Deal with salinity unit variants better about cmor

[doutriaux1]

thanks @durack1 I was thinking about this last night. Glad to open the issue before me.

[durack1]

We'll have to get a blessing from @taylor13 (if in fact Karl is actually a contributor on here..)

[taylor13]

one option would be to get the udunits folks to add it, then we wouldn't
have to do anything. Right?
Karl

On 11/20/14, 9:48 AM, Paul J. Durack wrote:

We'll have to get a blessing from @taylor13
https://github.com/taylor13 (if in fact Karl is actually a
contributor on here..)

—
Reply to this email directly or view it on GitHub
#12 (comment).

[durack1]

@taylor13 I'm one step ahead of you Unidata/UDUNITS-2#27

[doutriaux1]

yes we would still need to update the minimum requirements 😉

[durack1]

Looks like we'll have to go it alone on this one: Unidata/UDUNITS-2#27 (comment)

As we're not adding salinities, but just rewriting these in CMOR we won't hit the issue that was raised in the comment above.

[doutriaux1]

@durack1 it is in metrics will copycat what we did there in here at the C level.

[durack1]

@doutriaux1 I will need to test the metrics functionality.. We can talk more about metrics later in the week..

[dnadeau4]

@durack1 ping!!

[durack1]

@dnadeau4 ping indeed.. @doutriaux1 what is/was the status of the salinity fudges for CMOR2 is this something that can be pulled forward into CMOR3? I ask this as the inclusion of salinity units in UDUNITS was blocked

[dnadeau4]

@durack1 should we close this issue?

[doutriaux1]

@dnadeau4 no we should add all the units asked for by @durack1 in a similar fashion as what we did for psu

[durack1]

@dnadeau4, @doutriaux1 created the ability to add salinity "units" in CMOR, I had tried to get these into udunits upstream, but was blocked so instead wanted to catch this in the CMOR software. Has the salinity units work of @doutriaux1 been included in CMOR 3.2.x?

[dnadeau4]

What is the salinity works of @doutriaux1? I don't think it is in CMOR.

[durack1]

@doutriaux1 was that units stuff in a branch?

[dnadeau4]

@doutriaux1 where is you salinity code?

[dnadeau4]

PCMDI/pcmdi_metrics#477

[durack1]

This is the more directly relevant code block

import unidata
# Created scaled units for dimless
unidata.addScaledUnit("psu", .001, "dimless")
unidata.addScaledUnit("PSS-78", .001, "dimless")
unidata.addScaledUnit("Practical Salinity Scale 78", .001, "dimless")

And what we have in the current CMIP6_Omon.json table

"sos": {
   "modeling_realm": "ocean", 
   "standard_name": "sea_surface_salinity", 
   "units": "0.001", 
   "cell_methods": "time: mean", 
   "cell_measures": "area: areacello", 
   "long_name": "Sea Surface Salinity", 
   "comment": "", 
   "dimensions": "longitude latitude time", 
   "out_name": "sos", 
   "type": "real", 
   "positive": "", 
   "valid_min": "", 
   "valid_max": "", 
   "ok_min_mean_abs": "", 
   "ok_max_mean_abs": ""
}

So we may need to augment the above to deal with the "units" 0.001

[dnadeau4]

@durack1 I was able to create psu, PSS-78, Practical Salinity Scale 78. Here is the output:

	float sos(time, lat, lon) ;
		sos:standard_name = "sea_surface_salinity" ;
		sos:long_name = "Sea Surface Salinity" ;
		sos:units = "0.001" ;
		sos:original_units = "psu" ;
		sos:history = "2017-07-13T22:16:05Z altered by CMOR: Converted units from \'psu\' to \'0.001\'. 2017-07-13T22:16:05Z altered by CMOR: replaced missing value flag (0) with standard missing value (1e+20)." ;
		sos:cell_methods = "area: mean where sea time: mean" ;
		sos:cell_measures = "area: areacello" ;
		sos:missing_value = 1.e+20f ;
		sos:_FillValue = 1.e+20f ;

[durack1]

@dnadeau4 excellent, PSS-78 should be the units we aim for (as it is the correct notation).. not sure what CMIP6 specifies

[taylor13]

for CMIP6 model output the correct units are 0.001

[dnadeau4]

Actually udunits does not allow characters - and 0-9. The best I can do is "pss", which will also validate "pss-78".

Now, CMOR can convert PSS-78 to 0.001 which is what we want for CMIP6. This is good since you have the conversion in the history and the original_unis attributes.

If you want PSS-78 you need to talk to WCRP. I cannot change the tables.

[durack1]

@dnadeau4 what PR was this closed by? Actually found it, see #203