Geoelectrical Monitoring of Moisture Driven Processes in Natural and Engineered Slopes

Abstract

Landslides are major and common natural hazards. They endanger communities and critical infrastructure, and have caused more than 28,000 fatalities and more than $1.8 billion in direct damage within the last decade. Climate change, with more frequent weather extremes, is likely to increase the occurrence of shallow landslides worldwide. In order to mitigate the risk, there is a need to improve our understanding of shallow, rainfall-induced landslide dynamics, which are mostly controlled by hydrogeological processes. These processes are known to be spatially and temporally highly heterogeneous. Geophysical techniques, and geoelectric methods in particular, are capable of providing volumetric data that can address this spatial and temporal complexity of landslide processes. This thesis advances “landslide geophysics” by developing methodologies for a robust and cost-effective characterization and monitoring of subsurface landslide processes. In doing so it addresses some of the inherent difficulties of geophysical monitoring of landslides. Three key challenges are addressed: (i) accounting for electrode displacements, (ii) improved geoelectrical imaging through structurally constrained inversion using seismic data, and (iii) optimized survey design to improve image resolution whilst reducing installation costs. Geoelectrical imaging requires the exact knowledge of the electrode locations. This is a limiting factor for remote landslide monitoring, where continuous monitoring of electrode locations is not feasible. Here, an approach is introduced using GPS positions of a temporally and spatially sparse set of benchmark points to estimate the locations of the electrodes. Application to synthetic data shows that electrode locations can be estimated with an accuracy of ~3% of the initial electrode spacing, which is sufficient to virtually remove inversion artefacts caused by incorrect locations. This methodology was then applied to a real landslide at Hollin Hill, Northern Yorkshire, UK, at a long-term landslide observatory. This showed that electrode movements can be estimated with an accuracy of ~10% the initial electrode spacing. To fully understand the control of rainfall and moisture dynamics on the landslide movements, geotechnical and environmental data were investigated before attempting to incorporate the geophysical data. This shows that the landslide deformations correlate with pore water pressures causing a characteristic ‘S’-shape of movements. High pore water pressures cause an acceleration of movements and low pore water pressures a deceleration. As slope instabilities are not only caused by the hydraulic properties, but also by the elastic parameters of the subsurface, a seismic characterization of the landslide was performed. This highlights the very weak nature of the material forming Hollin Hill (Shear modulus ~25 kPa and Young’s modulus 100 kPa). Minima of the elastic moduli were found at the actively moving parts of the landslide, emphasizing the reduced strength of the material leading to mass movements at shallow slope angles. By incorporating estimated electrode locations and the results of the seismic characterization into a time-lapse inversion, the first application of long-term 3D geoelectrical monitoring of an actively moving landslide was achieved within this research. Crucially, thresholds for movement were defined, which are not solely based on rainfall, but on the actual cause of slope instabilities - i.e., subsurface moisture dynamics. Thereby providing a potentially more robust indicator for critical slope conditions. This step-change in geophysical capabilities has the potential to significantly advance our understanding of the landslide hydrological processes and prediction of failure events. Lastly, to increase the efficiency of ERT monitoring, a survey optimization methodology is proposed that reduces the number of electrodes while increasing image resolution within a target horizon. On a synthetic example it is shown that despite using > 50% fewer electrodes, a target is imaged with higher resolution than using a standard survey design. Translating this methodology to a laboratory example simulating an engineered slope shows a similar result, for which a target is imaged at higher resolution compared to a standard survey despite using > 20% fewer electrodes. This has the potential to improve the efficacy and lower the costs of ERT landslide monitoring installations. The achievements presented in this thesis should inspire future research to further develop geoelectrical methods to be routinely and robustly applied in landslide early warning systems, thus helping to reduce landslide risk and protecting communities and critical infrastructure.