Tracking Groundwater Levels Using the Ambient Seismic Field

Abstract

Aquifers are vital groundwater reservoirs for residential, agricultural, and industrial activities worldwide. Tracking their state with high temporal and spatial resolution is critical for water resource management at the regional scale yet is rarely achieved from a single type of groundwater data.Here we show that variations in groundwater levels can be mapped using perturbations in seismic velocity (dv/v). We measure dv/v in the San Gabriel Valley, California, from cross correlation of the ambient seismic field. dv/v reproduces the groundwater level changes that are marked by the multiyear depletions and rapid recharges, typical of California's cycles of droughts and floods. dv/v correlates spatially with vertical surface displacements and deformation measured with the Global Positioning System (GPS). Our results successfully predict the volume of water lost in the San Gabriel Valley during the 2012-2016 drought and thus provide a new, complementary approach to monitor groundwater storage. Plain Language SummaryThe largest reservoir of freshwater worldwide is stored beneath our feet in groundwater aquifers. Monitoring the availability of water stored in the ground is critical for communities that are susceptible to drought, as groundwater can supplement a lack of surface fresh water. We propose a new way to monitor groundwater levels in aquifers using weak seismic waves repeatedly generated by the ocean, wind, and human activities. We exploit the fact that the speed of seismic waves is sensitive to the amount of the water in the ground. Using a network of seismometers, we measure the change in seismic velocity throughout the San Gabriel Valley Basin, California. We find that with the change in seismic velocity we can estimate the flux of water in the San Gabriel Valley over the last three droughts and subsequent recoveries that have affected Southern California over the last two decades. These results provide a new approach to monitor groundwater storage.

Groundwater Variations From Autocorrelation and Receiver Functions

“…The relationship between GWL and dv/v reported here is opposite in sign to that found by Clements & Denolle (2018) using cross-correlations in the San Gabriel Valley, CA. We attribute this difference to the fact that inter-station correlations are more sensitive to S-wave rather than P-wave velocities, because they are dominated by Rayleigh wave energy.…”

“…We attribute this difference to the fact that inter-station correlations are more sensitive to S-wave rather than P-wave velocities, because they are dominated by Rayleigh wave energy. This is indeed what has been observed: Clements and Denolle (2018) report that a 16.8-m GWL increase decreased V S by 0.125% while we find that a 15-m GWL increase increased V P by 1.2%. This is indeed what has been observed: Clements and Denolle (2018) report that a 16.8-m GWL increase decreased V S by 0.125% while we find that a 15-m GWL increase increased V P by 1.2%.…”

“…Instead, we attribute the sudden increase observed in both dv/v and PSD of RFs to three consecutive hurricanes (i.e., Gustav, Ike, and Norbert; red symbols in Figures 3c and S1) that occurred less than a month apart prior to the observed peak during 2008-2010. We also note similar dv/v changes along with GWL rise in the San Gabriel Valley, CA, due to 1 m of total precipitation, the highest amount of rainfall in 100 years in Los Angeles (Clements & Denolle, 2018). However, we report the opposite sign of dv/v changes with one order of magnitude larger here in Texas and will provide a physical explanation for this difference later in the section.…”

“…GWL changes induce correlated changes in V P , but anticorrelated, much smaller-amplitude changes in V S , regardless of the assumed porosity (Figures 4b and S5). Previous studies have used empirically determined scaling factors to relate GWL changes to V S variations (Clements & Denolle, 2018); our physical model, though simple, offers a way of inferring GWL variations from V P and/or V S changes, if the porosity of the aquifer is known. Previous studies have used empirically determined scaling factors to relate GWL changes to V S variations (Clements & Denolle, 2018); our physical model, though simple, offers a way of inferring GWL variations from V P and/or V S changes, if the porosity of the aquifer is known.…”

“…Conceptual diagrams depicting changes in GWL (blue) within an aquifer and the proposed seismic methods for detecting them. Of particular relevance to our study is work relating changes in seismic velocity to groundwater recharge by precipitation (Sens-Schönfelder & Wegler, 2011) and a study showing that local variation of the GWL in the San Gabriel Valley, CA, can be tracked using the inter-station correlation of ambient noise (Clements & Denolle, 2018). Changes in the GWL can be monitored by measuring dv/v from ambient noise correlation (i.e., computing −dt/t from a dashed trace with a solid trace) and power spectral density (PSD) of receiver functions (RFs) from teleseismic arrivals (in red).…”