Using model2roms gives ESMF errors due to grid

Discussion about analysis, visualization, and collaboration tools and techniques

Moderators: arango, robertson

Post Reply
Message
Author
iriskeizer

Using model2roms gives ESMF errors due to grid

#1 Unread post by iriskeizer »

Hello fellow ROMS / model2roms users,


I'm using model2roms (https://github.com/trondkr/model2roms) and Glorys12V1 reanalysis data to create lateral forcing files for ROMS. However, the lateral forcing output is not correct.


=================== Below here is detailed info the sketch the problem =====================
The input ROMS grid file (attached) I use has a resolution of 1/4°. The domain is rectangular from -20°E to 10°E and 36°N to 62°N (glorys has the same domain but 1/12° resolution). Because the domain is rectangular, xi_rho directly translates to longitude points and eta_rho to latitude points (see figures below).
grid_lon_of_rho_points.png
grid_lat_of_rho_points.png
Hereby also the bathymetry of the grid file:
NorthSea4.png


When running model2roms using the described glorys reanalysis and ROMS grid file as input, I obtain lateral forcing files that seem wrong. Here, I show the results for 'zeta' and 'temp' of the initial file (and zos from glorys to compare):

zeta_init.png
zos_glorys.png
Compared to the glorys plot, not only is the resolution different. The plot of zeta for the initial file seems a lot more homogeneous including these horizontal bands that cannot be explained and are also seen in the temperature plots:
temp_init.png


=================== Above here is detailed info the sketch the problem =====================

When I look into the log files (PET0.ESMF_LogFile), this is shown:
EMSF_error_log.png

This makes me believe that the problem has something to do with ESMF not being able to do the horizontal interpolation using the ROMS input grid that is kind of 'rectangular' resulting in many 'duplicate' longitude values. However, I don't see others reporting on the internet about this error. When using xesmf myself to interpolate glorys data to this ROMS grid, there is no problem.


Another striking thing is that, when I run runM2R.py, when it loops over time, for each time step it says about the data range for zos:
"INFO:root:[M2R_model2roms] Data range of zos just after extracting from netcdf file: -9989999999999999496648810221797376.000-0.377"
This suggests that the fill value that is denoted in the file grd.py is used for each time step. Could this also be a problem?
Schermafbeelding 2022-08-25 om 16.51.51.png




Does anyone know what to alter in the 'model2roms' code to resolve this problem/ error message? Or does someone has another idea of what might be the problem causing the strange forcing files?
Attachments
NorthSea4_smooth01_sponge_nudg.nc
(6.55 MiB) Downloaded 133 times

Post Reply