
    aes(dist, depth),
    shape = 4
  ) +
  scale_y_reverse() +
  facet_grid(data_source~.)

Past versions of global_section_absolute_delta_RSS_additive_uncertainty-8.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25

10.2 Observations

dcant_global_section_all_inv <- dcant_global_section_all_ensemble %>%
  select(data_source, period, depth, lat = dist, basin_AIP = band,
         dcant, dcant_pos, Version_ID_group, MLR_basins) %>% 
  mutate(lon = "dummy") %>% 
  group_by(data_source, period, Version_ID_group, MLR_basins) %>%
  nest() %>%
  mutate(inv = map(.x = data, ~m_dcant_inv(.x))) %>%
  select(-data) %>%
  unnest(inv) %>% 
  rename(dist = lat, band = basin_AIP) %>% 
  select(-lon)

# dcant_global_section_all_inv %>% 
#   filter(period %in% two_decades) %>% 
#   ggplot(aes(dist, dcant, col = as.factor(inv_depth))) +
#   geom_path(size = 2) +
#   scale_color_viridis_d(direction = -1) +
#   coord_cartesian(ylim = c(0,20)) +
#   facet_grid(data_source ~ period)

dcant_global_section_all_inv %>% 
  filter(period %in% two_decades,
         inv_depth == 3000) %>% 
  ggplot(aes(dist, dcant, col = MLR_basins, group = interaction(Version_ID_group, MLR_basins))) +
  geom_path() +
  scale_color_brewer(palette = "Dark2") +
  coord_cartesian(ylim = c(0,20)) +
  facet_grid(data_source ~ period)

dcant_global_section_all_surface <- dcant_global_section_all_ensemble %>%
  select(data_source, period, depth, dist, band, dcant, Version_ID_group, MLR_basins) %>% 
  group_by(data_source) %>% 
  mutate(surface_depth = min(depth, na.rm = TRUE)) %>% 
  ungroup() %>% 
  filter(depth == surface_depth) %>% 
  select(-surface_depth) %>% 
  rename(dcant_surface = dcant)

dcant_global_section_all_surface %>% 
  filter(period %in% two_decades) %>% 
  arrange(dist) %>% 
  ggplot(aes(dist, dcant_surface, col = Version_ID_group, group = interaction(Version_ID_group, MLR_basins))) +
  geom_path() +
  scale_color_brewer(palette = "Dark2") +
  # coord_cartesian(ylim = c(0,20)) +
  facet_grid(data_source ~ period)

dcant_global_section_all_penetration <- 
  full_join(dcant_global_section_all_surface,
            dcant_global_section_all_inv)

dcant_global_section_all_penetration <-
  dcant_global_section_all_penetration %>% 
  mutate(dcant_surface = dcant_surface * 1e3 * 1.025 / 1e6,
         depth =  dcant / dcant_surface)

dcant_global_section_all_penetration <-
  dcant_global_section_all_penetration %>%
  filter(period %in% two_decades,
         inv_depth == 3000) %>%
  arrange(dist)

dcant_global_section_all_penetration %>% 
  ggplot(aes(dist, depth, col = Version_ID_group, group = interaction(Version_ID_group, MLR_basins))) +
  geom_path() +
  scale_y_reverse(limits = c(3000,0)) +
  scale_color_brewer(palette = "Dark2") +
  # coord_cartesian(ylim = c(0,20)) +
  facet_grid(data_source ~ period)

dcant_global_section_all_penetration_uncertainty <- 
  additive_uncertainty(dcant_global_section_all_penetration %>% 
          select(-c(dcant, dcant_pos, dcant_surface, inv_depth)) %>% 
            group_by(data_source, period, dist, Version_ID_group, MLR_basins) %>% 
            summarise(dcant = mean(depth, na.rm = TRUE)) %>% 
            ungroup())

delta_dcant_global_section_all_penetration_uncertainty <- dcant_global_section_all_penetration_uncertainty[[3]]
dcant_global_section_all_penetration_uncertainty <- dcant_global_section_all_penetration_uncertainty[[1]]


dcant_global_section_all_penetration_uncertainty %>%
  ggplot(aes(dist,
             std_case,
             ymin = std_case - sd,
             ymax = std_case + sd,
             fill = period)) +
  geom_ribbon(alpha = 0.3) +
  geom_path(aes(col = period)) +
  scale_y_reverse() +
  scale_color_brewer(palette = "Dark2") +
  scale_fill_brewer(palette = "Dark2") +
  coord_cartesian(ylim = c(3000,0)) +
  facet_grid(data_source ~ .)

delta_dcant_global_section_all_penetration_uncertainty %>%
  ggplot(aes(dist,
             delta_dcant,
             ymin = delta_dcant - RSS,
             ymax = delta_dcant + RSS)) +
  geom_hline(yintercept = 0) +
  geom_ribbon(alpha = 0.3) +
  geom_path() +
  scale_y_reverse() +
  scale_color_brewer(palette = "Dark2") +
  scale_fill_brewer(palette = "Dark2") +
  coord_cartesian(ylim = c(1000,-1000)) +
  facet_grid(data_source ~ period)

p_dcant_global_section_dec1 <- dcant_global_section_all %>%
  filter(data_source == "obs",
         period %in% "1994 - 2004") %>%
  p_section_global(
    var = "dcant",
    title_text = "A   1994 – 2004",
    df_uncertainty = dcant_global_section_all_uncertainty %>%
      filter(period %in% "1994 - 2004",
             data_source == "obs"),
    # df_penetration = dcant_global_section_all_penetration %>%
    #   filter(data_source == "obs",
    #          period %in% "1994 - 2004"),
    col_uncertainty = "grey",
    subtitle_text = NULL,
    plot_slabs = "n",
    contour_level = 5
  ) &
  theme(axis.title.x = element_blank(),
        axis.text.x.bottom = element_blank())

p_dcant_global_section_dec1

Past versions of global_sections_obs-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

p_dcant_global_section_dec2 <-
  dcant_global_section_all %>%
  filter(data_source == "obs",
         period %in% "2004 - 2014") %>%
  p_section_global(
    var = "dcant",
    title_text = "B   2004 – 2014",
    subtitle_text = NULL,
    plot_slabs = "n",
    contour_level = 5,
    df_uncertainty = dcant_global_section_all_uncertainty %>%
      filter(period %in% "2004 - 2014",
             data_source == "obs"),
    col_uncertainty = "grey",
    # df_penetration = dcant_global_section_all_penetration %>%
    #   filter(data_source == "obs",
    #          period %in% "2004 - 2014")
  )&
  theme(legend.position = "none")

p_dcant_global_section_offset <-
  dcant_global_section_all %>%
  filter(data_source == "obs",
         period %in% two_decades) %>%
  arrange(depth, dist) %>%
  select(depth, dist, dcant, period, gamma_mean) %>%
  drop_na() %>%
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(delta_dcant = `2004 - 2014` - `1994 - 2004`) %>%
  p_section_global(
    var = "delta_dcant",
    df_uncertainty = delta_dcant_global_section_all_uncertainty %>%
      filter(data_source == "obs"),
    plot_slabs = "y",
    col = "div",
    title_text = "C   Decadal difference",
    subtitle_text = NULL
  )

wrap_plots(
  p_dcant_global_section_dec1,
  p_dcant_global_section_dec2 &
    theme(axis.title.x = element_blank(),
          axis.text.x = element_blank()
          ),
  p_dcant_global_section_offset &
    theme(
      axis.text.x.top = element_blank()
      ),
  ncol = 1
)

Past versions of global_sections_obs-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05

ggsave(path = here::here("output/publication"),
       filename = "Fig_dcant_global_section.png",
       height = 10,
       width = 8,
       dpi = 600)

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_global_section.png",
       height = 9,
       width = 10,
       dpi = 600)

rm(p_dcant_global_section_dec1, p_dcant_global_section_dec2,
   p_dcant_global_section_offset)

10.3 Synthetic data

p_dcant_global_section_dec1 <-
  dcant_global_section_all %>%
  filter(data_source == "mod",
         period %in% "1994 - 2004") %>%
  p_section_global(var = "dcant",
                   title_text = "1994 - 2004",
                   df_uncertainty = dcant_global_section_all_uncertainty %>% 
                     filter(period %in% "1994 - 2004",
                            data_source == "mod"),
                   col_uncertainty = "grey",
                   subtitle_text = NULL,
                   plot_slabs = "n",
                   contour_level = 5) +
  theme(
    axis.title.x = element_blank(),
    axis.text.x.bottom = element_blank()
  )


p_dcant_global_section_dec2 <-
  dcant_global_section_all %>%
  filter(data_source == "mod",
         period %in% "2004 - 2014") %>%
  p_section_global(
    var = "dcant",
    title_text = NULL,
    subtitle_text = NULL,
    plot_slabs = "n",
    contour_level = 5,
    df_uncertainty = dcant_global_section_all_uncertainty %>%
      filter(period %in% "2004 - 2014",
                            data_source == "mod"),
    col_uncertainty = "grey"
  ) &
  theme(
    legend.position = "none"
  )

p_dcant_global_section_offset <-
  dcant_global_section_all %>%
  filter(data_source == "mod",
         period %in% two_decades) %>%
  arrange(depth, dist) %>%
  select(depth, dist, dcant, period, gamma_mean) %>%
  drop_na() %>%
  group_by(depth, dist) %>% 
  mutate(gamma_mean = mean(gamma_mean)) %>% 
  ungroup() %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(delta_dcant = `2004 - 2014` - `1994 - 2004`) %>%
  p_section_global(
    var = "delta_dcant",
    df_uncertainty = delta_dcant_global_section_all_uncertainty %>%
      filter(data_source == "mod"),
    col = "div",
    plot_slabs = "y",
    title_text = NULL,
    subtitle_text = NULL
  )



wrap_plots(
  p_dcant_global_section_dec1,
  p_dcant_global_section_dec2 &
    theme(axis.title.x = element_blank(),
          axis.text.x = element_blank()),
  p_dcant_global_section_offset &
    theme(axis.title.x.top = element_blank(),
          axis.text.x.top = element_blank()),
  ncol = 1
)

Past versions of global_sections_mod-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_global_section_synthetic_data.png",
       height = 10,
       width = 8,
       dpi = 600)

rm(p_dcant_global_section_dec1, p_dcant_global_section_dec2,
   p_dcant_global_section_offset)

11 Zonal sections

11.1 Additive uncertainty

delta_dcant_zonal_all <-
  dcant_zonal_all %>%
  filter(data_source %in% c("obs", "mod"),
         period %in% two_decades) %>%
  select(data_source, lat, depth, period, basin_AIP, dcant) %>%
  drop_na() %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(dcant_offset = `2004 - 2014` - `1994 - 2004`,
         period = "(2004 - 2014) - (1994 - 2004)")

dcant_zonal_all_uncertainty <-
  dcant_zonal_all_ensemble %>%
  filter(data_source %in% c("obs", "mod"),
         period %in% two_decades) %>%
  select(lat,
         depth,
         period,
         basin_AIP,
         data_source,
         MLR_basins,
         Version_ID_group,
         dcant) %>%
  mutate(lat_grid = as.numeric(as.character(cut(
    lat, seq(-90, 90, 10), seq(-85, 85, 10)
  ))),
  # depth_grid = depth
  depth_grid = as.numeric(as.character(cut(
    depth,
    c(seq(0, 500, 100), seq(1000, 10000, 500)),
    c(seq(50, 450, 100), seq(750, 10000, 500)),
    right = FALSE
  )))
  ) %>%
  group_by(lat_grid,
           depth_grid,
           data_source,
           period,
           basin_AIP,
           MLR_basins,
           Version_ID_group) %>%
  summarise(dcant = mean(dcant)) %>%
  ungroup() %>%
  rename(lat = lat_grid,
         depth = depth_grid)

dcant_zonal_all_uncertainty <-
  additive_uncertainty(dcant_zonal_all_uncertainty)


delta_dcant_zonal_all_uncertainty <- dcant_zonal_all_uncertainty[[3]]
dcant_zonal_all_uncertainty <- dcant_zonal_all_uncertainty[[1]]

dcant_zonal_all_uncertainty  %>% 
  filter(data_source == "obs") %>% 
  p_section_zonal_continous_depth(
    var = "std_case",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  ) + 
  facet_grid(period~basin_AIP)

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

dcant_zonal_all_uncertainty  %>%
  filter(data_source == "obs") %>%
  p_section_zonal_continous_depth(
    var = "sd",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  ) +
  facet_grid(period ~ basin_AIP)

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

ggplot() +
  geom_point(
    data = dcant_zonal_all_uncertainty %>%
      filter(signif_single == 0,
             data_source == "obs"),
    aes(lat, depth),
    shape = 4
  ) +
  facet_grid(period ~ basin_AIP) +
  scale_y_reverse()

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

ggplot() +
  geom_point(
    data = dcant_zonal_all_uncertainty %>%
      filter(signif_double == 0,
             data_source == "obs"),
    aes(lat, depth),
    shape = 4
  ) +
  facet_grid(period ~ basin_AIP) +
  scale_y_reverse()

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-4.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

delta_dcant_zonal_all_uncertainty  %>%
  filter(data_source == "obs") %>%
  p_section_zonal_continous_depth(
    var = "delta_dcant",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  ) +
  facet_grid(. ~ basin_AIP)

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-5.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

delta_dcant_zonal_all_uncertainty  %>% 
  filter(data_source == "obs") %>% 
  select(-delta_dcant) %>% 
  rename(delta_dcant = RSS) %>% 
  p_section_zonal_continous_depth(
    var = "delta_dcant",
    plot_slabs = "n",
    title_text = NULL,
    subtitle_text = NULL,
    col = "div"
  ) +
  facet_grid(. ~ basin_AIP)

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-6.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

ggplot() +
  geom_point(
    data = delta_dcant_zonal_all_uncertainty %>%
      filter(signif_single == 0),
    aes(lat, depth),
    shape = 4
  ) +
  scale_y_reverse() +
  facet_grid(data_source~basin_AIP)

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-7.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

ggplot() +
  geom_point(
    data = delta_dcant_zonal_all_uncertainty %>%
      filter(signif_double == 0),
    aes(lat, depth),
    shape = 4
  ) +
  scale_y_reverse() +
  facet_grid(data_source~basin_AIP)

Past versions of zonal_section_absolute_delta_RSS_additive_uncertainty-8.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28

11.2 dcant - absolut

p_dcant_zonal_absolute <-
  dcant_zonal_all %>%
  filter(
    data_source == "obs",
    period %in% two_decades,
    depth <= params_global$inventory_depth_standard
  ) %>%
  p_section_zonal_continous_depth(var = "dcant",
                                  plot_slabs = "n",
                                  title_text = NULL) +
  geom_contour(aes(lat, depth, z = dcant),
               breaks = 5,
               col = "white") +
  # geom_point(
  #   data = dcant_zonal_all_uncertainty %>% filter(signif_single == 0),
  #   aes(lat, depth),
  #   shape = 20,
  #   size = 1,
  #   col = "grey"
  # ) +
  # geom_point(
  #   data = dcant_zonal_all_uncertainty %>% filter(signif_double == 0),
  #   aes(lat, depth),
  #   shape = 20,
  #   size = 0.1,
  #   col = "grey"
  # )+
  # geom_text_contour(
  #   aes(lat, depth, z = dcant),
  #   breaks = 5,
  #   stroke = 0.2,
  #   stroke.colour = "white",
  #   rotate = FALSE,
  #   label.placer = label_placer_n(n = 1)
  # ) +
  facet_grid(period ~ basin_AIP) +
  theme(
    axis.text.x = element_blank(),
    axis.title.x = element_blank(),
    legend.position = "top"
  ) +
  guides(fill = guide_colorsteps(
    barheight = unit(0.5, "cm"),
    barwidth = unit(7, "cm"),
    show.limits = FALSE
  ))

p_dcant_zonal_absolute

Past versions of zonal_sections-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26
e2adac6	jens-daniel-mueller	2022-07-22
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
ea46812	jens-daniel-mueller	2022-07-13
16dc3af	jens-daniel-mueller	2022-06-27
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
09b0780	jens-daniel-mueller	2022-05-24
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
094bfa0	jens-daniel-mueller	2022-02-18
d2191ad	jens-daniel-mueller	2022-02-04
5f2aed0	jens-daniel-mueller	2022-01-27
9753eb8	jens-daniel-mueller	2022-01-26
b1d7720	jens-daniel-mueller	2022-01-21
e572075	jens-daniel-mueller	2022-01-21

11.3 dcant - absolute delta

dcant_zonal_all %>%
  filter(data_source %in% c("mod", "obs"),
         period != "1994 - 2014") %>%
  select(data_source, lat, depth, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(delta_dcant = `2004 - 2014` - `1994 - 2004`) %>%
  group_by(data_source) %>%
  group_split() %>%
  # head(1) %>%
  map(
    ~ p_section_zonal_continous_depth(
      df = .x,
      var = "delta_dcant",
      plot_slabs = "n",
      subtitle_text = unique(.x$data_source),
      col = "bias"
    ) +
      facet_grid(basin_AIP ~ .)
  )

[[1]]

Past versions of inventory_maps_absolute_delta_zonal-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05

[[2]]

Past versions of inventory_maps_absolute_delta_zonal-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05

p_dcant_zonal_delta <-
  dcant_zonal_all %>%
  filter(data_source %in% c("obs"),
         period != "1994 - 2014") %>%
  select(data_source, lat, depth, basin_AIP, period, gamma_mean, dcant) %>%
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(delta_dcant = `2004 - 2014` - `1994 - 2004`) %>%
  p_section_zonal_continous_depth(
    var = "delta_dcant",
    plot_slabs = "y",
    drop_slabs = 1,
    subtitle_text = NULL,
    title_text = NULL,
    col = "bias"
  ) +
  # geom_point(
  #   data = delta_dcant_zonal_all_uncertainty %>% filter(signif_single == 0),
  #   aes(lat, depth),
  #   shape = 20,
  #   size = 1,
  #   col = "black"
  # ) +
  # geom_point(
  #   data = delta_dcant_zonal_all_uncertainty %>% filter(signif_double == 0),
  #   aes(lat, depth),
  #   shape = 20,
  #   size = 0.1,
  #   col = "black"
  # ) +
  facet_grid("Dec. diff." ~ basin_AIP) +
  theme(
    strip.text.x = element_blank(),
    strip.background.x = element_blank(),
    legend.position = "bottom"
  ) +
  guides(fill = guide_colorsteps(
    barheight = unit(0.5, "cm"),
    barwidth = unit(10, "cm"),
    show.limits = FALSE
  ))

p_dcant_zonal_absolute / p_dcant_zonal_delta + 
  plot_layout(heights = c(2,1)) +
  plot_annotation(tag_levels = 'A')

Past versions of inventory_maps_absolute_delta_zonal-3.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_zonal_mean_section.png",
       height = 6,
       width = 8,
       dpi = 600)

rm(p_dcant_zonal_absolute, p_dcant_zonal_delta)

11.4 Bias

p_dcant_zonal_absolute_mod <-
  dcant_zonal_all %>%
  filter(
    data_source != "obs",
    period %in% two_decades,
    depth <= params_global$inventory_depth_standard
  ) %>%
    mutate(data_source = case_when(
    data_source == "mod" ~ "Reconstruction",
    data_source == "mod_truth" ~ "Model truth"
  )) %>%  
  p_section_zonal_continous_depth(var = "dcant",
                                  plot_slabs = "n",
                                  title_text = NULL) +
  geom_contour(aes(lat, depth, z = dcant),
               breaks = 5,
               col = "white") +
  facet_nested(data_source+period ~ basin_AIP) +
  theme(
    axis.text.x = element_blank(),
    axis.title.x = element_blank(),
    legend.position = "top"
  ) +
  guides(fill = guide_colorsteps(
    barheight = unit(0.5, "cm"),
    barwidth = unit(7, "cm"),
    show.limits = FALSE
  ))

p_dcant_zonal_absolute_mod

Past versions of zonal_sections_mod-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
5b77c4f	jens-daniel-mueller	2022-08-10

p_dcant_zonal_bias <-
  dcant_zonal_all %>%
  filter(data_source != "obs",
         period %in% two_decades) %>%
  select(data_source, lat, depth, basin_AIP, period, dcant) %>%
  pivot_wider(names_from = data_source,
              values_from = dcant) %>%
  mutate(dcant_bias = mod - mod_truth) %>%
  p_section_zonal_continous_depth(
    var = "dcant_bias",
    plot_slabs = "n",
    drop_slabs = 1,
    subtitle_text = NULL,
    title_text = NULL,
    col = "bias"
  ) +
  facet_grid(period ~ basin_AIP) +
  theme(
    strip.text.x = element_blank(),
    strip.background.x = element_blank(),
    legend.position = "bottom"
  ) +
  guides(fill = guide_colorsteps(
    barheight = unit(0.5, "cm"),
    barwidth = unit(10, "cm"),
    show.limits = FALSE
  ))

p_dcant_zonal_absolute_mod / p_dcant_zonal_bias + 
  plot_layout(heights = c(2,1)) +
  plot_annotation(tag_levels = 'A')

Past versions of inventory_maps_absolute_bias-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
5b77c4f	jens-daniel-mueller	2022-08-10

ggsave(
  path = here::here("output/publication"),
  filename = "FigS_dcant_zonal_mean_section_synthetic_data.png",
  height = 9,
  width = 8,
  dpi = 600
)

rm(p_dcant_zonal_absolute_mod, 
   p_dcant_zonal_bias)

12 Penetration depth spatial

dcant_penetration_depth_all_lat_mean_all_ensemble %>%
  filter(dcant_lim == 5,
         period %in% two_decades,
         data_source != "obs") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Reconstruction\nensemble",
      data_source == "mod_truth" ~ "Model truth"
    ),
    data_source = fct_reorder(data_source, desc(data_source))
  ) %>%
  ggplot(aes(
    lat,
    depth_mean,
    group = interaction(MLR_basins, Version_ID_group, data_source)
  )) +
  geom_path(aes(col = data_source)) +
  geom_path(
    data = dcant_penetration_depth_all_lat_mean_all %>%
      filter(dcant_lim == 5,
             period %in% two_decades,
             data_source == "mod") %>%
      mutate(
        data_source = case_when(data_source == "mod" ~ "Standard case"),
        data_source = fct_reorder(data_source, desc(data_source))
      ),
    aes(lat,
        depth_mean,
        col = data_source)
  ) +
  scale_colour_manual(values = c("red", "grey", "black")) +
  labs(y = "Depth (m)",
       x = "Latitude (°N)") +
  theme(legend.title = element_blank()) +
  scale_y_reverse() +
  facet_grid(basin_AIP ~ period, scales = "free_x") +
  coord_cartesian(ylim = c(1500,0))

Past versions of penetration_depth-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
8105380	jens-daniel-mueller	2022-10-07

# ggsave(
#   path = here::here("output/publication"),
#   filename = "FigS_dcant_penetration_depth_synthetic_data.png",
#   height = 9,
#   width = 8
# )


dcant_penetration_depth_all_lat_mean_all_ensemble %>%
  filter(dcant_lim == 5,
         period %in% two_decades,
         data_source == "mod") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Reconstruction\nensemble",
      data_source == "mod_truth" ~ "Model truth"
    ),
    data_source = fct_reorder(data_source, desc(data_source))
  ) %>%
  ggplot(aes(
    lat,
    depth_mean,
    group = interaction(MLR_basins, Version_ID_group, data_source)
  )) +
  geom_path(aes(col = MLR_basins)) +
  geom_path(
    data = dcant_penetration_depth_all_lat_mean_all_ensemble %>%
      filter(
        dcant_lim == 5,
        period %in% two_decades,
        data_source == "mod_truth"
      ),
    aes(
      lat,
      depth_mean,
      group = interaction(MLR_basins, Version_ID_group, data_source)
    ),
    col = "black"
  ) +
  scale_color_vibrant() +
  labs(y = "Depth (m)",
       x = "Latitude (°N)") +
  theme(legend.title = element_blank()) +
  scale_y_reverse() +
  facet_grid(basin_AIP ~ period, scales = "free_x") +
  coord_cartesian(ylim = c(1500, 0))

Past versions of penetration_depth-2.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10
8105380	jens-daniel-mueller	2022-10-07

# ggsave(
#   path = here::here("output/publication"),
#   filename = "FigS_dcant_penetration_depth_synthetic_data_MLR_basins.png",
#   height = 9,
#   width = 8
# )


dcant_penetration_depth_all_lat_mean_all_ensemble %>%
  filter(dcant_lim == 5,
         period %in% two_decades,
         data_source != "mod_truth") %>%
  mutate(
    data_source = case_when(
      data_source == "mod" ~ "Synthetic data",
      data_source == "obs" ~ "Observations"
    ),
    data_source = fct_reorder(data_source, desc(data_source))
  ) %>%
  ggplot(aes(
    lat,
    depth_mean,
    group = interaction(MLR_basins, Version_ID_group, data_source)
  )) +
  geom_path(aes(col = data_source)) +
  scale_colour_highcontrast() +
  labs(y = "Depth (m)",
       x = "Latitude (°N)") +
  theme(legend.title = element_blank()) +
  scale_y_reverse() +
  facet_grid(basin_AIP ~ period, scales = "free_x") +
  coord_cartesian(ylim = c(1500,0))

Past versions of penetration_depth-3.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10

# ggsave(
#   path = here::here("output/publication"),
#   filename = "FigS_dcant_penetration_depth_synthetic_data_vs_observations.png",
#   height = 9,
#   width = 8
# )

dcant_penetration_depth_all_lat_mean_all_ensemble %>%
  filter(dcant_lim == 5,
         period %in% two_decades,
         data_source == "obs") %>%
  mutate(
    data_source = case_when(
      data_source == "obs" ~ "Observations"
    ),
    data_source = fct_reorder(data_source, desc(data_source))
  ) %>%
  ggplot(aes(
    lat,
    depth_mean,
    group = interaction(MLR_basins, Version_ID_group, data_source)
  )) +
  geom_path(aes(col = MLR_basins)) +
  scale_color_vibrant() +
  labs(y = "Depth (m)",
       x = "Latitude (°N)") +
  theme(legend.title = element_blank()) +
  scale_y_reverse() +
  facet_grid(basin_AIP ~ period, scales = "free_x") +
  coord_cartesian(ylim = c(1500,0))

Past versions of penetration_depth-4.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
46c163a	jens-daniel-mueller	2022-10-10

13 Concentration profiles

13.1 dCant, absolute

13.1.1 Standard case

dcant_profile_basin_MLR_all %>%
  arrange(depth) %>%
  filter(period != "1994 - 2014",
         depth <= params_global$inventory_depth_standard) %>%
  group_split(data_source) %>%
  # head(1) %>%
  map(
    ~ ggplot(data = .x,
             aes(dcant,
                 depth)) +
      geom_hline(yintercept = params_global$inventory_depth_standard) +
      geom_vline(xintercept = 0) +
      geom_ribbon(
        aes(
          xmin = dcant - dcant_sd,
          xmax = dcant + dcant_sd,
          fill = period
        ),
        alpha = 0.3
      ) +
      geom_path(aes(col = period)) +
      scale_y_continuous(trans = trans_reverser("sqrt"),
                         breaks = c(0, 100, 500, seq(1500, 5000, 1000))) +
      coord_cartesian(expand = 0) +
      scale_color_brewer(
        palette = "Set1",
        name = "mean \u00B1 regional SD",
        direction = -1
      ) +
      scale_fill_brewer(
        palette = "Set1",
        name = "mean \u00B1 regional SD",
        direction = -1
      ) +
      labs(
        title = paste("data_source", unique(.x$data_source)),
        y = "Depth (m)",
        x = dcant_umol_label
      ) +
      facet_wrap( ~ basin, ncol = 3, dir = "v") +
      theme(legend.position = c(0.8, 0.2))
  )

[[1]]

Past versions of concentration_profiles_MLR_basins-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
d2191ad	jens-daniel-mueller	2022-02-04
4673df5	jens-daniel-mueller	2022-02-02
60727e6	jens-daniel-mueller	2022-02-02
913e42f	jens-daniel-mueller	2022-02-01
189de95	jens-daniel-mueller	2022-02-01
b62308d	jens-daniel-mueller	2022-01-31
5f2aed0	jens-daniel-mueller	2022-01-27
c6fe495	jens-daniel-mueller	2022-01-26
9753eb8	jens-daniel-mueller	2022-01-26
b1d7720	jens-daniel-mueller	2022-01-21

[[2]]

Past versions of concentration_profiles_MLR_basins-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
318fefe	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
748aa43	jens-daniel-mueller	2022-06-27
16dc3af	jens-daniel-mueller	2022-06-27
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
d2191ad	jens-daniel-mueller	2022-02-04
4673df5	jens-daniel-mueller	2022-02-02
60727e6	jens-daniel-mueller	2022-02-02
913e42f	jens-daniel-mueller	2022-02-01
189de95	jens-daniel-mueller	2022-02-01
b62308d	jens-daniel-mueller	2022-01-31
5f2aed0	jens-daniel-mueller	2022-01-27
c6fe495	jens-daniel-mueller	2022-01-26
9753eb8	jens-daniel-mueller	2022-01-26
b1d7720	jens-daniel-mueller	2022-01-21

[[3]]

Past versions of concentration_profiles_MLR_basins-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
318fefe	jens-daniel-mueller	2022-08-11

13.1.2 Ensemble members

dcant_profile_basin_MLR_all_ensemble %>%
  arrange(depth) %>%
  filter(
    period != "1994 - 2014",
    depth <= params_global$inventory_depth_standard,
    data_source == "obs"
  ) %>%
  group_split(data_source) %>%
  map(
    ~ ggplot() +
      geom_hline(yintercept = params_global$inventory_depth_standard) +
      geom_vline(xintercept = 0) +
      geom_path(
        data = .x,
        aes(
          dcant,
          depth,
          col = period,
          group = Version_ID,
          linetype = "Ensemble member",
          size = "Ensemble member",
          alpha = "Ensemble member"
        )
      ) +
      geom_path(
        data = .x %>% filter(Version_ID %in% Version_IDs),
        aes(
          dcant,
          depth,
          col = period,
          group = Version_ID,
          linetype = "Standard case",
          alpha = "Standard case",
          size = "Standard case"
        )
      ) +
      geom_path(
        data = .x %>% filter(Version_ID %in% Version_IDs),
        aes(
          dcant,
          depth,
          group = Version_ID
        )
      ) +
      scale_size_manual(values = c(0.7, 1.5), name=" ") +
      scale_linetype_manual(values = c(1, 1), name=" ") +
      scale_alpha_manual(values = c(0.4, 1), name=" ") +
      scale_y_continuous(trans = trans_reverser("sqrt"),
                         breaks = c(0, 100, 500, seq(1500, 5000, 1000))) +
      coord_cartesian(expand = 0) +
      scale_color_brewer(
        palette = "Set1",
        direction = -1
      ) +
      scale_fill_brewer(
        palette = "Set1",
        direction = -1
      ) +
      labs(
        # title = paste("data_source", unique(.x$data_source)),
        y = "Depth (m)",
        x = dcant_umol_label
      ) +
      facet_wrap( ~ basin, ncol = 3, dir = "v") +
      theme(legend.position = "top")
  )

[[1]]

Past versions of concentration_profiles_MLR_basins_ensemble_individual-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
f1d7f80	jens-daniel-mueller	2022-07-13

13.1.3 Surface dCant

mean_surface_dcant <- dcant_profile_basin_MLR_all_ensemble %>%
  arrange(depth) %>%
  filter(depth <= 50) %>%
  group_by(period, basin, Version_ID_group, MLR_basins, data_source) %>%
  summarise(mean_surface_dcant = mean(dcant)) %>% 
  ungroup()


mean_surface_dcant %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = period,
             y = mean_surface_dcant,
             fill = Version_ID_group)) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_muted() +
  facet_wrap(~basin)

Past versions of surface_dcant-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25

full_join(dcant_profile_basin_MLR_all_ensemble %>%
            filter(depth <= 50),
          mean_surface_dcant) %>%
  group_by(basin, period, data_source) %>%
  summarise(
    mean_surface_dcant = mean(mean_surface_dcant, na.rm = TRUE),
    sd_surface_dcant = sd(dcant) * sd_factor
  ) %>%
  ungroup() %>%
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

basin	period	data_source	mean_surface_dcant	sd_surface_dcant
Global	1994 - 2004	mod	9.986285	1.0945634
Global	1994 - 2004	mod_truth	9.119623	0.0942101
Global	1994 - 2004	obs	10.027440	2.1875373
Global	1994 - 2014	mod	20.349763	1.7671640
Global	1994 - 2014	mod_truth	19.512012	0.1485833
Global	1994 - 2014	obs	20.616197	3.8921063
Global	2004 - 2014	mod	10.478444	0.9027755
Global	2004 - 2014	mod_truth	10.392390	0.0545695
Global	2004 - 2014	obs	10.657403	1.7478204
Indian	1994 - 2004	mod	9.618109	1.7588451
Indian	1994 - 2004	mod_truth	9.186885	0.0580122
Indian	1994 - 2004	obs	9.394741	2.2003655
Indian	1994 - 2014	mod	19.877479	3.1468737
Indian	1994 - 2014	mod_truth	19.386323	0.1379031
Indian	1994 - 2014	obs	20.198908	2.8316950
Indian	2004 - 2014	mod	10.400052	1.5129513
Indian	2004 - 2014	mod_truth	10.199438	0.0805878
Indian	2004 - 2014	obs	10.888760	1.5095440
N. Pacific	1994 - 2004	mod	10.098490	1.3675247
N. Pacific	1994 - 2004	mod_truth	9.221239	0.1727115
N. Pacific	1994 - 2004	obs	10.030856	2.7076018
N. Pacific	1994 - 2014	mod	20.590696	2.1681197
N. Pacific	1994 - 2014	mod_truth	19.988747	0.2795669
N. Pacific	1994 - 2014	obs	20.602712	4.8778119
N. Pacific	2004 - 2014	mod	10.518470	0.9349842
N. Pacific	2004 - 2014	mod_truth	10.767508	0.1072571
N. Pacific	2004 - 2014	obs	10.642882	2.2265069
S. Pacific	1994 - 2004	mod	9.609298	1.2072769
S. Pacific	1994 - 2004	mod_truth	8.898910	0.0712817
S. Pacific	1994 - 2004	obs	9.807174	2.0420842
S. Pacific	1994 - 2014	mod	19.847672	1.8235284
S. Pacific	1994 - 2014	mod_truth	19.346396	0.0775971
S. Pacific	1994 - 2014	obs	19.859225	3.6926131
S. Pacific	2004 - 2014	mod	10.235104	1.1423519
S. Pacific	2004 - 2014	mod_truth	10.447486	0.0106044
S. Pacific	2004 - 2014	obs	10.128636	1.6929087
N. Atlantic	1994 - 2004	mod	11.219982	1.1475039
N. Atlantic	1994 - 2004	mod_truth	9.693375	0.0623285
N. Atlantic	1994 - 2004	obs	11.093514	2.4026173
N. Atlantic	1994 - 2014	mod	21.737565	0.7119938
N. Atlantic	1994 - 2014	mod_truth	20.378481	0.0628299
N. Atlantic	1994 - 2014	obs	22.459459	4.4267368
N. Atlantic	2004 - 2014	mod	10.563675	1.4511103
N. Atlantic	2004 - 2014	mod_truth	10.685107	0.0125663
N. Atlantic	2004 - 2014	obs	11.422295	2.1258690
S. Atlantic	1994 - 2004	mod	10.109842	0.4621263
S. Atlantic	1994 - 2004	mod_truth	8.791151	0.0908767
S. Atlantic	1994 - 2004	obs	10.572092	2.2335592
S. Atlantic	1994 - 2014	mod	20.559925	1.2162979
S. Atlantic	1994 - 2014	mod_truth	18.459719	0.1557427
S. Atlantic	1994 - 2014	obs	21.288251	4.0332443
S. Atlantic	2004 - 2014	mod	10.978043	3.6754168
S. Atlantic	2004 - 2014	mod_truth	9.668568	0.0651916
S. Atlantic	2004 - 2014	obs	10.747927	1.8425424

13.1.4 Halving depth

dcant_profile_basin_MLR_all_ensemble_penetration_depth <-
full_join(dcant_profile_basin_MLR_all_ensemble,
          mean_surface_dcant) %>%
  filter(Version_ID_group != "Surface eMLR(C*)") %>%
  arrange(depth) %>%
  filter(dcant >= mean_surface_dcant / 2) %>%
  group_by(period, basin, Version_ID_group, MLR_basins, data_source) %>%
  summarise(max_depth = max(depth)) %>%
  ungroup()

dcant_profile_basin_MLR_all_ensemble_penetration_depth %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = period,
             y = max_depth,
             fill = MLR_basins)) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_muted() +
  facet_wrap(~basin)

Past versions of halving_depth_of_surface_dcant-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25

dcant_profile_basin_MLR_all_ensemble_penetration_depth_uncertainty <- 
additive_uncertainty(
  dcant_profile_basin_MLR_all_ensemble_penetration_depth %>% 
    rename(dcant = max_depth))

dcant_profile_basin_MLR_all_ensemble_penetration_depth_uncertainty <-
  dcant_profile_basin_MLR_all_ensemble_penetration_depth_uncertainty[[1]]

dcant_profile_basin_MLR_all_ensemble_penetration_depth_uncertainty %>% 
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

period	basin	data_source	sd	std_case	signif_single	signif_double
1994 - 2004	Global	mod	57.15453	327	1	1
1994 - 2004	Global	mod_truth	0.00000	351	1	1
1994 - 2004	Global	obs	100.00000	300	1	1
1994 - 2004	Indian	mod	179.17466	409	1	1
1994 - 2004	Indian	mod_truth	0.00000	483	1	1
1994 - 2004	Indian	obs	318.74755	400	1	0
1994 - 2004	N. Pacific	mod	58.37157	285	1	1
1994 - 2004	N. Pacific	mod_truth	0.00000	268	1	1
1994 - 2004	N. Pacific	obs	70.71068	250	1	1
1994 - 2004	S. Pacific	mod	79.50748	351	1	1
1994 - 2004	S. Pacific	mod_truth	0.00000	409	1	1
1994 - 2004	S. Pacific	obs	223.60680	300	1	0
1994 - 2004	N. Atlantic	mod	29.50525	285	1	1
1994 - 2004	N. Atlantic	mod_truth	0.00000	327	1	1
1994 - 2004	N. Atlantic	obs	100.00000	400	1	1
1994 - 2004	S. Atlantic	mod	31.24100	327	1	1
1994 - 2004	S. Atlantic	mod_truth	0.00000	351	1	1
1994 - 2004	S. Atlantic	obs	100.49876	300	1	1
1994 - 2014	Global	mod	54.35476	327	1	1
1994 - 2014	Global	mod_truth	0.00000	351	1	1
1994 - 2014	Global	obs	100.00000	300	1	1
1994 - 2014	Indian	mod	141.26302	443	1	1
1994 - 2014	Indian	mod_truth	0.00000	443	1	1
1994 - 2014	Indian	obs	101.98039	400	1	1
1994 - 2014	N. Pacific	mod	47.00468	268	1	1
1994 - 2014	N. Pacific	mod_truth	0.00000	251	1	1
1994 - 2014	N. Pacific	obs	113.57817	200	1	0
1994 - 2014	S. Pacific	mod	53.47560	378	1	1
1994 - 2014	S. Pacific	mod_truth	0.00000	409	1	1
1994 - 2014	S. Pacific	obs	215.40659	300	1	0
1994 - 2014	N. Atlantic	mod	25.86426	305	1	1
1994 - 2014	N. Atlantic	mod_truth	0.00000	351	1	1
1994 - 2014	N. Atlantic	obs	200.00000	400	1	1
1994 - 2014	S. Atlantic	mod	36.41758	327	1	1
1994 - 2014	S. Atlantic	mod_truth	0.00000	378	1	1
1994 - 2014	S. Atlantic	obs	153.62292	400	1	1
2004 - 2014	Global	mod	58.07753	327	1	1
2004 - 2014	Global	mod_truth	0.00000	351	1	1
2004 - 2014	Global	obs	100.00000	300	1	1
2004 - 2014	Indian	mod	175.84186	409	1	1
2004 - 2014	Indian	mod_truth	0.00000	443	1	1
2004 - 2014	Indian	obs	112.24972	300	1	1
2004 - 2014	N. Pacific	mod	38.27584	268	1	1
2004 - 2014	N. Pacific	mod_truth	0.00000	251	1	1
2004 - 2014	N. Pacific	obs	112.24972	200	1	0
2004 - 2014	S. Pacific	mod	79.52258	409	1	1
2004 - 2014	S. Pacific	mod_truth	0.00000	378	1	1
2004 - 2014	S. Pacific	obs	101.98039	300	1	1
2004 - 2014	N. Atlantic	mod	14.00000	327	1	1
2004 - 2014	N. Atlantic	mod_truth	0.00000	351	1	1
2004 - 2014	N. Atlantic	obs	224.49944	500	1	1
2004 - 2014	S. Atlantic	mod	45.00000	351	1	1
2004 - 2014	S. Atlantic	mod_truth	0.00000	378	1	1
2004 - 2014	S. Atlantic	obs	256.12497	600	1	1

dcant_profile_basin_MLR_all_ensemble_penetration_depth_uncertainty %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = period,
             y = std_case,
             ymin = std_case - sd,
             ymax = std_case + sd)) +
  geom_pointrange() +
  facet_wrap(~basin)

Past versions of halving_depth_of_surface_dcant-2.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25

rm(dcant_profile_basin_MLR_all_ensemble_penetration_depth,
   dcant_profile_basin_MLR_all_ensemble_penetration_depth_uncertainty)

13.1.5 Additive uncertainty

# dcant_profile_basin_MLR_all_ensemble %>% distinct(data_source)

dcant_profile_basin_MLR_all_uncertainty <- dcant_profile_basin_MLR_all_ensemble %>%
  select(-c(dcant_sd, Version_ID, tref1, tref2)) %>% 
  filter(period %in% two_decades)

dcant_profile_basin_MLR_all_uncertainty <- additive_uncertainty(dcant_profile_basin_MLR_all_uncertainty)

delta_dcant_profile_basin_MLR_all_uncertainty <- dcant_profile_basin_MLR_all_uncertainty[[3]]
dcant_profile_basin_MLR_all_uncertainty_contributions <- dcant_profile_basin_MLR_all_uncertainty[[2]]
dcant_profile_basin_MLR_all_uncertainty <- dcant_profile_basin_MLR_all_uncertainty[[1]]


p_dcant_profiles <-
  dcant_profile_basin_MLR_all_uncertainty %>%
  arrange(depth) %>%
  filter(
    depth <= params_global$inventory_depth_standard,
    data_source == "obs"
  ) %>%
  ggplot() +
  geom_vline(xintercept = 0, size = 0.1) +
  geom_ribbon(
    aes(
      xmin = std_case - sd*2,
      xmax = std_case + sd*2,
      y = depth,
      fill = period
    ),
    alpha = 0.2
  ) +
  geom_ribbon(
    aes(
      xmin = std_case - sd,
      xmax = std_case + sd,
      y = depth,
      fill = period
    ),
    alpha = 0.4
  ) +
  geom_path(
    aes(std_case,
        depth,
        col = period),
    size = 1
  ) +
  scale_y_reverse(breaks = seq(0,3000,500))+
  # scale_y_continuous(
  #   trans = trans_reverser("sqrt"),
  #   breaks = c(0, 100, 500, seq(1000, 5000, 1000)))+
  coord_cartesian(expand = 0, xlim = c(-2,14)) +
  scale_color_manual(values = c("#EE7733", "#009988"), name = "Mean \u00B1 1σ/2σ") +
  scale_fill_manual(values = c("#EE7733", "#009988"), name = "Mean \u00B1 1σ/2σ") +
  labs(y = "Depth (m)",
       x = dcant_umol_label) +
  facet_wrap(~ basin, ncol = 3, dir = "v") +
  theme(legend.position = c(0.9, 0.1))

p_dcant_profiles

Past versions of concentration_profiles_MLR_basins_additive_uncertainty-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14

p_dcant_profiles_mod <-
  dcant_profile_basin_MLR_all_uncertainty %>%
  arrange(depth) %>%
  filter(
    depth <= params_global$inventory_depth_standard,
    data_source != "obs"
  ) %>%
    mutate(data_source = case_when(
    data_source == "mod" ~ "Reconstruction",
    data_source == "mod_truth" ~ "Model truth"
  )) %>%  
  ggplot() +
  geom_vline(xintercept = 0, size = 0.1) +
  # geom_hline(yintercept = 1000, size = 0.1) +
  geom_ribbon(
    aes(
      xmin = std_case - sd*2,
      xmax = std_case + sd*2,
      y = depth,
      fill = data_source
    ),
    alpha = 0.2
  ) +
  geom_ribbon(
    aes(
      xmin = std_case - sd,
      xmax = std_case + sd,
      y = depth,
      fill = data_source
    ),
    alpha = 0.4
  ) +
  geom_path(
    aes(std_case,
        depth,
        col = data_source),
    size = 1
  ) +
  scale_y_reverse(breaks = seq(0,3000,500)) +
  # scale_y_continuous(trans = trans_reverser("sqrt"),
  #                    breaks = c(0, 100, 500, seq(1000, 5000, 1000))) +
  coord_cartesian(expand = 0, xlim = c(-2,14)) +
  scale_color_manual(values = c("#009988", "#EE7733"), name = "Mean \u00B1 1σ/2σ") +
  scale_fill_manual(values = c("#009988", "#EE7733"), name = "Mean \u00B1 1σ/2σ") +
  labs(y = "Depth (m)",
       x = dcant_umol_label) +
  facet_grid(period ~ basin) +
  theme(legend.position = "top")

p_dcant_profiles_mod

Past versions of concentration_profiles_MLR_basins_additive_uncertainty_mod-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14

dcant_profile_basin_MLR_all_ensemble %>%
  arrange(depth) %>%
  filter(
    period != "1994 - 2014",
    depth <= params_global$inventory_depth_standard,
    data_source == "obs"
  ) %>%
  group_split(data_source) %>%
  map(
    ~ ggplot() +
      geom_hline(yintercept = params_global$inventory_depth_standard) +
      geom_vline(xintercept = 0) +
      geom_path(
        data = .x,
        aes(dcant,
            depth,
            col = MLR_basins,
            group = Version_ID)
      ) +
      scale_size_manual(values = c(0.5, 2), name = "x") +
      scale_linetype_manual(values = c(2, 1), name = "x") +
      scale_y_continuous(trans = trans_reverser("sqrt"),
                         breaks = c(0, 100, 500, seq(1500, 5000, 1000))) +
      coord_cartesian(expand = 0) +
      scale_color_brewer(palette = "Dark2", name = "regional\nclustering") +
      labs(y = "Depth (m)",
           x = dcant_umol_label) +
      facet_grid(basin ~ period)
  )

[[1]]

Past versions of concentration_profiles_MLR_basins_ensemble_basin_separation-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
232909e	jens-daniel-mueller	2022-07-01

dcant_profile_basin_MLR_all_ensemble %>%
  arrange(depth) %>%
  filter(
    period != "1994 - 2014",
    depth <= params_global$inventory_depth_standard,
    data_source != "mod_truth"
  ) %>%
  group_split(data_source) %>%
  # head(1) %>%
  map(
    ~ ggplot() +
      geom_hline(yintercept = params_global$inventory_depth_standard) +
      geom_vline(xintercept = 0) +
      geom_path(
        data = .x,
        aes(
          dcant,
          depth,
          col = Version_ID_group,
          group = Version_ID
        )) +
      scale_y_continuous(trans = trans_reverser("sqrt"),
                         breaks = c(0, 100, 500, seq(1500, 5000, 1000))) +
      coord_cartesian(expand = 0) +
      scale_color_muted(
        name = "configuration"
      ) +
      labs(
        y = "Depth (m)",
        x = dcant_umol_label,
        title = paste("data_source", unique(.x$data_source))
      ) +
      facet_grid(basin ~ period)
  )

[[1]]

Past versions of concentration_profiles_MLR_basins_ensemble_member-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
6be73e0	jens-daniel-mueller	2022-06-30

[[2]]

Past versions of concentration_profiles_MLR_basins_ensemble_member-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
318fefe	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
6be73e0	jens-daniel-mueller	2022-06-30

bind_rows(
  dcant_profile_basin_MLR_all_ensemble %>%
    filter(
      period %in% two_decades,
      depth <= params_global$inventory_depth_standard,
      data_source == "obs",
      Version_ID_group == "Standard case"
    ),
  dcant_profile_basin_MLR_all_sensitivity %>%
    filter(
      period %in% two_decades,
      depth <= params_global$inventory_depth_standard,
      data_source == "obs",
      Version_ID_group == "No data adjustments"
    )
) %>%
  group_by(depth,
           period,
           basin,
           Version_ID_group) %>% 
  summarise(dcant_sd = sd(dcant),
            dcant = mean(dcant)) %>% 
  ungroup() %>% 
  filter(basin == "N. Pacific" & period == "2004 - 2014" |
         basin == "Indian" & period == "1994 - 2004") %>%
  ggplot() +
  geom_hline(yintercept = params_global$inventory_depth_standard) +
  geom_vline(xintercept = 0) +
  geom_ribbon(
    aes(
      xmin = dcant - dcant_sd*2,
      xmax = dcant + dcant_sd*2,
      y = depth,
      fill = Version_ID_group
    ),
    alpha = 0.2
  )+
  geom_ribbon(
    aes(
      xmin = dcant - dcant_sd,
      xmax = dcant + dcant_sd,
      y = depth,
      fill = Version_ID_group
    ),
    alpha = 0.4
  )+
  geom_path(aes(dcant,
                depth,
                col = Version_ID_group,
                group = Version_ID_group),
    size = 1) +
  scale_y_continuous(trans = trans_reverser("sqrt"),
                     breaks = c(0, 100, 500, seq(1500, 5000, 1000))) +
  coord_cartesian(expand = FALSE, xlim = c(-1,13)) +
  scale_color_manual(values = colour("high contrast", names = FALSE)(3)[c(1,3)],
                     name = "Mean \u00B1 1σ/2σ") +
  scale_fill_manual(values = colour("high contrast", names = FALSE)(3)[c(1,3)],
                    name = "Mean \u00B1 1σ/2σ") +
  labs(y = "Depth (m)",
       x = dcant_umol_label) +
  facet_nested_wrap(~ basin + period) +
  theme(legend.position = c(0.85,0.15))

Past versions of concentration_profiles_data_adjustments-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
46c163a	jens-daniel-mueller	2022-10-10

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_profiles_data_adjustments.png",
       height = 6,
       width = 8,
       dpi = 600)

13.2 dCant - delta

delta_dcant_profile_basin_MLR_all_uncertainty %>%
  group_split(data_source) %>%
  # head(3) %>%
  map(
    ~ ggplot(data = .x,
             aes(delta_dcant,
                 depth)) +
      geom_hline(yintercept = params_global$inventory_depth_standard) +
      geom_vline(xintercept = 0) +
      geom_ribbon(
        aes(
          xmin = delta_dcant - RSS,
          xmax = delta_dcant + RSS
        ),
        alpha = 0.3
      ) +
      geom_path() +
      scale_y_continuous(trans = trans_reverser("sqrt"),
                         breaks = c(0, 100, 500, seq(1500, 5000, 1000))) +
      coord_cartesian(expand = 0) +
      scale_color_brewer(palette = "Set1", name = "mean \u00B1 sd") +
      scale_fill_brewer(palette = "Set1", name = "mean \u00B1 sd") +
      labs(
        title = paste("data_source", unique(.x$data_source)),
        y = "Depth (m)",
        x = ddcant_umol_label
      ) +
      facet_wrap(~ basin, ncol = 3)
  )

[[1]]

Past versions of concentration_profiles_MLR_basins_offset-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
d2191ad	jens-daniel-mueller	2022-02-04
913e42f	jens-daniel-mueller	2022-02-01
189de95	jens-daniel-mueller	2022-02-01
b62308d	jens-daniel-mueller	2022-01-31
5f2aed0	jens-daniel-mueller	2022-01-27
eccd82b	jens-daniel-mueller	2022-01-26
c6fe495	jens-daniel-mueller	2022-01-26

[[2]]

Past versions of concentration_profiles_MLR_basins_offset-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
5520de8	jens-daniel-mueller	2022-11-15
7275091	jens-daniel-mueller	2022-11-05
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
318fefe	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
d2191ad	jens-daniel-mueller	2022-02-04
913e42f	jens-daniel-mueller	2022-02-01
189de95	jens-daniel-mueller	2022-02-01
b62308d	jens-daniel-mueller	2022-01-31
5f2aed0	jens-daniel-mueller	2022-01-27
eccd82b	jens-daniel-mueller	2022-01-26
c6fe495	jens-daniel-mueller	2022-01-26

[[3]]

Past versions of concentration_profiles_MLR_basins_offset-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
318fefe	jens-daniel-mueller	2022-08-11

14 Inventories

dcant_budget_basin_MLR_all_uncertainty_sensitivity <-
  dcant_budget_basin_MLR_all_sensitivity %>%
  filter(data_source != "mod_truth",
         period != "1994 - 2014") %>%
  select(-c(inv_depth, Version_ID, tref1, tref2))

dcant_budget_basin_MLR_all_uncertainty_sensitivity <- additive_uncertainty(dcant_budget_basin_MLR_all_uncertainty_sensitivity)

delta_dcant_budget_basin_MLR_all_uncertainty_sensitivity <- dcant_budget_basin_MLR_all_uncertainty_sensitivity[[3]]
dcant_budget_basin_MLR_all_uncertainty_contributions_sensitivity <- dcant_budget_basin_MLR_all_uncertainty_sensitivity[[2]]
dcant_budget_basin_MLR_all_uncertainty_sensitivity <- dcant_budget_basin_MLR_all_uncertainty_sensitivity[[1]]

dcant_budget_basin_MLR_all_uncertainty <-
  dcant_budget_basin_MLR_all_ensemble %>%
  filter(data_source != "mod_truth",
         period == "1994 - 2014") %>%
  select(-c(inv_depth, Version_ID, tref1, tref2))

dcant_budget_basin_MLR_all_uncertainty <- additive_uncertainty(dcant_budget_basin_MLR_all_uncertainty)

delta_dcant_budget_basin_MLR_all_uncertainty <- dcant_budget_basin_MLR_all_uncertainty[[3]]
dcant_budget_basin_MLR_all_uncertainty_contributions <- dcant_budget_basin_MLR_all_uncertainty[[2]]
dcant_budget_basin_MLR_all_uncertainty <- dcant_budget_basin_MLR_all_uncertainty[[1]]

dcant_budget_basin_MLR_all_uncertainty <-
  dcant_budget_basin_MLR_all_ensemble %>%
  filter(data_source != "mod_truth",
         period != "1994 - 2014") %>%
  select(-c(inv_depth, Version_ID, tref1, tref2))

dcant_budget_basin_MLR_all_uncertainty <- additive_uncertainty(dcant_budget_basin_MLR_all_uncertainty)

delta_dcant_budget_basin_MLR_all_uncertainty <- dcant_budget_basin_MLR_all_uncertainty[[3]]
dcant_budget_basin_MLR_all_uncertainty_contributions <- dcant_budget_basin_MLR_all_uncertainty[[2]]
dcant_budget_basin_MLR_all_uncertainty <- dcant_budget_basin_MLR_all_uncertainty[[1]]

14.1 Deep Ocean

dcant_budget_basin_MLR_all_ensemble_10000 <-
  bind_rows(
    dcant_budget_basin_MLR_all_ensemble,
    dcant_budget_basin_MLR_all_ensemble_10000)

dcant_budget_basin_MLR_all_ensemble_10000 %>% 
  ggplot(aes(period,dcant, fill = as.factor(inv_depth))) +
  geom_jitter(shape = 21, alpha = 0.8) +
  facet_grid(basin ~ data_source, scales = "free_y") +
  scale_fill_highcontrast(name = "Integration\ndepth (m)") +
  scale_y_continuous(limits = c(0,NA))

Past versions of deep_cant_accumulation-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15

dcant_budget_basin_MLR_all_ensemble_10000 %>% 
  arrange(inv_depth) %>% 
  group_by(basin, period, data_source, MLR_basins, Version_ID_group) %>% 
  summarise(deep_dcant = 100*((dcant / lag(dcant))-1)) %>% 
  ungroup() %>% 
  ggplot(aes(period,deep_dcant, fill = Version_ID_group)) +
  geom_hline(yintercept = 0) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_muted() +
  facet_grid(basin ~ data_source)

Past versions of deep_cant_accumulation-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15

dcant_budget_basin_MLR_all_ensemble_10000 %>% 
  arrange(inv_depth) %>% 
  group_by(basin, period, data_source, MLR_basins, Version_ID_group) %>% 
  summarise(deep_dcant = 100*((dcant / lag(dcant))-1)) %>% 
  ungroup() %>% 
  ggplot(aes(period,deep_dcant, fill = MLR_basins)) +
  geom_hline(yintercept = 0) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_bright() +
  facet_grid(basin ~ data_source)

Past versions of deep_cant_accumulation-3.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15

# inventory scaling

dcant_budget_basin_MLR_all_ensemble <-
  dcant_budget_basin_MLR_all_ensemble %>%
  mutate(dcant = if_else(basin == "Global",
                         dcant * dcant_scaling,
                         dcant))

dcant_budget_basin_MLR_all <- dcant_budget_basin_MLR_all %>%
  mutate(dcant = if_else(basin == "Global",
                         dcant * dcant_scaling,
                         dcant))

dcant_budget_basin_MLR_all_sensitivity <-
  dcant_budget_basin_MLR_all_sensitivity %>%
  mutate(dcant = if_else(basin == "Global",
                         dcant * dcant_scaling,
                         dcant))

# uncertainty scaling

scaling_uncertainty <-
  dcant_budget_basin_MLR_all_uncertainty %>%
  select(-c(sd, starts_with("signif_"))) %>%
  filter(basin == "Global") %>%
  mutate(
    sd = std_case * (dcant_scaling - 1) * dcant_scaling_uncertainty,
    Version_ID_group = "Scaling uncertainty"
  ) %>%
  select(-std_case)


dcant_budget_basin_MLR_all_uncertainty_contributions <-
  bind_rows(dcant_budget_basin_MLR_all_uncertainty_contributions,
            scaling_uncertainty)

rm(scaling_uncertainty)

dcant_budget_basin_MLR_all_uncertainty <-
  dcant_budget_basin_MLR_all_uncertainty %>%
  mutate(
    sd = if_else(basin == "Global",
                 sqrt(
                   sd ^ 2 +
                     (std_case * (dcant_scaling - 1) * dcant_scaling_uncertainty) ^
                     2
                 ),
                 sd),
    std_case = if_else(basin == "Global",
                       std_case * dcant_scaling,
                       std_case)
  )





dcant_budget_basin_MLR_all_uncertainty_sensitivity <-
  dcant_budget_basin_MLR_all_uncertainty_sensitivity %>%
  mutate(
    sd = if_else(basin == "Global",
                 sqrt(
                   sd ^ 2 +
                     (std_case * (dcant_scaling - 1) * dcant_scaling_uncertainty) ^
                     2
                 ),
                 sd),
    std_case = if_else(basin == "Global",
                       std_case * dcant_scaling,
                       std_case)
  )



delta_dcant_budget_basin_MLR_all_uncertainty <-
  delta_dcant_budget_basin_MLR_all_uncertainty %>%
  mutate(
    RSS = if_else(basin == "Global",
                  RSS * dcant_scaling,
                  RSS),
    delta_dcant = if_else(basin == "Global",
                          delta_dcant * dcant_scaling,
                          delta_dcant),
  )

dcant_budget_basin_MLR_all_uncertainty_contributions %>%
  filter(data_source == "obs") %>%
  mutate(
    basin = fct_relevel(
      basin,
      "Global",
      "N. Pacific",
      "N. Atlantic",
      "Indian",
      "S. Pacific",
      "S. Atlantic"
    )
  ) %>%
  ggplot(aes("x", abs(sd), fill = Version_ID_group)) +
  geom_hline(yintercept = 0, col = "grey10") +
  geom_col(position = position_dodge(preserve = "single"),
           col = "grey10") +
  scale_fill_bright(name = "eMLR(C*)\nconfiguration\ncomponent") +
  labs(y = expression(Absolute~Delta*Cant~inventory~offset~from~standard~case~(Pg~C~dec^{-1}))) +
  theme(
    axis.title.x = element_blank(),
    axis.text.x = element_blank(),
    axis.ticks.x = element_blank()
  ) +
  guides(pattern = guide_legend(override.aes = list(fill = "white")),
         fill = guide_legend(override.aes = list(pattern = "none"))) +
  # expand_limits(y = c(0,2.2)) +
  # scale_y_continuous(expand = c(0,0)) +
  facet_nested_wrap( ~ basin + period, ncol = 6)

Past versions of dcant_inventory_uncertainty_components-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
f39bfd4	jens-daniel-mueller	2022-11-17
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_inventory_uncertainty_contributions.png",
       height = 6,
       width = 8,
       dpi = 600) 

dcant_budget_basin_MLR_all_uncertainty_contributions_sensitivity %>%
  filter(data_source == "obs",
         Version_ID_group != "Regional clustering") %>%
  mutate(
    basin = fct_relevel(
      basin,
      "Global",
      "N. Pacific",
      "N. Atlantic",
      "Indian",
      "S. Pacific",
      "S. Atlantic"
    )
  ) %>%
  ggplot(aes("x", abs(sd), fill = Version_ID_group)) +
  geom_hline(yintercept = 0, col = "grey10") +
  geom_col(position = position_dodge(preserve = "single"),
           col = "grey10") +
  scale_fill_bright(name = "eMLR(C*)\nconfiguration\ncomponent") +
  labs(y = expression(Absolute~Delta*Cant~inventory~offset~from~standard~case~(Pg~C~dec^{-1}))) +
  theme(
    axis.title.x = element_blank(),
    axis.text.x = element_blank(),
    axis.ticks.x = element_blank()
  ) +
  guides(pattern = guide_legend(override.aes = list(fill = "white")),
         fill = guide_legend(override.aes = list(pattern = "none"))) +
  # expand_limits(y = c(0,2.2)) +
  # scale_y_continuous(expand = c(0,0)) +
  facet_nested_wrap( ~ basin + period, ncol = 6)

Past versions of dcant_inventory_uncertainty_components_sensitivity_cases-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_inventory_uncertainty_contributions_sensitivity.png",
       height = 6,
       width = 8,
       dpi = 600) 

# estimate area scaling factor to achieve identical coverage

tcant_inv <-
  inner_join(tcant_inv, basinmask)

dcant_inv_all <-
  inner_join(dcant_inv_all, basinmask)

area_scaling <-
  bind_rows(
    tcant_inv %>% distinct(lat, lon, basin) %>%
      mutate(source = "SO4"),
    dcant_inv_all %>% distinct(lat, lon, basin) %>%
      mutate(source = "M22")
  )


map +
  geom_tile(data = area_scaling,
            aes(lon, lat, fill = basin)) +
  facet_wrap( ~ source)

Past versions of budgets_basins_hemisphere_tcant_for_scatterplot-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
2345935	jens-daniel-mueller	2022-10-29
5b77c4f	jens-daniel-mueller	2022-08-10
d2ae54c	jens-daniel-mueller	2022-07-18

area_scaling <- area_scaling %>% 
  mutate(area = earth_surf(lat, lon)) %>% 
  group_by(basin, source) %>% 
  summarise(area_total = sum(area)) %>% 
  ungroup() %>% 
  pivot_wider(values_from = area_total,
              names_from = source) %>% 
  mutate(scaling_factor = M22 / SO4)


tcant_budget_basin_MLR <-
  tcant_inv %>%
  mutate(method = "layer",
         data_source = "obs") %>% 
  group_by(basin) %>%
  nest() %>% 
  mutate(budget = map(.x = data, ~m_dcant_budget(.x))) %>% 
  select(-data) %>%
  unnest(budget)

tcant_budget_basin_MLR <-
  full_join(tcant_budget_basin_MLR,
            area_scaling %>% select(basin, scaling_factor)) %>%
  mutate(value = value * scaling_factor) %>%
  select(-scaling_factor)


tcant_budget_basin_MLR <-
  left_join(tcant_budget_basin_MLR %>%
              mutate(period = "1800 - 1994"),
            delta_pCO2_atm)


tcant_budget_basin_MLR <- tcant_budget_basin_MLR %>% 
  filter(estimate == "dcant") %>% 
  select(-estimate) %>% 
  rename(dcant = value)

tcant_budget_basin_MLR <- tcant_budget_basin_MLR %>% 
  mutate(beta_1994 = dcant / delta_pCO2) %>% 
  select(basin, beta_1994)

tcant_budget_basin_MLR <-
  bind_rows(tcant_budget_basin_MLR,
            tibble(basin = "Global", beta_1994 = 118 / 78.3))

14.2 Scatterplots

14.2.1 Ensemble uncertainty

dcant_budget_basin_MLR_all_ensemble %>%
  ggplot(aes(period, dcant, col = Version_ID_group)) +
  geom_jitter(alpha = 0.5) +
  scale_color_brewer(palette = "Dark2") +
  facet_grid(basin ~ data_source, scales = "free_y")

Past versions of dcant_budget_basin_hemisphere_ensemble_uncertainty-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

tcant_budget_basin_MLR <- expand_grid(
  tcant_budget_basin_MLR,
  delta_pCO2_atm %>% filter(period %in% two_decades)
)

tcant_budget_basin_MLR <- tcant_budget_basin_MLR %>% 
  mutate(dcant = beta_1994 * delta_pCO2)

tcant_budget_basin_MLR <- tcant_budget_basin_MLR %>%
  mutate(
    basin = fct_relevel(
      basin,
      "Global",
      "Indian",
      "N. Pacific",
      "S. Pacific",
      "N. Atlantic",
      "S. Atlantic"
    )
  )


dcant_budget_basin_MLR_all_ensemble %>%
  filter(data_source == "obs",
         period != "1994 - 2014") %>%
  mutate(ensemble_role = "Uncertainty cases") %>% 
  ggplot(aes(period, dcant)) +
  # scale_fill_muted(name = "ensemble group") +
  scale_color_manual(values = c("grey", "black", "red"),
                     name = "Ensemble member") +
  geom_boxplot(
    width = .1,
    outlier.shape = NA,
    fill = "grey",
    alpha = 0.7,
    position = position_nudge(x = 0.3)
  ) +
  geom_point(
    data = dcant_budget_basin_MLR_all_sensitivity %>%
      mutate(ensemble_role = "Sensitivity cases") %>%
      filter(data_source == "obs",
         period != "1994 - 2014"),
    aes(fill = Version_ID_group,
        col = ensemble_role),
    size = 1.3,
    alpha = .7,
    shape = 21,
    position = position_dodge2(width = 0.3)
  ) +
  geom_point(
    aes(fill = Version_ID_group,
        col = ensemble_role),
    size = 1.3,
    alpha = .7,
    shape = 21,
    position = position_dodge2(width = 0.3)
  ) +
  scale_y_continuous(name = dcant_pgc_label) +
  facet_wrap(~ basin, ncol = 3, scales = "free_y")

Past versions of dcant_budget_basin_hemisphere_ensemble_uncertainty-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

14.2.2 Additive uncertainty

dcant_budget_basin_MLR_all_ensemble %>%
  filter(data_source == "obs",
         period != "1994 - 2014",
         !(Version_ID %in% Version_IDs),
         MLR_basins == unique(params_local_all$MLR_basins)) %>%
  ggplot(aes(period, dcant)) +
  geom_errorbar(
    data = tcant_budget_basin_MLR,
    aes(x = period,
        ymax = dcant, ymin = dcant,
        col = "Scaled 1994\nestimate"),
    width = 0.8,
    size = 1,
    linetype = 1,
    position = position_nudge(x = 0.05)
  ) +
  geom_jitter(
    data = dcant_budget_basin_MLR_all_ensemble %>%
      filter(data_source == "obs",
             period != "1994 - 2014",
             Version_ID_group == unique(params_local_all$Version_ID_group)),
    aes(fill = "Regional clustering",
        alpha = "Regional clustering"),
    shape = 21,
    position = position_jitter(width = 0.1, height = 0)
  ) +
  geom_point(
    aes(fill = "Configuration changes",
        alpha = "Configuration changes"),
    shape = 21,
    position = position_jitter(width = 0.1, height = 0)
  ) +
  geom_crossbar(
    data = dcant_budget_basin_MLR_all_uncertainty %>% filter(data_source == "obs"),
    aes(
      x = period,
      y = std_case,
      ymin = std_case - sd*2,
      ymax = std_case + sd*2,
      linetype = "Combined uncertainty"
    ),
    width = 0.1,
    fill = "#BBBBBB",
    alpha = 0.2,
    position = position_nudge(x = 0.2)
  ) +
  geom_crossbar(
    data = dcant_budget_basin_MLR_all_uncertainty %>% filter(data_source == "obs"),
    aes(
      x = period,
      y = std_case,
      ymin = std_case - sd,
      ymax = std_case + sd,
      alpha = "Combined uncertainty",
    ),
    width = 0.1,
    fill = "#BBBBBB",
    alpha = 0.4,
    position = position_nudge(x = 0.2)
  ) +
  geom_point(
    data = dcant_budget_basin_MLR_all_uncertainty %>% filter(data_source == "obs"),
    aes(x = period,
        y = std_case,
        fill = "Standard case",
        alpha = "Standard case"),
    size = 2,
    shape = 21,
    position = position_nudge(x = 0.2)
  ) +
  scale_fill_manual(name = "group", values = c("#BB5566", "#004488", "white")) +
  scale_color_manual(values = c("grey50"), label = expression(Scaled~C[ant]~1994)) +
  scale_linetype(name = "X") +
  scale_alpha_manual(name = "group", values = c(0.7,0.7,1)) +
  scale_shape_manual(values = 95) +
  scale_y_continuous(name = dcant_pgc_label) +
  guides(
    fill = guide_legend(order = 1),
    linetype = guide_legend(order = 2),
    alpha = guide_legend(order = 1)
  ) +
  facet_wrap(~ basin, ncol = 3, dir = "v", scales = "free_y") +
  theme(legend.title = element_blank(),
        axis.title.x = element_blank())

Past versions of dcant_budget_basin_hemisphere_additive_uncertainty-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
0b939ca	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05

ggsave(path = here::here("output/publication"),
       filename = "Fig_dcant_inventory.png",
       height = 6,
       width = 8,
       dpi = 600)

14.2.3 Beta

# determine scaling factor from dcant -> beta inventories
# including scaling for: delta pCO2 atm, basin area, PgC -> molC

beta_scaling <-
  bind_rows(
    area_scaling,
    area_scaling %>%
      summarise(M22 = sum(M22) * dcant_scaling) %>%
      mutate(basin = "Global")
  ) %>% 
  select(basin, area = M22)


beta_scaling <- 
  full_join(delta_pCO2_atm,
            beta_scaling,
            by = character())

beta_scaling <- beta_scaling %>% 
  mutate(scaling_factor = 1e15 / 12 / delta_pCO2 / area) %>% 
  select(period, basin, scaling_factor, area)

beta_scaling <- beta_scaling %>%
  mutate(
    basin = fct_relevel(
      basin,
      "Global",
      "Indian",
      "N. Pacific",
      "S. Pacific",
      "N. Atlantic",
      "S. Atlantic"
    )
  )

beta_budget_basin_MLR_all_sensitivity <-
  left_join(dcant_budget_basin_MLR_all_sensitivity,
            beta_scaling) %>%
  mutate(beta = dcant * scaling_factor) %>% 
  select(-scaling_factor)

beta_budget_basin_MLR_all_ensemble <-
  left_join(dcant_budget_basin_MLR_all_ensemble,
            beta_scaling) %>%
  mutate(beta = dcant * scaling_factor) %>% 
  select(-scaling_factor)


tcant_budget_basin_MLR <-
  left_join(tcant_budget_basin_MLR,
            beta_scaling) %>%
  mutate(beta_1994 = beta_1994 * 1e15 / 12 / area) %>% 
  select(-c(area,scaling_factor))

beta_budget_basin_MLR_all_uncertainty <- 
  left_join(dcant_budget_basin_MLR_all_uncertainty,
            beta_scaling) %>% 
  mutate(
    sd = sd * scaling_factor,
    std_case = std_case * scaling_factor
    ) %>% 
  select(-scaling_factor)

beta_budget_basin_MLR_all_uncertainty_contributions <- 
  left_join(dcant_budget_basin_MLR_all_uncertainty_contributions,
            beta_scaling) %>% 
  mutate(
    sd = sd * scaling_factor
    ) %>% 
  select(-scaling_factor)

dcant_beta_basin_MLR_all_uncertainty <-
  bind_rows(
    dcant_budget_basin_MLR_all_uncertainty %>% mutate(estimate = "dcant"),
    beta_budget_basin_MLR_all_uncertainty %>% mutate(estimate = "beta")
  ) %>%
  select(-c(area, starts_with("signif_")))
  

dcant_beta_basin_MLR_all_uncertainty_significance <- 
  dcant_beta_basin_MLR_all_uncertainty %>% 
  arrange(period) %>% 
  group_by(estimate, basin, data_source) %>%
  mutate(
    delta = std_case - lag(std_case),
    RSS = sqrt(sd ^ 2 + lag(sd ^ 2))
  ) %>%
  ungroup() %>% 
  filter(!is.na(delta))
  

dcant_beta_basin_MLR_all_uncertainty_significance <-
  full_join(
    dcant_beta_basin_MLR_all_uncertainty %>%
      pivot_wider(
        names_from = period,
        values_from = c(sd, std_case),
        names_sep = " "
      ),
    dcant_beta_basin_MLR_all_uncertainty_significance %>% 
      select(-c(sd, std_case))
  )

dcant_beta_basin_MLR_all_uncertainty_significance <-
  dcant_beta_basin_MLR_all_uncertainty_significance %>% 
  mutate(across(where(is.numeric), signif, 2))

dcant_beta_basin_MLR_all_uncertainty_significance %>% 
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

basin	data_source	estimate	sd 1994 - 2004	sd 2004 - 2014	std_case 1994 - 2004	std_case 2004 - 2014	period	delta	RSS
Global	mod	dcant	1.600	1.600	20.00	25.00	2004 - 2014	4.8000	2.300
Global	obs	dcant	2.500	2.500	29.00	27.00	2004 - 2014	-1.9000	3.600
Indian	mod	dcant	0.800	0.700	5.10	6.60	2004 - 2014	1.5000	1.100
Indian	obs	dcant	0.860	0.640	7.20	5.70	2004 - 2014	-1.4000	1.100
N. Pacific	mod	dcant	0.260	0.300	3.10	3.40	2004 - 2014	0.2300	0.400
N. Pacific	obs	dcant	0.780	1.800	2.90	3.20	2004 - 2014	0.3200	1.900
S. Pacific	mod	dcant	0.800	0.440	5.50	7.40	2004 - 2014	1.9000	0.920
S. Pacific	obs	dcant	1.200	0.960	8.60	7.40	2004 - 2014	-1.2000	1.500
N. Atlantic	mod	dcant	0.180	0.280	2.20	2.40	2004 - 2014	0.2200	0.330
N. Atlantic	obs	dcant	0.230	0.360	4.80	3.90	2004 - 2014	-0.9000	0.430
S. Atlantic	mod	dcant	0.230	0.240	2.70	3.40	2004 - 2014	0.7100	0.330
S. Atlantic	obs	dcant	0.530	0.620	3.90	5.40	2004 - 2014	1.5000	0.820
Global	mod	beta	0.020	0.018	0.25	0.28	2004 - 2014	0.0340	0.027
Global	obs	beta	0.032	0.029	0.37	0.31	2004 - 2014	-0.0540	0.043
Indian	mod	beta	0.049	0.039	0.31	0.37	2004 - 2014	0.0540	0.063
Indian	obs	beta	0.052	0.036	0.44	0.32	2004 - 2014	-0.1200	0.063
N. Pacific	mod	beta	0.015	0.016	0.18	0.18	2004 - 2014	-0.0033	0.022
N. Pacific	obs	beta	0.045	0.091	0.17	0.17	2004 - 2014	0.0023	0.100
S. Pacific	mod	beta	0.037	0.019	0.26	0.32	2004 - 2014	0.0580	0.042
S. Pacific	obs	beta	0.055	0.041	0.40	0.31	2004 - 2014	-0.0880	0.069
N. Atlantic	mod	beta	0.020	0.028	0.24	0.25	2004 - 2014	0.0010	0.034
N. Atlantic	obs	beta	0.025	0.036	0.53	0.39	2004 - 2014	-0.1400	0.044
S. Atlantic	mod	beta	0.023	0.022	0.27	0.31	2004 - 2014	0.0410	0.032
S. Atlantic	obs	beta	0.053	0.057	0.39	0.49	2004 - 2014	0.0990	0.078

dcant_beta_basin_MLR_all_uncertainty_significance_table <-
dcant_beta_basin_MLR_all_uncertainty_significance %>%
  filter(data_source == "obs") %>% 
  select(-data_source) %>% 
  mutate(
    `1994 - 2004` = paste(`std_case 1994 - 2004`, `sd 1994 - 2004`, sep = " ± "),
    `2004 - 2014` = paste(`std_case 2004 - 2014`, `sd 2004 - 2014`, sep = " ± "),
    `Decadal change` = paste0(delta, " ± ", RSS, 
                              if_else(abs(delta) > RSS, " *",""), 
                              if_else(abs(delta) > RSS*2, "*",""))
  ) %>%
  select(Region = basin,
         Estimate = estimate,
         `1994 - 2004`,
         `2004 - 2014`,
         `Decadal change`) %>%
  arrange(Region) 

dcant_beta_basin_MLR_all_uncertainty_significance_table %>% 
  write_csv(here::here("output/publication/Table_inventory_stats.csv"))

beta_budget_basin_MLR_all_ensemble %>%
  filter(
    data_source == "obs",
    period != "1994 - 2014",
    !(Version_ID %in% Version_IDs),
    MLR_basins == unique(params_local_all$MLR_basins)
  ) %>%
  ggplot(aes(period, beta)) +
  geom_errorbar(
    data = tcant_budget_basin_MLR %>%
      rename(beta = beta_1994),
    aes(x = period,
        ymax = beta, ymin = beta,
        col = "Scaled 1994\nestimate"),
    width = 0.8,
    size = 1,
    linetype = 1,
    position = position_nudge(x = 0.05)
  ) +
  geom_jitter(
    data = beta_budget_basin_MLR_all_ensemble %>%
      filter(data_source == "obs",
             period != "1994 - 2014",
             Version_ID_group == unique(params_local_all$Version_ID_group)),
    aes(fill = "Regional clustering",
        alpha = "Regional clustering"),
    shape = 21,
    position = position_jitter(width = 0.1, height = 0)
  ) +
  geom_point(
    aes(fill = "Configuration changes",
        alpha = "Configuration changes"),
    shape = 21,
    position = position_jitter(width = 0.1, height = 0)
  ) +
  geom_crossbar(
    data = beta_budget_basin_MLR_all_uncertainty %>% filter(data_source == "obs"),
    aes(
      x = period,
      y = std_case,
      ymin = std_case - sd*2,
      ymax = std_case + sd*2,
      linetype = "Combined uncertainty"
    ),
    width = 0.1,
    fill = "#BBBBBB",
    alpha = 0.2,
    position = position_nudge(x = 0.2)
  ) +
  geom_crossbar(
    data = beta_budget_basin_MLR_all_uncertainty %>% filter(data_source == "obs"),
    aes(
      x = period,
      y = std_case,
      ymin = std_case - sd,
      ymax = std_case + sd,
      alpha = "Combined uncertainty",
    ),
    width = 0.1,
    fill = "#BBBBBB",
    alpha = 0.4,
    position = position_nudge(x = 0.2)
  ) +
  geom_point(
    data = beta_budget_basin_MLR_all_uncertainty %>% filter(data_source == "obs"),
    aes(x = period,
        y = std_case,
        fill = "Standard case",
        alpha = "Standard case"),
    size = 2,
    shape = 21,
    position = position_nudge(x = 0.2)
  ) +
  scale_fill_manual(name = "group", values = c("#BB5566", "#004488", "white")) +
  scale_color_manual(values = c("grey50"), label = expression(Scaled~C[ant]~1994)) +
  scale_linetype(name = "X") +
  scale_alpha_manual(name = "group", values = c(0.7,0.7,1)) +
  scale_shape_manual(values = 95) +
  scale_y_continuous(name = beta_inv_label) +
  guides(
    fill = guide_legend(order = 1),
    linetype = guide_legend(order = 2),
    alpha = guide_legend(order = 1)
  ) +
  facet_wrap(~ basin, ncol = 3, dir = "v") +
  theme(legend.title = element_blank(),
        axis.title.x = element_blank())

Past versions of beta_dcant_inventory_uncertainty-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
8e1702d	jens-daniel-mueller	2022-11-16
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14

ggsave(path = here::here("output/publication"),
       filename = "FigS_beta_inventory.png",
       height = 6,
       width = 8,
       dpi = 600)

14.2.4 Biases

dcant_budget_basin_MLR_all_bias <-
  dcant_budget_basin_MLR_all_ensemble %>%
  filter(data_source %in% c("mod", "mod_truth")) %>%
  pivot_wider(names_from = data_source,
              values_from = dcant) %>%
  mutate(dcant_bias = mod - mod_truth,
         dcant_bias_rel = 100 * dcant_bias / mod_truth) %>%
  filter(period %in% two_decades)


dcant_budget_basin_MLR_all_bias <-
  dcant_budget_basin_MLR_all_bias %>%
  select(basin, period, MLR_basins, Version_ID_group, dcant = dcant_bias)

dcant_budget_basin_MLR_all_bias_uncertainty <-
  additive_uncertainty(dcant_budget_basin_MLR_all_bias)

delta_dcant_budget_basin_MLR_all_bias_uncertainty <-
  dcant_budget_basin_MLR_all_bias_uncertainty[[3]]
dcant_budget_basin_MLR_all_bias_uncertainty_contributions <-
  dcant_budget_basin_MLR_all_bias_uncertainty[[2]]
dcant_budget_basin_MLR_all_bias_uncertainty <-
  dcant_budget_basin_MLR_all_bias_uncertainty[[1]]



dcant_budget_basin_MLR_all_bias  %>%
  filter(period != "1994 - 2014",
         MLR_basins == unique(params_local_all$MLR_basins)) %>%
  ggplot(aes(period, dcant)) +
  geom_hline(yintercept = 0) +
  geom_jitter(
    data = dcant_budget_basin_MLR_all_bias %>%
      filter(period != "1994 - 2014",
             Version_ID_group == unique(params_local_all$Version_ID_group)),
    aes(fill = "Regional clustering",
        alpha = "Regional clustering"),
    shape = 21,
    position = position_jitter(width = 0.1, height = 0)
  ) +
  geom_point(
    aes(fill = "Configuration changes",
        alpha = "Configuration changes"),
    shape = 21,
    position = position_jitter(width = 0.1, height = 0)
  ) +
  geom_crossbar(
    data = dcant_budget_basin_MLR_all_bias_uncertainty,
    aes(
      x = period,
      y = std_case,
      ymin = std_case - sd*2,
      ymax = std_case + sd*2,
      linetype = "Combined uncertainty"
    ),
    width = 0.1,
    fill = "#BBBBBB",
    alpha = 0.2,
    position = position_nudge(x = 0.2)
  ) +
  geom_crossbar(
    data = dcant_budget_basin_MLR_all_bias_uncertainty,
    aes(
      x = period,
      y = std_case,
      ymin = std_case - sd,
      ymax = std_case + sd,
      alpha = "Combined uncertainty",
    ),
    width = 0.1,
    fill = "#BBBBBB",
    alpha = 0.4,
    position = position_nudge(x = 0.2)
  ) +
  geom_point(
    data = dcant_budget_basin_MLR_all_bias_uncertainty,
    aes(x = period,
        y = std_case,
        fill = "Standard case",
        alpha = "Standard case"),
    size = 2,
    shape = 21,
    position = position_nudge(x = 0.2)
  ) +
  scale_fill_manual(name = "group", values = c("#BB5566", "#004488", "white")) +
  scale_color_manual(values = c("grey50"), label = expression(Scaled~C[ant]~1994)) +
  scale_linetype(name = "X") +
  scale_alpha_manual(name = "group", values = c(0.7,0.7,1)) +
  scale_shape_manual(values = 95) +
  scale_y_continuous(name = dcant_bias_pgc_label) +
  guides(
    fill = guide_legend(order = 1),
    linetype = guide_legend(order = 2),
    alpha = guide_legend(order = 1)
  ) +
  facet_wrap(~ basin, ncol = 3, dir = "v", scales = "free_y") +
  theme(legend.title = element_blank(),
        axis.title.x = element_blank())

Past versions of budget_biases_additive-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_bias_inventory_synthetic_data.png",
       height = 6,
       width = 8,
       dpi = 600)

14.3 Flow chart

dcant_budget_basin_MLR_all <- 
  dcant_budget_basin_MLR_all %>% 
  filter(basin != "Global")

dcant_budget_basin_MLR_all <- left_join(dcant_budget_basin_MLR_all,
                           co2_atm %>% rename(tref1 = year,
                                              pCO21 = pCO2))

dcant_budget_basin_MLR_all <- left_join(dcant_budget_basin_MLR_all,
                           co2_atm %>% rename(tref2 = year,
                                              pCO22 = pCO2))


dcant_budget_basin_MLR_all_plot <- dcant_budget_basin_MLR_all %>%
  filter(period != "1994 - 2014",
         data_source == "obs")

tcant_budget_basin_MLR <-
  tcant_inv %>%
  mutate(method = "layer",
         data_source = "obs") %>% 
  group_by(basin) %>%
  nest() %>% 
  mutate(budget = map(.x = data, ~m_dcant_budget(.x))) %>% 
  select(-data) %>%
  unnest(budget)

tcant_budget_basin_MLR <-
  full_join(tcant_budget_basin_MLR,
            area_scaling %>% select(basin, scaling_factor)) %>%
  mutate(value = value * scaling_factor) %>%
  select(-scaling_factor)


tcant_budget_basin_MLR <- tcant_budget_basin_MLR %>% 
  mutate(pCO21 = 281,
         tref1 = 1800,
         tref2 = 1994)

tcant_budget_basin_MLR <- left_join(tcant_budget_basin_MLR,
                                    co2_atm %>% rename(tref2 = year,
                                                       pCO22 = pCO2))

tcant_budget_basin_MLR <- tcant_budget_basin_MLR %>% 
  mutate(period = paste(tref1, tref2, sep = " - ")) %>% 
  filter(estimate == "dcant") %>% 
  select(-estimate) %>% 
  rename(dcant = value)

dcant_budget_basin_MLR_all_plot <- bind_rows(
  dcant_budget_basin_MLR_all_plot,
  tcant_budget_basin_MLR)

dcant_budget_basin_MLR_all_plot <-
  left_join(dcant_budget_basin_MLR_all_plot,
            dcant_budget_basin_MLR_all_ensemble_summary)

14.3.1 Absolute

g1 <- dcant_budget_basin_MLR_all_plot %>%
  filter(period != "1800 - 1994") %>% 
  ggplot(aes(
    y = dcant,
    x = period,
    alluvium = basin,
    fill  = basin,
    stratum = basin
  )) +
  stat_alluvium(decreasing = FALSE) +
  stat_stratum(decreasing = FALSE) +
  stat_stratum(geom = "text",
               decreasing = FALSE,
               aes(label = paste(
                 round(after_stat(max-min),1)
                 # 100*round(after_stat(prop), 2), "%"
                 ))) +
  ggrepel::geom_label_repel(
    data = dcant_budget_basin_MLR_all_plot %>% filter(period == "2004 - 2014"),
    stat = "stratum",
    size = 4,
    nudge_x = .5,
    point.padding = 3,
    aes(fill = basin, label = basin),
    decreasing = FALSE
  )+
  scale_fill_brewer(palette = "Paired", guide = "none") +
  scale_y_continuous(limits = c(0, 32), expand = c(0, 0)) +
  labs(y = dcant_pgc_label) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        axis.title.x = element_blank()) +
  theme_classic()


newdat <- tibble(layer_data(g1))

change <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  pivot_wider(names_from = x,
              values_from = count) %>% 
  mutate(dcant_change = round(100*(`2` - `1`) / `1`)) %>% 
  select(alluvium, dcant_change)

coord <- newdat %>%
  filter(x == 2) %>% 
  select(x, y, alluvium)

new_layer <- full_join(
  change,
  coord
)

new_layer <- new_layer %>% 
  mutate(dcant_change = as.character(dcant_change),
         dcant_change = if_else(str_detect(dcant_change, "-"),
                                dcant_change,
                                paste0("+", dcant_change)),
         dcant_change = paste(dcant_change, "%"))

g1 +
  geom_text(data = new_layer,
            aes(
              x = x - 0.3,
              y = y,
              label = dcant_change
            ),
            inherit.aes = FALSE)

g2 <- dcant_budget_basin_MLR_all_plot %>%
    filter(period != "1800 - 1994") %>%
  ggplot(aes(
    y = dcant,
    x = period,
    alluvium = basin,
    fill  = basin,
    stratum = basin,
    label = basin
  )) +
  geom_alluvium() +
  geom_stratum() +
  stat_stratum(geom = "text",
               aes(label = paste(
                 round(after_stat(count),1)
                 # 100*round(after_stat(prop), 2), "%"
                 ))) +
  ggrepel::geom_label_repel(
    data = dcant_budget_basin_MLR_all_plot %>% filter(period == "2004 - 2014"),
    stat = "stratum",
    size = 4,
    nudge_x = .5,
    point.padding = 3,
    aes(fill = basin)
  )+
  scale_fill_brewer(palette = "Paired", guide = "none") +
  scale_color_brewer(palette = "Paired", guide = "none") +
  scale_y_continuous(limits = c(0, 33), expand = c(0, 0)) +
  guides(y = "none") +
  labs(title = dcant_pgc_label) +
  theme_classic() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        plot.title = element_text(hjust = 0.5))


newdat <- tibble(layer_data(g2))

change_basin <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  pivot_wider(names_from = x,
              values_from = count) %>% 
  mutate(dcant_change = round(100*(`2` - `1`) / `1`)) %>% 
  select(alluvium, dcant_change)

coord_basin <- newdat %>%
  filter(x == 2) %>% 
  select(x, y, alluvium)

new_layer_basin <- full_join(
  change_basin,
  coord_basin
)

new_layer_basin <- new_layer_basin %>% 
  mutate(dcant_change = as.character(dcant_change),
         dcant_change = if_else(str_detect(dcant_change, "-"),
                                dcant_change,
                                paste0("+", dcant_change)),
         dcant_change = paste(dcant_change, "%"))

new_layer_total <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  group_by(x) %>% 
  summarise(dcant_change = sum(count)) %>% 
  ungroup()

new_layer_total <- new_layer_total %>% 
  mutate(y = dcant_change,
         dcant_change = as.character(round(dcant_change,1)),
         dcant_change = paste("global:",dcant_change))

g2 +
  geom_text(data = new_layer_basin,
            aes(
              x = x - 0.3,
              y = y,
              label = dcant_change
            ),
            inherit.aes = FALSE) +
  geom_label(data = new_layer_total,
            aes(
              x = x,
              y = y + 1,
              label = dcant_change
            ),
            inherit.aes = FALSE)

14.3.2 Scaled

dcant_budget_basin_MLR_all_plot <- left_join(dcant_budget_basin_MLR_all_plot,
                           co2_atm %>% rename(tref1 = year,
                                              pCO21 = pCO2))

dcant_budget_basin_MLR_all_plot <- left_join(dcant_budget_basin_MLR_all_plot,
                           co2_atm %>% rename(tref2 = year,
                                              pCO22 = pCO2))

dcant_budget_basin_MLR_all_plot <- dcant_budget_basin_MLR_all_plot %>% 
  mutate(delta_pCO2 = pCO22 - pCO21,
         beta = dcant / delta_pCO2,
         dcant_sd_scaled = dcant_sd / delta_pCO2) %>% 
  select(-starts_with("pCO2"))


dcant_budget_global_all_ensemble_summary <- left_join(dcant_budget_global_all_ensemble_summary,
                           co2_atm %>% rename(tref1 = year,
                                              pCO21 = pCO2))

dcant_budget_global_all_ensemble_summary <- left_join(dcant_budget_global_all_ensemble_summary,
                           co2_atm %>% rename(tref2 = year,
                                              pCO22 = pCO2))

dcant_budget_global_all_ensemble_summary <- dcant_budget_global_all_ensemble_summary %>% 
  mutate(delta_pCO2 = pCO22 - pCO21,
         beta = dcant_mean / delta_pCO2,
         dcant_sd_scaled = dcant_sd / delta_pCO2) %>% 
  select(-starts_with("pCO2"))

g2 <- dcant_budget_basin_MLR_all_plot %>%
  filter(period != "1800 - 1994") %>% 
  ggplot(aes(
    y = beta,
    x = period,
    alluvium = basin,
    fill  = basin,
    stratum = basin,
    label = basin
  )) +
  geom_alluvium() +
  geom_stratum() +
  stat_stratum(geom = "text",
               aes(label = paste(
                 round(after_stat(count),2)
                 # 100*round(after_stat(prop), 2), "%"
                 ))) +
  ggrepel::geom_label_repel(
    data = dcant_budget_basin_MLR_all_plot %>% filter(period == "2004 - 2014"),
    stat = "stratum",
    size = 4,
    nudge_x = .5,
    point.padding = 3,
    aes(fill = basin)
  )+
  scale_fill_brewer(palette = "Paired", guide = "none") +
  scale_color_brewer(palette = "Paired", guide = "none") +
  # scale_y_continuous(limits = c(0, 33), expand = c(0, 0)) +
  guides(y = "none") +
  labs(title = beta_inv_label) +
  theme_classic() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        plot.title = element_text(hjust = 0.5))


newdat <- tibble(layer_data(g2))

change_basin <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  pivot_wider(names_from = x,
              values_from = count) %>% 
  mutate(dcant_change = round(100*(`2` - `1`) / `1`)) %>% 
  select(alluvium, dcant_change)

coord_basin <- newdat %>%
  filter(x == 2) %>% 
  select(x, y, alluvium)

new_layer_basin <- full_join(
  change_basin,
  coord_basin
)

new_layer_basin <- new_layer_basin %>% 
  mutate(dcant_change = as.character(dcant_change),
         dcant_change = if_else(str_detect(dcant_change, "-"),
                                dcant_change,
                                paste0("+", dcant_change)),
         dcant_change = paste(dcant_change, "%"))

new_layer_total <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  group_by(x) %>% 
  summarise(dcant_change = sum(count)) %>% 
  ungroup()

new_layer_total <- new_layer_total %>% 
  mutate(y = dcant_change,
         dcant_change = as.character(round(dcant_change,2)),
         dcant_change = paste("global:",dcant_change))

g2 +
  geom_text(data = new_layer_basin,
            aes(
              x = x - 0.3,
              y = y,
              label = dcant_change
            ),
            inherit.aes = FALSE) +
  geom_label(data = new_layer_total,
            aes(
              x = x,
              y = y + 0.05,
              label = dcant_change
            ),
            inherit.aes = FALSE)

g2 <- dcant_budget_basin_MLR_all_plot %>%
  ggplot(aes(
    y = beta,
    x = period,
    alluvium = basin,
    fill  = basin,
    stratum = basin,
    label = basin
  )) +
  geom_alluvium() +
  geom_stratum() +
  stat_stratum(geom = "text",
               aes(label = paste(
                 round(after_stat(count),2)
                 # 100*round(after_stat(prop), 2), "%"
                 ))) +
  ggrepel::geom_label_repel(
    data = dcant_budget_basin_MLR_all_plot %>% filter(period == "2004 - 2014"),
    stat = "stratum",
    size = 4,
    nudge_x = .5,
    point.padding = 3,
    aes(fill = basin)
  )+
  scale_fill_brewer(palette = "Paired", guide = "none") +
  scale_color_brewer(palette = "Paired", guide = "none") +
  scale_y_continuous(limits = c(0, 1.7), expand = c(0, 0)) +
  guides(y = "none") +
  labs(title = beta_inv_label) +
  theme_classic() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        plot.title = element_text(hjust = 0.5))


newdat <- tibble(layer_data(g2))

change_basin <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  group_by(alluvium) %>% 
  arrange(x) %>% 
  mutate(dcant_change = round(100*(count - lag(count)) / lag(count))) %>% 
  ungroup()

coord_basin <- newdat %>%
  select(x, y, alluvium)

new_layer_basin <- left_join(
  change_basin %>% drop_na(),
  coord_basin
)

new_layer_basin <- new_layer_basin %>% 
  mutate(dcant_change = as.character(dcant_change),
         dcant_change = if_else(str_detect(dcant_change, "-"),
                                dcant_change,
                                paste0("+", dcant_change)),
         dcant_change = paste(dcant_change, "%"))

new_layer_total <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  group_by(x) %>% 
  summarise(dcant_change = sum(count)) %>% 
  ungroup()

new_layer_total <- new_layer_total %>% 
  mutate(y = dcant_change,
         dcant_change = as.character(round(dcant_change,2)),
         dcant_change = paste("global:",dcant_change))

g2 +
  geom_text(data = new_layer_basin,
            aes(
              x = x - 0.3,
              y = y,
              label = dcant_change
            ),
            inherit.aes = FALSE) +
  geom_label(data = new_layer_total,
            aes(
              x = x,
              y = y + 0.1,
              label = dcant_change
            ),
            inherit.aes = FALSE)

g2 <- dcant_budget_basin_MLR_all_plot %>%
  ggplot(aes(
    y = beta,
    x = period,
    alluvium = basin,
    fill  = basin,
    stratum = basin,
    label = basin
  )) +
  geom_alluvium() +
  geom_stratum() +
  ggrepel::geom_label_repel(
    data = dcant_budget_basin_MLR_all_plot %>% filter(period == "2004 - 2014"),
    stat = "stratum",
    size = 4,
    nudge_x = .5,
    point.padding = 3,
    aes(fill = basin)
  )+
  scale_fill_brewer(palette = "Paired", guide = "none") +
  scale_color_brewer(palette = "Paired", guide = "none") +
  scale_y_continuous(limits = c(0, 1.6), expand = c(0, 0)) +
  guides(y = "none") +
  labs(title = beta_inv_label) +
  theme_classic() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        plot.title = element_text(hjust = 0.5))


newdat <- tibble(layer_data(g2))

# construct budget labels

budget_basin <- 
  newdat %>% 
  select(x, y, alluvium, count)

budget_layer_basin <- budget_basin %>% 
  mutate(dcant = as.character(round(count,2)))

# construct uncertainty labels

uncertainty_basin <- 
  dcant_budget_basin_MLR_all_plot %>% 
  mutate(x = fct_recode(period,
                        "2" = "1994 - 2004",
                        "3" = "2004 - 2014")) %>% 
  select(x, alluvium = basin, dcant_sd_scaled) %>% 
  drop_na() %>% 
  arrange(x, alluvium)
  

coord_basin <- newdat %>%
  select(x, y, alluvium) %>% 
  filter(x != 1)

uncertainty_layer_basin <- bind_cols(
  coord_basin,
  uncertainty_basin %>% select(dcant_sd_scaled),
)

# construct change labels

change_basin <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  group_by(alluvium) %>% 
  arrange(x) %>% 
  mutate(dcant_change_abs = count - lag(count),
         dcant_change = round(100*(dcant_change_abs) / lag(count))) %>% 
  ungroup()

coord_basin <- newdat %>%
  select(x, y, alluvium)

new_layer_basin <- left_join(
  change_basin %>% drop_na(),
  coord_basin
)

new_layer_basin <- new_layer_basin %>% 
  mutate(dcant_change = as.character(dcant_change),
         dcant_change = if_else(str_detect(dcant_change, "-"),
                                dcant_change,
                                paste0("+", dcant_change)),
         dcant_change = paste0(dcant_change, "%"))

# construct global budget labels

new_layer_total <- 
  newdat %>% 
  select(x, alluvium, count) %>% 
  group_by(x) %>% 
  summarise(dcant_change = sum(count)) %>% 
  ungroup()

dcant_sd_scaled <- round(c(
  0.24,
  dcant_budget_global_all_ensemble_summary %>%
    pull(dcant_sd_scaled)
), 2)

new_layer_total <- bind_cols(new_layer_total, dcant_sd_scaled = dcant_sd_scaled)

new_layer_total <- new_layer_total %>% 
  mutate(y = dcant_change,
         dcant_sd_scaled_rel = round(100 * dcant_sd_scaled/dcant_change),
         dcant_change = as.character(round(dcant_change,2)),
         dcant_sd_scaled = round(dcant_sd_scaled,1),
         dcant_change = paste0(dcant_change, "\n(±", dcant_sd_scaled, ")"))

# basin change + uncertainty

budget_layer_basin <- full_join(budget_layer_basin,
                                uncertainty_layer_basin)

budget_layer_basin <- 
  budget_layer_basin %>% 
  mutate(dcant_sd_scaled_rel = 100*dcant_sd_scaled/count,
         dcant_sd_scaled_rel = as.character(round(dcant_sd_scaled_rel)),
         dcant_sd_scaled_numeric = dcant_sd_scaled,
         dcant_sd_scaled = round(dcant_sd_scaled, 2),
         dcant_sd_scaled = paste0("(±", dcant_sd_scaled, ")"),
         dcant_label = if_else(dcant_sd_scaled != "(±NA)",
                               paste0(dcant, "\n", dcant_sd_scaled),
                               dcant))


# uncertainty changes

uncertainty_changes <- budget_layer_basin %>% 
  select(x, alluvium, dcant_sd_scaled = dcant_sd_scaled_numeric) %>% 
  group_by(alluvium) %>% 
  mutate(dcant_change_uncert = sqrt(dcant_sd_scaled^2 + lag(dcant_sd_scaled)^2)) %>% 
  ungroup()


new_layer_basin <- left_join(new_layer_basin, uncertainty_changes) %>% 
  mutate(dcant_change_uncert_rel = 100*dcant_change_uncert/abs((count + lag(count))/2),
         dcant_change_label = if_else(!is.na(dcant_change_uncert),
                                      paste0(dcant_change, "\n(±", round(dcant_change_uncert_rel), "%)"),
                                      dcant_change))


g2 +
  geom_text(data = budget_layer_basin,
            aes(
              x = x,
              y = y,
              label = dcant_label
            ),
            size = 3,
            inherit.aes = FALSE) +
  geom_text(data = new_layer_basin,
            aes(
              x = x - 0.3,
              y = y,
              label = dcant_change_label
            ),
            size = 4,
            inherit.aes = FALSE)
  # geom_label(data = new_layer_total,
  #           aes(
  #             x = x,
  #             y = y + 0.13,
  #             label = dcant_change
  #           ),
  #           size = 4,
  #           inherit.aes = FALSE)



# ggsave(path = here::here("output/publication"),
#        filename = "Fig_budget_beta_basin_hemisphere.png",
#        height = 6,
#        width = 8)

14.4 Depth layer

dcant_budget_change <- dcant_budget_basin_MLR_layer_all %>%
  filter(estimate == "dcant",
         period != "1994 - 2014",
         inv_depth <= 3000,
         data_source == "obs") %>%
  rename(dcant = value) %>%
  select(-c(tref1, tref2, Version_ID)) %>%
  pivot_wider(names_from = period,
              values_from = dcant) %>%
  mutate(sign = if_else(`2004 - 2014` - `1994 - 2004` > 0,
                        "increase",
                        "decrease"))

dcant_budget_layer <- dcant_budget_basin_MLR_layer_all %>%
  filter(estimate == "dcant",
         period != "1994 - 2014",
         inv_depth <= 3400,
         data_source == "obs") %>%
  rename(dcant = value) %>%
  group_by(basin, period, data_source) %>% 
  mutate(dcant = if_else(is.na(lead(dcant)), 888, dcant),
         dcant = na_if(dcant, 888)) %>%
  fill(dcant) %>% 
  ungroup()

dcant_budget_layer %>%
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_rect(
    data = dcant_budget_change,
    aes(
      xmin = inv_depth - 250,
      xmax = inv_depth + 250,
      ymin = `1994 - 2004`,
      ymax = `2004 - 2014`,
      fill = sign
    ),
    alpha = 0.4
  ) +
  geom_step(aes(inv_depth - 250, dcant, col = period), direction = "vh",
            size = 1) +
  coord_flip() +
  scale_x_reverse(breaks = seq(0, 3000, 500)) +
  scale_color_brewer(palette = "Set1", direction = -1, name = "period") +
  scale_fill_brewer(palette = "Set1", direction = -1, name = "") +
  facet_wrap(~ basin, ncol = 3, dir = "v") +
  labs(y = dcant_layer_budget_label,
       x = "Depth (m)") +
      theme(legend.position = c(0.8,0.2))

Past versions of depth_layer_budget_basin_MLR_standard_case-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
26e9496	jens-daniel-mueller	2022-07-12

dcant_budget_basin_MLR_layer_all_uncertainty <-
  dcant_budget_basin_MLR_layer_all_ensemble %>%
  filter(estimate == "dcant",
         period != "1994 - 2014",
         inv_depth <= 3400) %>%
  rename(dcant = value) %>%
  select(-c(method, estimate, Version_ID, tref1, tref2)) %>%
  group_by(basin, period, data_source, MLR_basins, Version_ID_group) %>%
  mutate(dcant = if_else(is.na(lead(dcant)), 888, dcant),
         dcant = na_if(dcant, 888)) %>%
  fill(dcant) %>%
  ungroup()

dcant_budget_basin_MLR_layer_all_uncertainty <- additive_uncertainty(dcant_budget_basin_MLR_layer_all_uncertainty)

delta_dcant_budget_basin_MLR_layer_all_uncertainty <- dcant_budget_basin_MLR_layer_all_uncertainty[[3]]
dcant_budget_basin_MLR_layer_all_uncertainty_contributions <- dcant_budget_basin_MLR_layer_all_uncertainty[[2]]
dcant_budget_basin_MLR_layer_all_uncertainty <- dcant_budget_basin_MLR_layer_all_uncertainty[[1]]

p_dcant_budget_depth_layer <- 
dcant_budget_basin_MLR_layer_all_uncertainty %>%
  filter(data_source == "obs") %>% 
  ggplot() +
  geom_hline(yintercept = 0, size = 0.1) +
  geom_rect(
    aes(
      xmin = inv_depth - 250,
      xmax = inv_depth + 250,
      ymin = std_case - sd*2,
      ymax = std_case + sd*2,
      fill = period
    ),
    alpha = 0.2
  ) +
  geom_rect(
    aes(
      xmin = inv_depth - 250,
      xmax = inv_depth + 250,
      ymin = std_case - sd,
      ymax = std_case + sd,
      fill = period
    ),
    alpha = 0.4
  ) +
  geom_step(aes(inv_depth - 250, std_case, col = period), direction = "vh",
            size = 1) +
  coord_flip() +
  scale_x_reverse(breaks = seq(0, 3000, 500), limits = c(3000, 0), expand = c(0,0)) +
  scale_color_manual(values = c("#EE7733", "#009988"), name = "Mean \u00B1 1σ/2σ") +
  scale_fill_manual(values = c("#EE7733", "#009988"), name = "Mean \u00B1 1σ/2σ") +
  labs(y = dcant_layer_budget_label,
       x = "Depth (m)") +
  facet_wrap(~ basin, ncol = 3, dir = "v", scales = "free_x") +
  theme(legend.position = c(0.9, 0.1))

p_dcant_budget_depth_layer

Past versions of depth_layer_budget_basin_MLR_ensemble-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d803308	jens-daniel-mueller	2022-07-19
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
26e9496	jens-daniel-mueller	2022-07-12

p_dcant_profiles + p_dcant_budget_depth_layer +
  plot_layout(ncol = 1) +
  plot_annotation(tag_levels = 'A')

Past versions of depth_layer_budget_basin_MLR_ensemble-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d803308	jens-daniel-mueller	2022-07-19

ggsave(path = here::here("output/publication"),
       filename = "Fig_dcant_profiles_and_depth_layer_inventories.png",
       height = 12,
       width = 8,
       dpi = 600)

p_dcant_budget_depth_layer_mod <- 
dcant_budget_basin_MLR_layer_all_uncertainty %>%
  filter(data_source != "obs") %>% 
  mutate(data_source = case_when(
    data_source == "mod" ~ "Reconstruction",
    data_source == "mod_truth" ~ "Model truth"
  )) %>%  
  ggplot() +
  geom_hline(yintercept = 0, size = 0.1) +
  geom_rect(
    aes(
      xmin = inv_depth - 250,
      xmax = inv_depth + 250,
      ymin = std_case - sd,
      ymax = std_case + sd,
      fill = data_source
    ),
    alpha = 0.2
  ) +
  geom_rect(
    aes(
      xmin = inv_depth - 250,
      xmax = inv_depth + 250,
      ymin = std_case - sd*2,
      ymax = std_case + sd*2,
      fill = data_source
    ),
    alpha = 0.4
  ) +
  geom_step(aes(inv_depth - 250, std_case, col = data_source), direction = "vh",
            size = 1) +
  coord_flip() +
  scale_x_reverse(breaks = seq(0, 3000, 500), limits = c(3000, 0), expand = c(0,0)) +
  scale_color_manual(values = c("#009988", "#EE7733"), name = "Mean \u00B1 2σ") +
  scale_fill_manual(values = c("#009988", "#EE7733"), name = "Mean \u00B1 2σ") +
  labs(y = dcant_layer_budget_label,
       x = "Depth (m)") +
  facet_grid(period ~ basin, scales = "free_x") +
  theme(legend.position = "none")



p_dcant_budget_depth_layer_mod

Past versions of depth_layer_budget_basin_MLR_ensemble_mod-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
318fefe	jens-daniel-mueller	2022-08-11

p_dcant_profiles_mod + p_dcant_budget_depth_layer_mod +
  plot_layout(ncol = 1) +
  plot_annotation(tag_levels = 'A')

Past versions of depth_layer_budget_basin_MLR_ensemble_mod-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
318fefe	jens-daniel-mueller	2022-08-11

ggsave(path = here::here("output/publication"),
       filename = "FigS_dcant_profiles_and_depth_layer_inventories_synthetic_data.png",
       height = 10,
       width = 7,
       dpi = 600)

dcant_budget_change <- dcant_budget_basin_MLR_layer_all %>%
  filter(estimate == "dcant",
         period != "1994 - 2014",
         inv_depth <= 3000,
         data_source == "obs") %>%
  rename(dcant = value) %>%
  arrange(inv_depth) %>% 
  group_by(data_source, basin, period) %>% 
  mutate(dcant_cum = cumsum(dcant)) %>% 
  ungroup() %>%
  select(-c(tref2, Version_ID, dcant)) %>%
  pivot_wider(names_from = period,
              values_from = dcant_cum) %>%
  mutate(sign = if_else(`2004 - 2014` - `1994 - 2004` > 0,
                        "increase",
                        "decrease"))

dcant_budget_layer <- dcant_budget_basin_MLR_layer_all %>%
  filter(estimate == "dcant",
         period != "1994 - 2014",
         inv_depth <= 3400,
         data_source == "obs") %>%
  rename(dcant = value) %>%
  arrange(inv_depth) %>% 
  group_by(data_source, basin, period) %>% 
  mutate(dcant_cum = cumsum(dcant)) %>% 
  ungroup() %>%
  group_by(basin, period, data_source) %>% 
  mutate(dcant_cum = if_else(is.na(lead(dcant_cum)), 888, dcant_cum),
         dcant_cum = na_if(dcant_cum, 888)) %>%
  fill(dcant_cum) %>% 
  ungroup()



dcant_budget_layer %>%
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_rect(
    data = dcant_budget_change,
    aes(
      xmin = inv_depth - 250,
      xmax = inv_depth + 250,
      ymin = `1994 - 2004`,
      ymax = `2004 - 2014`,
      fill = sign
    ),
    alpha = 0.3
  ) +
  geom_step(aes(inv_depth - 250, dcant_cum, col = period), direction = "vh",
            size = 1) +
  coord_flip() +
  scale_x_reverse(breaks = seq(0, 3000, 500)) +
  scale_color_brewer(palette = "Set1",
                     direction = -1,
                     name = "Decade") +
  scale_fill_brewer(palette = "Set1",
                    direction = -1,
                    name = "Decadal change") +
  
  facet_wrap( ~ basin, ncol = 3, dir = "v") +
  labs(y = dcant_layer_budget_label,
       x = "Depth (m)",
       title = "Cumulative layer inventory change") +
      theme(legend.position = c(0.8,0.2))

Past versions of depth_layer_budget_basin_MLR_cumulative-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
5520de8	jens-daniel-mueller	2022-11-15
ebc85e2	jens-daniel-mueller	2022-11-14
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
26e9496	jens-daniel-mueller	2022-07-12
37f56b3	jens-daniel-mueller	2022-07-01
df21d31	jens-daniel-mueller	2022-07-01
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
d2191ad	jens-daniel-mueller	2022-02-04
32e9682	jens-daniel-mueller	2022-02-02
913e42f	jens-daniel-mueller	2022-02-01
189de95	jens-daniel-mueller	2022-02-01
ab001eb	jens-daniel-mueller	2022-01-31
b62308d	jens-daniel-mueller	2022-01-31
5f2aed0	jens-daniel-mueller	2022-01-27

15 Penetration depth

mean_column_inventory <- beta_budget_basin_MLR_all_ensemble %>% 
  mutate(dcant_area = dcant * 1e15 / 12 / area)

mean_column_inventory %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = period,
             y = dcant_area,
             fill = MLR_basins)) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_muted() +
  facet_wrap(~basin)

Past versions of pentration_depth_of_surface_dcant-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25

mean_surface_dcant %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = period,
             y = mean_surface_dcant,
             fill = MLR_basins)) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_muted() +
  facet_wrap(~basin)

Past versions of pentration_depth_of_surface_dcant-2.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25

dcant_mean_penetration_depth <-
full_join(mean_column_inventory,
          mean_surface_dcant) %>%
  filter(Version_ID_group != "Surface eMLR(C*)") %>%
  mutate(mean_surface_dcant = mean_surface_dcant * 1e3 * 1.025 / 1e6,
         penetration_depth = dcant_area/mean_surface_dcant)

dcant_mean_penetration_depth %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = period,
             y = penetration_depth,
             fill = MLR_basins)) +
  geom_jitter(shape = 21, alpha = 0.8) +
  scale_fill_muted() +
  facet_wrap(~basin)

Past versions of pentration_depth_of_surface_dcant-3.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25

dcant_mean_penetration_depth_uncertainty <- 
additive_uncertainty(
  dcant_mean_penetration_depth %>% 
    filter(period %in% two_decades) %>% 
    select(data_source, period, Version_ID_group, MLR_basins, basin, penetration_depth) %>% 
    rename(dcant = penetration_depth))

delta_dcant_mean_penetration_depth_uncertainty <-
  dcant_mean_penetration_depth_uncertainty[[3]]
dcant_mean_penetration_depth_uncertainty <-
  dcant_mean_penetration_depth_uncertainty[[1]]

dcant_mean_penetration_depth_uncertainty %>% 
  filter(data_source == "obs") %>% 
  select(-c(data_source, starts_with("signif_"))) %>% 
  arrange(basin, period) %>% 
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

period	basin	sd	std_case
1994 - 2004	Global	195.1579	635.7219
2004 - 2014	Global	128.5704	564.4056
1994 - 2004	Indian	220.8476	834.0676
2004 - 2014	Indian	101.2224	559.8123
1994 - 2004	N. Pacific	100.3504	279.0525
2004 - 2014	N. Pacific	174.3395	300.0561
1994 - 2004	S. Pacific	227.9048	708.5043
2004 - 2014	S. Pacific	145.4462	596.1376
1994 - 2004	N. Atlantic	273.0824	829.9681
2004 - 2014	N. Atlantic	169.2454	668.0443
1994 - 2004	S. Atlantic	208.4508	633.5340
2004 - 2014	S. Atlantic	230.6747	870.9505

delta_dcant_mean_penetration_depth_uncertainty %>% 
  filter(data_source == "obs") %>% 
  select(-c(data_source, sd, std_case)) %>% 
  arrange(basin, period) %>% 
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

period	basin	delta_dcant	RSS	signif_single	signif_double
2004 - 2014	Global	-71.31634	233.7027	0	0
2004 - 2014	Indian	-274.25531	242.9396	1	0
2004 - 2014	N. Pacific	21.00356	201.1578	0	0
2004 - 2014	S. Pacific	-112.36662	270.3612	0	0
2004 - 2014	N. Atlantic	-161.92382	321.2756	0	0
2004 - 2014	S. Atlantic	237.41652	310.9060	0	0

dcant_mean_penetration_depth_uncertainty %>% 
  filter(period %in% two_decades,
         data_source == "obs") %>% 
  ggplot(aes(x = basin,
             y = std_case,
             ymin = std_case - sd,
             ymax = std_case + sd,
             col = period)) +
  scale_y_reverse(limits = c(1000,0)) +
  geom_pointrange(position = position_dodge(width = 0.3))

Past versions of pentration_depth_of_surface_dcant-4.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
445761e	jens-daniel-mueller	2022-12-02
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25

rm(mean_surface_dcant,
   dcant_mean_penetration_depth,
   dcant_mean_penetration_depth_uncertainty)

16 Time series

16.1 eMLR only

dcant_budget_global_all_ensemble <-
  dcant_budget_basin_MLR_all_ensemble %>% 
  filter(basin == "Global")

dcant_budget_global_all_ensemble %>% 
  filter(data_source == "obs",
         period != "1994 - 2014") %>% 
  ggplot(aes(tref2, dcant, group = interaction(MLR_basins, Version_ID_group),
             col=Version_ID_group)) +
  geom_path() +
  geom_point()

Past versions of dcant_budget_time_series_member_connections-1.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
d20faeb	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
232909e	jens-daniel-mueller	2022-07-01

dcant_budget_global_all_ensemble %>% 
  filter(data_source == "obs") %>% 
  select(MLR_basins, Version_ID_group, period, dcant) %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>% 
  ggplot(aes(`1994 - 2004`, `2004 - 2014`,
             col=MLR_basins)) +
  coord_equal() +
  geom_point() +
  facet_wrap(~ Version_ID_group)

Past versions of dcant_budget_time_series_member_connections-2.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
d20faeb	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
232909e	jens-daniel-mueller	2022-07-01

dcant_budget_global_all_ensemble %>% 
  filter(data_source == "obs") %>% 
  select(MLR_basins, Version_ID_group, period, dcant) %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>% 
  ggplot(aes(`1994 - 2004`, `2004 - 2014`,
             col=Version_ID_group)) +
  coord_equal() +
  geom_point() +
  facet_wrap(~ MLR_basins)

Past versions of dcant_budget_time_series_member_connections-3.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
d20faeb	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
232909e	jens-daniel-mueller	2022-07-01

dcant_budget_global_all_ensemble %>% 
  filter(data_source == "obs") %>% 
  select(MLR_basins, Version_ID_group, period, dcant) %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>% 
  ggplot(aes(`1994 - 2004`, `1994 - 2014`,
             col=Version_ID_group)) +
  coord_equal() +
  geom_point()

Past versions of dcant_budget_time_series_member_connections-4.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
d20faeb	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
232909e	jens-daniel-mueller	2022-07-01

dcant_budget_global_all_ensemble %>% 
  filter(data_source == "obs") %>% 
  select(MLR_basins, Version_ID_group, period, dcant) %>% 
  pivot_wider(names_from = period,
              values_from = dcant) %>% 
  ggplot(aes(`1994 - 2004` + `2004 - 2014`, `1994 - 2014`,
             col=Version_ID_group)) +
  geom_abline(intercept = 0, slope = 1) +
  coord_equal() +
  scale_x_continuous(breaks = seq(0,100,1)) +
  scale_y_continuous(breaks = seq(0,100,1)) +
  geom_point()

Past versions of dcant_budget_time_series_member_connections-5.png

Version	Author	Date
8ded995	jens-daniel-mueller	2023-01-05
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
534ee74	jens-daniel-mueller	2022-11-25
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
fea41c1	jens-daniel-mueller	2022-07-20
b1f7ab3	jens-daniel-mueller	2022-07-18

dcant_budget_global_ts <- dcant_budget_basin_MLR_all %>%
  filter(basin == "Global",
         data_source == "obs",
         period  %in% two_decades) %>% 
  select(year = tref2, dcant_mean = dcant)


dcant_budget_global_all_ensemble_summary <-
  dcant_budget_basin_MLR_all_uncertainty %>%
  filter(data_source == "obs",
         basin == "Global") %>%
  mutate(
    tref2 = as.numeric(str_sub(period, 8, 11)),
    dcant_sd = sqrt(sd ^ 2 + lag(sd, default = 0) ^ 2),
    dcant_sd_int = sd
  ) %>%
  select(data_source, tref2, dcant_sd, dcant_sd_int)

# dcant_budget_global_all_ensemble_summary <-
#   dcant_budget_basin_MLR_all_uncertainty_two_decades %>%
#   filter(period %in% c("1994 - 2004", "1994 - 2014"),
#          data_source == "obs",
#          basin == "Global") %>%
#   select(data_source, tref2, period, sd) %>%
#   mutate(dcant_sd = sd*sd_factor)

# dcant_budget_global_all_ensemble_summary <-
#   dcant_budget_basin_MLR_all_ensemble %>%
#   filter(period %in% c("1994 - 2004", "1994 - 2014"),
#          data_source == "obs",
#          basin == "Global") %>%
#   group_by(data_source, tref2, period) %>%
#   summarise(#dcant_mean = mean(dcant),
#             dcant_sd = sd(dcant)*sd_factor) %>%
#   ungroup()

dcant_budget_global_ts <- left_join(dcant_budget_global_ts,
                                    dcant_budget_global_all_ensemble_summary %>% 
                                      select(year = tref2, dcant_sd, dcant_sd_int))


tcant_S04 <- bind_cols(year = 1994, dcant_mean = 118, dcant_sd = 19)

tcant_ts <- full_join(dcant_budget_global_ts, tcant_S04)

tcant_ts <- left_join(tcant_ts, co2_atm)

co2_atm_pi <- bind_cols(pCO2 = 280, dcant_mean = 0, year = 1800, dcant_sd = NA)

tcant_ts <- full_join(tcant_ts, co2_atm_pi)


tcant_ts <- tcant_ts %>%
  arrange(year) %>%
  mutate(
    tcant = cumsum(dcant_mean),
    tcant_sd = sqrt(dcant_sd ^ 2 + nth(dcant_sd,2) ^ 2)
  ) %>%
  mutate(
    tcant_sd = case_when(
      year == 1800 ~ 0,
      year == 1994 ~ dcant_sd,
      TRUE ~ tcant_sd
    ),
    dcant_sd_int = case_when(
      year == 1994 ~ dcant_sd,
      TRUE ~ dcant_sd_int
    ),
    dcant_sd = case_when(
      year == 1994 ~ 0,
      TRUE ~ dcant_sd
    )
  )

beta <- 1.57
co2_atm_pi_value <- 283
atm_co2_col <- "#CC3311"
tcant_col <- "#004488"

p_tcant_year <- tcant_ts %>%
  ggplot(aes(year, tcant)) +
  geom_line(data = co2_atm_reccap2 %>% filter(year > 1790),
            aes(year, (pCO2 - co2_atm_pi_value) * beta),
            col = atm_co2_col) +
  geom_linerange(data = tcant_ts %>% filter(year != 1994),
                 aes(
    ymin = tcant - dcant_sd_int*2,
    ymax = tcant + dcant_sd_int*2,
    col = "Central estimate\n \u00B1 1σ/2σ uncertainty"
  ), linewidth = 0.5) +
  geom_linerange(aes(
    ymin = tcant - dcant_sd_int,
    ymax = tcant + dcant_sd_int,
    col = "Central estimate\n \u00B1 1σ/2σ uncertainty"
  ), linewidth = 1) +
  geom_point(fill = "white", shape = 21,
             aes(col = "Central estimate\n \u00B1 1σ/2σ uncertainty")) +
  scale_y_continuous(name = expression(Total ~ oceanic ~ C[ant] ~ (Pg~C)),
                     sec.axis = sec_axis( ~ (. / beta + co2_atm_pi_value),
                                          name = expression(Atm. ~ pCO[2] ~ (µatm)))) +
  scale_x_continuous(name = "Year", breaks = seq(1800, 2000, 50)) +
  scale_color_manual(values = tcant_col) +
  theme(
    axis.title.y.right = element_text(color = atm_co2_col),
    axis.text.y.right = element_text(color = atm_co2_col),
    axis.ticks.y.right = element_line(color = atm_co2_col), 
    axis.title.y.left = element_text(color = tcant_col),
    axis.text.y.left = element_text(color = tcant_col),
    axis.ticks.y.left = element_line(color = tcant_col), 
    legend.background = element_rect(fill = "transparent"),
    legend.position = c(0.85, 0.2),
    legend.title = element_blank()
  )

p_tcant_year

Past versions of mean_tcant_over_time-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
9bbc6a4	jens-daniel-mueller	2022-07-20
fea41c1	jens-daniel-mueller	2022-07-20
d803308	jens-daniel-mueller	2022-07-19
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
8ab4a87	jens-daniel-mueller	2022-06-29
7629c78	jens-daniel-mueller	2022-06-29
f6786c8	jens-daniel-mueller	2022-06-29
0825298	jens-daniel-mueller	2022-06-28
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
8dca96a	jens-daniel-mueller	2022-04-12
209c9b6	jens-daniel-mueller	2022-04-10
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
5a6be34	jens-daniel-mueller	2022-03-22
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
251c7cf	jens-daniel-mueller	2022-02-17
565224d	jens-daniel-mueller	2022-02-17
2116dd3	jens-daniel-mueller	2022-02-09
6fe70a1	jens-daniel-mueller	2022-02-05
a6b33aa	jens-daniel-mueller	2022-02-04

atm_co2_col <- "transparent"

p_tcant_year_trans <- tcant_ts %>%
  ggplot(aes(year, tcant)) +
  geom_line(data = co2_atm_reccap2 %>% filter(year > 1790),
            aes(year, (pCO2 - co2_atm_pi_value) * beta),
            col = atm_co2_col) +
  geom_linerange(data = tcant_ts %>% filter(year != 1994),
                 aes(
    ymin = tcant - dcant_sd_int*2,
    ymax = tcant + dcant_sd_int*2,
    col = "Central estimate\n \u00B1 1σ/2σ uncertainty"
  ), linewidth = 0.5) +
  geom_linerange(aes(
    ymin = tcant - dcant_sd_int,
    ymax = tcant + dcant_sd_int,
    col = "Central estimate\n \u00B1 1σ/2σ uncertainty"
  ), linewidth = 1) +
  geom_point(fill = "white", shape = 21,
             aes(col = "Central estimate\n \u00B1 1σ/2σ uncertainty")) +
  scale_y_continuous(name = expression(Total ~ oceanic ~ C[ant] ~ (Pg~C)),
                     sec.axis = sec_axis( ~ (. / beta + co2_atm_pi_value),
                                          name = expression(Atm. ~ pCO[2] ~ (µatm)))) +
  scale_x_continuous(name = "Year", breaks = seq(1800, 2000, 50)) +
  scale_color_manual(values = tcant_col) +
  theme(
    axis.title.y.right = element_text(color = atm_co2_col),
    axis.text.y.right = element_text(color = atm_co2_col),
    axis.ticks.y.right = element_line(color = atm_co2_col),
    axis.title.y.left = element_text(color = tcant_col),
    axis.text.y.left = element_text(color = tcant_col),
    axis.ticks.y.left = element_line(color = tcant_col),
    legend.background = element_rect(fill = "transparent"),
    legend.position = c(0.85, 0.2),
    legend.title = element_blank()
  )

p_tcant_year_trans

Past versions of mean_tcant_over_time-2.png

Version	Author	Date
fdf5f22	jens-daniel-mueller	2023-01-11
2cca4eb	jens-daniel-mueller	2023-01-10
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
d803308	jens-daniel-mueller	2022-07-19
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
8ab4a87	jens-daniel-mueller	2022-06-29
7629c78	jens-daniel-mueller	2022-06-29
f6786c8	jens-daniel-mueller	2022-06-29
0825298	jens-daniel-mueller	2022-06-28
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
8dca96a	jens-daniel-mueller	2022-04-12
209c9b6	jens-daniel-mueller	2022-04-10
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
5a6be34	jens-daniel-mueller	2022-03-22
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
251c7cf	jens-daniel-mueller	2022-02-17
565224d	jens-daniel-mueller	2022-02-17
2116dd3	jens-daniel-mueller	2022-02-09

atm_co2_col <- "#CC3311"

p_tcant_atm_co2 <- tcant_ts %>%
  ggplot(aes(pCO2, tcant)) +
  geom_ribbon(aes(
    ymin = tcant - tcant_sd,
    ymax = tcant + tcant_sd,
    fill = "1800 - 2014",
    alpha = "1800 - 2014"
  )) +
  geom_ribbon(aes(
    ymin = tcant - dcant_sd,
    ymax = tcant + dcant_sd,
    fill = "1994 - 2014",
    alpha = "1994 - 2014"
  )) +
  geom_line(color = tcant_col) +
  geom_point(shape = 21, fill = "white", color = tcant_col) +
  scale_x_continuous(breaks = seq(280, 400, 30)) +
  scale_fill_manual(values = c(tcant_col, tcant_col), name = "Cumulative\n1σ-uncertainty") +
  scale_alpha_manual(values = c(0.2, 0.6), name = "Cumulative\n1σ-uncertainty") +
  geom_text(aes(label = year), nudge_x = -6, nudge_y = 6) +
  labs(x = expression(Atmospheric ~ pCO[2] ~ (µatm)),
       y = expression(Total ~ oceanic ~ C[ant] ~ (Pg~C))) +
  theme(
    axis.title.y.left = element_text(color = tcant_col),
    axis.text.y.left = element_text(color = tcant_col),
    axis.ticks.y.left = element_line(color = tcant_col),
    legend.background = element_rect(fill = "transparent"),
    legend.position = c(0.85, 0.25),
    axis.title.x = element_text(color = atm_co2_col),
    axis.text.x = element_text(color = atm_co2_col),
    axis.ticks.x = element_line(color = atm_co2_col)
  )

p_tcant_year / p_tcant_atm_co2 + 
  plot_annotation(tag_levels = 'A')

Past versions of mean_tcant_over_atm_co2-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
9bbc6a4	jens-daniel-mueller	2022-07-20
a80b59b	jens-daniel-mueller	2022-07-20
fea41c1	jens-daniel-mueller	2022-07-20
d803308	jens-daniel-mueller	2022-07-19
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
8ab4a87	jens-daniel-mueller	2022-06-29
7629c78	jens-daniel-mueller	2022-06-29
f6786c8	jens-daniel-mueller	2022-06-29
0825298	jens-daniel-mueller	2022-06-28
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
8dca96a	jens-daniel-mueller	2022-04-12
209c9b6	jens-daniel-mueller	2022-04-10
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
5a6be34	jens-daniel-mueller	2022-03-22
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
251c7cf	jens-daniel-mueller	2022-02-17
565224d	jens-daniel-mueller	2022-02-17
2116dd3	jens-daniel-mueller	2022-02-09
6fe70a1	jens-daniel-mueller	2022-02-05
a6b33aa	jens-daniel-mueller	2022-02-04
d2191ad	jens-daniel-mueller	2022-02-04
0d0f790	jens-daniel-mueller	2022-02-02
60727e6	jens-daniel-mueller	2022-02-02
c7b4984	jens-daniel-mueller	2022-02-02
7fb28a2	jens-daniel-mueller	2022-02-02
7655085	jens-daniel-mueller	2022-02-02
ec7fe7e	jens-daniel-mueller	2022-01-31

ggsave(path = here::here("output/publication"),
       filename = "Fig_dcant_inventory_global_vs_atm_pCO2.png",
       height = 6,
       width = 7,
       dpi = 600)


p_tcant_year / p_tcant_atm_co2

Past versions of mean_tcant_over_atm_co2-2.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
0f2c9d5	jens-daniel-mueller	2022-12-08
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
8ab4a87	jens-daniel-mueller	2022-06-29
7629c78	jens-daniel-mueller	2022-06-29
f6786c8	jens-daniel-mueller	2022-06-29
0825298	jens-daniel-mueller	2022-06-28
87e9eb8	jens-daniel-mueller	2022-06-27
b52b159	jens-daniel-mueller	2022-06-27
1d73ec9	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
8dca96a	jens-daniel-mueller	2022-04-12
209c9b6	jens-daniel-mueller	2022-04-10
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
5a6be34	jens-daniel-mueller	2022-03-22
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
251c7cf	jens-daniel-mueller	2022-02-17
565224d	jens-daniel-mueller	2022-02-17
2116dd3	jens-daniel-mueller	2022-02-09

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_budget_global_vs_atm_pCO2.png",
       height = 6,
       width = 7)

p_tcant_year_trans / p_tcant_atm_co2

Past versions of mean_tcant_over_atm_co2-3.png

Version	Author	Date
fdf5f22	jens-daniel-mueller	2023-01-11
2cca4eb	jens-daniel-mueller	2023-01-10
3a7bab0	jens-daniel-mueller	2022-11-28
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_budget_global_vs_atm_pCO2_trans.png",
       height = 6,
       width = 7)

tcant_ts %>% 
  mutate(beta = dcant_mean / (pCO2-lag(pCO2)))

# A tibble: 4 × 8
   year dcant_mean dcant_sd dcant_sd_int  pCO2 tcant tcant_sd  beta
  <dbl>      <dbl>    <dbl>        <dbl> <dbl> <dbl>    <dbl> <dbl>
1  1800        0      NA           NA     280     0       0   NA   
2  1994      118       0           19     358.  118      19    1.51
3  2004       29.3     2.53         2.53  377.  147.     19.2  1.57
4  2014       27.3     3.59         2.55  397.  175.     19.3  1.34

16.2 incl fluxes

Ocean_Sink <- Ocean_Sink %>%
  filter(product != "Watson")

Ocean_Sink %>%
  filter(year >= 2010, year <= 2019) %>%
  group_by(type) %>%
  summarise(GtC_yr_mean = mean(GtC_yr)) %>%
  ungroup()

# A tibble: 2 × 2
  type          GtC_yr_mean
  <chr>               <dbl>
1 data_products        3.02
2 models               2.46

Ocean_Sink_cum_1994 <-
  Ocean_Sink %>%
  filter(year >= 1994, year <= 2007) %>%
  mutate(fgco2_glob = if_else(year == 1994, 0, GtC_yr),
         fgco2_glob = fgco2_glob) %>% 
  arrange(year) %>%
  group_by(product, type) %>%
  mutate(fgco2_glob_cum = cumsum(fgco2_glob)) %>%
  ungroup()

Ocean_Sink_cum_1994 <-
  left_join(Ocean_Sink_cum_1994,
            co2_atm) 


Ocean_Sink_cum_1994 %>%
  group_split(type) %>%
  map(
    ~ ggplot(data = .x, aes(pCO2, fgco2_glob_cum, col = product)) +
      geom_line() +
      geom_point(shape = 21, fill = "white") +
      theme(legend.position = "bottom") +
      facet_grid(. ~ type)
  )

[[1]]

Past versions of GCB_carbon_sink_cum_1994_2007_reference-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
91052ae	jens-daniel-mueller	2022-10-20
8105380	jens-daniel-mueller	2022-10-07
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26

[[2]]

Past versions of GCB_carbon_sink_cum_1994_2007_reference-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
8105380	jens-daniel-mueller	2022-10-07

Ocean_Sink_cum_1994 <- Ocean_Sink_cum_1994 %>% 
  mutate(fgco2_glob_cum_total = fgco2_glob_cum + 118) %>%
  select(year, pCO2, product, fgco2_glob_cum, fgco2_glob_cum_total, type)

Ocean_Sink_cum_1994 <- Ocean_Sink_cum_1994 %>% 
  mutate(type = case_when(
    type == "data_products" ~ "Observation-based\nsurface flux\nproducts",
    TRUE ~ "Global Ocean\nBiogeochemical Models\n(GOBM)"
  ))

Ocean_Sink_cum_1994 %>% 
  filter(year == 2007) %>% 
  group_by(type) %>% 
  summarise(dcant_sd = sd(fgco2_glob_cum)*sd_factor) %>% 
  ungroup()

# A tibble: 2 × 2
  type                                          dcant_sd
  <chr>                                            <dbl>
1 "Global Ocean\nBiogeochemical Models\n(GOBM)"     6.99
2 "Observation-based\nsurface flux\nproducts"       6.66

Ocean_Sink_cum_1994 <-
  Ocean_Sink %>%
  filter(year >= 1994, year <= 2014) %>%
  mutate(fgco2_glob = if_else(year == 1994, 0, GtC_yr),
         fgco2_glob = fgco2_glob) %>% 
  arrange(year) %>%
  group_by(product, type) %>%
  mutate(fgco2_glob_cum = cumsum(fgco2_glob)) %>%
  ungroup()

Ocean_Sink_cum_1994 <-
  left_join(Ocean_Sink_cum_1994,
            co2_atm) 

Ocean_Sink_cum_1994 %>%
  group_split(type) %>%
  map(
    ~ ggplot(data = .x, aes(pCO2, fgco2_glob_cum, col = product)) +
      geom_line() +
      geom_point(shape = 21, fill = "white") +
      theme(legend.position = "bottom") +
      facet_grid(. ~ type)
  )

[[1]]

Past versions of GCB_carbon_sink_cum_1994-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
91052ae	jens-daniel-mueller	2022-10-20
8105380	jens-daniel-mueller	2022-10-07
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26
bd24a0f	jens-daniel-mueller	2022-07-15

[[2]]

Past versions of GCB_carbon_sink_cum_1994-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
8105380	jens-daniel-mueller	2022-10-07

Ocean_Sink_cum_1994 <- Ocean_Sink_cum_1994 %>% 
  mutate(fgco2_glob_cum_total = fgco2_glob_cum + 118) %>%
  select(year, pCO2, product, fgco2_glob_cum, fgco2_glob_cum_total, type)

Ocean_Sink_cum_1994 <- Ocean_Sink_cum_1994 %>% 
  mutate(type = case_when(
    type == "data_products" ~ "Observation-based\nsurface flux\nproducts",
    TRUE ~ "Global Ocean\nBiogeochemical Models\n(GOBM)"
  ))

tcant_2004 <- tcant_ts %>% 
  filter(year == 2004) %>% 
  pull(tcant)

Ocean_Sink_cum_2004 <-
  Ocean_Sink %>%
  filter(year >= 2004, year <= 2014) %>%
  mutate(fgco2_glob = if_else(year == 2004, 0, GtC_yr)) %>% 
  arrange(year) %>%
  group_by(product, type) %>%
  mutate(fgco2_glob_cum = cumsum(fgco2_glob)) %>%
  ungroup()

Ocean_Sink_cum_2004 <-
  left_join(Ocean_Sink_cum_2004,
            co2_atm) 

Ocean_Sink_cum_2004 %>%
  group_split(type) %>%
  map(
    ~ ggplot(data = .x, aes(pCO2, fgco2_glob_cum, col = product)) +
      geom_line() +
      geom_point(shape = 21, fill = "white") +
      theme(legend.position = "bottom") +
      facet_grid(. ~ type)
  )

[[1]]

Past versions of GCB_carbon_sink_cum_2004-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
91052ae	jens-daniel-mueller	2022-10-20
8105380	jens-daniel-mueller	2022-10-07
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26
bd24a0f	jens-daniel-mueller	2022-07-15

[[2]]

Past versions of GCB_carbon_sink_cum_2004-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
8105380	jens-daniel-mueller	2022-10-07

Ocean_Sink_cum_2004 <- Ocean_Sink_cum_2004 %>% 
  mutate(fgco2_glob_cum_total = fgco2_glob_cum + tcant_2004) %>% 
  select(year, pCO2, product, fgco2_glob_cum, fgco2_glob_cum_total, type)

Ocean_Sink_cum_2004 <- Ocean_Sink_cum_2004 %>% 
  mutate(type = case_when(
    type == "data_products" ~ "Observation-based\nsurface flux\nproducts",
    TRUE ~ "Global Ocean\nBiogeochemical Models\n(GOBM)"
  ))

tcant_ts_zoom <- tcant_ts %>%
  filter(year != 1800)

tcant_ts_zoom <- 
  tcant_ts_zoom %>%
  mutate(tcant = tcant - 118)

tcant_min = tcant_ts_zoom %>% pull(tcant) %>% min()
tcant_max =
  tcant_ts_zoom %>% pull(tcant) %>% max() +
  tcant_ts_zoom %>% pull(dcant_sd) %>% max()*2

Ocean_Sink_cum_1994 <- 
  Ocean_Sink_cum_1994 %>% 
  mutate(fgco2_glob_cum_total = fgco2_glob_cum_total - 118)

Ocean_Sink_cum_2004 <- 
  Ocean_Sink_cum_2004 %>% 
  mutate(fgco2_glob_cum_total = fgco2_glob_cum_total - 118)


p_tcant_ts_zoom <-
  tcant_ts_zoom %>%
  ggplot(aes(pCO2, tcant)) +
  geom_ribbon(
    aes(ymin = tcant - dcant_sd*2,
        ymax = tcant + dcant_sd*2),
    alpha = 0.2,
    fill = tcant_col
  ) +
  geom_ribbon(
    aes(
      ymin = tcant - dcant_sd,
      ymax = tcant + dcant_sd
    ),
    alpha = 0.3,
    fill = tcant_col
  ) +
  geom_line(data = Ocean_Sink_cum_1994,
            aes(pCO2, fgco2_glob_cum_total, group = product, col = type)) +
  geom_point(
    data = Ocean_Sink_cum_1994,
    aes(pCO2, fgco2_glob_cum_total, group = product, col = type),
    size = 0.3
  ) +
  geom_point(
    data = Ocean_Sink_cum_1994 %>% filter(year %in% c(2004, 2014)),
    aes(
      pCO2,
      fgco2_glob_cum_total,
      col = type
    ),
    fill = "white",
    shape = 21
  ) +
  geom_line(aes(col = "eMLR(C*)")) +
  geom_point(aes(col = "eMLR(C*)"),
             shape = 21,
             fill = "white") +
  geom_text(aes(label = year), nudge_x = -2, nudge_y = 3) +
  scale_color_highcontrast(name = "Data source") +
  scale_fill_highcontrast(name = "Data source") +
  scale_y_continuous(limits = c(tcant_min, tcant_max)) +
  labs(x = expression(Atmospheric ~ pCO[2] ~ (µatm)),
       y = expression(Cumulative ~ ocean ~ carbon ~ storage ~ since ~ 1994 ~ (Pg~C))) +
  guides(colour = guide_legend(order = 1),
         fill = "none") +
  theme_classic() +
  theme(
    legend.background = element_rect(fill = "transparent"),
    legend.position = c(0.25, 0.83),
    legend.title = element_blank(),
    legend.box.just = "top",
    legend.box = "horizontal",
    legend.key.size = unit(2.5, 'lines'),
    panel.grid.major.y = element_line(colour = "black",
                                      size = 0.1)
  )

p_tcant_ts_zoom

Past versions of mean_tcant_over_atm_co2_incl_fluxes-1.png

Version	Author	Date
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
8e1702d	jens-daniel-mueller	2022-11-16
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20
8105380	jens-daniel-mueller	2022-10-07
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26
4af29f9	jens-daniel-mueller	2022-07-24
b18b250	jens-daniel-mueller	2022-07-20
a80b59b	jens-daniel-mueller	2022-07-20
fea41c1	jens-daniel-mueller	2022-07-20
d803308	jens-daniel-mueller	2022-07-19
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
8ab4a87	jens-daniel-mueller	2022-06-29
7629c78	jens-daniel-mueller	2022-06-29
f6786c8	jens-daniel-mueller	2022-06-29
9393c07	jens-daniel-mueller	2022-06-28
0825298	jens-daniel-mueller	2022-06-28
fb59a6f	jens-daniel-mueller	2022-06-27
3c1100c	jens-daniel-mueller	2022-05-16
2ca0109	jens-daniel-mueller	2022-05-02
b018a9a	jens-daniel-mueller	2022-04-29
e09320d	jens-daniel-mueller	2022-04-12
8dca96a	jens-daniel-mueller	2022-04-12
209c9b6	jens-daniel-mueller	2022-04-10
acad2e2	jens-daniel-mueller	2022-04-09
3d81135	jens-daniel-mueller	2022-04-07
a74e341	jens-daniel-mueller	2022-04-04
bd9e11d	jens-daniel-mueller	2022-03-22
2501978	jens-daniel-mueller	2022-03-21
c3a6238	jens-daniel-mueller	2022-03-08
094bfa0	jens-daniel-mueller	2022-02-18
ba2d62e	jens-daniel-mueller	2022-02-17
192504c	jens-daniel-mueller	2022-02-17
d2191ad	jens-daniel-mueller	2022-02-04
4c5b079	jens-daniel-mueller	2022-02-03
4077397	jens-daniel-mueller	2022-02-03

dcant_type_box <-
  bind_rows(
    dcant_budget_basin_MLR_all_ensemble %>%
      filter(
        data_source == "obs",
        basin == "Global",
        !(Version_ID %in% Version_IDs),
        MLR_basins == unique(params_local_all$MLR_basins) |
        Version_ID_group == unique(params_local_all$Version_ID_group)
      ) %>%
      select(period, dcant) %>%
      mutate(type = "eMLR(C*)",
             dcant = dcant,
             dcant = if_else(period == "2004 - 2014",
                             dcant + tcant_2004 - 118,
                             dcant)),
    Ocean_Sink_cum_1994 %>%
      filter(year %in% c(2004, 2014)) %>%
      mutate(period = if_else(year == 2004,
                              "1994 - 2004",
                              "1994 - 2014")) %>%
      select(period, dcant = fgco2_glob_cum_total, type),
    Ocean_Sink_cum_2004 %>%
      filter(year %in% c(2014)) %>%
      mutate(period = if_else(year == 2014,
                              "2004 - 2014",
                              "other")) %>%
      select(period, dcant = fgco2_glob_cum_total, type)
  )

dcant_type_box_stat <- dcant_type_box %>% 
  group_by(period, type) %>% 
  summarise(dcant_mean = mean(dcant),
            dcant_sd = sd(dcant)) %>% 
  ungroup()

dcant_type_box_stat <-
left_join(
  dcant_type_box_stat,
  dcant_budget_basin_MLR_all_uncertainty %>%
    filter(data_source == "obs",
           basin == "Global") %>% 
    mutate(
      std_case = if_else(period == "2004 - 2014",
                      std_case + tcant_2004 - 118,
                      std_case)
    ) %>%
    select(period, std_case, sd)
)

dcant_type_box_stat <- dcant_type_box_stat %>% 
  mutate(dcant_mean = if_else(type == "eMLR(C*)",
                              std_case,
                              dcant_mean)) %>% 
  mutate(dcant_sd = if_else(type == "eMLR(C*)",
                              sd,
                              dcant_sd)) %>% 
  select(-c(std_case, sd))

p_tcant_box <-
dcant_type_box %>%
  filter(period == "1994 - 2014") %>%
  ggplot(aes(period, dcant)) +
 geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period == "1994 - 2014"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd*2,
      ymax = dcant_mean + dcant_sd*2,
      col = type,
      fill = type
    ), width = 0.7, alpha = 0.2
  ) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period == "1994 - 2014"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd,
      ymax = dcant_mean + dcant_sd,
      col = type,
      fill = type
    ), width = 0.7, alpha = 0.4
  ) +
  geom_point(aes(fill = type),
              position = position_jitterdodge(
                dodge.width = 1,
                jitter.width = 0.1),
              shape = 21,
              alpha = 0.5) +
  scale_fill_highcontrast(
    name = "Data source",
    guide = "none"
  ) +
  scale_color_highcontrast(
    name = "Data source",
    guide = "none"
  ) +
  scale_y_continuous(limits = c(tcant_min, tcant_max)) +
  labs(x = "20yr period") +
  theme_classic() +
  theme(
    axis.line.y.left = element_blank(),
    axis.text.y = element_blank(),
    axis.ticks.y = element_blank(),
    axis.title.y = element_blank(),
    panel.grid.major.y = element_line(colour = "black",
                                      size = 0.1)
  )

name_sec_axis <- expression(
  atop(Cumulative ~ ocean ~ carbon ~ storage,
       since ~ 2004 ~ (Pg ~ C)))

p_tcant_box_10 <-
  dcant_type_box %>%
  filter(period != "1994 - 2014") %>%
  ggplot(aes(period, dcant)) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period != "1994 - 2014"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd*2,
      ymax = dcant_mean + dcant_sd*2,
      col = type,
      fill = type
    ),
    width = 0.7,
    alpha = 0.2
  ) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period != "1994 - 2014"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd,
      ymax = dcant_mean + dcant_sd,
      col = type,
      fill = type
    ),
    width = 0.7,
    alpha = 0.4
  ) +
  geom_point(
    aes(fill = type),
    position = position_jitterdodge(dodge.width = 1,
                                    jitter.width = 0.1),
    shape = 21,
    alpha = 0.5
  ) +
  geom_point(
    data = dcant_type_box_stat %>%
      filter(period != "1994 - 2014"),
    position = position_dodge(width = 1),
    aes(x = period,
        y = dcant_mean,
        col = type),
    fill = "white",
    shape = 21,
    size = 2
  ) +
  scale_fill_highcontrast(name = "Data source",
                          guide = "none") +
  scale_color_highcontrast(name = "Data source",
                           guide = "none") +
  scale_y_continuous(
    limits = c(tcant_min, tcant_max),
    sec.axis = sec_axis(
      ~ . - (tcant_2004 - 118),
      name = name_sec_axis,
      guide = guide_axis_truncated(trunc_lower = tcant_2004 - 118,
                                   trunc_upper = tcant_2004 - 118 + 35),
      breaks = seq(0, 30, 10)
    )
  )+
  labs(x = "Decades")+
  theme_classic() +
  # guides(y.right = "axis_truncated") +
  theme(
    axis.line.y.left = element_blank(),
    axis.text.y.left = element_blank(),
    axis.ticks.y.left = element_blank(),
    axis.title.y.left = element_blank(),
    panel.grid.major.y = element_line(colour = "black",
                                      size = 0.1)
  )

p_tcant_box_10

Past versions of mean_tcant_over_atm_co2_incl_fluxes_boxplot-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
b9d5a73	jens-daniel-mueller	2022-12-15
0f2c9d5	jens-daniel-mueller	2022-12-08
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
8e1702d	jens-daniel-mueller	2022-11-16
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
0b40366	jens-daniel-mueller	2022-10-26
91052ae	jens-daniel-mueller	2022-10-20
77e6a80	jens-daniel-mueller	2022-09-08
051b911	jens-daniel-mueller	2022-09-07
75bd58c	jens-daniel-mueller	2022-09-07
1c6de0d	jens-daniel-mueller	2022-09-07
78dbb27	jens-daniel-mueller	2022-08-11
302d289	jens-daniel-mueller	2022-08-11
318fefe	jens-daniel-mueller	2022-08-11
5b77c4f	jens-daniel-mueller	2022-08-10
9c3be27	jens-daniel-mueller	2022-08-09
a691b29	jens-daniel-mueller	2022-08-09
910f96e	jens-daniel-mueller	2022-08-08
f68fcfa	jens-daniel-mueller	2022-07-26
f2615fb	jens-daniel-mueller	2022-07-26
4af29f9	jens-daniel-mueller	2022-07-24
b18b250	jens-daniel-mueller	2022-07-20
a80b59b	jens-daniel-mueller	2022-07-20
fea41c1	jens-daniel-mueller	2022-07-20
d803308	jens-daniel-mueller	2022-07-19
b1f7ab3	jens-daniel-mueller	2022-07-18
d2ae54c	jens-daniel-mueller	2022-07-18
2695085	jens-daniel-mueller	2022-07-17
535196a	jens-daniel-mueller	2022-07-17
d20faeb	jens-daniel-mueller	2022-07-17
afb27ad	jens-daniel-mueller	2022-07-15
b492b46	jens-daniel-mueller	2022-07-15
bd24a0f	jens-daniel-mueller	2022-07-15
17cd1d1	jens-daniel-mueller	2022-07-13
26e9496	jens-daniel-mueller	2022-07-12
8fb595c	jens-daniel-mueller	2022-07-12
003b161	jens-daniel-mueller	2022-07-12
8ab4a87	jens-daniel-mueller	2022-06-29
7629c78	jens-daniel-mueller	2022-06-29
f6786c8	jens-daniel-mueller	2022-06-29
9393c07	jens-daniel-mueller	2022-06-28
0825298	jens-daniel-mueller	2022-06-28

p_tcant_ts_zoom + p_tcant_box + p_tcant_box_10 +
  plot_layout(ncol = 3, widths = c(6,1,1.8)) + 
  plot_annotation(tag_levels = 'A')

Past versions of mean_tcant_over_atm_co2_incl_fluxes_boxplot-2.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
b9d5a73	jens-daniel-mueller	2022-12-15
0f2c9d5	jens-daniel-mueller	2022-12-08
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
8e1702d	jens-daniel-mueller	2022-11-16
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
0b40366	jens-daniel-mueller	2022-10-26
91052ae	jens-daniel-mueller	2022-10-20

# ggsave(path = here::here("output/publication"),
#        filename = "Fig_dcant_inventory_global_vs_GCB_ocean_sink.png",
#        height = 6,
#        width = 9)

p_tcant_ts_zoom + p_tcant_box_10 +
  plot_layout(ncol = 2, widths = c(6,1.8))

Past versions of mean_tcant_over_atm_co2_incl_fluxes_boxplot-3.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

ggsave(path = here::here("output/presentation"),
       filename = "Fig_dcant_budget_global_vs_GCB_ocean_sink.png",
       height = 6,
       width = 9)

dcant_type_significance <- dcant_type_box_stat %>%
  mutate(dcant_mean = dcant_mean,
         dcant_mean = if_else(period == "2004 - 2014",
                         dcant_mean - (tcant_2004 - 118),
                         dcant_mean)) %>%
  group_by(period) %>%
  mutate(
    dcant_diff = dcant_mean - first(dcant_mean),
    dcant_diff_rel = 100 * dcant_diff / first(dcant_mean),
    dcant_diff_sd = sqrt(dcant_sd ^ 2 + first(dcant_sd) ^ 2),
    dcant_diff_sd_comparison = abs(dcant_diff_sd) - abs(dcant_diff)
  ) %>%
  ungroup()

dcant_type_significance %>% 
  drop_na() %>% 
  kable() %>%
  kable_styling() %>%
  scroll_box(height = "300px")

period	type	dcant_mean	dcant_sd	dcant_diff	dcant_diff_rel	dcant_diff_sd	dcant_diff_sd_comparison
1994 - 2004	eMLR(C*)	29.27668	2.527296	0.0000000	0.000000	3.574136	3.5741362
1994 - 2004	Global Ocean Biogeochemical Models (GOBM)	19.52578	2.701502	-9.7508981	-33.306028	3.699370	-6.0515281
1994 - 2004	Observation-based surface flux products	21.41458	2.798867	-7.8620944	-26.854464	3.771058	-4.0910360
2004 - 2014	eMLR(C*)	27.34720	2.549551	0.0000000	0.000000	3.605609	3.6056091
2004 - 2014	Global Ocean Biogeochemical Models (GOBM)	22.75899	2.788027	-4.5882109	-16.777624	3.778002	-0.8102086
2004 - 2014	Observation-based surface flux products	26.48067	1.306376	-0.8665297	-3.168623	2.864756	1.9982262

p_tcant_box_10_dec1 <-
  dcant_type_box %>%
  filter(period == "1994 - 2004") %>%
  ggplot(aes(period, dcant)) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period == "1994 - 2004"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd*2,
      ymax = dcant_mean + dcant_sd*2,
      col = type,
      fill = type
    ),
    width = 0.7,
    alpha = 0.2
  ) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period == "1994 - 2004"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd,
      ymax = dcant_mean + dcant_sd,
      col = type,
      fill = type
    ),
    width = 0.7,
    alpha = 0.4
  ) +
  geom_point(
    aes(fill = type),
    position = position_jitterdodge(dodge.width = 1,
                                    jitter.width = 0.1),
    shape = 21,
    alpha = 0.5
  ) +
  geom_point(
    data = dcant_type_box_stat %>%
      filter(period == "1994 - 2004"),
    position = position_dodge(width = 1),
    aes(x = period,
        y = dcant_mean,
        col = type),
    fill = "white",
    shape = 21,
    size = 2
  ) +
  scale_fill_highcontrast(name = "Data source",
                          guide = "none") +
  scale_color_highcontrast(name = "Data source",
                           guide = "none") +
  scale_y_continuous(
    limits = c(tcant_min, tcant_max),
    breaks = seq(0,40,20)
  )+
  labs(x = "Decades")+
  theme_classic() +
  # guides(y.right = "axis_truncated") +
  theme(
    axis.line.y.left = element_blank(),
    axis.text.y.left = element_blank(),
    axis.title.x = element_blank(),
    axis.ticks.y.left = element_blank(),
    axis.title.y.left = element_blank(),
    panel.grid.major.y = element_line(colour = "black",
                                      size = 0.1)
  )

p_tcant_box_10_dec1

Past versions of panel_per_decade-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

p_tcant_box_10_dec2 <-
  dcant_type_box %>%
  filter(period == "2004 - 2014") %>%
  ggplot(aes(period, dcant)) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period == "2004 - 2014"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd*2,
      ymax = dcant_mean + dcant_sd*2,
      col = type,
      fill = type
    ),
    width = 0.7,
    alpha = 0.2
  ) +
  geom_crossbar(
    data = dcant_type_box_stat %>%
      filter(period == "2004 - 2014"),
    position = position_dodge(width = 1),
    aes(
      x = period,
      y = dcant_mean,
      ymin = dcant_mean - dcant_sd,
      ymax = dcant_mean + dcant_sd,
      col = type,
      fill = type
    ),
    width = 0.7,
    alpha = 0.4
  ) +
  geom_point(
    aes(fill = type),
    position = position_jitterdodge(dodge.width = 1,
                                    jitter.width = 0.1),
    shape = 21,
    alpha = 0.5
  ) +
  geom_point(
    data = dcant_type_box_stat %>%
      filter(period == "2004 - 2014"),
    position = position_dodge(width = 1),
    aes(x = period,
        y = dcant_mean,
        col = type),
    fill = "white",
    shape = 21,
    size = 2
  ) +
  scale_fill_highcontrast(name = "Data source",
                          guide = "none") +
  scale_color_highcontrast(name = "Data source",
                           guide = "none") +
  scale_y_continuous(
    limits = c(tcant_min, tcant_max),
    breaks = seq(tcant_2004 - 118, tcant_2004 - 118 + 30, 20),
    sec.axis = sec_axis(
      ~ . - (tcant_2004 - 118),
      name = name_sec_axis,
      guide = guide_axis_truncated(trunc_lower = tcant_2004 - 118,
                                   trunc_upper = tcant_2004 - 118 + 35),
      breaks = seq(0, 30, 20)
    )
  )+
  theme_classic() +
  # guides(y.right = "axis_truncated") +
  theme(
    axis.line.y.left = element_blank(),
    axis.text.y.left = element_blank(),
    axis.title.y.right = element_text(angle = 90,
                                      hjust = 0.9),
    axis.title.x = element_blank(),
    axis.ticks.y.left = element_blank(),
    axis.title.y.left = element_blank(),
    panel.grid.major.y = element_line(colour = "black",
                                      size = 0.1)
  )

p_tcant_box_10_dec2

Past versions of panel_per_decade-2.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10

Ocean_Sink_cum_1994_ensemble_mean <- Ocean_Sink_cum_1994 %>% 
  group_by(year, pCO2, type) %>% 
  summarise(fgco2_glob_cum_total_mean = mean(fgco2_glob_cum_total),
            fgco2_glob_cum_total_sd = sd(fgco2_glob_cum_total)) %>% 
  ungroup()

p_tcant_ts_zoom <-
tcant_ts_zoom %>%
  ggplot() +
  geom_ribbon(aes(pCO2, tcant,
                  ymin = tcant - dcant_sd,
                  ymax = tcant + dcant_sd,
              fill = "eMLR(C*)"),
              alpha = 0.4) +
  geom_ribbon(
    aes(pCO2, tcant,
        ymin = tcant - dcant_sd*2,
        ymax = tcant + dcant_sd*2,
              fill = "eMLR(C*)"),
    alpha = 0.2
  ) +
  geom_line(data = Ocean_Sink_cum_1994,
            aes(pCO2, fgco2_glob_cum_total, group = product, col = type), size = 0.3) +
  geom_line(data = Ocean_Sink_cum_1994_ensemble_mean,
            aes(pCO2, fgco2_glob_cum_total_mean, col = type), size = 2) +
  geom_point(
    data = Ocean_Sink_cum_1994_ensemble_mean,
    aes(pCO2, fgco2_glob_cum_total_mean, col = type), size = 1,
    shape = 21,
    fill = "white"
  ) +
  geom_point(
    data = Ocean_Sink_cum_1994_ensemble_mean %>% filter(year %in% c(2004, 2014)),
    aes(pCO2, fgco2_glob_cum_total_mean, col = type), size = 2,
    shape = 21,
    fill = "white"
  ) +
  geom_line(aes(pCO2, tcant, col = "eMLR(C*)"), size = 2) +
  geom_point(aes(pCO2, tcant, col = "eMLR(C*)"), size = 2,
             shape = 21,
             fill = "white") +
  geom_text(aes(pCO2, tcant, label = year), nudge_x = -2, nudge_y = 3) +
  scale_color_highcontrast(name = "Data source") +
  scale_fill_highcontrast(name = "Data source") +
  scale_y_continuous(limits = c(tcant_min, tcant_max)) +
  # coord_cartesian(expand = 0) +
  labs(
    x = expression(Atmospheric ~ pCO[2] ~ (µatm)),
    y = expression(Cumulative ~ ocean ~ carbon ~ storage ~ since ~ 1994~(Pg~C))
  ) +
  guides(colour = guide_legend(order = 1),
         fill = "none") +
  theme_classic() +
  theme(
    legend.background = element_rect(fill = "transparent"),
    legend.position = c(0.25, 0.83),
    legend.title = element_blank(),
    legend.box.just = "top",
    legend.box = "horizontal",
    legend.key.size = unit(2.5, 'lines'),
    panel.grid.major.y = element_line(colour = "black",
                                      size = 0.1)
  )

p_tcant_ts_zoom

Past versions of mean_tcant_over_atm_co2_incl_fluxes_mean-1.png

Version	Author	Date
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
8e1702d	jens-daniel-mueller	2022-11-16
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

p_tcant_ts_zoom + p_tcant_box_10_dec1 + p_tcant_box_10_dec2 +
  plot_layout(ncol = 3, widths = c(8,1,1)) + 
  plot_annotation(tag_levels = 'A')

Past versions of mean_tcant_over_atm_co2_incl_fluxes_boxplot_ensemble-1.png

Version	Author	Date
2cca4eb	jens-daniel-mueller	2023-01-10
b9d5a73	jens-daniel-mueller	2022-12-15
0f2c9d5	jens-daniel-mueller	2022-12-08
322b1c1	jens-daniel-mueller	2022-11-25
491f8b1	jens-daniel-mueller	2022-11-25
4da4d66	jens-daniel-mueller	2022-11-25
534ee74	jens-daniel-mueller	2022-11-25
e924c80	jens-daniel-mueller	2022-11-17
8e1702d	jens-daniel-mueller	2022-11-16
ea34027	jens-daniel-mueller	2022-11-15
8ac0da0	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
375300e	jens-daniel-mueller	2022-11-14
ebc85e2	jens-daniel-mueller	2022-11-14
6edad9b	jens-daniel-mueller	2022-11-11
16beb51	jens-daniel-mueller	2022-11-08
b92657d	jens-daniel-mueller	2022-11-08
7275091	jens-daniel-mueller	2022-11-05
99610ed	jens-daniel-mueller	2022-11-01
7060801	jens-daniel-mueller	2022-10-31
ecb84c1	jens-daniel-mueller	2022-10-30
2345935	jens-daniel-mueller	2022-10-29
0b40366	jens-daniel-mueller	2022-10-26
91052ae	jens-daniel-mueller	2022-10-20

# p_tcant_ts_zoom + p_tcant_box_10 +
#   plot_layout(ncol = 2, widths = c(6,1.8)) + 
#   plot_annotation(tag_levels = 'A')

ggsave(path = here::here("output/publication"),
       filename = "Fig_dcant_inventory_global_vs_GCB_ocean_sink.png",
       height = 6,
       width = 8,
       dpi = 600)

17 Revelle changes

delta_revelle <- co2_atm %>%
  mutate(revelle = seacarb::buffer(24,
                                   pCO2,
                                   2300 * 1e-6)$BetaD,
         DIC = seacarb::carb(24,
                                   pCO2,
                                   2300 * 1e-6)$DIC,
         delta_DIC = (1/pCO2) * DIC * (1/revelle) *1e6,
         delta_DIC_rel = delta_DIC / mean(delta_DIC),
         delta_DIC_rel_cum = delta_DIC_rel - first(delta_DIC_rel),
         delta_year = year - first(year))
  
  
delta_revelle %>% 
  ggplot(aes(delta_year, pCO2)) +
  geom_path() +
  geom_point()

Past versions of revelle_changes-1.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
084a41c	jens-daniel-mueller	2022-10-13
fea41c1	jens-daniel-mueller	2022-07-20

delta_revelle %>% 
  ggplot(aes(delta_year,1/revelle)) +
  geom_smooth(method = "lm", aes(col="lm"), se = FALSE) +
  geom_path() +
  geom_point()

Past versions of revelle_changes-2.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
084a41c	jens-daniel-mueller	2022-10-13

delta_revelle %>% 
  ggplot(aes(delta_year,delta_DIC_rel)) +
  geom_smooth(method = "lm", aes(col="lm"), se = FALSE) +
  geom_path() +
  geom_point()

Past versions of revelle_changes-3.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
084a41c	jens-daniel-mueller	2022-10-13

delta_revelle %>% 
  ggplot(aes(delta_year,delta_DIC_rel_cum)) +
  geom_smooth(method = "lm", aes(col="lm"), se = FALSE) +
  geom_path() +
  geom_point()

Past versions of revelle_changes-4.png

Version	Author	Date
3a7bab0	jens-daniel-mueller	2022-11-28
084a41c	jens-daniel-mueller	2022-10-13

min(delta_revelle$delta_DIC_rel_cum / max(delta_revelle$delta_year))

[1] -0.006078176

18 GCB implications

18.1 Land sink

flux_label <- expression(CO[2]~flux~(Pg~C~yr^{-1}))

flux_components <- Global_Carbon_Budget %>% 
  distinct(estimate) %>% 
  pull()

emission_components <- flux_components[c(1,2,6)]
sink_components <- flux_components[3:5]

estimates <- Global_Carbon_Budget %>%
  select(-GtC) %>%
  filter(year >= 1990) %>% 
  pivot_wider(names_from = estimate,
              values_from = GtC_yr) %>%
  mutate(`fossil emissions` = 
           `fossil emissions excluding carbonation` - `cement carbonation sink`) %>%
  select(-c(`fossil emissions excluding carbonation`, `cement carbonation sink`)) %>% 
  pivot_longer(-year,
               names_to = "estimate",
               values_to = "GtC_yr")

estimates <- estimates %>%
  filter(!(estimate %in% c("budget imbalance"))) %>%
  mutate(
    sign = if_else(estimate %in% sink_components,
                   "sinks",
                   "emissions"),
    GtC_yr = if_else(sign == "sinks",
                     -GtC_yr,
                     GtC_yr)
    # GtC_yr = if_else(estimate == "cement carbonation sink",
    #                  -GtC_yr,
    #                  GtC_yr)
  )

emissions_sinks <- estimates %>% 
  group_by(year, sign) %>% 
  summarise(GtC_yr = sum(GtC_yr)) %>% 
  ungroup()

estimates %>% distinct(estimate) %>% pull()

[1] "land-use change emissions" "atmospheric growth"       
[3] "ocean sink"                "land sink"                
[5] "fossil emissions"         

p_stacked_fluxes <- estimates %>%
  mutate(
    estimate = fct_relevel(
      estimate,
      "fossil emissions",
      "land-use change emissions",
      "atmospheric growth",
      "land sink",
      "ocean sink"
    )
  ) %>%
  ggplot() +
  geom_area(aes(year, GtC_yr, fill = estimate),
            col = "black",
            alpha = 0.7,
            size = 0.5) +
  # geom_line(data = emissions_sinks %>%   mutate(GtC_yr = if_else(sign == "emissions",
  #                                                                -GtC_yr,
  #                                                                GtC_yr)),
  #           aes(year, GtC_yr, linetype = sign)) +
  scale_fill_manual(values = c("#BBBBBB", "#CCBB44", "#66CCEE", "#228833", "#4477AA"),
                    name = "flux component") +
  geom_hline(yintercept = 0) +
  scale_x_continuous(expand = c(0,0)) +
  labs(y = flux_label,
       title = "Carbon emissions and sinks",
       subtitle = "Global Carbon Budget | Friedlingstein et al. (2021)") +
  theme(axis.title.x = element_blank())

p_stacked_fluxes

Past versions of GCB_land_sink_assesment-1.png

Version	Author	Date
0f2c9d5	jens-daniel-mueller	2022-12-08
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
91052ae	jens-daniel-mueller	2022-10-20

ggsave(path = here::here("output/presentation"),
       filename = "Fig_GCB_emissions_sinks.png",
       height = 4,
       width = 7)

p_individual_fluxes <- estimates %>%
  ggplot(aes(year, GtC_yr, col = estimate)) +
  geom_hline(yintercept = 0) +
  geom_path() +
  geom_point(size = 0.5) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_bright(name = "flux component") +
  labs(y = flux_label,
       title = "Individual flux components from GCB") +
  theme(axis.title.x = element_blank())

p_individual_fluxes

Past versions of GCB_land_sink_assesment-2.png

Version	Author	Date
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
91052ae	jens-daniel-mueller	2022-10-20

BIM <- emissions_sinks %>% 
  pivot_wider(names_from = sign,
              values_from = GtC_yr) %>% 
  mutate(BIM = emissions + sinks)

p_BIM <- ggplot() +
  geom_line(data = Global_Carbon_Budget %>%
              filter(estimate == "budget imbalance",
                     year >= 1994,
                     year < 2014),
            aes(year, GtC_yr, col = "reported")) +
  geom_line(data = BIM, aes(year, BIM, col = "recalculated")) +
  geom_hline(yintercept = 0) +
  scale_color_brewer(palette = "Set1") +
  scale_x_continuous(expand = c(0, 0)) +
  labs(y = flux_label,
       title = "BIM from GCB") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_BIM

Past versions of GCB_land_sink_assesment-3.png

Version	Author	Date
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
91052ae	jens-daniel-mueller	2022-10-20

p_land_sink_gcb <- estimates %>%
  select(year, estimate, GtC_yr) %>%
  pivot_wider(names_from = estimate,
              values_from = GtC_yr) %>%
  mutate(`land sink residual` = 
           `fossil emissions` + 
           `land-use change emissions` + 
           `atmospheric growth` + 
           `ocean sink`) %>%
  ggplot() +
  geom_line(aes(year, -`land sink residual`, col = "residual")) +
  geom_line(aes(year, `land sink`, col = "reported")) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Set1", name = "land sink\nestimate") +
  labs(y = flux_label,
       title = "Land sink estimates from GCB",
       subtitle = "Note: The offset between reported and residual estimate equals BIM") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_land_sink_gcb

Past versions of GCB_land_sink_assesment-4.png

Version	Author	Date
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
91052ae	jens-daniel-mueller	2022-10-20

# Ocean_Sink %>% distinct(product)

Ocean_Sink_annual_mean <- Ocean_Sink %>%
  filter(year >= 1990) %>% 
  group_by(year, type) %>%
  summarise(GtC_yr = -mean(GtC_yr)) %>%
  ungroup()

Ocean_Sink_annual_mean <- Ocean_Sink_annual_mean %>% 
  pivot_wider(names_from = type,
              values_from = GtC_yr,
              names_prefix = "ocean_sink_")

estimates_wide <- estimates %>%
  select(year, estimate, GtC_yr) %>%
  pivot_wider(names_from = estimate,
              values_from = GtC_yr)

estimates_wide <-
  full_join(estimates_wide, Ocean_Sink_annual_mean)

dcant_interior <- tcant_ts_zoom %>% 
  filter(year > 2000) %>%
  mutate(dcant_mean = dcant_mean/10) %>% 
  pull(dcant_mean)

cnat <-
  dcant_type_significance %>%
  filter(type == "Observation-based\nsurface flux\nproducts" ,
         period %in% two_decades) %>%
  mutate(dcant_diff = dcant_diff/10) %>% 
  pull(dcant_diff)

estimates_wide <- estimates_wide %>% 
  mutate(ocean_sink_interior_Cant = case_when(
    between(year, 1995, 2004) ~ -dcant_interior[1],
    between(year, 2005, 2014) ~ -dcant_interior[2],
    TRUE ~ NaN
  )) %>% 
  mutate(ocean_sink_interior_Cnet = case_when(
    between(year, 1995, 2004) ~ -dcant_interior[1]-cnat[1],
    between(year, 2005, 2014) ~ -dcant_interior[2]-cnat[2],
    TRUE ~ NaN
  )) %>% 
  mutate(ocean_sink_interior_Cnat = case_when(
    between(year, 1995, 2004) ~ -cnat[1],
    between(year, 2005, 2014) ~ -cnat[2],
    TRUE ~ NaN
  ))


p_ocean_sinks <- estimates_wide %>%
  select(year, starts_with("ocean")) %>% 
  rename(GCB_mean = `ocean sink`) %>% 
  pivot_longer(-year,
               names_to = "estimate",
               names_prefix = "ocean_sink_",
               values_to = "GtC_yr") %>% 
  mutate(estimate = str_replace(estimate, "_", " ")) %>% 
  ggplot(aes(year, GtC_yr, col = estimate)) +
  geom_path() +
  geom_point(size = 0.5) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Dark2", name = "ocean sink\nestimate") +
  labs(y = flux_label,
       title = "Ocean sink estimates",
       subtitle = "Note: Watson et al. (2020) is excluded in data products and GCB mean") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_ocean_sinks

Past versions of GCB_land_sink_assesment-5.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

estimates_long <- estimates_wide %>%
  select(year, starts_with("ocean")) %>% 
  rename(GCB_mean = `ocean sink`) %>% 
  pivot_longer(-year,
               names_to = "estimate",
               names_prefix = "ocean_sink_",
               values_to = "GtC_yr") %>% 
  filter(estimate %in% c("data_products", "models"),
         year >= 1990)

p_ocean_sinks_presentation <- estimates_long %>%
  # mutate(estimate = str_replace(estimate, "_", " ")) %>%
  ggplot() +
  geom_path(data = Ocean_Sink %>% filter(year >= 1990),
            aes(year, GtC_yr, group = product, col = type), size = 0.3) +
  geom_path(aes(year, -GtC_yr, col = estimate), size = 2) +
  geom_point(aes(year, -GtC_yr, col = estimate), size = 2, fill = "white", shape = 21) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_manual(values = c("#BB5566", "#DDAA33"),
                     name = "ocean sink\nestimate") +
  labs(y = flux_label,
       title = "Ocean Carbon sink",
       subtitle = "Global Carbon Budget | Friedlingstein et al. (2021)") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_ocean_sinks_presentation

Past versions of GCB_land_sink_assesment-6.png

Version	Author	Date
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
91052ae	jens-daniel-mueller	2022-10-20

ggsave(path = here::here("output/presentation"),
       filename = "Fig_GCB_ocean_sinks.png",
       height = 4,
       width = 6)

p_ocean_sinks_presentation <-
  estimates_long %>%
  # mutate(estimate = str_replace(estimate, "_", " ")) %>%
  ggplot() +
  geom_path(
    data = Ocean_Sink %>% filter(year >= 1990),
    aes(year, GtC_yr, group = product, col = type),
    size = 0.3
  ) +
  geom_path(aes(year, -GtC_yr, col = estimate), size = 2) +
  geom_point(
    aes(year, -GtC_yr, col = estimate),
    size = 2,
    fill = "white",
    shape = 21
  ) +
  geom_path(
    data =             tibble(
      year = c(1994, 2007),
      estimate = "ocean_interior",
      GtC_yr = -2.6
    ),
    aes(year, -GtC_yr, col = estimate),
    size = 2
  )+
  geom_rect(
    data =             tibble(
      year_min = 1994,
      year_max = 2007,
      estimate = "ocean_interior",
      GtC_yr_min = -2.6 - 0.3,
      GtC_yr_max = -2.6 + 0.3,
    ),
    aes(xmax = year_max, xmin = year_min, ymin = -GtC_yr_min, ymax = -GtC_yr_max, col = estimate),
    fill = "transparent"
  )+
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_manual(values = c("#BB5566", "#DDAA33","#004488"),
                     name = "ocean sink\nestimate") +
  labs(y = flux_label,
       title = "Ocean Carbon sink",
       subtitle = "Global Carbon Budget | Friedlingstein et al. (2021)") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_ocean_sinks_presentation

Past versions of GCB_land_sink_assesment-7.png

Version	Author	Date
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
91052ae	jens-daniel-mueller	2022-10-20

ggsave(path = here::here("output/presentation"),
       filename = "Fig_GCB_ocean_sinks_eMLR.png",
       height = 4,
       width = 6)

land_sink <-
  estimates_wide %>%
  mutate(emissions_atmosphere = `fossil emissions` +
           `land-use change emissions` +
           `atmospheric growth`) %>%
  rename(ocean_sink_gcb_mean = `ocean sink`) %>%
  select(year, emissions_atmosphere, starts_with("ocean")) %>%
  pivot_longer(
    starts_with("ocean_sink"),
    names_to = "ocean_sink_estimate",
    values_to = "ocean_sink",
    names_prefix = "ocean_sink_"
  ) %>%
  mutate(land_sink = -emissions_atmosphere - ocean_sink)


p_land_sink <- land_sink %>%
  mutate(ocean_sink_estimate = 
           str_replace(ocean_sink_estimate, "_", " ")) %>% 
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_line(aes(year, land_sink, col = ocean_sink_estimate, linetype = "residual\nestimate")) +
  geom_line(data = estimates_wide,
            aes(year, `land sink`, linetype = "direct\nestimate", col = "reported land sink")) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_linetype_manual(values = c(2, 1), name = "land sink\nestimate") +
  scale_color_brewer(palette = "Dark2",
                     name = "ocean sink\nestimate\nunderlying\nresidual\nestimate") +
  guides(linetype = guide_legend(order = 1),
         color = guide_legend(order = 2)) +
  labs(y = flux_label,
       title = "Land sink estimates",
       subtitle = "Residuals estimates are determined for a range of ocean sink estimates") +
  theme(axis.title.x = element_blank())

p_land_sink

Past versions of GCB_land_sink_assesment-8.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

land_sink_long <-
  full_join(
    land_sink %>% 
      select(year, ocean_sink_estimate, land_sink),
    estimates_wide %>%
      select(year, land_sink = `land sink`) %>% 
      mutate(ocean_sink_estimate = "reported_land_sink")
  )

land_sink_long_decade <- land_sink_long %>%
  mutate(period = cut(year, c(1993, 2003, 2013),
                      c("1994 - 2004", "2004 - 2014"))) %>%
  filter(!is.na(period)) %>%
  group_by(ocean_sink_estimate, period) %>%
  summarise(across(starts_with("land_sink"), list(mean = mean, sd = sd))) %>%
  ungroup() %>%
  mutate(ocean_sink_estimate =
           str_replace_all(ocean_sink_estimate, "_", " "))

land_sink_long_decade

# A tibble: 14 × 4
   ocean_sink_estimate period      land_sink_mean land_sink_sd
   <chr>               <fct>                <dbl>        <dbl>
 1 data products       1994 - 2004          -1.96        1.19 
 2 data products       2004 - 2014          -2.96        0.789
 3 gcb mean            1994 - 2004          -2.07        1.19 
 4 gcb mean            2004 - 2014          -3.16        0.800
 5 interior Cant       1994 - 2004         NaN          NA    
 6 interior Cant       2004 - 2014          -2.80        0.837
 7 interior Cnat       1994 - 2004         NaN          NA    
 8 interior Cnat       2004 - 2014          -5.71        0.876
 9 interior Cnet       1994 - 2004         NaN          NA    
10 interior Cnet       2004 - 2014          -2.96        0.860
11 models              1994 - 2004          -2.15        1.21 
12 models              2004 - 2014          -3.32        0.818
13 reported land sink  1994 - 2004          -2.33        0.892
14 reported land sink  2004 - 2014          -2.94        0.578

p_land_sink_decade <-   
  land_sink_long_decade %>% 
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_pointrange(
    aes(
      period,
      land_sink_mean,
      ymin = land_sink_mean - land_sink_sd,
      ymax = land_sink_mean + land_sink_sd,
      col = ocean_sink_estimate
    ),
    position = position_dodge2(width = 0.5)
  ) +
  scale_color_brewer(palette = "Dark2",
                     name = "ocean sink\nestimate\nunderlying\nresidual\nestimate") +
  guides(shape = guide_legend(order = 1),
         color = guide_legend(order = 2)) +
  theme(axis.title.x = element_blank()) +
  labs(y = flux_label,
       title = "Decadal mean land sink estimates",
       subtitle = "Errorbars indicate the StDev of the interannual variability") +
  theme(axis.title.x = element_blank())

p_land_sink_decade

Past versions of GCB_land_sink_assesment-9.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

land_sink <-
  full_join(
    land_sink,
    estimates_wide %>%
      select(year, land_sink_reported = `land sink`)
  )

p_land_sink_residual <- land_sink %>%
  mutate(ocean_sink_estimate =
           str_replace(ocean_sink_estimate, "_", " ")) %>%
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_line(aes(year, land_sink - land_sink_reported, col = ocean_sink_estimate)) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Dark2",
                     name = "ocean sink\nestimate\nunderlying\nresidual") +
  labs(y = flux_label,
       title = "Land sink estimates (reported - residual)",
       subtitle = "Residuals are determined for a range of ocean sink estimates") +
  theme(axis.title.x = element_blank())

p_land_sink_residual

Past versions of GCB_land_sink_assesment-10.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29
3a7bab0	jens-daniel-mueller	2022-11-28
ea34027	jens-daniel-mueller	2022-11-15
1dad51a	jens-daniel-mueller	2022-11-15
6edad9b	jens-daniel-mueller	2022-11-11
7275091	jens-daniel-mueller	2022-11-05
2345935	jens-daniel-mueller	2022-10-29
91052ae	jens-daniel-mueller	2022-10-20

p_land_sink_residual_decade <- land_sink %>%
  mutate(period = cut(year, c(1993, 2003, 2013),
                      c("1994 - 2004", "2004 - 2014"))) %>%
  filter(!is.na(period)) %>%
  mutate(land_sink_residual = land_sink - land_sink_reported) %>% 
  select(period, ocean_sink_estimate, land_sink_residual) %>% 
  group_by(ocean_sink_estimate, period) %>%
  summarise(across(starts_with("land_sink"), list(mean = mean, sd = sd))) %>%
  ungroup() %>%
  mutate(ocean_sink_estimate =
           str_replace(ocean_sink_estimate, "_", " ")) %>%
  ggplot() +
    geom_hline(yintercept = 0) +
  geom_pointrange(
    aes(
      period,
      land_sink_residual_mean,
      ymin = land_sink_residual_mean - land_sink_residual_sd,
      ymax = land_sink_residual_mean + land_sink_residual_sd,
      col = ocean_sink_estimate
    ),
    position = position_dodge2(width = 0.5)
  ) +
  scale_color_brewer(palette = "Dark2",
                     name = "ocean sink\nestimate\nunderlying\nresidual") +
  theme(axis.title.x = element_blank()) +
  labs(y = flux_label,
       title = "Decadal mean land sink estimates",
       subtitle = "Errorbars indicate the StDev of the interannual variability") +
  theme(axis.title.x = element_blank())

p_land_sink_residual_decade

Past versions of GCB_land_sink_assesment-11.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
9048dc4	jens-daniel-mueller	2022-11-29

BIM_ocean <- land_sink %>% 
  mutate(BIM = emissions_atmosphere + ocean_sink + land_sink_reported)


p_BIM_ocean <- BIM_ocean %>%
  mutate(ocean_sink_estimate =
           str_replace(ocean_sink_estimate, "_", " ")) %>%
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_line(aes(year, BIM, col = ocean_sink_estimate)) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Dark2",
                     name = "ocean sink\nestimate\nunderlying\nBIM") +
  labs(y = flux_label,
       title = "BIM estimates",
       subtitle = "BIM estimates are determined for a range of ocean sink estimates") +
  theme(axis.title.x = element_blank())

p_BIM_ocean

Past versions of GCB_land_sink_assesment-12.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
ef50aa6	jens-daniel-mueller	2022-11-30
439cb85	jens-daniel-mueller	2022-11-29
9048dc4	jens-daniel-mueller	2022-11-29

BIM_ocean_decade <- BIM_ocean %>%
  mutate(period = cut(year, c(1993, 2003, 2013),
                      c("1994 - 2004", "2004 - 2014"))) %>%
  filter(!is.na(period)) %>%
  group_by(ocean_sink_estimate, period) %>%
  summarise(across(starts_with("BIM"), list(mean = mean, sd = sd))) %>%
  ungroup() %>%
  mutate(ocean_sink_estimate =
           str_replace_all(ocean_sink_estimate, "_", " "))

BIM_ocean_decade

# A tibble: 12 × 4
   ocean_sink_estimate period      BIM_mean BIM_sd
   <chr>               <fct>          <dbl>  <dbl>
 1 data products       1994 - 2004  -0.361   0.703
 2 data products       2004 - 2014   0.0208  0.530
 3 gcb mean            1994 - 2004  -0.254   0.709
 4 gcb mean            2004 - 2014   0.215   0.535
 5 interior Cant       1994 - 2004 NaN      NA    
 6 interior Cant       2004 - 2014  -0.143   0.570
 7 interior Cnat       1994 - 2004 NaN      NA    
 8 interior Cnat       2004 - 2014   2.77    0.573
 9 interior Cnet       1994 - 2004 NaN      NA    
10 interior Cnet       2004 - 2014   0.0134  0.561
11 models              1994 - 2004  -0.177   0.734
12 models              2004 - 2014   0.379   0.546

p_BIM_ocean_decade <-   
  BIM_ocean_decade %>% 
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_pointrange(
    aes(
      period,
      BIM_mean,
      ymin = BIM_mean - BIM_sd,
      ymax = BIM_mean + BIM_sd,
      col = ocean_sink_estimate
    ),
    position = position_dodge2(width = 0.5)
  ) +
  scale_color_brewer(palette = "Dark2",
                     name = "ocean sink\nestimate\nunderlying\nBIM") +
  guides(shape = guide_legend(order = 1),
         color = guide_legend(order = 2)) +
  theme(axis.title.x = element_blank()) +
  labs(y = flux_label,
       title = "Decadal mean BIM estimates",
       subtitle = "Errorbars indicate the StDev of the interannual variability") +
  theme(axis.title.x = element_blank())

p_BIM_ocean_decade

Past versions of GCB_land_sink_assesment-13.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02
9048dc4	jens-daniel-mueller	2022-11-29

pdf(
  file = here::here("output/publication/GCB_land_sink_and_BIM.pdf"),
  height = 4,
  width = 8
)

# p_stacked_fluxes
p_individual_fluxes
# p_BIM
# p_land_sink_gcb
p_ocean_sinks
p_land_sink
p_land_sink_decade
# p_land_sink_residual
# p_land_sink_residual_decade
p_BIM_ocean
p_BIM_ocean_decade

dev.off()

png 
  2 

rm(
  p_stacked_fluxes,
  p_individual_fluxes,
  p_BIM,
  p_land_sink_gcb,
  p_ocean_sinks,
  p_land_sink,
  p_land_sink_decade,
  p_land_sink_residual,
  p_land_sink_residual_decade,
  p_BIM_ocean,
  p_BIM_ocean_decade,
  estimates,
  # estimates_wide,
  land_sink,
  land_sink_long,
  land_sink_long_decade,
  Ocean_Sink_annual_mean,
  flux_components,
  flux_label,
  emission_components,
  sink_components
)

18.1.1 Decadal Budgets

Decadal_GCB_budgets <- estimates_wide %>%
  pivot_longer(-year,
               names_to = "CO2_sources_sinks",
               values_to = "GtC_yr") %>% 
  mutate(period = cut(year, c(1994, 2004, 2014), c("1994 - 2004", "2004 - 2014"))) %>%
  drop_na() %>%
  group_by(period, CO2_sources_sinks) %>%
  summarise(GtC = sum(GtC_yr)) %>%
  ungroup()

Decadal_GCB_budgets <- Decadal_GCB_budgets %>%
  filter(
    !(CO2_sources_sinks %in% c("ocean_sink_data_products", "ocean sink", "ocean_sink_models"))
  ) %>% 
  mutate(CO2_sources_sinks = str_remove(CO2_sources_sinks, "_interior"))

total_emissions <- Decadal_GCB_budgets %>% 
  filter(
    CO2_sources_sinks %in% c("fossil emissions", "land-use change emissions")
  ) %>% 
  group_by(period) %>%
  summarise(GtC = sum(GtC)) %>%
  ungroup() %>% 
  mutate(CO2_sources_sinks = "total emissions")

Decadal_GCB_budgets <- full_join(Decadal_GCB_budgets, total_emissions)


ocean_stats <- dcant_type_box_stat %>% 
  filter(period %in% two_decades) %>% 
  mutate(type = case_when(
    type == "eMLR(C*)" ~ "ocean_sink_Cant",
    type == "Observation-based\nsurface flux\nproducts" ~ "ocean_sink_Cnet",
    TRUE ~ "x"
    )) %>% 
  filter(type != "x") %>% 
  select(period, CO2_sources_sinks = type, GtC_sigma = dcant_sd)

Decadal_GCB_budgets <- 
  full_join(Decadal_GCB_budgets,
            ocean_stats)

rm(ocean_stats)

ocean_stats <- dcant_type_significance %>% 
  filter(period %in% two_decades) %>% 
  mutate(type = case_when(
    type == "Observation-based\nsurface flux\nproducts" ~ "ocean_sink_Cnat",
    TRUE ~ "x"
    )) %>% 
  filter(type != "x") %>% 
  select(period, CO2_sources_sinks = type, GtC_sigma = dcant_diff_sd)

library(rqdatatable)

Decadal_GCB_budgets <-
  natural_join(
    Decadal_GCB_budgets,
    ocean_stats,
    by = c("period", "CO2_sources_sinks"),
    jointype = "FULL"
  )

rm(ocean_stats)

Decadal_GCB_budgets <- Decadal_GCB_budgets %>% 
  mutate(GtC_sigma = case_when(
    CO2_sources_sinks == "fossil emissions" ~ GtC * 0.4 / 7.7,
    CO2_sources_sinks == "land-use change emissions" ~ GtC * 0.7 / 1.2,
    CO2_sources_sinks == "total emissions" ~ GtC * 0.8 / 9.0,
    CO2_sources_sinks == "atmospheric growth" ~ GtC * 0.02 / 4.0,
    CO2_sources_sinks == "land sink" ~ GtC * 0.5 / 2.6,
    TRUE ~ GtC_sigma
  )) %>% 
  mutate(GtC_sigma = abs(GtC_sigma))

Decadal_GCB_budgets %>% 
  ggplot(aes(period, GtC,
             ymin = GtC - GtC_sigma,
             ymax = GtC + GtC_sigma,
             col = CO2_sources_sinks)) +
  geom_hline(yintercept = 0) +
  scale_color_muted() +
  geom_pointrange(position = position_dodge2(width = 0.5),
                  shape = 1)

Past versions of decadal_budgets-1.png

Version	Author	Date
445761e	jens-daniel-mueller	2022-12-02

Decadal_GCB_fractions <- Decadal_GCB_budgets

Decadal_GCB_budgets <- Decadal_GCB_budgets %>% 
  mutate(across(where(is.numeric), round, 1)) %>% 
  mutate(GtC = paste(GtC, GtC_sigma, sep = " ± ")) %>% 
  select(-GtC_sigma) %>% 
  pivot_wider(names_from = period,
              values_from = GtC) %>% 
  mutate(CO2_sources_sinks = str_replace_all(CO2_sources_sinks, "_", " ")) 

Decadal_GCB_budgets %>% 
  write_csv(here::here("output/publication/Table_GCB.csv"))

Decadal_GCB_fractions <- Decadal_GCB_fractions %>% 
  filter(CO2_sources_sinks %in% c("atmospheric growth", "land sink", 
                                  "ocean_sink_Cnet", "ocean_sink_Cant",
                                  "total emissions")) %>% 
  group_by(period) %>% 
  mutate(fraction = abs(GtC / last(GtC))*100,
         GtC_rel = abs(GtC_sigma / GtC),
         fraction_rel = sqrt(GtC_rel^2 + last(GtC_rel)^2),
         fraction_sigma = fraction_rel * fraction)  %>% 
  ungroup() %>% 
  select(period, CO2_sources_sinks, fraction, fraction_sigma) %>% 
  mutate(across(where(is.numeric), round, 0)) %>% 
  mutate(fraction = paste0(fraction, " ± ", fraction_sigma)) %>% 
  select(-fraction_sigma) %>% 
  filter(CO2_sources_sinks != "total emissions") %>% 
  pivot_wider(names_from = period,
              values_from = fraction) %>% 
  mutate(CO2_sources_sinks = str_replace_all(CO2_sources_sinks, "_", " ")) 

Decadal_GCB_fractions

# A tibble: 4 × 3
  CO2_sources_sinks  `1994 - 2004` `2004 - 2014`
  <chr>              <chr>         <chr>        
1 atmospheric growth 48 ± 4        44 ± 4       
2 land sink          31 ± 6        30 ± 6       
3 ocean sink Cant    36 ± 4        27 ± 4       
4 ocean sink Cnet    26 ± 4        26 ± 3       

Decadal_GCB_fractions %>% 
  write_csv(here::here("output/publication/Table_GCB_fraction.csv"))

18.2 Net land sink

flux_label <- expression(flux~(Pg~C~yr^{-1}))

Global_Carbon_Budget_net_land_sink <- Global_Carbon_Budget %>%
  select(year, estimate, GtC_yr) %>%
  pivot_wider(names_from = estimate,
              values_from = GtC_yr) %>%
  mutate(`net land sink` = `land sink` - `land-use change emissions`) %>%
  select(-c(`land sink`, `land-use change emissions`)) %>%
  pivot_longer(-year,
               names_to = "estimate",
               values_to = "GtC_yr")

flux_components <- Global_Carbon_Budget_net_land_sink %>% 
  distinct(estimate) %>% 
  pull()

emission_components <- flux_components[c(1,4)]
sink_components <- flux_components[c(2,3,6)]

estimates <- Global_Carbon_Budget_net_land_sink %>%
  filter(!(estimate %in% c("budget imbalance")),
         year >= 1990) %>%
  mutate(
    sign = if_else(estimate %in% sink_components,
                   "sinks",
                   "emissions"),
    GtC_yr = if_else(sign == "sinks",
                     -GtC_yr,
                     GtC_yr),
    GtC_yr = if_else(estimate == "cement carbonation sink",
                     -GtC_yr,
                     GtC_yr)
  )

emissions_sinks <- estimates %>% 
  group_by(year, sign) %>% 
  summarise(GtC_yr = sum(GtC_yr)) %>% 
  ungroup()


p_stacked_fluxes <- estimates %>%
  ggplot() +
  geom_area(aes(year, GtC_yr, fill = estimate),
            col = "white",
            alpha = 0.7,
            size = 0.5) +
  geom_line(data = emissions_sinks %>%   mutate(GtC_yr = if_else(sign == "emissions",
                                                                 -GtC_yr,
                                                                 GtC_yr)),
            aes(year, GtC_yr, linetype = sign)) +
  scale_fill_bright(name = "flux component") +
  geom_hline(yintercept = 0) +
  scale_x_continuous(expand = c(0,0)) +
  labs(y = flux_label,
       title = "Stacked emission and sink components from GCB",
       subtitle = "Note: The cement carbonation sink is accounted for in emissions") +
  theme(axis.title.x = element_blank())

p_stacked_fluxes

p_individual_fluxes <- estimates %>%
  ggplot(aes(year, GtC_yr, col = estimate)) +
  geom_hline(yintercept = 0) +
  geom_path() +
  geom_point(size = 0.5) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_bright(name = "flux component") +
  labs(y = flux_label,
       title = "Individual flux components from GCB") +
  theme(axis.title.x = element_blank())

p_individual_fluxes


BIM <- emissions_sinks %>% 
  pivot_wider(names_from = sign,
              values_from = GtC_yr) %>% 
  mutate(BIM = emissions + sinks)

p_BIM <- ggplot() +
  geom_line(data = Global_Carbon_Budget %>%
              filter(estimate == "budget imbalance",
                     year >= 1990),
            aes(year, GtC_yr, col = "reported")) +
  geom_line(data = BIM, aes(year, BIM, col = "recalculated")) +
  geom_hline(yintercept = 0) +
  scale_color_brewer(palette = "Set1") +
  scale_x_continuous(expand = c(0, 0)) +
  labs(y = flux_label,
       title = "BIM from GCB") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_BIM

p_land_sink_gcb <- estimates %>%
  select(year, estimate, GtC_yr) %>%
  pivot_wider(names_from = estimate,
              values_from = GtC_yr) %>%
  mutate(`net land sink residual` = 
           `fossil emissions excluding carbonation` + 
           `cement carbonation sink` + 
           `atmospheric growth` + 
           `ocean sink`) %>%
  ggplot() +
  geom_line(aes(year, -`net land sink residual`, col = "residual")) +
  geom_line(aes(year, `net land sink`, col = "reported")) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Set1", name = "land sink\nestimate") +
  labs(y = flux_label,
       title = "Land sink estimates from GCB",
       subtitle = "Note: The offset between reported and residual estimate equals BIM") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_land_sink_gcb

Ocean_Sink_annual_mean <- Ocean_Sink %>%
  filter(year >= 1990) %>%
  group_by(year, type) %>%
  summarise(GtC_yr = -mean(GtC_yr)) %>%
  ungroup()

Ocean_Sink_annual_mean <- Ocean_Sink_annual_mean %>% 
  pivot_wider(names_from = type,
              values_from = GtC_yr,
              names_prefix = "ocean_sink_")

estimates_wide <- estimates %>%
  select(year, estimate, GtC_yr) %>%
  pivot_wider(names_from = estimate,
              values_from = GtC_yr)

estimates_wide <-
  full_join(estimates_wide, Ocean_Sink_annual_mean)

dcant_interior <- tcant_ts_zoom %>% 
  filter(year > 2000) %>%
  mutate(dcant_mean = dcant_mean/10) %>% 
  pull(dcant_mean)

cnat <-
  dcant_type_significance %>%
  filter(type == "Observation-based\nsurface flux\nproducts" ,
         period %in% two_decades) %>%
  mutate(dcant_diff = dcant_diff/10) %>% 
  pull(dcant_diff)

estimates_wide <- estimates_wide %>% 
  mutate(ocean_sink_interior_Cant = case_when(
    between(year, 1994, 2003) ~ -dcant_interior[1],
    between(year, 2004, 2013) ~ -dcant_interior[2],
    TRUE ~ NaN
  )) %>% 
  mutate(ocean_sink_interior_net = case_when(
    between(year, 1994, 2003) ~ -dcant_interior[1]-cnat[1],
    between(year, 2004, 2013) ~ -dcant_interior[2]-cnat[2],
    TRUE ~ NaN
  ))


p_ocean_sinks <- estimates_wide %>%
  select(year, starts_with("ocean")) %>% 
  rename(GCB_mean = `ocean sink`) %>% 
  pivot_longer(-year,
               names_to = "estimate",
               names_prefix = "ocean_sink_",
               values_to = "GtC_yr") %>% 
  mutate(estimate = str_replace(estimate, "_", " ")) %>% 
  ggplot(aes(year, GtC_yr, col = estimate)) +
  geom_path() +
  geom_point(size = 0.5) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Set1", name = "ocean sink\nestimate") +
  labs(y = flux_label,
       title = "Ocean sink estimates",
       subtitle = "Note: Watson et al. (2020) is included in data products and GCB mean") +
  theme(axis.title.x = element_blank(),
        legend.title = element_blank())

p_ocean_sinks

land_sink <-
  estimates_wide %>%
  mutate(emissions_atmosphere = `fossil emissions excluding carbonation` +
           `atmospheric growth`) %>%
  rename(ocean_sink_gcb_mean = `ocean sink`) %>%
  select(year, emissions_atmosphere, starts_with("ocean")) %>%
  pivot_longer(
    starts_with("ocean_sink"),
    names_to = "ocean_sink_estimate",
    values_to = "ocean_sink",
    names_prefix = "ocean_sink_"
  ) %>%
  mutate(net_land_sink = -emissions_atmosphere - ocean_sink)


p_land_sink_residual <- land_sink %>%
  mutate(ocean_sink_estimate = 
           str_replace(ocean_sink_estimate, "_", " ")) %>% 
  ggplot() +
  geom_line(aes(year, net_land_sink, col = ocean_sink_estimate, linetype = "residual")) +
  geom_line(data = estimates_wide %>%
              filter(year >= 1990),
            aes(year, `net land sink`, linetype = "reported")) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_linetype_manual(values = c(2, 1), name = "net land sink\nestimate") +
  scale_color_brewer(palette = "Set1",
                     name = "ocean sink\nestimate\nunderlying\nresidual") +
  guides(linetype = guide_legend(order = 1),
         color = guide_legend(order = 2)) +
  labs(y = flux_label,
       title = "Net land sink estimates",
       subtitle = "Note: Residuals are determined for a range of ocean sink estimates") +
  theme(axis.title.x = element_blank())

p_land_sink_residual

land_sink <-
  full_join(
    land_sink,
    estimates_wide %>%
      select(year, net_land_sink_reported = `net land sink`)
  )


p_land_sink_residual_delta <- land_sink %>%
  mutate(ocean_sink_estimate =
           str_replace(ocean_sink_estimate, "_", " ")) %>%
  ggplot() +
  geom_hline(yintercept = 0) +
  geom_line(aes(year, net_land_sink - net_land_sink_reported, col = ocean_sink_estimate)) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_color_brewer(palette = "Set1",
                     name = "ocean sink\nestimate\nunderlying\nresidual") +
  labs(y = flux_label,
       title = "Let land sink estimates (reported - residual)",
       subtitle = "Note: Residuals are determined for a range of ocean sink estimates") +
  theme(axis.title.x = element_blank())

p_land_sink_residual_delta

p_land_sink_residual_decade <- land_sink %>%
  mutate(period = cut(year, c(1993, 2003, 2013),
                      c("1994 - 2004", "2004 - 2014"))) %>%
  filter(!is.na(period)) %>%
  group_by(ocean_sink_estimate, period) %>%
  summarise(across(starts_with("net_land_sink"), list(mean = mean, sd = sd))) %>%
  ungroup() %>%
  mutate(ocean_sink_estimate =
           str_replace(ocean_sink_estimate, "_", " ")) %>%
  ggplot() +
  geom_pointrange(
    aes(
      period,
      net_land_sink_reported_mean,
      ymin = net_land_sink_reported_mean - net_land_sink_reported_sd,
      ymax = net_land_sink_reported_mean + net_land_sink_reported_sd,
      shape = "reported"
    ),
    position = position_dodge2(width = 0.6)
  ) +
  geom_pointrange(
    aes(
      period,
      net_land_sink_mean,
      ymin = net_land_sink_mean - net_land_sink_sd,
      ymax = net_land_sink_mean + net_land_sink_sd,
      col = ocean_sink_estimate,
      shape = "residual"
    ),
    position = position_dodge2(width = 0.5)
  ) +
  scale_color_brewer(palette = "Set1",
                     name = "ocean sink\nestimate\nunderlying\nresidual") +
  scale_shape_manual(values = c(17, 16), name = "land sink\nestimate") +
  guides(shape = guide_legend(order = 1),
         color = guide_legend(order = 2)) +
  theme(axis.title.x = element_blank()) +
  labs(y = flux_label,
       title = "Decadal mean land sink estimates",
       subtitle = "Note: Errorbars indicate the StDev of the IAV") +
  theme(axis.title.x = element_blank())

p_land_sink_residual_decade

pdf(
  file = here::here("output/publication/GCB_net_land_sink_residual.pdf"),
  height = 6,
  width = 9
)

p_stacked_fluxes
p_individual_fluxes
p_BIM
p_land_sink_gcb
p_ocean_sinks
p_land_sink_residual
p_land_sink_residual_delta
p_land_sink_residual_decade

dev.off()

rm(
  p_stacked_fluxes,
  p_individual_fluxes,
  p_BIM,
  p_land_sink_gcb,
  p_ocean_sinks,
  p_land_sink_residual,
  p_land_sink_residual_delta,
  p_land_sink_residual_decade,
  estimates,
  estimates_wide,
  land_sink,
  Ocean_Sink_annual_mean,
  flux_components,
  flux_label,
  emission_components,
  sink_components
)

19 Write output files

19.1 Ocean interior dcant

Version_IDs_out_vector <- Version_IDs_out %>% 
  pull(Version_ID)


for (i_Version_IDs in Version_IDs_out_vector) {
  # i_Version_IDs <- Version_IDs_out_vector[1]
  
  path_version_data     <-
    paste(path_observations,
          i_Version_IDs,
          "/data/",
          sep = "")
  
  # load and join data files
  
  dcant_3d <-
    read_csv(paste(path_version_data,
                   "dcant_3d.csv",
                   sep = ""))
  
  
  dcant_3d <- dcant_3d %>%
    filter(data_source == "obs") %>%
    select(lon, lat, depth, dcant) %>%
    mutate(Version_ID = i_Version_IDs)
  
  if (exists("dcant_3d_all")) {
    dcant_3d_all <- bind_rows(dcant_3d_all, dcant_3d)
  }
  
  if (!exists("dcant_3d_all")) {
    dcant_3d_all <- dcant_3d
  }
  
}

dcant_3d_all <- full_join(dcant_3d_all, Version_IDs_out) %>% 
  select(-Version_ID)


# convert from tibble to stars object
dcant_3d_out_stars <- st_as_stars(
  dcant_3d_all,
  dims = c(
    "lon",
    "lat",
    "depth",
    "period",
    "MLR_basins",
    "Version_ID_group"
  )
)

# assign coordinate system to horizontal (lon, lat) coordinates
# st_crs(dcant_inv_out_stars) <- "+proj=longlat"
st_crs(dcant_3d_out_stars) <- st_crs(4326)

# plot variable per depth level
ggplot() +
  geom_stars(data = dcant_3d_out_stars %>% 
               filter(period == "1994 - 2004",
                      depth == 50)) +
  facet_grid(MLR_basins~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")


file_out <- paste0(path_out, "dcant_emlr_cstar_mueller_94-04-14.nc")

# write stars object to .nc file
dcant_3d_out_stars %>%
  write_mdim(file_out)

# reopen .nc file with all available stars functions
dcant_3d_read_ncdf <-
  read_ncdf(file_out)
dcant_3d_read_mdim <-
  read_mdim(file_out)

# plot variable in object reopened

ggplot() +
  geom_stars(data = dcant_3d_read_ncdf %>% 
               filter(period == "1994 - 2004",
                      depth == 50)) +
  facet_grid(MLR_basins~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")

ggplot() +
  geom_stars(data = dcant_3d_read_mdim %>% 
               filter(period == "1994 - 2004",
                      depth == 50)) +
  facet_grid(MLR_basins~Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")


dcant_3d_read_tidync <-
  tidync::tidync(file_out)

dcant_3d_read_tidync %>%
  tidync::hyper_tibble() %>%
  filter(period == "1994 - 2004",
         depth == 50) %>%
  ggplot(aes(lon, lat, fill = dcant)) +
  geom_raster() +
  facet_grid(MLR_basins ~ Version_ID_group) +
  scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
                       na.value = "transparent")


# library(ncdf4)
# nc <- nc_open(file_out)
# nc_close(nc)


volume_mask <- dcant_3d_all %>% 
  filter(MLR_basins == "3",
         Version_ID_group == "Standard case",
         period == "1994 - 2004") %>% 
  select(lat, lon, depth)

volume_mask <- m_layer_thickness(volume_mask)

volume_mask <- volume_mask %>% 
  mutate(surface_area = earth_surf(lat = lat, lon = lon),
         volume = surface_area * layer_thickness) %>% 
  select(lon, lat, depth, volume)

# volume_mask %>% 
#   summarise(
#     total_surface_area = sum(surface_area),
#     total_volume = sum(volume))



# convert from tibble to stars object
volume_mask_out_stars <- st_as_stars(
  volume_mask,
  dims = c(
    "lon",
    "lat",
    "depth"
  )
)

# assign coordinate system to horizontal (lon, lat) coordinates
# st_crs(dcant_inv_out_stars) <- "+proj=longlat"
st_crs(volume_mask_out_stars) <- st_crs(4326)


file_out <- paste0(path_out, "volume_mask_emlr_cstar_mueller_94-04-14.nc")

# write stars object to .nc file
volume_mask_out_stars %>%
  write_mdim(file_out)

# reopen .nc file with all available stars functions
volume_mask_read_mdim <-
  read_mdim(file_out)

# plot variable in object reopened

# ggplot() +
#   geom_stars(data = dcant_3d_read_ncdf %>% 
#                filter(period == "1994 - 2004",
#                       depth == 50)) +
#   facet_grid(MLR_basins~Version_ID_group) +
#   scale_fill_viridis_b(breaks = c(-Inf, seq(0,16,2), Inf),
#                        na.value = "transparent")

20 basinmask

basinmask_all <- basinmask_all %>%
  mutate(
    MLR_basins = case_when(
      MLR_basins == "AIP" ~ "3",
      MLR_basins == "SO_AIP" ~ "3+SO",
      MLR_basins == "SO_5" ~ "5+SO",
      TRUE ~ MLR_basins
    )
  )


basinmask_all %>% 
  ggplot(aes(lon, lat, fill = basin)) +
  geom_raster() +
  facet_wrap(~ MLR_basins)

basinmask_all <- basinmask_all %>%
  filter(MLR_basins != "SO_2")

basinmask_all %>% distinct(MLR_basins, basin)

# A tibble: 23 × 2
   basin        MLR_basins
   <chr>        <chr>     
 1 global       1         
 2 Indo-Pacific 2         
 3 Atlantic     2         
 4 S_Pacific    5         
 5 S_Atlantic   5         
 6 Indian       5         
 7 N_Atlantic   5         
 8 N_Pacific    5         
 9 S_Pacific    5+SO      
10 S_Atlantic   5+SO      
# … with 13 more rows

basinmask_all <- basinmask_all %>% 
  select(lon, lat, basin_mask = MLR_basins, basin_name = basin)

basinmask_all <- basinmask_all %>% 
  group_by(basin_mask) %>% 
  mutate(basin_name = as.numeric(as.factor(basin_name))) %>% 
  ungroup()

# convert from tibble to stars object
basinmask_all_out_stars <- st_as_stars(
  basinmask_all,
  dims = c(
    "lon",
    "lat",
    "basin_mask"
  )
)

# assign coordinate system to horizontal (lon, lat) coordinates
# st_crs(dcant_inv_out_stars) <- "+proj=longlat"
st_crs(basinmask_all_out_stars) <- st_crs(4326)

file_out <- paste0(path_out, "basin_mask_emlr_cstar_mueller_94-04-14.nc")

# write stars object to .nc file
basinmask_all_out_stars %>%
  write_mdim(file_out)

# reopen .nc file with all available stars functions
basinmask_all_read_mdim <-
  read_mdim(file_out)

# plot variable in object reopened

ggplot() +
  geom_stars(data = basinmask_all_read_mdim,
             aes(fill = as.factor(basin_name))) +
  facet_wrap(~ basin_mask) +
  scale_fill_muted(na.value = "transparent")

surface_area_mask <- basinmask_all %>% 
  filter(basin_mask == "1") %>% 
  select(-starts_with("basin")) %>% 
  mutate(surface_area = earth_surf(lat = lat, lon = lon))
  

surface_area_mask_out_stars <- st_as_stars(
  surface_area_mask,
  dims = c(
    "lon",
    "lat"
  )
)

# assign coordinate system to horizontal (lon, lat) coordinates
# st_crs(dcant_inv_out_stars) <- "+proj=longlat"
st_crs(surface_area_mask_out_stars) <- st_crs(4326)

file_out <- paste0(path_out, "surface_area_mask_emlr_cstar_mueller_94-04-14.nc")

# write stars object to .nc file
surface_area_mask_out_stars %>%
  write_mdim(file_out)

21 Map test color palettes

breaks <- c(-Inf, seq(0, 16, 2), Inf)

legend_title <- expression(atop(Delta * C["ant"],
                                (mol ~ m ^ {
                                  -2
                                })))

breaks_n <- length(breaks) - 1

dcant_inv_all_color_test <- dcant_inv_all %>%
  filter(data_source %in% c("obs"),
         period %in% two_decades) %>%
  mutate(dcant_int = cut(dcant,
                         breaks,
                         right = FALSE))

scico_continous_palettes <- c(
  "acton",
  "bamako",
  "batlow",
  "bilbao",
  "buda",
  "davos",
  "devon",
  "grayC",
  "hawaii",
  "imola",
  "lajolla",
  "lapaz",
  "nuuk",
  "oslo",
  "tokyo",
  "turku"
)

p_test_scico <- function(df, i_palette) {
  p_reg <- map +
    geom_raster(data = df,
                aes(lon, lat, fill = dcant_int)) +
    scale_fill_scico_d(
      palette = i_palette,
      drop = FALSE,
      name = legend_title,
      guide = "none"
    ) +
    labs(title = i_palette) +
    theme(axis.text = element_blank(),
          axis.ticks = element_blank()) +
    facet_grid(.~period)
  
  p_rev <- map +
    geom_raster(data = df,
              aes(lon, lat, fill = dcant_int)) +
    scale_fill_scico_d(
      palette = i_palette,
      drop = FALSE,
      name = legend_title,
      direction = -1,
      guide = "none"
    ) +
    labs(title = paste(i_palette, "rev")) +
    theme(axis.text = element_blank(),
          axis.ticks = element_blank()) +
    facet_grid(.~period)
  
  
  p_comb <- p_reg / p_rev
  
  return(p_comb)
}


all_plots_scico <- scico_continous_palettes %>% 
  # head(1) %>% 
  map(~p_test_scico(df = dcant_inv_all_color_test,
                      i_palette = .x))

viridis_continous_palettes <- c(
  "viridis",
  "magma",
  "inferno",
  "plasma"
)

p_test_viridis <- function(df, i_palette) {
  p_reg <- map +
    geom_raster(data = df,
                aes(lon, lat, fill = dcant_int)) +
    scale_fill_viridis_d(
      option = i_palette,
      drop = FALSE,
      name = legend_title,
      guide = "none"
    ) +
    labs(title = i_palette) +
    theme(axis.text = element_blank(),
          axis.ticks = element_blank()) +
    facet_grid(.~period)
  
  
  p_rev <- map +
    geom_raster(data = df,
              aes(lon, lat, fill = dcant_int)) +
    scale_fill_viridis_d(
      option = i_palette,
      drop = FALSE,
      name = legend_title,
      direction = -1,
      guide = "none"
    ) +
    labs(title = paste(i_palette, "rev")) +
    theme(axis.text = element_blank(),
          axis.ticks = element_blank()) +
    facet_grid(.~period)
  
  
  p_comb <- p_reg / p_rev
  
  return(p_comb)
}

all_plots_viridis <- viridis_continous_palettes %>% 
  # head(1) %>% 
  map(~p_test_viridis(df = dcant_inv_all_color_test,
                      i_palette = .x))



library(colorspace)

colorspace_continous_palettes <- c(
  # "Purple - Blue",
  # "Red - Purple",
  # "Blue - Yellow",
  # "Heat",
  # "Heat 2",
  "Rocket"
  # "Mako",
  # "Dark",
  # "Mint",
  # "Mint",
  # "BluGrn",
  # "Teal",
  # "Sunset",
  # "SunsetDark",
  # "ag_Sunset",
  # "YlOrRd",
  # "YlGnBu",
  # "RdPu"
)

p_test_colorspace <- function(df, i_palette) {
  
  i_palette <<- i_palette
  
  p_reg <- map +
    geom_raster(data = df,
                aes(lon, lat, fill = dcant_int)) +
    scale_fill_discrete_sequential(
      palette = i_palette,
      drop = FALSE,
      name = legend_title,
      guide = "none"
    ) +
    labs(title = i_palette) +
    theme(axis.text = element_blank(),
          axis.ticks = element_blank()) +
    facet_grid(.~period)
  
  
  p_rev <- map +
    geom_raster(data = df,
                aes(lon, lat, fill = dcant_int)) +
    scale_fill_discrete_sequential(
      palette = i_palette,
      drop = FALSE,
      name = legend_title,
      rev = FALSE,
      guide = "none"
    ) +
    labs(title = paste(i_palette, "rev")) +
    theme(axis.text = element_blank(),
          axis.ticks = element_blank()) +
    facet_grid(.~period)
  

  p_comb <- p_reg / p_rev
  
  return(p_comb)
  rm(i_palette)
  
}

all_plots_colorspace <- colorspace_continous_palettes %>%
  # head(2) %>%
  map(~p_test_colorspace(df = dcant_inv_all_color_test,
                           i_palette = .x))

all_plots_colorspace

all_plots <- c(all_plots_colorspace, all_plots_viridis, all_plots_scico)

pdf(here::here("output/publication/test_color_scales.pdf"),
    width = 8, height = 4)
all_plots
dev.off()

Session information

sessionInfo()

R version 4.2.2 (2022-10-31)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: openSUSE Leap 15.4

Matrix products: default
BLAS:   /usr/local/R-4.2.2/lib64/R/lib/libRblas.so
LAPACK: /usr/local/R-4.2.2/lib64/R/lib/libRlapack.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] rqdatatable_1.3.1   rquery_1.4.9        wrapr_2.0.9        
 [4] stars_0.6-0         abind_1.4-5         ggpattern_1.0.1    
 [7] ggh4x_0.2.3         khroma_1.9.0        kableExtra_1.3.4   
[10] ggdist_3.2.0        ggalluvial_0.12.3   terra_1.6-41       
[13] sf_1.0-9            rnaturalearth_0.1.0 geomtextpath_0.1.1 
[16] colorspace_2.0-3    marelac_2.1.10      shape_1.4.6        
[19] ggforce_0.4.1       metR_0.13.0         scico_1.3.1        
[22] patchwork_1.1.2     collapse_1.8.9      forcats_0.5.2      
[25] stringr_1.4.1       dplyr_1.0.10        purrr_0.3.5        
[28] readr_2.1.3         tidyr_1.2.1         tibble_3.1.8       
[31] ggplot2_3.4.0       tidyverse_1.3.2     workflowr_1.7.0    

loaded via a namespace (and not attached):
  [1] readxl_1.4.1            backports_1.4.1         systemfonts_1.0.4      
  [4] lwgeom_0.2-10           plyr_1.8.8              splines_4.2.2          
  [7] sp_1.5-1                digest_0.6.30           htmltools_0.5.3        
 [10] fansi_1.0.3             magrittr_2.0.3          checkmate_2.1.0        
 [13] memoise_2.0.1           googlesheets4_1.0.1     tzdb_0.3.0             
 [16] modelr_0.1.10           vroom_1.6.0             svglite_2.1.0          
 [19] timechange_0.1.1        rvest_1.0.3             textshaping_0.3.6      
 [22] haven_2.5.1             xfun_0.35               callr_3.7.3            
 [25] crayon_1.5.2            jsonlite_1.8.3          zoo_1.8-11             
 [28] glue_1.6.2              polyclip_1.10-4         gtable_0.3.1           
 [31] gargle_1.2.1            webshot_0.5.4           distributional_0.3.1   
 [34] scales_1.2.1            DBI_1.1.3               Rcpp_1.0.9             
 [37] isoband_0.2.6           viridisLite_0.4.1       units_0.8-0            
 [40] bit_4.0.5               proxy_0.4-27            httr_1.4.4             
 [43] seacarb_3.3.1           RColorBrewer_1.1-3      ellipsis_0.3.2         
 [46] pkgconfig_2.0.3         farver_2.1.1            sass_0.4.4             
 [49] dbplyr_2.2.1            utf8_1.2.2              here_1.0.1             
 [52] tidyselect_1.2.0        labeling_0.4.2          rlang_1.0.6            
 [55] later_1.3.0             munsell_0.5.0           cellranger_1.1.0       
 [58] tools_4.2.2             cachem_1.0.6            cli_3.4.1              
 [61] generics_0.1.3          broom_1.0.1             evaluate_0.18          
 [64] fastmap_1.1.0           yaml_2.3.6              ragg_1.2.4             
 [67] oce_1.7-10              processx_3.8.0          knitr_1.41             
 [70] bit64_4.0.5             fs_1.5.2                nlme_3.1-160           
 [73] whisker_0.4             xml2_1.3.3              compiler_4.2.2         
 [76] rstudioapi_0.14         e1071_1.7-12            reprex_2.0.2           
 [79] tweenr_2.0.2            bslib_0.4.1             stringi_1.7.8          
 [82] highr_0.9               ps_1.7.2                lattice_0.20-45        
 [85] Matrix_1.5-3            classInt_0.4-8          vctrs_0.5.1            
 [88] pillar_1.8.1            lifecycle_1.0.3         jquerylib_0.1.4        
 [91] gsw_1.1-1               data.table_1.14.6       httpuv_1.6.6           
 [94] R6_2.5.1                promises_1.2.0.1        KernSmooth_2.23-20     
 [97] codetools_0.2-18        MASS_7.3-58.1           assertthat_0.2.1       
[100] rprojroot_2.0.3         withr_2.5.0             SolveSAPHE_2.1.0       
[103] mgcv_1.8-41             parallel_4.2.2          hms_1.1.2              
[106] grid_4.2.2              rnaturalearthdata_0.1.0 class_7.3-20           
[109] rmarkdown_2.18          googledrive_2.0.0       git2r_0.30.1           
[112] getPass_0.2-2           lubridate_1.9.0