← Back to Sessions

Error Estimates for Seismic Body Wave Delay Times in the International Seismological Centre's Bulletin

Description: 

A cumulative half century of openly accessible global seismic phase arrival time data at the International Seismological Centre (ISC) presents an opportunity to independently asses their errors and how these might vary with epoch, epicentral distance, hypocentral depth, phase identity, and other factors. To separate the effects of Earth structure on the errors from measurement errors we constructed a large tomographic matrix for each of nearly 20 thousand earthquake clusters, using delay time data from the years 1964-2018 in the most recent version of the ISC-EHB bulletin. We applied a singular value decomposition to these matrices and estimated errors from the transformed data vectors belonging to the least significant singular values, i.e. those not related to Earth structure.
Using a robust initial estimate of the standard deviation of the clustered delay times, we removed a small fraction of outliers before calculating the ultimate errors.
We found that the errors depend on the phase identity, and hypocentral depth (crust or mantle), slowness, as well as on the precision with which the arrival times were reported. The epoch in which the arrival time was reported does not significantly affect the error estimates. Using these parameters, we distinguish 45 different classes of delay times for 11 different types of body waves. The errors for each class generally are normally distributed with means that range from 0.32 s for PKPbc waves from mantle earthquakes to 2.82 s for S waves from shallow earthquakes bottoming in the upper mantle. The uncertainties in the errors themselves are around one third of the estimated error, with two outliers at 10 % and 50 %, rendering the estimated errors representative in formal statistical quantification of the quality of fit in tomographic experiments.

Session: Network Seismology: Recent Developments, Challenges and Lessons Learned

Type: Oral

Date: 4/20/2023

Presentation Time: 08:30 AM (local time)

Presenting Author: Suzan van der Lee

Student Presenter: No

Invited Presentation: 

Authors