Usually, it is necessary to use radar observations to verify the WRF prediction skills, maybe it is a good way to interpolate the observations to the model grids.
You can use this example here: A Typical Workflow For Radar-Based Rainfall Estimation — wradlib
The ‘grid_xy’ variable can be any grid you like. Take what ever gridpoints you have from WRF and interpolate the radar field to your custom grid. I.e. use the wrl.comp.togrid() function with the custom grid as second parameter (the ‘trg’ keyword at wradlib.comp.togrid — wradlib).
Thanks for your helpful advice! This function looks very convenient.