Analysis of a Mesoscale Gravity Wave Event Using Empirical Orthogonal Functions

A method is detailed that filters mesoscale gravity wave signals from synoptic-level observation data using empirical orthogonal functions (EOFs). Similar EOF analyses have been used to study many oceanographic and meteorological features by allowing the examination of the variance associated with the principal orthogonal components of a time series in both spatial and temporal formats. Generally, EOF components are tied only to the underlying physical phenomena driving the observations when they represent a significantly large portion of the cumulative EOF variance. This work demonstrates a case in which a physically significant gravity wave event is recovered from the synoptic signal using EOF components that represent a small percentage of the total signal variance. In this case this EOF filtering technique appears to offer several advantages over more traditional digital filtering methods; namely, it appears to capture more of the gravity wave amplitude, it requires less preconditioning of the time series data, and it provides filtered solutions at the first time step.