Modelling shallow landslides: the importance of hydrological controls and lateral reinforcement

Abstract

Shallow landslides are important as geomorphic agents of erosion, sources of catchment sediment and potential hazards to life and infrastructure. The importance of these mass movements is difficult to define using solely field- based approaches because these are often too limited in both duration and resolution to fully determine the magnitude and frequency of these processes. Modelling is a powerful alternative tool for providing insight into underlying processes governing shallow landslides and for testing new hypotheses regarding environmental and land-use change impacts. The explanatory power of models is a function of their process representation and predictive ability. Current models suitable for catchment-scale application provide valuable probabilistic information on failure, but not detailed deterministic predictions. Using the English Lake District as a study area, this thesis addresses three issues necessary to provide the process-basis of these probabilistic analyses. First, poorly constrained or spatially variable input parameters such as soil depth, root reinforcement or material properties are often used to explain the locations of failure within a larger area that has a high, sometimes equal, probability of failure. The thesis develops rigorous new methods to quantify and minimise error in these parameters, representing them as distributions to capture both their natural variability and the error in their measurement. Results suggest that lateral root reinforcement even for grasses and shrubs may provide important additional strength (as much as 6 kPa) in the top 0.5 m of the soil. Second, infinite slope stability analysis neglects important additional lateral friction and root reinforcement effects at the margins of an unstable block. More sophisticated three-dimensional stability analyses can represent this process but are limited in their applicability by computational and data resolution requirements. This thesis derives from first principles a set of analytical governing equations for three-dimensional analysis; tests these against benchmark geotechnical methods; and applies them to establish key landslide scaling relationships. Third, shallow landslides in the UK are almost exclusively hydrologically triggered, resulting from local high pore water pressures. In line with the current paradigm existing stability models assume that the topography plays a dominant role in defining the spatial pattern of soil moisture and therefore pore water pressures in the landscape. This hypothesis is tested: first at the hillslope scale (10(^1) km(^2)) with a network of ֊100 wells; then the catchment scale (10(^2) km(^2)) using high resolution orthorectified aerial photographs to identify vegetation indicative of wet habitats and applying these as a proxy for soil moisture. These studies indicate that, for the case-study, wet areas are controlled at the landscape scale by a set of broad topographic limits in terms of slope and contributing area. Within these there is considerable scatter, resulting from the interplay of local factors such as: bedrock topography, preferential flow and soil stratification. Lateral root cohesion represents an important source of additional strength which can be included within analytical stability equations to create a threshold dependence on landslide size. Patterns of instability will then depend on the spatial pattern of other influencing factors (e.g. soil strength and pore pressure). At present the limits to available data and our understanding of hillslope hydrology constrain our ability to predict slope instability in environments like the Lake District. Future research might usefully identify landscape scale controls on this predictability.