Resampling Methods, Lomb-Scargle Analysis, and Empirical Orthogonal Functions: A Combined Approach to Gappy Data in the Maritime Continent

Lomb-Scargle time series methods show promise as a new way to derive empirical orthogonal functions (EOFs) from gappy datasets. Among other oscillations, this set of techniques is able to discern the diurnal cycle and harmonics in the high-frequency range, as well as dispersive oscillations usually hidden by the noise floor. Notably, these techniques are capable of providing phase data, which provide high-quality information about regional propagation patterns. The time series results can be further refined by resampling methods, but the nature of gappy datasets implies that relatively common resampling techniques can create significant impediments to calculating EOFs, especially the phase results of complex-valued EOFs. Preliminary results show that an ordinary time series analysis can be significantly refined by examining the self-coherence of the resampled time series analyses, and punishing high-amplitude signals with self-incoherent phases. The current state of this resampling technique is presented by analyzing TRMM 3B42 rainfall data over the Maritime Continent, an area rich with a variety of natural oscillations across the observable frequency band.