there was an update to OOPS this morning requiring us to implement an empty VariableChange.
there was an update to SABER fixing a bug in land masking, this resulted in tiny changes to the reference answers
The correlation lengths required by SABER are defined as the cuttoff distance of a Gaspari-Cohn function. This is not very intuitive, since we tend to think of correlation lengths as the sigma of a Gaussian. I have changed Covariance so that it converts the correlation length we give from Gaussian to GC cutoff. The ctest correlation lengths were changed accordingly from 3000km to 840km so that the effective correlation length remains the same.
VariableChange.Covarianceso that it converts the correlation length we give from Gaussian to GC cutoff. The ctest correlation lengths were changed accordingly from 3000km to 840km so that the effective correlation length remains the same.