Aquifer Susceptibility to Earthquake-Induced Water-Level Changes: Towards a Probabilistic Model of Fluid Pressure Changes During Earthquakes

Session: Environmental and Near Surface Seismology: From Glaciers and Rivers to Engineered Structures and Beyond
Type: Oral
Date: 4/29/2020
Time: 11:15 AM
Room: 110 + 140
Description: 

New Zealand has experienced several large earthquakes in the last decade that have induced widespread hydrological changes in groundwater systems. This provides a rare opportunity to develop a probabilistic model of aquifer susceptibility to earthquake-induced groundwater changes as a function of shaking intensity. Persistent groundwater-level changes, or absences of change, have been quantified in 495 monitoring wells in response to one or more of 11 New Zealand earthquakes larger than M_W 5.4 between 2008 and 2017. We apply a binary logistic regression model with random effects to the dataset and examine the effects of seismic factors (peak ground velocity, peak ground acceleration, epicentral distance, shaking frequency and seismic energy density) and hydrogeological factors (degree of hydrogeological confinement and rock strength). The model also accounts for variations in monitoring wells’ susceptibilities to earthquake-induced persistent water-level changes. We calculate marginal probabilities as a function of three predictors — earthquake shaking (peak ground velocity), degree of hydrogeological confinement (monitoring well depth) nd rock strength (site-average shear-wave velocity) — and of Modified Mercalli Intensity (MMI). This enables us to quantify the likelihood of persistent water-level changes for MMI levels II–VIII. Our study constitutes a regional scale, multi-site, multi-earthquake investigation of the occurrence and absence of hydraulic responses to large earthquakes spanning almost a decade of seismic shaking. It is a first attempt at incorporating seismic and hydrogeological factors in a common probabilistic description of earthquake-induced groundwater level changes. Although focused to date solely on New Zealand data, the modeling framework we have developed provides a more generalizable approach to quantifying hydrological responses to earthquakes than alternative metrics based on epicentral distance, magnitude and seismic energy density and one which we anticipate being of use to both the scientific and engineering communities alike.

Presenting Author: John Townend

Authors