A WAVELET COHERENCE Array Processing Procedure for Mapping Long-Period (LP) and Very Long-Period (VLP) Seismicity at Active Volcanoes

Abstract

Volcanic signals are one of the most difficu lt subject of study for seismologists.The lack of clearbody-wave phase arrivals and therapid loss of signal coherence with distance make generally impossible application of traditional location techniques based on travel times inversion.Over the past ten years, however,the wide spread application of multichannel techniques has allowed for robust and reliable locations of these signals over abroad range of frequencies. Consequent to the largely increased computing capabilities,array processing in Volcano Observatories is now developing to ward thereal-time detectionand tracking of volcanic sources.However,the energy of volcanic signals may spread over abroad frequency interval(Fig.1).The successful detection and measurement of these signals requirest here for esubsequent band-pass filtering operations,and exact tayloring of the time window lengths to the central frequency of each filter.Exploting the localisation properties of Continuos Wavelet Transforms(CWT)in the time and frequency/scaledomains,in this workweuse multichannel Wavelet Coherence(WCO)to perform signal detection from local correlation information. For the time/scaleregions of high multichannel coherence,wethenextend the MUltiple Signal Classification(MUSIC) method to the spatial covariance matrice sobtained from the complex-valued wavelet coefficient timeseries