Denoising seismic noise cross correlations | INSTITUT DE PHYSIQUE DU GLOBE DE PARIS


Aller au compte twitter

  Denoising seismic noise cross correlations

Type de publication:

Journal Article


Journal of Geophysical Research-Solid Earth, Volume 114, p.B08310 (2009)



Numéro d'accès:




UMR 7154 ; Sismologie ; data adaptive filtering ; surface wave traveltimes ; volcanic precursors


Seismic noise cross correlations have become a novel way of probing the elastic structure of the Earth without relying on an often highly nonuniform and sporadic distribution of earthquakes. By circumventing this restriction, one can determine the elastic Green's function between any two points where instruments exist. For tomography, this will allow for a larger distribution of crossing paths and therefore better resolution in the inversions. One can also monitor the same station pair Green's functions for changes in the state of the Earth, an application that has been employed in volcanic monitoring. One limitation of this cross-correlation technique is that the input time series are frequently very long to recover high-fidelity signals. We present two time-frequency stacking algorithms to denoise the correlated signals and to alleviate this problem; increasing signal-to-noise ratios allows for high-fidelity Green's functions to be constructed from shorter time series. We demonstrate the increase in signal fidelity by applying these routines to seismic data, first to ambient noise across southern California and then to data from le Piton de la Fournaise volcano on La Reunion Island. In the former, we find that denoising the data allows for more traveltimes to be measured, particularly at longer station separations, across all passbands examined except for long-period Love waves, where no data are recovered. In the latter, we apply a time-frequency denoising algorithm to resolve subtle shifts in phase in cross correlations between seismic stations that occur before eruptions: we see a clear precursor to the June 2000 eruption.


Baig, A. M. Campillo, M. Brenguier, F.