A thermo-hydro-mechanical approach to soil slope stability under climate change

Landslide initiation and dynamics are approached with different methods according to the scale of investigation. Individual landslides are typically described mechanistically, relying on the known coupling between the hydrological input – via atmosphere-soil interaction – and the mechanical response. As the scale increases, fully-coupled hydro-mechanical approaches gradually give way to simpler sequential couplings of hydrological and infinite slope or rigid block models, and finally to geostatistical models. At any scale, while the role of temperature in controlling evapotranspiration and thus the hydrological balance is well recognised, direct thermo-mechanical couplings are systematically neglected unless changes in water phase are involved. This contrasts with abundant experimental evidence of a fully-coupled thermo-hydro-mechanical behaviour of most geomaterials, even in ranges of temperature naturally experienced at the ground surface or in the near subsurface. Here, we review temperature-dependent processes that are potentially relevant to slope stability, focusing in particular on clayey slopes in temperate and warm regions. Our thought-provoking hypothesis is that temperature fluctuations and trends induced by climate change may exert, in short to long terms, a hydro-mechanical forcing on slopes (by altering permeability, water retention capacity, compressibility, shear strength). Together with other known effects (such as altered precipitation patterns and changes in land use), they could affect landslide activity and the distribution and frequency of slope failures. To verify this hypothesis across the scales, systematic field monitoring of temperature-related variables is necessary, together with geostatistical analyses entailing thermal remote sensing products. At the same time, fully-coupled approaches need to be upscaled to permit physically-based catchment- or regional-scale studies accounting for appropriate temperature-related variables and the inherent heterogeneity in materials and boundary conditions.