Magnitude Station Corrections to Improve Initial Magnitude Estimates for ShakeAlert
Description:
In the early stages of the ShakeAlert earthquake early warning (EEW) system, it was considered a success to analyze, locate, and report an earthquake warning promptly. The initial magnitude estimate of the earthquake, within the first few seconds, has a large impact on the area of alert. We have investigated how to better estimate the initial magnitude, including (a) using machine learning to train the dataset; (b) re-evaluating the initial magnitude scaling relationship estimate using a larger dataset; and (c) including amplitude station corrections to account for non-linear characteristics of station effects.
ShakeAlert does currently not use station corrections in normal automatic location and magnitude detection. We propose to include magnitude station corrections, which are an amplitude correction factor applied to the magnitude initial estimate. We have also developed station correction factors using vertical amplitudes to match the initial P-wave magnitude estimate model more closely. We began by developing a comprehensive list of magnitude station corrections for ShakeAlert sites in the West Coast of the US including UCB, Caltech, PNSN and NRCan network. We have made the assumption that the methodology used to develop the amplitude correction factor is not important since most of the networks use different methods in the development of the datasets.
At UCB, dML station corrections were developed using the methodology of Uhrhammer et al. 2011 which uses the horizontal amplitudes selected on synthetic Wood-Anderson time series. For ShakeAlert, it is important to have a correction for the P-wave (or vertical data) since the initial magnitude estimates use the first few seconds of event data in the time domain. Therefore, we expand the Uhrhammer study to vertical component data of ShakeAlert sites. For vertical data we filter, pick station amplitudes, and determine station corrections.
Preliminary results show the impact of the inclusion of station corrections on the initial magnitude estimates as shown by case histories for the Oct 2023 Isleton and Aug 2023 Ojai earthquakes.
Session: End-to-End Advancements in Earthquake Early Warning Systems [Poster Session]
Type: Poster
Date: 5/3/2024
Presentation Time: 08:00 AM (local time)
Presenting Author: Fabia
Student Presenter: No
Invited Presentation:
Authors
Fabia Terra Presenting Author Corresponding Author terra@berkeley.edu University of California, Berkeley |
Peter Lombard lombard@berkeley.edu Retired, University of California, Berkeley |
Amy Williamson amy.williamson@berkeley.edu University of California, Berkeley |
Robert Uhrhammer ruhrhammer@gmail.com Retired, University of California, Berkeley |
Taka’aki Taira taira@seismo.berkeley.edu University of California, Berkeley |
Angie Lux angie.lux@berkeley.edu University of California, Berkeley |
|
|
|
Magnitude Station Corrections to Improve Initial Magnitude Estimates for ShakeAlert
Category
End-to-End Advancements in Earthquake Early Warning Systems