Conversion from gridtype spectral to gaussian_reduced


I'm struggeling a bit with bringing together different parameters from ERA5 available as grib files as surface values or on model levels. For example, I have the specific cloud liquid water content (clwc) or specific humidity (q) on a gaussian_reduced grid (N320).  The temperature instead is given as spectral coefficients with a triangular truncation of T639.  My question would be now how to bring temperature on the same grid as all other parameters.

Ideally, this would be within python after reading the grib files into the script xarray or cfgrib. But any  thoughts or ideas that help me to understand the problem are welcome.

All best