Wavelet analysis of a microbarograph network

This paper presents a wavelet-based algorithm for the detection, identification, and extraction of gravity waves from atmospheric pressure traces. The main data processing tool is a nonlinear adaptive filter based on the selective reconstruction of a waveform from its wavelet coefficients. The time-frequency localization of the wavelet transform provides an ideal framework for the decomposition of long-period gravity waves (30 min–6 h), which are characterized by a generally broad spectrum and few oscillation cycles. The procedure is iterative and allows the exhaustive processing of all the events present in a fixed time period. The waveform of each disturbance is reconstructed with high accuracy. This minimizes the influence of the data-processing technique on the estimate of horizontal speed and direction of propagation, obtained by maximization of the cross-correlation functions between the reconstructed waveforms at the different stations. The introduction of coherency criteria through the network of seven stations allows us to separate the events into two classes. The first includes the events that propagate with very small distortion through the network, while the second includes less coherent but still highly energetic events. The size of the network and the algorithm developed for the analysis is well suited for the identification and the extraction of those mesoscale disturbances that have a particularly strong influence on the weather as well as on the forecast.