Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm

The New Zealand GeoNet P-phase picker's performance can be significantly improved, by up to 70%, when Mexican hat and Haar wavelet scale thresholding (WST) is implemented in the preprocessing stage of manually picked events. In total, 811 events with the required spatial and quality criteria fr...

Full description

Bibliographic Details
Main Authors: Jafarzadeh Rastin, Sepideh, Unsworth, C., Gledhill, K.
Format: Journal Article
Published: Seismological Society of America 2014
Online Access:http://hdl.handle.net/20.500.11937/51790
_version_ 1848758771686309888
author Jafarzadeh Rastin, Sepideh
Unsworth, C.
Gledhill, K.
author_facet Jafarzadeh Rastin, Sepideh
Unsworth, C.
Gledhill, K.
author_sort Jafarzadeh Rastin, Sepideh
building Curtin Institutional Repository
collection Online Access
description The New Zealand GeoNet P-phase picker's performance can be significantly improved, by up to 70%, when Mexican hat and Haar wavelet scale thresholding (WST) is implemented in the preprocessing stage of manually picked events. In total, 811 events with the required spatial and quality criteria from the 2008 Matata swarm were studied. The WST was applied to manually processed waveforms recorded by key stations, including 48 waveforms recorded by URZ; 47 by stations LIRZ, MARZ, and OPRZ; 64 by EDRZ; 33 by KARZ; and 47 by OPRZ. The signal-to-noise ratios of these key stations were dominated by additive noise rather than weaknesses introduced by the radiation pattern. We found that the Mexican hat WST produced the largest amount of retrieved and revised P onsets for station EDRZ (70.3%). The next best station was KARZ, which had 66.7% of picks showing improvement and only 3% of picks adversely affected. The main best scales (i.e., the wavelet's degree of compression) for stations EDRZ and KARZ were scales 1 and 2, respectively. Haar WST for the OPRZ recordings resulted in a total improvement greater than 50%, with the lowest adversely affected picked onsets of 2.1%. Similarly, the Haar WST retrieved a considerable number of missed onsets at the TGRZ and LIRZ stations (25.5% and 20%, respectively), with a minor effect on the performance. The results confirmed that WST served to significantly enhance the performance of the GeoNet P picker for the chosen Matata earthquakes due to the frequency localization of the WSR compared to the current Butterworth filtering scheme.
first_indexed 2025-11-14T09:49:17Z
format Journal Article
id curtin-20.500.11937-51790
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T09:49:17Z
publishDate 2014
publisher Seismological Society of America
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-517902017-09-13T15:37:22Z Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm Jafarzadeh Rastin, Sepideh Unsworth, C. Gledhill, K. The New Zealand GeoNet P-phase picker's performance can be significantly improved, by up to 70%, when Mexican hat and Haar wavelet scale thresholding (WST) is implemented in the preprocessing stage of manually picked events. In total, 811 events with the required spatial and quality criteria from the 2008 Matata swarm were studied. The WST was applied to manually processed waveforms recorded by key stations, including 48 waveforms recorded by URZ; 47 by stations LIRZ, MARZ, and OPRZ; 64 by EDRZ; 33 by KARZ; and 47 by OPRZ. The signal-to-noise ratios of these key stations were dominated by additive noise rather than weaknesses introduced by the radiation pattern. We found that the Mexican hat WST produced the largest amount of retrieved and revised P onsets for station EDRZ (70.3%). The next best station was KARZ, which had 66.7% of picks showing improvement and only 3% of picks adversely affected. The main best scales (i.e., the wavelet's degree of compression) for stations EDRZ and KARZ were scales 1 and 2, respectively. Haar WST for the OPRZ recordings resulted in a total improvement greater than 50%, with the lowest adversely affected picked onsets of 2.1%. Similarly, the Haar WST retrieved a considerable number of missed onsets at the TGRZ and LIRZ stations (25.5% and 20%, respectively), with a minor effect on the performance. The results confirmed that WST served to significantly enhance the performance of the GeoNet P picker for the chosen Matata earthquakes due to the frequency localization of the WSR compared to the current Butterworth filtering scheme. 2014 Journal Article http://hdl.handle.net/20.500.11937/51790 10.1785/0120130120 Seismological Society of America restricted
spellingShingle Jafarzadeh Rastin, Sepideh
Unsworth, C.
Gledhill, K.
Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm
title Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm
title_full Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm
title_fullStr Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm
title_full_unstemmed Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm
title_short Applying Haar and Mexican hat wavelets to significantly improve the performance of the New Zealand GeoNet P-phase picker for the 2008 Matata region swarm
title_sort applying haar and mexican hat wavelets to significantly improve the performance of the new zealand geonet p-phase picker for the 2008 matata region swarm
url http://hdl.handle.net/20.500.11937/51790