Performs an n-dimensional optimal interpolation.
Every elements in f corresponds to a data point (observation) at location the x(:,1), x(:,2),... x(:,n) in a n-dimensional space with the error variance var
Each element of the n by 1 vector len is the correlation length in the corresponding direction.
m represents the number of influential points.
xi(:,1), xi(:,2) ... xi(:,n) are the coordinates of the points where the field is interpolated. fi is the interpolated field and vari is its error variance.
The background field of the optimal interpolation is zero. For a different background field, the background field must be subtracted from the observation, the difference is mapped by OI onto the background grid and finally the background is added back to the interpolated field.
The error variance of the background field is assumed to have a error variance of one. The error variances of the observations need to be scaled accordingly.
See also: optiminterp1,optiminterp2,optiminterp3,optiminterp4.
Package: optiminterp