[Questions] potential_evapotranspiration with FAO providing net radiation directly

[saschahofmann]

Setup Information

• Xclim version: 0.51.0

Context

I am trying to compute potential_evapotranspiration with the FAO_PM98 method but I don't have the 4 radiation parts separated instead I have ssr which I think is the net radiation already. I think I can hack my way around it by providing, rsds=0 and the rest as 0s but I was wondering whether it would make sense to either provide the 4 components or the net radiation directly or to extract the net_radiation computation and have a single argument for it to the pet function.

I have an additional caveat that I only have monthly climatologies for my radiation but I guess that's a completely different problem.

Steps To Reproduce

$ pip install foo --bar

[coxipi]

I think it's reasonable, other PET formulas also use ssr. It will make the function even messier, but it's alright. I will open a PR.

I only have monthly climatologies for my radiation

Yes, the FAO-PM98 is an empirical formula, in xclim it's currently hard coded for daily timeseries. I think it would only be a matter of computing a different

G = 0  # Daily soil heat flux density [MJ m-2 d-1]

the monthly one will depend on some variables.

[saschahofmann]

I now have the hourly era5 data for net solar radiation (ssr) and net thermal radiation (str). My plan is to provide them as rsus and rlus and the other two parts as 0 but I am not sure whether I should do sum or mean aggregation for the daily values?

[coxipi]

Hmm wait, ssr is Surface Solar Radiation? So it's the net shortwave radiation? Do you know a variable name for net radiation?

EDIT: Ok, according to: https://cfconventions.org/Data/cf-standard-names/current/build/cf-standard-name-table.html,
rss is surface_net_downward_shortwave_flux

ssr is used in ERA5 for "surface_net_downward_shortwave_flux" as well.

[saschahofmann]

Yeah. I am just checking whether something like this is available in ERA5 but they only have have longwave (thermal) in the MARS catalogue

[coxipi]

Monthly ERA5-Land has: rsds, rlds, rss, rls

https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-land-monthly-means?tab=form

[saschahofmann]

Yeah. I am also reading through the Allen 98 paper, linked in the docs and learning a lot. If I understand correctly, the solar flux arrives in short-waves hence only that is available, it will be transformed to thermal/long-wave. So I guess with these the net radiation should be Surface net short-wave (solar) radiation - Surface net long-wave (thermal) radiation, assuming that the thermal radiation represents an energy loss (=upwards flux) ?

CDS servers seem to be pretty busy right now but I am going to download some of these datasets and test my hypotheses.

[saschahofmann]

Another question, above you're suggesting to use monthly era5 land data in that case I would need to resample my other datasets to monthly as well right?

[coxipi]

Sorry, I didn't see your questions until now.

1. Short == solar; long == thermal. They are used as synonyms.

I guess with these the net radiation should be Surface net short-wave (solar) radiation - Surface net long-wave (thermal) radiation

2. No I don't think that's right. Net>0 if it points downwards et Net<0 if it point up. I don't think there's a difference between shortwave/longwave on that matter, but you should double check to be sure.
   Net radiation = Net shortwaves + Net longwaves = (down_shortwave - up_shortwave) + (down_longwave - up_longwave)

above you're suggesting to use monthly era5 land

I think I was just working with your constraint: "I have an additional caveat that I only have monthly climatologies for my radiation." so I was searching an example similar to yours. If you can find daily radiations that suit your needs, that's good. But, indeed, if you only find monthly radiations, you should work with variables that are all monthly in the PET formula.

[saschahofmann]

Stumbled over this repo https://github.com/woodcrafty/PyETo that has great implementations of the pet functions a little closer to the paper.