search

NorESMhub / noresm2cmor

A command line tool for cmorizing NorESM output
http://noresmhub.github.io/noresm2cmor/
5 stars 16 forks source link

[ERROR] sign error in rsutaf and rsutafcs #312

Open DirkOlivie opened 2 years ago

DirkOlivie commented 2 years ago

Describe the error There is an error in the fields rsutaf and rsutafcs. They should have the inverse sign from what they currently have.

Way to correct Is there a place where we can share this information for people using NorESM2 data from ESGF? (saying these two fields should be multiplied by -1 to have the correct physical meaning)

At a later state, we can maybe recmorize these two fields, but that can/should possibly wait.

YanchunHe commented 2 years ago

Hi Dirk, does this apply for all datasets for rsutaf and rsutafcs, right?

@monsieuralok should be able to report an errtum at ESGF doc.

DirkOlivie commented 2 years ago

Hi Yanchun, yes, it applies to all datasets.

adagj commented 2 years ago

Hi, I made a new issue, but now I see that this issue already exists so I'll add here and close the other one. I don't agree that -1 should be multiplied to the fields, I think the problem rater is related to that the cmorized AERmon variables rsutaf and rsutcsaf are the difference from rsdt and not the actual variables. E.g. rsutaf (cmorized in noresm) => rsdt - rsutaf

This is different from what is outputted in other models and not described in the netcdf file

rsutaf:standard_name = "toa_outgoing_shortwave_flux" ; rsutaf:long_name = "TOA Outgoing Aerosol-Free Shortwave Radiation" ; rsutaf:comment = "Flux corresponding to rsut resulting from aerosol-free call to radiation, following Ghan (ACP, 2013)" ;

Way to correct: To cmorize the actual field and not the difference from rsdt

YanchunHe commented 1 year ago

Hi @adagj, what is the actually fields name for these two variables rsutaf and rsutcsaf?

Are there some conclusion for the solution?

adagj commented 1 year ago

Hi @YanchunHe , I think the best solution is to retract the files and publish new files. The cmorized rsutaf should just be the outgoing shortwave flux for aerosol free air, and not the net radiation minus the outgoing shortwave flux for aerosol free air.

From the NorESM2 rawdata:

And the total column absorbed solar flux is not the same as TOA outgoing aerosol-free shortwave radiation. I think what should be cmorized is the difference from the net, so you should cmorize FSNT_DRF - FSNT, and the same with rsutafcs -> cmorize FSNTCDRF - FSNTC

But I think we need @DirkOlivie or @oyvindseland to approve that is the correct method :-)

adagj commented 1 year ago

Hi @YanchunHe . The calculation turned out a bit more complicated than first assumed, at least by me. The cmorized variables should be top-of-atmosphere fluxes, and not top-of-model fluxes which I think FSNT_DRF is, even if the long name suggests differently. So the question on how to deal with these variables remain unanswered and I don't know what is the best method. I don't know how to calculate the TOA fluxes from the TOM output we have available at the moment, but maybe @DirkOlivie or @oyvindseland have a good suggestion...

You can also see a relevant discussion started by Dirk here: https://github.com/NorESMhub/NorESM/discussions/418

adagj commented 1 year ago

@DirkOlivie , I checked what variables CESM2 uses to calculate TOA fluxes, but unfortunately the original name is not provided. However, CMCC uses CAM, as far as I understand, and uses different (cmor) calculations than us (however I can not find any aerosol free variables, but I checked clear sky)

YanchunHe commented 1 year ago

Any conclusion here?

YanchunHe commented 1 year ago

Could any one summerize what to do for this?

Retract the data? any fix for the cmorization?

YanchunHe commented 1 year ago

So, are the following expression correct?

rsutaf  = FSNT_DRF-FSNT
rsutcsaf = FSNTCDRF-FSNTC
YanchunHe commented 1 year ago

Please retract all rsutaf and rsutcsaf, @monsieuralok

monsieuralok commented 1 year ago

@YanchunHe should we remove rsutaf and rsutcsaf from all experiments, also from LM and MM and all versions?

YanchunHe commented 1 year ago

@YanchunHe should we remove rsutaf and rsutcsaf from all experiments, also from LM and MM and all versions?

I think so, thanks!

adagj commented 1 year ago

@YanchunHe @monsieuralok @DirkOlivie

Hi, I had a brief discussion with Dirk and we suggest not to retract the data. The data is not wrong, if used correct. But we probably should add a description or an errata and explain how the data should be used (multiplied with -1 and only look at differences from the control simulation). Do you know where we can add such a description?

Best, Ada

monsieuralok commented 1 year ago

@adagj We need to add description in errata and provide list of all affected datasets. @adagj could you add exact description what should we mention? if I am correct you want to say rsutaf = rsutaf -1 and rsutcsaf=rsutcsaf-1 . For control simulations data are correct or ..? or comparing control simulations from other data providers?

adagj commented 1 year ago

Hi, I think it is best if @DirkOlivie write such a description since he knows these data best. I can provide a list of all affected datasets

adagj commented 1 year ago

The cases are:

NorESM2-MM :

and for NorESM2-LM :

monsieuralok commented 6 months ago

@DirkOlivie @YanchunHe Please could you provide description for these datasets only mentioning calculate rsutaf and rsutcsaf like below would be sufficient rsutaf = FSNT_DRF-FSNT rsutcsaf = FSNTCDRF-FSNTC

or it should be rsutaf = (rsutaf)x-1 and rsutcsaf=(rsutcsaf)x-1