Seismoelectric signals from earthquakes : Detection, analysis and interpretation

When seismic pressure waves propagate through the subsurface they can excite ‘seismoelectric’ (SE) signals via electrokinetic coupling. These SE signals are sensitive towards poroelastic as well as electrical properties while providing the resolutions of seismic measurements. They might therefore add information to conventional geophysical measurements. SE signals are generally divided into coseismic waves and interface responses (IRs). The coseismic waves are bound to seismic waves and can therefore only reveal information about the direct vicinity of the recording stations. Interface Responses (IRs) are excited when the (co)seismic waves cross electric or hydraulic discontinuities. They are sensitive to the rocks' porosity and permeability as well as the salinity and viscosity of the pore fluids. As they propagate independently from the seismic fields they can contain information from interfaces further away from the recording stations. Therefore, IRs are usually the main focus of SE investigations. The goal of this thesis is to investigate SE signals generated by earthquakes with a special focus on the detection, analysis and interpretation of these signals. While coseismic signals have been observed during earthquakes, IRs have yet to be detected. In the first part of this thesis distance-magnitude-thresholds are established that quantify the possibility to observe SE signals from earthquakes. Based on over 16,000 events we found that 67 % of earthquakes with magnitude M ≥ 2.3·log10(r/km) at hypocentral distance r showed clear SE signals if the noise level is below 0.1 μV/m. Furthermore we confirmed the site-dependency of SE earthquake signals recorded in the Atacama Desert of Northern Chile by comparing seismic and SE envelopes and introducing SE spectral ratios (SESRs) as a new concept to evaluate the influence of IRs on SE signals in the frequency domain. The concept of SESRs is explained in the second part of this thesis in more detail. We approximated a SESR calculated from field data through synthetic modelling with the help of an inversion algorithm. The calculations implied that the shape of the SESR is primarily influenced by IRs generated in the first few hundred meters of the subsurface and that a sensitivity exists towards to the rock's porosity, the pore-fluid's salinity and the depth of the interfaces that generate IRs. Last but not least, the response of numerical SE signals to a multi-layered subsurface model was analysed, simulating a possible passive SE field experiment on the Armutlu Peninsula in Turkey. The results showed that the chance to identify single IRs in a SE wavetrain is highest if the recording stations are placed close to the epicenter of the earthquakes. Through the analysis of SESRs and SE envelopes, porosity and salinity changes in the upper subsurface could be detected. Nevertheless, depending on the local noise level the estimated small amplitudes of the field data will provide a challenge. All in all, the investigations of this thesis showed that SE signals can be detected in a limited radius from the hypocenter if the magnitudes of the earthquakes are large enough for the generated SE amplitudes to overcome the anthropogenic noise-level. The detection radius thereby depends on the observation target (IRs or coseismic signals). If the signal-to-noise ratio is sufficient, SESRs as well as envelope analysis could be used to detect porosity and salinity changes in the first few hundred meters of the subsurface, although the results have probably to be constrained by previous information. Field measurements are now required to verify the findings of this thesis.