Investigating well water level oscillations caused by seismic waves using automatically detected responses

Abstract

Understanding the interaction between earthquakes and groundwater is crucial for elucidating variations in subsurface properties as well as fault structure and activity. For this purpose, we develop an automatic detection program to identify coseismic water level oscillations, which integrates time- and frequency-domain detection methods, achieving a precision of 92.8 % and a recall of 89.2 %. Application to a decade’s worth of high-rate pore-pressure data recorded near Parkfield shows that the coseismic water-level oscillations are primarily influenced by the seismic energy density, and the well-bore construction, aquifer hydraulic properties and dilation of seismic waves can also influence water-level responses. The seismic energy density can be well approximated by the square of the peak ground velocity, or by the existing empirical relation if surface-wave magnitude is used. The seismic energy density threshold to cause water level oscillation is ∼ 5 × 10−9 J/m3 based on our dataset. The coseismic water level oscillations are more sensitive to seismic waves with energy at low frequency band. We find that in the existing empirical relation using surface-wave magnitude would be more appropriate, so we propose an updated empirical relation log10e=-3log10r+1.5MW-4.82 among the seismic energy e(J/m3), hypocentral distance r(km), and moment magnitude MW. The predicted seismic energies using this relation are consistent with the observed values within two orders of magnitude.Plain language summary: Understanding how well water levels respond to seismic waves is important to studying the interaction between earthquakes and groundwater. An extensive catalog of water-level responses can significantly improve such investigations. Therefore, we develop an automatic detection program to efficiently detect the well water-level oscillations caused by seismic waves in both time and frequency domains. In our catalog, the amplitudes of the coseismic water-level oscillations show a good linear relationship with the seismic wave energy densities, which can be approximated by the squares of peak ground velocity. The minimum seismic energy density required to trigger water-level oscillations is ∼ 5 × 10−9 J/m3. However, reaching this threshold does not always cause well water-level oscillations, and the energy at low frequency is important to cause water-level response. We find that in the existing empirical relation using surface-wave magnitude would be more appropriate, and we update the empirical relation based on the moment magnitude. The differences between the predicted and observed seismic energy densities generally fall within two orders of magnitude. This suggests that when seismic observations are available nearby, the observed seismic energy estimated by the square of peak ground velocity would be more representative for studying hydrological responses to earthquakes.