Intraseasonal Oscillation Indices from Complex EOFs

Abstract Indices of oscillatory behavior are conveniently obtained by projecting the fields in question into a phase space few (mostly just two) dimensions; empirical orthogonal functions (EOFs) or other, more dynamical, modes typically used for projection. If sufficiently coherent and quadrature, projected variables simply describe rotating vector space, which then serves as basis predictions. Using boreal summer intraseasonal oscillation (BSISO) test case, an alternative procedure is introduced: it augments original with their Hilbert transform (HT) to form complex series projects onto its (single) dominant EOF. The real imaginary parts corresponding pattern index compared those (real) new explains slightly less variance physical than original, but much coherent, partly from use future information HT. Because latter way real-time monitoring, can only be cases predicted fields, promises superior. By developing causal approximation HT, variant whose coherency comparable noncausal version, smaller explained fields. In compares well other indices BSISO. potential using both discussed.