Hello,
After a year-long hiatus, I have started running WRF simulations again. Last time I used WRF, I downloaded data from the Copernicus CDS, and ungribbed it using the Vtable.ECMWF. I used the cdsapi python script to download the data. All worked fine. The WRF version was 4.3.3, and I ran it on a supercomputer.
I once again started working on the supercomputer, again using WRF 4.3.3 and the same Vtable. I downloaded the pressure-level and single-level data (same data as previously, although the python API seems to have changed somewhat in the meantime), one file for 3D data (pressure level) and one for the 2D data (single-level data). I ran ungrib and metgrid, producing met_em files.
However, running real.exe produces the following:
What seems to have happened is that some of the met_em files have 1/3 of the size of others. More precisely, I took data from 21.09.2023. to 26.09.2023., 3-hourly spacing, and for the data for 22.09.2023., metgrid put the UU, VV, GHT, RH and TT fields into the 2D category, meaning only one level was read??
It seems to me that the issue is a change in how ERA5 data is interacting with WPS, as using data downloaded previously (in 2023) with WRF yields the correct results.
So, my question is - what changed and what is the correct way to use ERA5 data with WRF nowadays?
I will post the cdsapi requests to this thread shortly.
After a year-long hiatus, I have started running WRF simulations again. Last time I used WRF, I downloaded data from the Copernicus CDS, and ungribbed it using the Vtable.ECMWF. I used the cdsapi python script to download the data. All worked fine. The WRF version was 4.3.3, and I ran it on a supercomputer.
I once again started working on the supercomputer, again using WRF 4.3.3 and the same Vtable. I downloaded the pressure-level and single-level data (same data as previously, although the python API seems to have changed somewhat in the meantime), one file for 3D data (pressure level) and one for the 2D data (single-level data). I ran ungrib and metgrid, producing met_em files.
However, running real.exe produces the following:
What seems to have happened is that some of the met_em files have 1/3 of the size of others. More precisely, I took data from 21.09.2023. to 26.09.2023., 3-hourly spacing, and for the data for 22.09.2023., metgrid put the UU, VV, GHT, RH and TT fields into the 2D category, meaning only one level was read??
It seems to me that the issue is a change in how ERA5 data is interacting with WPS, as using data downloaded previously (in 2023) with WRF yields the correct results.
So, my question is - what changed and what is the correct way to use ERA5 data with WRF nowadays?
I will post the cdsapi requests to this thread shortly.