Abstract
We propose an effective method for treating singular value decomposition (SVD) analysis. This method is economical in the sense that the problem can be reduced to a matrix problem of the temporal size, not spatial size. Thus, this method is efficient when the spatial size is very large and the number of the ensemble is much smaller, as in a typical climatological problem. With this method, it is very simple to change the area of analysis by the use of weights, which are introduced to deal with the inhomogeneity of the data.