Seismoelectric monitoring of aquifers using local seismicity—a feasibility study

Abstract

SUMMARYSeismoelectric (SE) signals, accompanying seismic wave fields radiated from earthquakes, can be observed on records of magnetotelluric stations. Assuming that these SE signals are generated by electrokinetic coupling we investigate whether they can be used as a ‘pore-space monitoring’-tool. Regarding future field experiments we analyse synthetic SE waveforms calculated for a fully saturated base model consisting of five layers overlying a half-space, resembling the conditions of the Armutlu Peninsula (Turkey). This example site stands for a location with near-surface thermal aquifers exposed to tectonic stress and significant microseismicity. As expected, coseismic SE waves arrive simultaneously with the seismic onsets whereas interface response (IRs) SE waves arrive (shortly) before the generating seismic onsets. Therein, so-called evanescent IRs show a similar moveout as seismic phases and so-called radiation IRs travel with zero slowness. We found that the influence of IRs on the overall SE signal can be identified by envelope analysis of SE time series and by seismoelectric spectral ratios (SESRs) in the frequency domain. For a sensitivity analysis we added an extra layer to the base model with differing porosity, porefluid salinity and permeability values. At near-epicentral distances both trace-envelopes and SESRs are sensitive to the porosity and porefluid salinity changes in the simulated near-surface aquifer. The SESRs’ and SE envelopes’ amplitudes vary in the order of up to some 10 per cent in response to porosity and salinity increases of factor 2 and 100, respectively. In contrast, a decrease of the permeability value by the factor 100 leads to an SESR amplitude variation of less than 1–10 per cent. In the Armutlu model the largest relative changes of SE signals occur near the epicentre where the ratio between coseismic and IR amplitudes is close to 1. For 1–6 km deep source depth the SE detection swell at the earth surface is in the order of magnitude 2–3, depending on the ambient electromagnetic noise and hypocentral distance. This estimate assumes that SE signals are recorded with standard magnetotelluric stations. It can improve if array methods are applied.