Introduction of covariance components in slip inversion of geodetic data following a non-uniform spatial distribution and application to slip deficit rate estimation in the Nankai Trough subduction zone

This is a Preprint and has not been peer reviewed.

Downloads

Download Preprint

Authors

Ryoichiro Agata 

Abstract

When spatial distribution of observation stations has bias in geodetic slip inversion, modeling errors in the inversion scheme may result in significant unnatural short-wave components in estimated slip distribution, which overfit to data. Combined use of both land and seafloor geodetic data in slip inversion often leads to such situations. To avoid overfitting, I proposed a method to incorporate proper covariance components in the data covariance matrix originated by modeling errors in the geodetic slip inversion.
Because linearity of the inversion problem is retained, widely-known approaches to introduce prior constraints to the inversion, which assumes linear inversion, are easily applied to the proposed method.
Synthetic tests showed that the introduction of covariance components allowed for the estimation of slip distributions closer to the true one, avoiding overfitting to geodetic data in biasedly distributed observation stations, compared to a conventional approach that does not introduce covariance components. I applied the proposed method to the estimation of the slip deficit rate in the Nankai Trough subduction zone, using geodetic data of displacement rates provided by land GNSS stations and seafloor GNSS-Acoustic stations. The proposed method estimated a reasonably smooth distribution of slip deficit rate compared to the conventional approach. Spatial distribution of residuals in the displacement rates suggested that the proposed method successfully avoids overfitting to the block motion at the south of the Median Tectonic Line, which the physical model I assumed here cannot describe.

DOI

https://doi.org/10.31223/osf.io/p5q6d

Subjects

Earth Sciences, Geophysics and Seismology, Physical Sciences and Mathematics

Keywords

Dates

Published: 2019-08-02 13:30

Older Versions
License

GNU Lesser General Public License (LGPL) 2.1

Add a Comment

You must log in to post a comment.


Comments

There are no comments or no comments have been made public for this article.