Quick start guide on how to use the Argo profile files
Most users are interested in getting the vertical profile data from Argo floats. Vertical profile data are stored in Argo profile files. There is one profile file for every cycle. This quick start guide is designed to help you get started using Argo netCDF files available at the Argo GDACs (click here to learn how to get data from the GDACs). To learn about other ways of accessing Argo data, visit the Argo data products page, the data visualizations page or the software tools page. This guide will cover four main topics:
- Real time and delayed mode files
- Core and BGC files
- How to choose the correct variables to use for each parameter
- How to use quality control flags and errors
Real time profile files
Real time profile files are where Argo data are first made publicly available at the GDACs. Real time files on the GDACs all contain an “R” before the WMO number (e.g. R5900400_001.nc, BR5904179_001.nc). These real time profile files are available within 12 – 24 hours of the float completing its profile.
In the real time profile files, data measured by the float are recorded in the variables <PARAM>. For the list of Argo <PARAM>s, please refer to Reference Table 3 in the Argo Users Manual. For example, the basic vertical temperature and salinity data are stored in the <PARAM> = PRES, TEMP, PSAL.
Each <PARAM> has an associated variable called <PARAM>_QC which records the quality control flags. In the real time profile files, the quality control flags are set by a number of automatic tests which aim to detect gross errors.
Each <PARAM> also has another associated variable called <PARAM>_ADJUSTED. If scientific calibration is applied to the measured parameters in real time, the real time adjusted values will be recorded in <PARAM>_ADJUSTED.
If real time adjusted values are available, the variable DATA_MODE will record ‘A’. If real time adjusted values are not available, the variable DATA_MODE will record ‘R’.
In general these data should be consistent with ocean climatologies. Monthly climatology tests are performed, along with visual comparisons for profiles that fail the tests. However, for scientific applications that are sensitive to small pressure and salinity biases (e.g. calculations of global ocean heat content or mixed layer depth), it is not recommended to use “R” files.
To learn more about real time quality control tests, see the ADMT documentation page for core Argo and BGC-Argo quality control documentation.
Delayed mode files
Delayed mode profile files are the same as the real time profile files, except their file names on the GDACs all contain a “D” before the WMO number (e.g. D5900400_001.nc, BD5904179_001.nc). These profile files contain delayed mode adjusted data, which are recorded in the variable <PARAM>_ADJUSTED. The variable DATA_MODE will record ‘D’. Two other variables are also filled in delayed mode, which are <PARAM>_ADJUSTED_QC and <PARAM>_ADJUSTED_ERROR, which record the delayed mode quality control flags and the delayed mode adjustment uncertainty.
Core Argo delayed mode files are available 1 – 2 years after a profile is taken; sometimes earlier. These have been subjected to detailed scrutiny by oceanographic experts and the adjusted salinity has been estimated by comparison with high quality ship-based CTD data and Argo climatologies using the process described by Wong et al, 2003; Böhme and Send, 2005; Owens and Wong, 2009; Cabanes et al, 2016.
For BGC parameters, delayed mode files can be available within 5 – 6 cycles after deployment. This is because the BGC sensors often return data that are out of calibration, but early adjustment methodologies exist that can significantly improve their accuracy. Additional delayed mode quality control occurs when a longer record of float data is available.
To learn more about delayed mode quality control, read the papers on the methods linked above or see the ADMT documentation page for core Argo and BGC-Argo quality control documentation.
Core and BGC files
How to choose the correct variables to use for each parameter
For each parameter, there are two variables associated with it: a raw version and an adjusted version. The raw version can be found in the “PARAM” variable (e.g. TEMP, PRES, DOXY) and the adjusted version can be found in the “PARAM_ADJUSTED” variable (e.g. TEMP_ADJUSTED, PRES_ADJUSTED, DOXY_ADJUSTED).
Core files
Look at the ‘DATA_MODE’ variable.
If DATA_MODE(i) is ‘R’, there is no adjusted data. Use PARAM(i) only.
If DATA_MODE(i) is ‘D’ or ‘A’, there is adjusted data. Use PARAM_ADJUSTED(i) only.
BGC files
The ‘DATA_MODE’ variable is not enough in BGC files because the various BGC parameters can be adjusted at different times. Instead, look at the ‘PARAMETER_DATA_MODE’ variable to see which parameter has adjusted values available.
If PARAMETER_DATA_MODE(i,j) is ‘R’, there is no adjusted data in the ith profile for the jth parameter. Use PARAM(i,j) only.
If PARAMETER_DATA_MODE(i,j) is ‘D’ or ‘A’, there is adjusted data in the ith profile for the jth parameter. Use PARAM_ADJUSTED(i,j) only.
How to use quality control flags and errors
The Argo profile files provide both quality control (QC) flags and errors to help users decide if they want to use the data or not. There are flags for both the raw and adjusted data. The QC flags are as follows:
QC flag | Meaning | Real time comment | Delayed mode comment |
0 | No QC performed | No QC performed | No QC performed |
1 | Good data | All real time QC tests passed | The adjusted value is statistically consistent and a statistical error estimate is supplied. |
2 | Probably good data | Probably good | |
3 | Bad data that are potentially correctable | Test 15 or Test 16 or Test 17 failed and all other real-time QC tests passed. These data are not to be used without scientific correction. A flag ‘3’ may be assigned by an operator during additional visual QC for bad data that may be corrected in delayed mode. | An adjustment has been applied, but the value may still be bad. |
4 | Bad data | Data have failed one or more of the real-time QC tests, excluding Test 16. A flag ‘4’ may be assigned by an operator during additional visual QC for bad data that are not correctable. | Bad data. Not adjustable. |
5 | Value changed | Value changed | Value changed |
6 | Not currently used | Not currently used | Not currently used |
7 | Not currently used | Not currently used | Not currently used |
8 | Estimated | Estimated value (interpolated, extrapolated or other estimation). | Estimated value (interpolated, extrapolated or other estimation). |
9 | Missing value | Missing value | Missing value |
Many positions for under ice profiles are estimated, leading to JULD_QC and POSITION_QC flags of ‘8’.
Uncertaintly of the adjustments are provided, and are recorded in the <PARAM>_ADJUSTED_ERROR variables. See the Data FAQ page for more information on this error.
So, to get the cleanest and best possible data set, it is advised to use only delayed mode data with a QC flag of 1 and reasonably small errors. If that isn’t important to your study, you can relax the data state, QC flags and error requirements as necessary.