distance.mat matrix doesn't need to be just the physical distance, however it represent a measure of similarity between state variables in different sites.
Usage
Local.support(Pf, distance.mat, scalef = 1)
Arguments
- Pf
Forecast error covariance matrix
- distance.mat
is matrix of distances between sites.
- scalef
scalef is a numeric value that requires tuning and it controls the shape of the corrolation function
Value
It returns a localized covariance matrix by taking a Schur product between Pf and a corrolation function