Coseismic fluid-pressure response estimated from prediction-error filtering of tidal-band loading

Abstract

A methodology combining prediction-error filters (PEFs) and transfer functions was developed to identify the quasi-static fluid-pressure response observed in wells due to coseismic strain. Water levels in confined aquifers respond to long-term and seasonal trends, recharge events, barometric and ocean tide loading, tidal strain, and tectonic strain. Low-frequency features can be neglected from the quasistatic coseismic response estimation. Transfer functions were constructed to deconvolve the fluid-pressure response due to measured barometric loading. Because direct tidal strain and ocean tide loading measurements are rarely available, theoretical tidal loading is often calculated from astronomical data. However, the calculations are subject to many assumptions. Because tidal driving processes are cyclic, PEFs are a natural choice for removing the fluid-pressure response without assuming a theoretical forcing function in the tidal band. The method was applied to hourly fluid pressure data collected over a 3-year period from two wells in the villages of Gaocun and Tayuan, China. Results of this analysis yielded coseismic fluid pressure heads of - 1.6 X 10-2 and + 7.6 X 10-2 m for the respective wells in response to the Datong-Yanggao earthquake swarm mainshock (M(s) 6.1), 18-24 October 1989. Epicentral distances to the wells were about 200 km. The coseismic fluid-pressure response for each well was also predicted from dislocation model strain scaled by material-dependent volumetric strain sensitivity parameters. These parameters were determined from the static confined response to O1 and M2 earth-tide strain constituents. The predicted response was - 2.9 X 10-3 m for the Gaocun well and + 2.1 X 10-3 m for the Tayuan well. Although predicted and observed response phases were consistent, both predictions underestimated observed response amplitudes, as has been true in other reported instances.