Last updated: 2023-12-06

Checks: 7 0

Knit directory: bgc_argo_r_argodata/

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(20211008)

The command set.seed(20211008) 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: 83269d5

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 83269d5. 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:    output/

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/load_argo.Rmd) and HTML (docs/load_argo.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
html e99aaa1 ds2n19 2023-11-27 Build site.
Rmd b7df46c ds2n19 2023-11-27 Cluster under surface extreme.
html 5180ed0 ds2n19 2023-11-27 Build site.
Rmd a2cda31 ds2n19 2023-11-27 Cluster under surface extreme.
html 80c16c2 ds2n19 2023-11-15 Build site.
Rmd 3eba518 ds2n19 2023-11-15 Introduction of vertical alignment and cluster analysis to github website.
Rmd e07ffb5 ds2n19 2023-11-11 Updates after code reviewand additional documentation.
Rmd b18136a jens-daniel-mueller 2023-11-09 code review
html 6522bfa ds2n19 2023-11-04 Build site.
Rmd 6780853 ds2n19 2023-11-04 Addition of Chl-a to dataset.
html 093c4f3 ds2n19 2023-10-27 Build site.
Rmd 8fac6bb ds2n19 2023-10-27 maintaining file_id in all datasets
html 93b4545 ds2n19 2023-10-18 Build site.
Rmd 7af09d6 ds2n19 2023-10-18 standard range v climatology, season order resolved and count labels to
Rmd 8edb9f4 ds2n19 2023-10-17 BGC load process aligned to core load. Associated changes to pH and oxygen analysis.
html db6fc8a ds2n19 2023-10-17 Build site.
Rmd ed8e997 ds2n19 2023-10-17 Revised version of BGC load to match core load process
html 0ee3427 ds2n19 2023-10-16 Build site.
Rmd 82d027d ds2n19 2023-10-16 Revised version of BGC load to match core load process
Rmd 3f6800f ds2n19 2023-10-14 Changes to core load to create a simple qc flag summary
Rmd fad0b84 ds2n19 2023-10-04 Changed BGC Argo location folders
html 7b3d8c5 pasqualina-vonlanthendinenna 2022-08-29 Build site.
Rmd 8e81570 pasqualina-vonlanthendinenna 2022-08-29 load and add in core-argo data (1 month)
html bdd516d pasqualina-vonlanthendinenna 2022-05-23 Build site.
Rmd b41e65f pasqualina-vonlanthendinenna 2022-05-23 recreate data in bgc_argo_preprocessed_data
html dfe89d7 jens-daniel-mueller 2022-05-12 Build site.
Rmd f29dafc jens-daniel-mueller 2022-05-11 manual commit

Task

makes use of argodata to load BGC Argo profile related data, stages load index, data and metadata files. Cache files (saved in /nfs/kryo/work/datasets/ungridded/3d/ocean/floats/bgc_argo and dac subdirectory) are used, set option opt_refresh_cache to TRUE to force a refresh – This process takes considerable time. bgc_index data frame is created based on delayed mode files and a supplied date range. The index file is then used to load profile data into bgc_data and associated meta data into bgc_metadata.

Both bgc_data and bgc_metadata are filtered by n_prof == opt_n_prof_sel. Description of n_prof usage is provided at https://argo.ucsd.edu/data/data-faq/version-3-profile-files/ the following is from that page. The main Argo CTD profile is stored in N_PROF=1. All other parameters (including biogeochemical parameters) that are measured with the same vertical sampling scheme and at the same location and time as the main Argo CTD profile are also stored in N_PROF=1.

Both bgc_data and bgc_metadata initially contain a string field file that uniquely identifies the profile and links the two data frames. An additional data from bgc_fileid is created with a unique list file fields along with a numeric file_id field. The file fields in bgc_data and bgc_metadata are then replaced with file_id. An additional consolidated data frame bgc_merge is created to support prior analysis work.

A series of files are created that support prior analysis these are. bgc_merge_pH_qc_1, ph_merge_1x1, ph_surface_1x1, ph_surface_2x2, bgc_merge_temp_qc_1, bgc_merge_temp_AB, bgc_merge_flag_AB and bgc_merge_qc_1. In addition, a data frame bgc_measure_summary is created that is used for load level reporting figures.

Files are written to the BGC preprocessed folder for ongoing analysis.

Dependencies

Cache files - /nfs/kryo/work/datasets/ungridded/3d/ocean/floats/bgc_argo

Outputs (in BGC preprocessed folder)

bgc_index.rds – Copy of the index file that was used in the selection of data and meta data files.

bgc_data.rds – The main BGC profile data.

bgc_metadata.rds – The associated meta data information.

bgc_fileid.rds – A lookup from file_id to file.

The following files are created to maintain previous analysis but no longer needed.

bgc_merge.rd – A file that combines bgc_data and bgc_metadata.

bgc_merge_pH_qc_1.rds

bgc_merge_temp_qc_1.rds

bgc_merge_temp_AB.rds

bgc_merge_flag_AB.rds

bgc_merge_qc_1.rds

ph_merge_1x1.rds

ph_surface_1x1.rds

ph_surface_2x2.rds

Set load options

Determine if files are refreshed from dac or cache directory is used. Are metadata, temperature, salinity and BGC property year files renewed? Are the consolidated all year files created from the individual year files?

# opt_refresh_cache
#   FALSE = do not refresh cache.
#   TRUE  = refresh cache. (any none zero value will force a refresh)
opt_refresh_cache = FALSE

# opt_min_year and opt_max_year
#   year to be refreshed are set by opt_min_year and opt_max_year
opt_min_year = 2013
opt_max_year = 2023

# opt_review_mode
# if set (1) the processing will take place in a sub-directory opt_review_dir and only process 10 days of profiles per year to reduce size
# of output and processing time
opt_review_mode = FALSE
opt_review_dir = "/review_mode"

#if (opt_review_mode) {
#  path_argo_preprocessed <- paste0(path_argo_core, "/preprocessed_data", opt_review_dir)
#}
  
# opt_qc_only
# Avoids reprocessing files and ensures qc summary plots are created from a previous run!
#   FALSE = carry out reprocessing based on options set above and create QC summaries.
#   TRUE  = do NOT reprocessing files and just create QC summaries from previous loads.
opt_qc_only = TRUE

# opt_n_prof_sel
# The selection criteria that is used against n_prof, here set to 1
# Description of n_prof usage is provided at https://argo.ucsd.edu/data/data-faq/version-3-profile-files/ the next two lines are from that page.
#     The main Argo CTD profile is stored in N_PROF=1. All other parameters (including biogeochemical parameters) that are measured 
#     with the same vertical sampling scheme and at the same location and time as the main Argo CTD profile are also stored in N_PROF=1.
opt_n_prof_sel = 1

Set cache directory

Directory where the core-Argo profile files are stored. Either use the cached files or force a refresh from dac (long process)

if (!opt_qc_only) {

  # set cache directory
  argo_set_cache_dir(cache_dir = path_argo)
  
  # check cache directory
  argo_cache_dir()
  
  # check argo mirror
  argo_mirror()
  
  # age argument: age of the cached files to update in hours (Inf means always use the cached file, and -Inf means always download from the server) 
  # ex: max_global_cache_age = 5 updates files that have been in the cache for more than 5 hours, max_global_cache_age = 0.5 updates 
  # files that have been in the cache for more than 30 minutes, etc.
  if (opt_refresh_cache){
    argo_update_global(max_global_cache_age = -Inf)  
    argo_update_data(max_data_cache_age = -Inf)
  } else {
    argo_update_global(max_global_cache_age = Inf)  
    argo_update_data(max_data_cache_age = Inf)
  }
}

Data Load

Builds yearly files for temperature, salinity, BGC properties and metadata that can be consolidated in the next code chunk (consolidate_into_allyears). Load in the synthetic (merged core and bgc) index files (uses the data stored on the ifremer server by default), keeping only delayed-mode data.

File Types

A core-Argo profile contains the CTD sensor parameters (pressure, temperature, salinity) that are measured with the same vertical sampling scheme and at the same location and time. Additional parameters from other sensors are stored in the b-Argo profile files.

A b-Argo profile contains all the parameters from a float, except the core-Argo parameters temperature, pressure, and salinity. A float that performs only CTD measurements does not have a b-Argo file. The vertical level PRES is the simple and unambiguous link between the parameters in the core-Argo and b-Argo files. The same PRES is recorded in the core-Argo and b-Argo files. PRES is the only parameter duplicated in core-Argo and b-Argo profile files.

To facilitate the use of BGC-Argo data, the regional data centers merge each b-Argo file with its corresponding core-Argo file into one synthetic (s-Argo) file. The goal of a simplified s-Argo file is to co-locate as many BGC observations as possible while preserving the character of the sampling pattern, i.e., sample interval, number of samples, and approximate pressure locations. Data come from the single c- and b-Argo files. The synthetic pressure axis is constructed from the BGC sampling levels from each cycle. This means that there is no fixed vertical grid for all floats and all cycles.

The co-location takes different vertical attachments of BGC sensors into account by displacing the pressure location, which is not the case in the c- and b-files. The single-cycle s–file profiles contain all the c-file parameter observations in their original location and resolution.

The adjusted pressure parameter (pres_adjusted) is only available in the core- and s-Argo profile files. The variables profile_pres_qc, pres_adjusted, and pres_adjusted_error, are not duplicated in the b-Argo files.

Data Modes

Delayed-mode data are denoted by data_mode = 'D', and are quality-checked by PIs, who apply any necessary adjustments. For the core CTD data, delayed-mode data is generally available 12 months after the transmission of raw data, because the raw data is usually of good quality. Their delayed-mode assessment involves evaluation of the long-term sensor stability, which typically requires a float record of 12 months. Incorrect QC flag attribution and erroneous raw data not flagged during real-time procedures are corrected in delayed-mode.

Delayed-mode BGC data may be available as early as 5-6 cycles after initial data transmission as the raw data are typically unfit for scientific usage. Adjustments significantly increase the accuracy of these data. In b- and s-Argo profile files, the variable parameter_data_mode indicates the mode of each parameter. Biogeochemical parameters in the same file may receive their delayed-mode adjustments at different times.

Synthetic files info: https://archimer.ifremer.fr/doc/00445/55637/80863.pdf

Argo User Manual: https://archimer.ifremer.fr/doc/00187/29825/86414.pdf

Load Index

Load index file that forms the basis of subsequent sections that load data and metadata

if (!opt_qc_only) {

      # if working in reiew mode only consider first 10 days of opt_max_year
      if (opt_review_mode) {
        bgc_index <- argo_global_synthetic_prof() %>% 
          argo_filter_data_mode(data_mode = 'delayed') %>% 
          argo_filter_date(date_min = paste0(opt_max_year, "-01-01"),
                           date_max = paste0(opt_max_year, "-01-31"))
      } else {
        bgc_index <- argo_global_synthetic_prof() %>% 
          argo_filter_data_mode(data_mode = 'delayed') %>% 
          argo_filter_date(date_min = paste0(opt_min_year, "-01-01"),
                           date_max = paste0(opt_max_year, "-12-31"))
      }

}

Load data

Read in the adjusted bgc and core variables corresponding to the index files downloaded above, with their quality control flags. (can take a while)

if (!opt_qc_only) {

  bgc_data <- argo_prof_levels(
    path = bgc_index,
    vars =
      c(
        'PRES_ADJUSTED',
        'PRES_ADJUSTED_QC',
        'PRES_ADJUSTED_ERROR',
        'PSAL_ADJUSTED',
        'PSAL_ADJUSTED_QC',
        'PSAL_ADJUSTED_ERROR',
        'TEMP_ADJUSTED',
        'TEMP_ADJUSTED_QC',
        'TEMP_ADJUSTED_ERROR',
        'DOXY_ADJUSTED',
        'DOXY_ADJUSTED_QC',
        'DOXY_ADJUSTED_ERROR',
        'NITRATE_ADJUSTED',
        'NITRATE_ADJUSTED_QC',
        'NITRATE_ADJUSTED_ERROR',
        'PH_IN_SITU_TOTAL_ADJUSTED',
        'PH_IN_SITU_TOTAL_ADJUSTED_QC',
        'PH_IN_SITU_TOTAL_ADJUSTED_ERROR',
        'CHLA_ADJUSTED',
        'CHLA_ADJUSTED_QC',
        'CHLA_ADJUSTED_ERROR'
      ),
    quiet = TRUE
  ) 
  # read in the profiles (takes a while)

  # see option section above for rational of why we only want n_prof = 1 profiles.
  # Note as working with symthetic profiles has no impact.
  bgc_data <- bgc_data %>%
  filter(n_prof == opt_n_prof_sel)

}

Read meta data

Read in the corresponding metadata:

if (!opt_qc_only) {

  bgc_metadata <- argo_prof_prof(path = bgc_index)

  # see option section above for rational of why we only want n_prof = 1 profiles
  # Note as working with symthetic profiles has no impact.
  bgc_metadata <- bgc_metadata %>%
  filter(n_prof == opt_n_prof_sel)

  # Select just the columns we are interested in
  bgc_metadata <- bgc_metadata %>%
    select (
      file,
      date,
      latitude,
      longitude,
      platform_number, 
      cycle_number,
      position_qc,
      profile_pres_qc,
      profile_temp_qc,
      profile_psal_qc,
      profile_doxy_qc,
      profile_nitrate_qc,
      profile_ph_in_situ_total_qc,
      profile_chla_qc
    )

}

Join data

Join the metadata and data together into one dataset

if (!opt_qc_only) {

  bgc_merge <- full_join(bgc_data, bgc_metadata)
  
  bgc_merge <- bgc_merge %>%
    rename(lon = longitude,
           lat = latitude) %>%
    mutate(lon = if_else(lon < 20, lon + 360, lon)) %>%
    mutate(
      lat = cut(lat, seq(-90, 90, 1), seq(-89.5, 89.5, 1)),
      lat = as.numeric(as.character(lat)),
      lon = cut(lon, seq(20, 380, 1), seq(20.5, 379.5, 1)),
      lon = as.numeric(as.character(lon))
    ) %>%
    mutate(depth = gsw_z_from_p(pres_adjusted, latitude =  lat) * -1.0,
           .before = pres_adjusted)
  
  # Harmonise metadata
  bgc_metadata <- bgc_metadata %>%
    rename(lon = longitude,
           lat = latitude) %>%
    mutate(lon = if_else(lon < 20, lon + 360, lon)) %>%
    mutate(
      lat = cut(lat, seq(-90, 90, 1), seq(-89.5, 89.5, 1)),
      lat = as.numeric(as.character(lat)),
      lon = cut(lon, seq(20, 380, 1), seq(20.5, 379.5, 1)),
      lon = as.numeric(as.character(lon))
  )
  
  # Create fileid dataframe and link back to merge and metadata
  bgc_fileid <- unique(bgc_merge$file)
  bgc_fileid <- tibble(bgc_fileid)
  bgc_fileid <- bgc_fileid %>% select (file = bgc_fileid)
  bgc_fileid <- tibble::rowid_to_column(bgc_fileid, "file_id")

  # Change metadate, data and merge to have file_id
  bgc_metadata <- full_join(bgc_metadata, bgc_fileid)
  bgc_metadata <- bgc_metadata %>%
    select(-c(file))
  bgc_merge <- full_join(bgc_merge, bgc_fileid)
  bgc_merge <- bgc_merge %>%
    select(-c(file))
  bgc_data <- full_join(bgc_data, bgc_fileid)
  bgc_data <- bgc_data %>%
    select(-c(file))

  # Summary measurement QC data
  bgc_data_qc <- bgc_merge %>%
    select (
      'date',
      'pres_adjusted_qc',
      'psal_adjusted_qc',
      'temp_adjusted_qc',
      'doxy_adjusted_qc',
      'nitrate_adjusted_qc',
      'ph_in_situ_total_adjusted_qc',
      'chla_adjusted_qc'
            ) %>%
    mutate(year = year(date),
           .after = date)
  
  # Create the summary file that is used later for qc analysis
  bgc_measure_summary <- bgc_data_qc %>%
    group_by(
      year,
      measure_order = 1,
      measure = "Pressure",
      measure_qc = pres_adjusted_qc
    ) %>%
    summarise(count_measure = n())
  
  bgc_measure_summary <- rbind(
    bgc_measure_summary,
    bgc_data_qc %>%
      group_by(
        year,
        measure_order = 2,
        measure = "Temperature",
        measure_qc = temp_adjusted_qc
      ) %>%
      summarise(count_measure = n())
  )
  
  bgc_measure_summary <- rbind(
    bgc_measure_summary,
    bgc_data_qc %>%
      group_by(
        year,
        measure_order = 3,
        measure = "Salinity",
        measure_qc = psal_adjusted_qc
      ) %>%
      summarise(count_measure = n())
  )
  
  bgc_measure_summary <- rbind(
    bgc_measure_summary,
    bgc_data_qc %>%
      group_by(
        year,
        measure_order = 4,
        measure = "pH",
        measure_qc = ph_in_situ_total_adjusted_qc
      ) %>%
      summarise(count_measure = n())
  )
  
  bgc_measure_summary <- rbind(
    bgc_measure_summary,
    bgc_data_qc %>%
      group_by(
        year,
        measure_order = 5,
        measure = "Oxygen",
        measure_qc = doxy_adjusted_qc
      ) %>%
      summarise(count_measure = n())
  )
  
  bgc_measure_summary <- rbind(
    bgc_measure_summary,
    bgc_data_qc %>%
      group_by(
        year,
        measure_order = 6,
        measure = "Nitrate",
        measure_qc = nitrate_adjusted_qc
      ) %>%
      summarise(count_measure = n())
  )
  
  bgc_measure_summary <- rbind(
    bgc_measure_summary,
    bgc_data_qc %>%
      group_by(
        year,
        measure_order = 7,
        measure = "Chl-a",
        measure_qc = chla_adjusted_qc
      ) %>%
      summarise(count_measure = n())
  )
  
  rm(bgc_data_qc)
  bgc_measure_summary %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_measure_summary.rds"))
  
}

pH data

All pH data from BGC floats with QC flag 1 (good data) ph_surface: pH data in the top 20 m of the watercolumn with QC flag 1

if (!opt_qc_only) {

  bgc_merge_pH_qc_1 <- bgc_merge %>%
    filter(ph_in_situ_total_adjusted_qc == '1') %>%
    select(
      file_id,
      date,
      lat,
      lon,
      depth,
      psal_adjusted,
      temp_adjusted,
      temp_adjusted_qc,
      ph_in_situ_total_adjusted,
      ph_in_situ_total_adjusted_qc,
      platform_number,
      cycle_number,
      profile_ph_in_situ_total_qc,
      profile_temp_qc
    )
  
  # create a dataframe of full pH data (only good data) with corresponding CTD and metadata, in a 1x1º longitude/latitude grid
  ph_merge_1x1 <- bgc_merge %>%
    select(
      -c(doxy_adjusted:nitrate_adjusted_error),
      -c(profile_doxy_qc, profile_nitrate_qc)
    ) %>%
    filter(ph_in_situ_total_adjusted_qc == '1') %>%
    mutate(year = year(date),
           month = month(date),
           .after = n_prof)
  
  
  # create a dataframe of pH data in the surface ocean (upper 20 m of the watercolumn), in a 1x1º longitude/latitude grid
  ph_surface_1x1 <- ph_merge_1x1 %>%
    filter(between(depth, 0, 20))
  
  # create a dataframe of pH for the surface ocean (upper 20 m of the watercolumn) in a 2x2º longitude/latitude grid
  ph_surface_2x2 <- ph_surface_1x1 %>%
    mutate(
      lat = cut(lat, seq(-90, 90, 2), seq(-89, 89, 2)),
      lat = as.numeric(as.character(lat)),
      lon = cut(lon, seq(20, 380, 2), seq(21, 379, 2)),
      lon = as.numeric(as.character(lon))
    )   # regrid into 2x2º grid
  
}

Temperature data

All temperature data from BGC floats with QC flag 1

if (!opt_qc_only) {

  # Include bgc_merge_temp_qc == 8 as this is as good as 1 and matches the BGC data points
  bgc_merge_temp_qc_1 <- bgc_merge %>% 
    filter(temp_adjusted_qc == '1' | temp_adjusted_qc == '8') %>% 
    select(file_id, date, lat, lon, 
           depth, temp_adjusted,
           platform_number, cycle_number,
           temp_adjusted_qc, ph_in_situ_total_adjusted_qc,
           profile_temp_qc,
           profile_ph_in_situ_total_qc)
}

QC flag A and B temperature data

BGC-temperature data with QC flags A and B, irrespective of whether a corresponding pH measurement exists

if (!opt_qc_only) {
  
  bgc_merge_temp_AB <- bgc_merge %>% 
    filter(profile_temp_qc == 'A' | profile_temp_qc == 'B') %>% 
    filter(temp_adjusted_qc == '1' | temp_adjusted_qc == '8') %>% 
    select(file_id, date, lat, lon,
           depth, temp_adjusted, temp_adjusted_qc,
           platform_number, cycle_number,
           profile_temp_qc)

}

QC flag A and B pH & temperature data

pH and temperature data from floats where both variables have full profiles with QC flag A

if (!opt_qc_only) {

  # create a dataframe with temperature and pH profile flags A and B only 
  # keep only temperature observations where good pH data exists: 
  
  # using only complete profiles, and temperature data where pH measurements exist:
  
  bgc_merge_flag_AB <- bgc_merge %>%
    filter(profile_ph_in_situ_total_qc %in% c('A', 'B'),
           profile_temp_qc %in% c('A', 'B')) %>%
    select(file_id, depth,
           temp_adjusted:temp_adjusted_error,
           ph_in_situ_total_adjusted:ph_in_situ_total_adjusted_error,
           platform_number,
           cycle_number,
           date,
           lat, lon,
           profile_temp_qc,
          profile_ph_in_situ_total_qc) %>%
    filter(!is.na(ph_in_situ_total_adjusted))
  # no NA temperature values
  # 518 340 total observations 
}

QC flag 1 pH & temperature data

pH and temperature data where both variables have QC flags 1

if (!opt_qc_only) {

  # Include bgc_merge_temp_qc == 8 as this is as good as 1 and matches the BGC data points
  bgc_merge_qc_1 <- bgc_merge %>% 
    filter(ph_in_situ_total_adjusted_qc == '1' & (temp_adjusted_qc == '1' | temp_adjusted_qc == '8')) %>% 
    select(file_id, depth,
           temp_adjusted:temp_adjusted_error,
           ph_in_situ_total_adjusted:ph_in_situ_total_adjusted_error,
           platform_number,
           cycle_number,
           date,
           lat, lon,
           profile_temp_qc,
           profile_ph_in_situ_total_qc)
}

Write Argo data to files

if (!opt_qc_only) {
  bgc_index %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_index.rds"))
  
  bgc_data %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_data.rds"))
  
  bgc_merge %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_merge.rds"))
  
  bgc_fileid %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_fileid.rds"))
  
  bgc_merge_pH_qc_1 %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_merge_pH_qc_1.rds"))
  
  bgc_merge_temp_qc_1 %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_merge_temp_qc_1.rds"))
  
  bgc_merge_temp_AB %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_merge_temp_AB.rds"))
  
  bgc_merge_flag_AB %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_merge_flag_AB.rds"))
  
  bgc_merge_qc_1 %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_merge_qc_1.rds"))
  
  ph_merge_1x1 %>%
    write_rds(file = paste0(path_argo_preprocessed, "/ph_merge_1x1.rds"))
  
  bgc_metadata %>%
    write_rds(file = paste0(path_argo_preprocessed, "/bgc_metadata.rds"))
  
  ph_surface_1x1 %>%
    write_rds(file = paste0(path_argo_preprocessed, "/ph_surface_1x1.rds"))
  
  ph_surface_2x2 %>%
    write_rds(file = paste0(path_argo_preprocessed, "/ph_surface_2x2.rds"))
  
}

QC summary

profile QC flags (A-F)

Produce a summary of profile QC flags (A-F)

# Read metadata file and create profile summary table with a count for each year, measurement type and qc option
bgc_metadata <-
  read_rds(file = paste0(path_argo_preprocessed, "/bgc_metadata.rds"))

# Short version of replacing NA across columns
bgc_metadata <-
  bgc_metadata %>%
  mutate(across(starts_with("profile_"), ~ replace_na(., "")))

# bgc_metadata["profile_pres_qc"][is.na(bgc_metadata["profile_pres_qc"])] <- ""
# bgc_metadata["profile_temp_qc"][is.na(bgc_metadata["profile_temp_qc"])] <- ""
# bgc_metadata["profile_psal_qc"][is.na(bgc_metadata["profile_psal_qc"])] <- ""
# bgc_metadata["profile_doxy_qc"][is.na(bgc_metadata["profile_doxy_qc"])] <- ""
# bgc_metadata["profile_nitrate_qc"][is.na(bgc_metadata["profile_nitrate_qc"])] <- ""
# bgc_metadata["profile_ph_in_situ_total_qc"][is.na(bgc_metadata["profile_ph_in_situ_total_qc"])] <- ""
# bgc_metadata["profile_chla_qc"][is.na(bgc_metadata["profile_chla_qc"])] <- ""

# Short version for data wrangling and plotting below
bgc_metadata %>% 
  mutate(year = year(date)) %>% 
  select(year, starts_with("profile_")) %>% 
  pivot_longer(-year) %>% 
  mutate(name = str_remove(name, "profile_"),
         name = str_remove(name, "_qc")) %>% 
  filter(value != "") %>% 
  count(year, name, value) %>% 
  ggplot(aes(year, n, col = value)) +
  geom_path() +
  geom_point() +
  facet_wrap( ~ name) +
  labs(x = 'year',
       y = 'number of profiles',
       col = 'profile QC flag',
       title = 'Count of profile qc flags by year')

Version Author Date
80c16c2 ds2n19 2023-11-15
6522bfa ds2n19 2023-11-04
93b4545 ds2n19 2023-10-18
0ee3427 ds2n19 2023-10-16 
# bgc_profile_summary <- bgc_metadata %>%
#   filter (profile_chla_qc != "") %>%
#   group_by(
#     year = format(date, "%Y"),
#     measure = "Chl-a",
#     measure_order = 7,
#     profile_qc = profile_chla_qc
#   ) %>%
#   summarise(count_profiles = n())
# 
# bgc_profile_summary <- rbind(
#   bgc_profile_summary,
#   bgc_metadata %>%
#     filter (profile_ph_in_situ_total_qc != "") %>%
#     group_by(
#       year = format(date, "%Y"),
#       measure = "pH",
#       measure_order = 4,
#       profile_qc = profile_ph_in_situ_total_qc
#     ) %>%
#     summarise(count_profiles = n())
# )
# 
# bgc_profile_summary <- rbind(
#   bgc_profile_summary,
#   bgc_metadata %>%
#     filter (profile_doxy_qc != "") %>%
#     group_by(
#       year = format(date, "%Y"),
#       measure = "oxygen",
#       measure_order = 5,
#       profile_qc = profile_doxy_qc
#     ) %>%
#     summarise(count_profiles = n())
# )
# 
# # modify data frame to prepare for plotting
# bgc_profile_summary <- ungroup(bgc_profile_summary)
# bgc_profile_summary <-
#   bgc_profile_summary %>% group_by(measure_order)
# bgc_profile_summary <-
#   transform(bgc_profile_summary, year = as.numeric(year))
# 
# year_min <- min(bgc_profile_summary$year)
# year_max <- max(bgc_profile_summary$year)
# facet_label <-
#   as_labeller(c("7" = "Chl-a", "4" = "pH", "5" = "Oxygen"))
# 
# # draw plots for the separate parameters
# bgc_profile_summary %>%
#   ggplot(aes(
#     x = year,
#     y = count_profiles,
#     col = profile_qc,
#     group = profile_qc
#   )) +
#   geom_point() +
#   geom_line() +
#   facet_wrap(~ measure_order, labeller = facet_label) +
#   scale_x_continuous(breaks = seq(year_min, year_max, 2)) +
#   labs(x = 'year',
#        y = 'number of profiles',
#        col = 'profile QC flag',
#        title = 'Count of profile qc flags by year')

measurement QC flags (1-9)

Produce a summary of current measurement QC flags (1-9)

# Read temp and meta_data
bgc_measure_summary <-
  read_rds(file = paste0(path_argo_preprocessed, "/bgc_measure_summary.rds"))
bgc_measure_summary <- ungroup(bgc_measure_summary)

year_min <- min(bgc_measure_summary$year)
year_max <- max(bgc_measure_summary$year)
facet_label <-
  as_labeller(c("7" = "Chl-a", "4" = "pH", "5" = "Oxygen"))

# draw plots for the separate parameters
bgc_measure_summary %>%
  filter(measure_qc != " " & measure_order %in% c(7, 4, 5)) %>%
  ggplot(aes(
    x = year,
    y = count_measure,
    col = measure_qc,
    group = measure_qc
  )) +
  geom_point() +
  geom_line() +
  facet_wrap( ~ measure_order, labeller = facet_label) +
  scale_x_continuous(breaks = seq(year_min, year_max, 2)) +
  labs(x = 'year',
       y = 'number of measures',
       col = 'measure QC flag',
       title = 'Count of measure qc flags by year')

Version Author Date
6522bfa ds2n19 2023-11-04
93b4545 ds2n19 2023-10-18
0ee3427 ds2n19 2023-10-16

Data set description

Argo Columns

The resulting bgc_merge dataframe contains:

  • the file name (file column)

  • the sampling level (n_level column)

  • the number of profiles per file (n_prof column; Each single-cycle synthetic profile has the dimension n_prof = 1).

  • adjusted values for pressure (pres, in dbar), salinity (psal, in psu), temperature (temp, in degrees C), dissolved oxygen (doxy, in µmol kg-1), pH (ph_in_situ_total), and nitrate (nitrate, in µmol kg-1) (parameter_adjusted columns). This column is mandatory, so if no adjustment was performed (i.e. parameter_adjusted does not exist), FillValue is inserted (e.g., temp_adjusted:FillValue = 99999.f). If the raw value did not require adjustment in delayed-mode, then parameter_adjusted = parameter.

  • a quality control flag associated with these adjusted values (parameter_adjusted_qc columns). If an adjusted value does not exist (e.g., temp_adjusted = 99999.f), then FillValue is inserted (e.g., temp_adjusted_qc = " ").

  • an error estimate on the adjustment of the measurement (parameter_adjusted_error columns). If no adjusted value exists (e.g., temp_adjusted = 99999.f), then FillValue is inserted (e.g., temp_adjusted_error = 99999.f)

  • WMO float identifier (platform_number column)

  • name of the project in charge of the float (project_name column)

  • name of principal investigator in charge of the float (pi_name column)

  • float cycle number (cycle_number column; cycle 0 is the launch cycle during which technical data or configuration information is transmitted; cycle 1 is the first complete cycle)

  • descending (D) or ascending (A) profile (direction column). Profile measurements are taken on ascent, occasionally during descent (rarely both).

  • code for the data centre in charge of the float data management (data_centre column)

  • the type of float (platform_type column)

  • firmware version of the float (firmware_version column)

  • instrument type from the WMO code table 1770 (wmo_inst_type column)

  • the date and time at which the measurement was taken, in UTC (date column)

  • a quality control flag for the date and time value (date_qc column)

  • the date and time of the profile location (date_location column)

  • latitude in degrees N (latitude column)

  • longitude in degrees E (longitude column)

  • quality control flag on the position (position_qc column)

  • name of the system in charge of positioning the float locations (positioning_system column)

  • unique number of the mission to which this float belongs (config_mission_number column,

  • a global quality control flag on the profile of the parameter (profile_parameter_qc column; FillValue = ” “)

QC flags

QC flags for values (‘parameter_adjusted_qc’ columns) are between 1 and 8, where:
1 is ‘good’ data,
2 is ‘probably good’ data,
3 is ‘probably bad’ data,
4 is ‘bad’ data,
5 is ‘value changed’,
8 is ‘estimated value’,
9 is ‘missing value’ (data parameter will record FillValue)
(6 and 7 are not used).

Profile QC flags (‘profile_parameter_qc’ columns) are QC codes attributed to the entire profile, and indicate the number of depth levels (in %) where the value is considered to be good data (QC flags of 1, 2, 5, and 8; QC flags of 9 or ” ” are not used in the computation):
‘A’ means 100% of profile levels contain good data,
‘B’ means 75-<100% of profile levels contain good data,
‘C’ means 50-75% of profile levels contain good data,
‘D’ means 25-50% of profile levels contain good data,
‘E’ means >0-25% of profile levels contain good data,
‘F’ means 0% of profile levels contain good data.

Quality Control Tests

There are two levels of Argo data quality control:

  • The first level is the real-time system that performs a set of agreed automatic checks
  • The second level is the delayed-mode quality control system.

Real-time QC tests

  1. Platform identification test

If a float WMO ID cannot be matched to the correct float platform then none of the data will be distributed.

  1. Impossible date test

This test requires that the Julian Day of the float be later than 1st January 1997 and earlier than the current date of the check (in UTC time). If the date of a profile fails this test, the date of the profile should be flagged as bad data (‘4’) and none of the profile data is distributed.

  1. Impossible location test

This test requires that the observation latitude and longitude of a float be sensible, with latitude in the range -90 to 90º, and longitude in the range -180 to 180º.

If either latitude or longitude fails this test, the position is flagged as bad data (‘4’) and none of the profile data is distributed.

  1. Position on land test

This test requires that the observation latitude and longitude be located in an ocean. If a position cannot be located in an ocean, the position is flagged as bad data (‘4’) and none of the profile data is distributed.

  1. Impossible speed test

Drift speeds for floats can be generated given the positions and times of the floats when they are at the sea surface and between profiles. In all cases, we would not expect the drift speed to exceed 3 ms-1. If it does, it means either the positions or times are bad data, or a float is mislabeled. Using the multiple positions and times that are normally available for a float while at the sea surface, it is often possible to isolate the one position or time that is an error.

If an acceptable position and time can be used from the available suite, then the data can be distributed. Otherwise, the position, the time, or both, are flagged as bad data (‘4’) and the profile data is not distributed.

  1. Global range test

This test applies a gross filter on the values of TEMP, PRES, and PSAL. The ranges need to accommodate all of the expected extremes in the ocean.

  • Pressure cannot be less than -5 dbar. If PRES < -5, then PRES_QC = ‘4’, TEMP_QC = ‘4’, and PSAL_QC = ‘4’.
  • Pressure in the range -5 to -2.4 dbar should be considered ‘probably bad’ data. If -5 ≤ PRES ≤ -2.4, then PRES_QC = ‘3’, TEMP_QC = ‘3’, PSAL_QC = ‘3’.
  • Temperature should be in the range -2.5 to 40.0 ºC. Outside of this range TEMP_QC = ‘4’.
  • Salinity should be in the range 2 to 41.0 PSU. Outside of this range, PSAL_QC = ‘4’.
  • DOXY should be in the range -5 to 600 µmol kg-1. Outside of this range, DOXY_QC = '4'.
  • PH_IN_SITU_TOTAL should be in the range 7.3 to 8.5. Outside of this range, PH_IN_SITU_TOTAL_QC = '4'.
  1. Regional range test (CTD data)

This test applies to certain regions of the world where conditions can be further qualified. In this case, specific ranges for observations from the Mediterranean Sea and the Red Sea further restrict what can be accepted as reasonable values.

If a value fails this test, it is flagged as bad data (‘4’) and removed from the initial distribution. If temperature and salinity at the same pressure level both fail this test, both values are flagged as bad data (‘4’) and values for pressure, temperature, and salinity are removed from the distribution.

  1. Pressure increasing test (CTD data)

This test requires that the vertical profile has pressures that are monotonically increasing (assuming the pressure levels are ordered from smallest to largest).

If there is a region of constant pressure, all but the first of the consecutive constant pressure levels is flagged as bad data (‘4’). If there is a region where pressure reverses, all of the pressures in the reversed part of the profile are flagged as bad data (‘4’). All pressures flagged as bad data and associated temperatures and salinities are removed.

  1. Spike test

The difference between sequential measurements, where one measurement is significantly different from adjacent ones, is a spike in both size and gradient. This test does not consider differences in pressure, but assumes a sampling that adequately reproduces changes in temperature and salinity with pressure.

Test value = | V2 - (V3 + V1)/2 | - | (V3 - V1)/2 |

where V2 is the measurement being tested, and V1 and V3 are the values above and below.

Temperature: the V2 value is flagged when:

  • the test value exceeds 6.0 ºC for pressures less than 500 dbar, or
  • the test value exceeds 2.0 ºC for pressures equal to or greater than 500 dbar.

Salinity: the V2 value is flagged when:

  • the test value exceeds 0.9 PSU for pressures less than 500 dbar, or
  • the test value exceeds 0.3 PSU for pressures equal to or greater than 500 dbar.

DOXY: the V2 value is flagged when:

  • the test value exceeds 50 µmol kg-1 for pressures less than 500 dbar, or
  • the test value exceeds 25 µmol kg-1 for pressures equal to or greater than 500 dbar.

For pH:

Test value 2 = | V2 - median(V0, V1, V2, V3, V4) |

where the test value represents the anomaly of the observed pH from the median of the surrounding data. A pH data point is considered a spike and flagged as bad (‘4’) if Test Value 2 > 0.04pH

If the value V2 fails this test, it is flagged as bad data (‘4’) and is removed. If temperature and salinity both fail this test, both values are flagged as bad data (‘4’) and values for temperature, salinity and pressure are removed.

  1. Obsolete
  2. Gradient test (DOXY only)

This test is failed when the difference between vertically adjacent measurements is too steep. The test does not consider changes in depth, but assumes a sampling that adequately reproduces changes in DOXY with depth

Test value = | V2 - (V3 + V1)/2 |

where V2 is the value being tested as a spike, and V1 and V3 are the values above and below.

For DOXY, V2 is flagged when:

  • the test value exceeds 50 µmol kg-1 for pressures less than 500 dbar, or

  • the test value exceeds 25 µmol kg-1 for pressures equal to or greater than 500 dbar

  1. Digit rollover test (CTD data)

Only so many bits are allowed to store temperature and salinity values in a profiling float. This range is not always large enough to accommodate conditions which are encountered in the ocean. When the range is exceeded, stored values rollover to the lower end of the range. This rollover should be detected and compensated for when profiles are constructed from the data stream of the float. This test is used to make sure the rollover is properly detected.

  • Temperature difference between adjacent pressures > 10 ºC
  • Salinity difference between adjacent pressures > 5 PSU

If a value fails this test, it is flagged as bad data (‘4’) and removed from the initial distribution. If temperature and salinity at the same pressure level both fail this test, both values are flagged as bad data (‘4’) and values for pressure, temperature, and salinity are removed from the distribution.

  1. Stuck value test

This test looks for CTD and BGC measurements in the same profile being identical.

If this occurs, all of the values affected parameter are flagged as bad data (‘4’) and removed from the distribution. If both temperature and salinity are affected, then all observed values from the profile are flagged as bad data (‘4’).

  1. Density inversion test (CTD data)

This test compares potential density between valid measurements in a profile in both directions (i.e., from top to bottom, and from bottom to top). Values of temperature and salinity at the same pressure level Pi are used to compute potential density ρi ( or σi = ρi - 1000) kg m-3, referenced to the mid-point between Pi and the next valid pressure level. A threshold of 0.03 kg m-3 is allowed for small density inversions.

From top to bottom, if the potential density calculated at the greater pressure Pi+1 is less than that calculated at the lesser pressure Pi by more than 0.03 kg m-3, both the temperature and salinity values at pressure Pi are flagged as bad data (‘4’). From bottom to top, if the potential density calculated at the lesser pressure Pi-1 is greater than that calculated at the greater pressure Pi by more than 0.03 kg m-3, both the temperature and salinity values at pressure Pi are flagged as bad data (‘4’). Bad temperature and salinity are removed from the distribution.

  1. Grey list test

This test is implemented as a mechanism for data assembly centers (DACs) to flag, in real-time, sensors that are potentially not working correctly. Each DAC manages a grey list and sends it to the GDACs. The merged grey list from all DACs is available from the GDACs.

Naming convention: xxx_greylist.csv (where xxx is the DAC name, e.g., aoml_greylist.csv, coriolis_greylist.csv, etc).

Columns: PLATFORM, PARAMETER, START_DATE, END_DATE, QC, COMMENT, DAC

The decision to insert a float parameter in the grey list comes from the PI or the delayed-mode operator. A float parameter should be put in the grey list when the sensor drift is too big to be adjusted in real-time, or when the sensor is judged to be potentially not working correctly.

The grey list concerns only real-time files. When an anomalous float is dead and the offending parameter has been adjusted in delayed-mode, it is removed from the grey list. When an anomalous float is active and the offending parameter has been partially adjusted in delayed-mode, it will remain on the grey list if real-time adjustment is not adequate.

Grey-listed parameters are flagged as probably good (‘2’), probably bad (‘3’) or bad (‘4’) data, as determined by the PI or the delayed-mode operator.

  1. Gross salinity or temperature sensor drift test (CTD)

This test is implemented to detect a sudden and significant sensor drift. It calculates the average temperature and salinity from the deepest 100 dbar of a profile and the previous good profile. Only measurements with good QC are used.

For salinity, if the difference between the two average values is more than 0.5 PSU, then all the salinity values of the profile are flagged as probably bad data (‘3’). For temperature, if the difference between the two average values is more than 1 ºC, then all the temperature values from the profile are flagged as probably bad data (‘3’).

  1. Visual QC test

This is subjective visual inspection of float measurements by an operator. This test is not mandatory before real-time data distribution.

  1. Frozen profile test (CTD data)

This test is used to detect a float that produces the same profile (with very small deviations) over and over again. Typically the differences between two profiles are of the order of 0.001 PSU for salinity and of the order 0.01 ºC for temperature.

  1. Derive temperature and salinity profiles by averaging the original profiles to get mean values for each profile in 50 dbar slabs (T_prof, T_previous_prof, S_prof, S_previous_prof). This is necessary because the floats do not sample at the same level in each profile.

  2. Obtain absolute values of the difference between the averaged temperature and salinity profiles as follows:

  • deltaT = abs(T_prof - T_previous_prof)
  • deltaS = abs(S_prof - S_previous_prof)
  1. Find the maximum, minimum, and mean of the absolute values of the averaged differences between profiles for temperature and salinity:
  • mean(deltaT), max(deltaT), min(deltaT)
  • mean(deltaS), max(deltaS), min(deltaS)
  1. To fail the test, require that:
  • max(deltaT) < 0.3 ºC
  • min(deltaT) < 0.001 ºC
  • mean(deltaT) < 0.02 ºC
  • max(deltaS) < 0.3 PSU
  • min(deltaS) < 0.001 PSU
  • mean(deltaS) < 0.004 PSU

If a profile fails this test, all measurements from this profile are flagged as bad data (‘4’). If a float fails this test over 5 consecutive cycles, it is inserted in the grey list.

  1. Deepest pressure test

This test requires that a profile has pressures that are not greater than CONFIG_ProfilePressure_dbar plus 10%. The value of CONFIG_ProfilePressure_dbar is in the meta.nc file of the float.

If there is a region of incorrect pressures, those pressures and their corresponding temperature and salinity measurements are flagged as bad data (‘4’). Pressures flagged as bad data and their associated measurements should be removed from distribution.

  1. MEDian with a distance (MEDD) test (CTD)

This test is a set of algorithms based on three main steps:

  • First, the computation of a sliding median with some customizations
  • Then, limits are computed that are at relative 2-dimension distance d from the median
  • Finally, these limits are also computed for the density profile. There is a spike if both the density profile measurement profile are out of limits. If there is no conductivity sensor, then the spikes in temperature are evaluated using a bigger d value.

Temperature and salinity values that fail this test are flagged as bad data (‘4’).

  1. pH-specific real-time QC test (PH_IN_SITU_TOTAL only)

Currently, there is no pH-specific QC test. If one is established, it will be reported with the number ‘56’.

Real-time pH values which pass the real-time QC tests are assigned QC flags of ‘3’. The Argo goals for research-quality data require that pH values be adjusted to receive a quality flag of ‘1’.

  1. DOXY-specific real-time QC test (DOXY only)

Real-time unadjusted DOXYvalues receive QC flags of ‘3’. This is because the majority of oxygen sensors deployed on BGC Argo profiling floats are Aanderaa optodes that suffer from pre-deployment storage drift that can reduce accuracy by up to 20% or more. Because this is a known bias that affects the majority of oxygen sensors within the array, and because it can be corrected, DOXY_QC is set to ‘3’.

  1. Nitrate-specific real-time QC test (NITRATE only)

Not yet available

Test application order on vertical profiles

The Argo real-time QC tests on CTD data (temperature, salinity, pressure) are performed in the order described in the following table.

A CTD measurement with a QC flag ‘4’ is ignored by other QC tests. A measurement with QC flag ‘2’ or ‘3’ is tested by other QC tests.

A DOXY measurement with a QC flag ‘4’ or ‘3’ is ignored by other QC tests.

Note that the Test Number is different from the Application Order. The Test Number (n) is a number assigned permanently to each QC test. It is used to fill HISTORY_QCTEST in the Argo profile files. Therefore, each Test Number is uniquely associated to a QC test, and is never replaced, changed, or duplicated.

Each real-time QC test has a unique Binary ID (2n) of the unique Test Number (n) is used to record QC tests performed and failed in the variable HISTORY_QCTEST.

The QC flag assigned by a test cannot override a higher value assigned by a previous QC test.

e.g.: a QC flag ‘4’ (bad data) set by the Grey List Test cannot be decreased to QC flag ‘3’ (bad data that are potentially correctable) set by the Gross Salinity or Temperature Sensor Drift Test.

Application Order for CTD parameters Test Number (n) Binary ID (2n) Test Name
1 1 2 Platform Identification Test
2 2 4 Impossible Date Test
3 3 8 Impossible Location Test
4 4 16 Position on Land Test
5 5 32 Impossible Speed Test
6 15 32768 Grey List Test
7 19 524288 Deepest Pressure Test
8 6 64 Global Range Test
9 7 128 Regional Range Test
10 8 256 Pressure Increasing Test
11 9 512 Spike Test
12 25 33554432 MEDD Test
13 12 4096 Digit Rollover Test
14 13 8192 Stuck Value Test
15 14 16384 Density Inversion Test
16 16 65536 Gross Salinity or Temperature Sensor Drift Test
17 18 261144 Frozen Profile Test
18 17 131072 Visual QC Test

The real-time tests for BGC parameters are performed in the order described in the following table:

Application order for BGC parameters Test Number (n) Binary ID (2n) Test Name
1 19 524288 Deepest Pressure Test
2 1 2 Platform Identification Test
3 2 4 Impossible Date Test
4 3 8 Impossible Location Test
5 4 16 Position on Land Test
6 5 32 Impossible Speed Test
7 6 64 Global Range Test
8 7 128 Regional Range Test
9 9 512 Spike Test
10 11 2048 Gradient Test
11 12 4096 Digit Rollover Test
12 13 8192 Stuck Value Test
13 15 32768 Grey List Test
14 16 65536 Gross Temperature Sensor Drift Test (only for TEMP_DOXY)
15 18 261144 Frozen Profile Test
16 BGC parameter-specific tests
17 17 131072 Visual QC Test

Delayed-Mode Quality Control

The QC flags determined in delayed-mode replace those assigned in real-time because some bad data cannot be detected by the real-time tests, and some good data can be identified wrongly as bad by the real-time tests.

For vertical profile data, delayed-mode operators examine them for pointwise errors (such as spikes and jumps) and flag them appropriately. If an error is identified, both PARAM_QC and PARAM_ADJUSTED_QC record ‘4’. Conversely, if good data have wrongly been identified as bad by the real-time tests, then PARAM_QC and PARAM_ADJUSTED_QC record ‘1’.

In SD-files, the variables PROFILE_PARAMETER_QC, PARAMETER_ADJUSTED, PARAMETER_ADJUSTED_QC, and PARAMETER_ADJUSTED_ERROR are compulsory. If no adjustment in delayed-mode is necessary and if the flag is deemed assigned correctly, then PARAM_ADJUSTED = PARAMETER, PARAM_ADJUSTED_QC = PARAM_QC, and PARAM_ADJUSTED_ERROR is provided by the PI.

If no delayed-mode adjustment was performed, then PARAM_ADJUSTED = 99999.f, PARAM_ADJUSTED_QC = " ", PARAM_ADJUSTED_ERROR = 99999.f and PROFILE_PARAMETER_QC = " ".

If values are deemed unadjustable in delayed-mode, then PARAM_ADJUSTED_QC = '4', and PARAM_ADJUSTED = 99999.f and PARAM_ADJUSTED_ERROR = 99999.f.

The variable PROFILE_PARAMETER_QC is recomputed when PARAMETER_ADJUSTED_QC becomes available.

Dates

Delayed-mode operators check that the dates in the profile are in chronological order. Erroneous or missing dates are replaced with another telemetered value if available, or replaced with interpolated values and marked DATE_QC = '8'.

Location

Profile positions LONGITUDE, LATITUDE are checked for outliers. Erroneous or missing dates are replaced with another telemetered value if available, or replaced with interpolated values and marked POSITION_QC = '8'.

Pressure, Temperature, Salinity

Delayed-mode quality control of PRES and TEMP is done by subjective assessment of vertical profile plots of TEMP vs PRES and PSAL vs PRES and PRES vs TEMP and PSAL vs TEMP. This assessment is done in relation to measurements from the same float, as well as in relation to nearby floats and historical data. This assessment aims to identify: (a) erroneous data points that cannot be detected by real-time QC tests, and (b) vertical profiles that have the wrong shape.

Bad PRES data points identified by visual inspection from delayed-mode analysts are recorded with PRES_ADJUSTED_QC = '4' and PRES_QC = '4'. For these bad data points, TEMP_ADJUSTED_QC, TEMP_QC, PSAL_ADJUSTED_QC, and PSAL_QC are also set to ‘4’.

Bad TEMP data points are recorded with TEMP_ADJUSTED_QC = '4' and TEMP_QC = '4'. TEMP_ADJUSTED, TEMP_ADJUSTED_QC, TEMP_ADJUSTED_ERROR are filled even when the data is good and no adjustment is needed. In these cases, TEMP_ADJUSTED_ERROR can be the manufacturer’s quoted accuracy at deployment, which is 0.002 ºC.

Delayed-mode quality control of PSAL is done by checking for sensor offsets and drifts, as well as other instrument errors. Float salinity values that are considered adjustable in delayed-mode are compiled into time-series. Sufficiently long time-series are compared with statistical recommendations and uncertainties to check for sensor drift and offset.

After assessing all available information, the PI records PSAL_ADJUSTED, PSAL_ADJUSTED_QC, and PSAL_ADJUSTED_ERROR. Salinity data considered bad and unadjustable in delayed-mode are given PSAL_ADJUSTED_QC = '4', and PSAL_ADJUSTED and PSAL_ADJUSTED_ERROR are set to FillValue.

Oxygen

Raw DOXY values are adjusted in delayed-mode to account for sensor drift and bias. The errors associated with this calibration are recorded in DOXY_ADJUSTED_ERROR in µmol kg-1.

When DOXY for the whole profile is bad and cannot be adjusted, then DOXY_ADJUSTED = 99999.f, DOXY_ADJUSTED_ERROR = 99999.f, and DOXY_ADJUSTED_QC = '4'. The calibration information is recorded as SCIENTIFIC_CALIB_EQUATION = 'none', SCIENTIFIC_CALIB_EQUATION = 'none', and SCIENTIFIC_CALIB_COMMENT = 'Bad data; not adjustable'.

pH

The pH adjustment process depends on having an accurate model for pH below 1000 m, where temporal and spatial variability is minimal. pH values are adjusted using Multiple Linear Regression (MLR) methods, Linearly Interpolated Regression equations, and a neural network prediction system known as CANYON. The expected error in float pH measurements is derived from the uncertainty in the reference data as well as sensor uncertainties.

The empirical algorithms used in the adjustment process for pH are:

the MLR method of Williams et al. (2016)

the LIR method of Carter et al. (2018)

the CANYON method of Sauzede et al. (2017)

The method used for adjustment is recorded in SCIENTIFIC_CALIB_EQUATION, SCIENTIFIC_CALIB_COEFFICIENT, and SCIENTIFIC_CALIB_COMMENT

Quality control manuals

CTD data quality control: https://archimer.ifremer.fr/doc/00228/33951/32470.pdf (http://dx.doi.org/10.13155/33951)

Oxygen data quality control: https://archimer.ifremer.fr/doc/00354/46542/82301.pdf (http://dx.doi.org/10.13155/46542)

pH data quality control: https://archimer.ifremer.fr/doc/00460/57195/61336.pdf (https://doi.org/10.13155/57195)

BGC data quality control: https://archimer.ifremer.fr/doc/00298/40879/42267.pdf (http://dx.doi.org/10.13155/40879)


sessionInfo()
R version 4.2.2 (2022-10-31)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: openSUSE Leap 15.5

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] ggOceanMaps_1.3.4 ggspatial_1.1.7   oce_1.7-10        gsw_1.1-1        
 [5] sf_1.0-9          lubridate_1.9.0   timechange_0.1.1  argodata_0.1.0   
 [9] forcats_0.5.2     stringr_1.5.0     dplyr_1.1.3       purrr_1.0.2      
[13] readr_2.1.3       tidyr_1.3.0       tibble_3.2.1      ggplot2_3.4.4    
[17] tidyverse_1.3.2  

loaded via a namespace (and not attached):
 [1] fs_1.5.2            httr_1.4.4          rprojroot_2.0.3    
 [4] tools_4.2.2         backports_1.4.1     bslib_0.4.1        
 [7] utf8_1.2.2          R6_2.5.1            KernSmooth_2.23-20 
[10] rgeos_0.5-9         DBI_1.1.3           colorspace_2.0-3   
[13] raster_3.6-11       withr_2.5.0         sp_1.5-1           
[16] tidyselect_1.2.0    compiler_4.2.2      git2r_0.30.1       
[19] cli_3.6.1           rvest_1.0.3         RNetCDF_2.6-1      
[22] xml2_1.3.3          labeling_0.4.2      sass_0.4.4         
[25] scales_1.2.1        classInt_0.4-8      proxy_0.4-27       
[28] digest_0.6.30       rmarkdown_2.18      pkgconfig_2.0.3    
[31] htmltools_0.5.3     highr_0.9           dbplyr_2.2.1       
[34] fastmap_1.1.0       rlang_1.1.1         readxl_1.4.1       
[37] rstudioapi_0.15.0   farver_2.1.1        jquerylib_0.1.4    
[40] generics_0.1.3      jsonlite_1.8.3      googlesheets4_1.0.1
[43] magrittr_2.0.3      Rcpp_1.0.10         munsell_0.5.0      
[46] fansi_1.0.3         lifecycle_1.0.3     terra_1.7-39       
[49] stringi_1.7.8       whisker_0.4         yaml_2.3.6         
[52] grid_4.2.2          parallel_4.2.2      promises_1.2.0.1   
[55] crayon_1.5.2        lattice_0.20-45     haven_2.5.1        
[58] hms_1.1.2           knitr_1.41          pillar_1.9.0       
[61] codetools_0.2-18    reprex_2.0.2        glue_1.6.2         
[64] evaluate_0.18       modelr_0.1.10       vctrs_0.6.4        
[67] tzdb_0.3.0          httpuv_1.6.6        cellranger_1.1.0   
[70] gtable_0.3.1        assertthat_0.2.1    cachem_1.0.6       
[73] xfun_0.35           broom_1.0.5         e1071_1.7-12       
[76] later_1.3.0         class_7.3-20        googledrive_2.0.0  
[79] gargle_1.2.1        workflowr_1.7.0     units_0.8-0        
[82] ellipsis_0.3.2