UC Berkeley

Landslides and debris flows are among the most destructive natural hazards globally. Reliable assessment of slope hazard requires characterization of the subsurface at spatial scales that traditional site-specific geotechnical methods cannot efficiently achieve. Borehole drilling, laboratory testing, and contact-based geophysical surveys yield accurate but highly localized data. Regional remote sensing approaches primarily provide information on the surface rather than subsurface, and lack the resolution required to constrain the material properties governing slope stability. Bridging this scale gap is critical both for using site-specific measurements to make larger-scale observations as well as improving regional modeling efforts.This dissertation develops and evaluates a multi-scale framework for regional rock mass characterization using non-contact frequency-domain electromagnetic induction (FDEM), which can collect data at rates orders of magnitude faster than contact-based methods. By utilizing FDEM data as the primary connective tool, the utility of the multi-scale framework is demonstrated in three distinct settings: the Western Transverse Ranges of Southern California, the southwestern coast and central highlands of Puerto Rico, and the Melamchi Khola Valley of the Nepal Himalayas. Each of these projects represents an increasing level of complexity. A comprehensive theoretical and methodological review of electrical resistivity, electromagnetic induction, and the handheld GEM-2 FDEM instrument is provided first, as it is used as the primary tool for this dissertation. Several case studies are presented to demonstrate the method's applicability and limitations. The EM method is subsequently applied alongside ERT and MASW at 15 focus sites in a sedimentary rock sequence of the Topatopa Mountains of Southern California's Western Transverse Ranges, demonstrating that EM- and ERT-derived resistivity profiles capture similar subsurface structure in this environment. A systematic increase in resistivity and seismic velocity is observed with formation age, reflecting diagenetic strengthening. Further FDEM surveys conducted between and around the sites demonstrate that the resistivity trends observed at point locations persist at larger spatial scales, illustrating the scalability of non-contact EM data collection. This framework is then implemented in Puerto Rico across multiple lithologies, expanding from sedimentary units only to include igneous and metamorphic geologic units. The EM datasets are used to compare lithologic groups, identify topographic and environmental controls on subsurface resistivity, and develop a predictive data-driven modeling framework for extrapolating subsurface properties across unmeasured terrain. Finally, a culminating integration of EM with borehole logging, laboratory testing, and other geophysical methods within the Nepal Himalayas is presented. The multi-scale characterization approach enables a more robust characterization of rock mass properties and shows that weathering is becoming less pronounced northward along a ridge transect in Central Northern Nepal. Additionally, it shows that feldspar dissolution along fractures, rather than biotite oxidation, is the dominant weathering mechanism driving porosity development and mechanical degradation. These projects across diverse environmental conditions establish a multi-scale framework for regional rock mass characterization, demonstrate the viability of FDEM as a scalable tool, and provide insights into the relationships between climate, lithology, weathering/diagenesis, and rock mass strength. As extreme precipitation events, seismic activity, and human modification of hillslopes continue to drive landslide losses globally, methodological advances, such as the one presented in this dissertation, aim to advance regional characterization of the geoenvironment, which is critical for community risk reduction from landslides and slope instability.