swrad and lwrad not reading properly in ROMS

[fagundesmo]

Hi all,

I am using BULK FLUX and for some reason the way it's showing on ROMS print is not what I am forcing with (see attached lwrad.ps and swrad.ps). The same way I interpolated vwind, uwind, Tair, and Pair I interpolated swrad and lwrad. And as you can see from my output.txt my swrad and lwrad seems off. I got my atmospheric forcing from ERA5 and the values are in J/m² so I divided by 6*3600 to convert to W/m². I also have tried to use COAWST, I have tried to interpolate and multiply by my mask_rho, I have checked most of the Bulk flux problems in the forum but so far I have got nothing. So I come to you to ask for a help or to tell me with a different perspective what I am doing wrong that is making ROMS reading my data like that.

Thank you in advance,

Matheus

[ymamoutos]

Greetings,

Are you using 6hr timestep ERA5 data? if yes your conversion from J/m^2 to W/m^2 (1/(6*3600))
is valid. If you are using a different timestep then you must change it accordingly i.e 1hr 1/3600,
3hr 1/(3*3600) etc. Also are you downloading from ECMWF site directly or from CDS
(https://cds.climate.copernicus.eu)? If the source is CDS radiations (shortwave and
longwave) and fluxes (rainfall) fields with 1hr timestep are already accumulated.

Giannis

[fagundesmo]

Hi Giannis,

Thank you a lot for your input. I am indeed getting the data from (https://cds.climate.copernicus.eu) but only every 6 hours. Do you know if the values are shifted by 6 hours? Because I have noticed that the peak of swrad is around 18hrs in the data but it's actually supposed to be around 12hrs I am not mistaken.

Also, I think I figured it out why it is showing the values so different from my inputs (Min = 4.27741714E-06 Max = 1.09921240E-05) instead of (Min = 20 MAX = 50), for example. It is actually showing the flux (going through my output I noticed it is FLUX not only swrad) which is degC m/s like it shows in the ana_srflux.h script.


Thank you in advance

[ymamoutos]

Hi Matheus

check here

https://confluence.ecmwf.int/display/CK ... umentation

in Temporal frequency section. Also if you download, let's say, recent reanalysis data for example July 2020
some fields are not complete (i.e. total precipitation). There is a ~ 2 month delay in the dataset.
Also there is no need to interpolate the atmospheric data. ROMS can regrid them during the run if
the netcdf file has the correct structure.

Giannis

[fagundesmo]

Hi Giannis,

Thank you a lot for your answer. One more time really helpful.

Cheers,

Matheus

[wilkin]

Because I have noticed that the peak of swrad is around 18hrs in the data but it's actually supposed to be around 12hrs I am not mistaken.

When you convert data that are accumulations into rates, the natural time to use would be the center of the accumulation interval. So a quantity accumulated from 0000 to 0600, divided by 6 hours to get a rate, is best interpreted as the rate at 0300.

Perhaps this explains your apparent time shift.

[fagundesmo]

Hi Wilkin,

I didn't know that!!! Ok, now the values make sense when I make that shift in my time data.

Thank you a lot,

Matheus