R03: add min/max values to each parameter description

[vpaba]

On 09-Feb-2023. the @nvs-vocabs/avtt team decided to add the min/max value of each parameter to the description of each R03 concept.

The min/max value information can be found in the Argo physical parameters list: core-Argo and BGC-Argo, February 3rd 2023 - linked through the ADMT website: http://www.argodatamgt.org/Documentation

An very similar issue had already been opened under R03: https://github.com/nvs-vocabs/R03/issues/5

[gmaze]

Awesome decision ! Like in #56 I think this participates to the strategic objective of being able to create an Argo data model using webAPIs only It will be very useful for 3rd party APIs like argopy

[vpaba]

Thanks @gmaze. Is there a specific structure you would recommend for adding this information to the descriptions (e.g. valid min: 0.5f, valid max: 8.5f)? Or is it irrelevant what structure we use, as long as it's consistent? To note that not all parameters have a valid min/max associated with them in the original table.

[gmaze]

Yes, an attribute with a standard string providing how many digits should be displayed before and after the decimal point looks good. I would be curious to know which parameters have no valid min/max ... And may be the number of digits after the decimal point would further indicates the resolution of the parameter.

[vpaba]

Thanks @gmaze. The original list can be found on the ADMT website: http://www.argodatamgt.org/Documentation, under 'Argo data formats' in the linked Excel file named 'Argo physical parameters list: core-Argo and BGC-Argo, February 3rd 2023'. Columns H and I have valid_min and valid_max values against some of the parameters.

Good point about the digits after the decimal point as an indicator of the resolution - wondering if this is more sensor-dependent rather than an intrinsic attribute of the parameter? If so, we could move this information to the relevant sensor models in R27 maybe?

[gmaze]

I suspect the string format used in this table originates back to Fortran encoding or some legacy on the format, as pointed out by @apswong in #56 I don't know how to precisely interpret something like: "12000.f" for PRES. Is it 12000 before the decimal point and nothing after ? The PRES resolution is "1.f". For TEMP, the resolution is "0.001f" and valid min "-2.5f". The "f" is somehow confusing to me. May be @apswong or @tcarval you can recall what is the "f" for ?

wondering if this is more sensor-dependent rather than an intrinsic attribute of the parameter?

Good point, I guess it's determined by the sensor but can be modified in the encoding of the parameter

For core, all parameters have a "resolution" attribute:

        JULD:resolution = 0. ;
        JULD_LOCATION:resolution = 0. ;
        PRES:resolution = 1.f ;
        PRES_ADJUSTED:resolution = 1.f ;
        PRES_ADJUSTED_ERROR:resolution = 1.f ;
        TEMP:resolution = 0.001f ;
        TEMP_ADJUSTED:resolution = 0.001f ;
        TEMP_ADJUSTED_ERROR:resolution = 0.001f ;
        PSAL:resolution = 0.001f ;
        PSAL_ADJUSTED:resolution = 0.001f ;
        PSAL_ADJUSTED_ERROR:resolution = 0.001f ;

If resolution is added in R03, then the valid min and max values could be interpreted correctly, without ambiguity.

[apswong]

"%f" is a format specifier in C for floating point numbers. These control how the data will appear in standard output (e.g. ncdump, printf), e.g. "%7.1f" means the floating point number will appear in 7 characters, with 1 digit after the decimal point. In the Argo netCDF files, these are listed under a separate specification, e.g. PRES:C_format = %7.1f"

Having the "f" (not "%f") after the min/max specifications means they are floating point numbers (instead of integers). e.g. "12000.f" (not "%12000.f") means the floating point value of 12000; "8.5f" (not "%8.5f") means the floating point value of 8.5. I don't know the history behind using "f" in the min/max specifications. I hope @tcarval can tell us.

R03 should record the min/max values faithfully as they appear in the parameters xlsx, since they are used by the GDAC FileChecker.

The suggestion to add resolution to R03 is a good one, but it should be initiated by @catsch, who is in charge of the parameters xlsx. Once that's agreed by the ADMT and added to the parameters xlsx, then it can be added to R03.

Lastly, these min/max and resolution specifications are more parameter-dependent, but not necessarily sensor-dependent. Hence I would advise not to add them the sensors table R27.

[gmaze]

Thanks @apswong for the explanation !