qAny m by n matrix A can be factored into
q
q
q
qThe columns of U (m by m) are the EOFs
qThe columns of V (n by n) are the PCs.
qThe diagonal values of S are the eigenvalues represent the
amplitudes of the EOFs, but not the variance
explained by the EOF.
qThe square of the eigenvalue from the SVD is equal to the eigenvalue from the eigen analysis of the
covariance matrix.