Application of minimum entropy deconvolution to detect $pP$ phase in a seismogram

TL;DR

This paper introduces a method using minimum entropy deconvolution to detect the pP seismic phase in seismograms, improving depth estimation accuracy with a simple and effective approach.

Contribution

It applies minimum entropy deconvolution to identify impulse-like pP phases in seismograms, demonstrating effectiveness on real earthquake data.

Findings

9 out of 12 earthquakes had pP-P travel-time errors less than 2 seconds

Maximum error observed was 3.06 seconds

Technique effectively detects pP phase even in noisy single seismograms

Abstract

The hypocentral depth is a key requirement in seismology and earthquake engineering, but it is very difficult to be determined. The current accepted improvement is taking advantage of the depth phases, such as the pP, to constrain this parameter. However, it is not easy to pick such a phase in a seismogram from the other phases and the backgound noises. Here we propose the use of the minimum entropy deconvolution (MED) to detect it. Synthetic tests show that impulse(s) hidden in the seimic noises, eg. discrete unit impulses or the Gaussian mono impulses, can be detected completely. Further, we assume that the pP phase is an impulse-like signal buried in the Z component of the seismogram and applied this technique to 12 earthquakes in the International Association of Seismology and Physics (IASPEI) Ground Truth (GT) reference events list. Results show that 9 out of 12 earthquakes have absolute errors of less than 2.00 s for the travel-time differences of pP-P, and the maximum absolute error is 3.06 s . This demonstrate that the assumption above is reasonable, and this technique works well and effectively even for a single seismogram. Due to its little cost and effectiveness, this technique may be also useful in the starting points for other methods to detect pP phase.