Application of induced polarization imaging across different scales to understand surface and groundwater flow at the Hofermuehle landslide

Abstract

Understanding changes in hydraulic properties of the subsurface is critical to delineate areas susceptible for groundwater accumulation and the triggering of landslides. Laboratory studies have demonstrated the possibility to estimate the hydraulic conductivity from induced polarization (IP) measurements. However, to-date only rare studies have been applied at the field scale and none has evaluated the frequency-dependence in field IP imaging data. We show that the application of petrophysical models linking hydraulic conductivity (K) and IP at 1 Hz, resolves for the same estimations from frequency-domain measurements as well as from time-domain IP. Moreover, our IP images reveal an evident frequency-dependence between 0.1 and 240 Hz in the electrical properties, which is also observed in lab measurements conducted in soil samples. To account for this, we fit a Cole-Cole model to our multi-frequency IP results and evaluate models linking K and the normalized chargeability, gaining a subsurface model of the hydraulic conductivity with enhanced resolution. Our results reveal clear variations at depth in the electrical conductivity and polarization associated with the presence of a plane of instability and the sliding plane. We also resolve lateral changes delineating areas with K below 10-6 m/s, which may act as barriers for groundwater flow. To evaluate our results we also estimate the hydraulic conductivity using a joint inversion algorithm that directly solves for porosity and saturation through the simultaneous inversion of seismic refraction and electrical resistivity tomography data. We observed consistent hydraulic conductivity images obtained through the inversion of the IP data and the joint inversion, with such model being in agreement with existing information of the site. We resolve a 3D model of the hydraulic conductivity at the Hofermuehle landslide from IP data that reveals the geometry of areas with poor drainage that may lead to land sliding.