Regridding tool
Thanks so much for making this dataset available!
I was wondering what's your recommendation for regridding the icosohedral grid data to a regular lat/lon grid?
I would recommend the DWD official regridding code: https://github.com/DeutscherWetterdienst/regrid
We don't actually regrid the data for our purposes, and leave it in the native grid when inputting it into our models, so we don't have too much experience with that, sorry!
Thank you for your ongoing support. We aim to convert .zarr files to grib2 to apply CDO for icosahedral mesh interpolation as advised by DWD. In addition, the reason why we need gribs is because our system is integrated to use grib files since we are using data from GFS and ECMWF models.
Did anyone find a solution on how to regrid the files to a regular lat/lon grid?
I run into an error using cdo because they converted .zarr files from ocf don't have the information about the grid anymore.
I believe you should be able to use https://xesmf.readthedocs.io/en/latest/notebooks/Dataset.html to regrid the dataset to regular lat/lon, although I haven't tried it.
Yea, you can try using https://github.com/openclimatefix/pvnet_app/blob/main/pvnet_app/data/nwp.py#L29, this might help?
There is a way to regrid the data.
- Load the .zarr data and save a data_var to .nc file. eg. t2m.nc
- Run 'cdo -s -f nc -remap,/home/grid_icogl2world_0125.txt,/home/weights_icogl2world_0125.nc -setgrid,grid.grb input_ncfile output_ncfile'
ps.
a. You need to create grid_icogl2world_0125.txt and weights_icogl2world_0125.nc. Please refer to https://github.com/open-meteo/open-meteo/blob/main/Sources/App/Icon/Cdo.swift.
b. You can create grid.grb by 'cdo griddes icon_global_icosahedral_single-level_2024112000_000_T_2M.grib2 > grid.grb'