A New, Improved and Fully Automatic Method for Teleseismic Depth Estimation of Moderate Earthquakes (4.5

letort, jean;

The depth of an earthquake is difficult to estimate because of the trade-off between depth and origin time estimations, and because it can be biased by lateral Earth heterogeneities. To face this challenge, we have developed a new, blind and fully automatic teleseismic depth analysis. The results of this new method do not depend on epistemic uncertainties due to depth-phase picking and identification. The method consists of a modification of the cepstral analysis from
Letort et al. and Bonner et al., which aims to detect surface reflected (pP, sP) waves in a signal at teleseismic distances (30◦–90◦) through the study of the spectral holes in the shape of the signal spectrum. The ability of our automatic method to improve depth estimations is shown
by relocation of the recent moderate seismicity of the Guerrero subduction area (Mexico).We have therefore estimated the depth of 152 events using teleseismic data from the IRIS stations and arrays. One advantage of this method is that it can be applied for single stations (from IRIS) as well as for classical arrays. In the Guerrero area, our new cepstral analysis efficiently clusters event locations and provides an improved view of the geometry of the subduction.
Moreover, we have also validated our method through relocation of the same events using the new International Seismological Centre (ISC)-locator algorithm, as well as comparing our cepstral depths with the available Harvard–Centroid Moment Tensor (CMT) solutions and the three available ground thrust (GT5) events (where lateral localization is assumed to be well constrained with uncertainty <5 km) for this area. These comparisons indicate an overestimation of focal depths in the ISC catalogue for deeper parts of the subduction, and they show a systematic bias between the estimated cepstral depths and the ISC-locator depths.
Using information from the CMT catalogue relating to the predominant focal mechanism for
this area, this bias can be explained as a misidentification of sP phases by pP phases, which
shows the greater interest for the use of this new automatic cepstral analysis, as it is less
sensitive to phase identification.