Steps kind of data at the coast line in all domains, instead of smooth transitions

[athulcp]

I am trying to simulate WRF and getting a step-kind of transition near the coastlines of India (ECMWF seasonal simulation data as initial and boundary conditions). Attaching namelist.wps, namelist.input, and some screenshots of output variables (surface and pressure level variables). Please help me to resolve this problem.

While visualizing met_em files after metgrid.exe, we did not find any errors.

 

[Ming Chen]

What are the input data you used to drive this case? In your namelist.wps, I can see that your data are from different sources:

fg_name = 'FILE' , 'PSFC', 'SST',

 

[athulcp]

SST is from seasonal forecast 1 degree data interpolated to 0.25 degree and provided to the model as intermediate file.
Surface pressure psfc is derived using Temperature MSLP - hypsometric equations. Psfc is not directly available from the ecmwf. Hence derived using the equation and provided as intermediate file.
Both sst and psfc visuals are attached for a metem output file for your reference.

This noise is not present in the upper level variables and noticed in the surface such as T2, U10, V10 etc.. For 3D variables only first 3 to 5 layers near the surface is having problem. Found that only after 1st timestep this error is coming.

I am attaching the met_em files and wrfoutput screenshots, and after running the model, this problem is coming into the picture.

 

[Ming Chen]

This is apparently a land mask issue. Your SST data should have its own landsea mask, and you need to include this information in your WPS process. Please see more information at metgrid crashes: get_min(): No items left in the heap

 

[athulcp]

I run the model with the land sea mask file from the ERA5 seasonal simulation; still, the error is not rectified. Is there any other way I can rectify the error ?

 

[Ming Chen]

I don't think you use landsea mask correctly ....

As an alternative, do you have skintemp in your ERA5 data? if so, please try to replace SST by SKINTEMP. This will overcome the issue along the coast, while over the ocean points, SST and SKINTEMP are same.

 

[athulcp]

I corrected my LANDSEA mask files by giving them manually; still the error persists. The met_em files do not show any problem, but after running the WRF model, the output shows LU_INDEX = 15 (snow or ice) in the box regions near the coast. I am not sure from where this error is rising, which alters the whole model output. It will be really helpful if this can be resolved.