Numerical simulations to explain the coseismic electromagnetic signals: a case study for a M 5.4 aftershock of the 2016 Kumamoto earthquake

Coseismic electromagnetic (EM) signals that appear from the P arrival were observed in a volcanic area during the 2016 Kumamoto earthquake. In this study, we conduct numerical simulations to explain the coseismic EM signals observed for a M 5.4 aftershock of the earthquake. Initially, we adopt a water-saturated half-space model, and its simulation result for a receiver with a depth of 0.1 m suggests that the magnetic signals do not show up at the arrivals of P, refracted SV–P and Rayleigh waves because the evanescent EM waves just counterbalance the localized magnetic signals that accompany P, refracted SV–P and Rayleigh waves. Then, we conduct numerical simulations on a seven-layer half-space model in which the second layer corresponds to an aquifer analogy and the six other layers refer to air-saturated porous media. When only the electrokinetic effect is considered, the simulated coseismic magnetic signals still appear from the S arrival. The combination of electrokinetic effect and surface-charge assumption is also tested. We find that signals before the S arrival are missing on the transverse seismic, transverse electric, radial magnetic and vertical magnetic components, although the situation on horizontal magnetic components is improved to an extent. Then, we introduce an artificial scattering effect into our numerical simulations given that the scattering effect should exist in the volcanic area. New numerical result shows good agreement with the observation result on the signal appearance time. Hence, the combination of electrokinetic and scattering effects is a plausible explanation of coseismic EM signals. Further investigations indicate that coseismic electric and/or magnetic signals are more sensitive to the scattering effect and the aquifer thickness than seismic signals.