Physically based dynamic modelling of the effects of land use changes on shallow landslide initiation in the Western Ghats, Kerala, India

The highland region forming the western slopes of the Western Ghats of Kerala state, India, is increasingly affected by shallow landslides and consequent debris flows. Many researchers suspected that the spatio-temporal probabilities of shallow landslide initiation in this area are dependent on the land use which in turn determines the mechanical and hydrological effects of vegetation on slope stability. A physically-based spatially distributed shallow landslide initiation model named STARWARS (Storage and Redistribution of Water on Agricultural and Revegetated Slopes) + PROBSTAB (Probability of Stability) model was identified as the most suitable for this research. This model was calibrated and validated for the Aruvikkal catchment (~9 km2). Climatic and hydrological data necessary for this was acquired from 13 automated open stand pipe piezometers, one discharge station (stage gauge) and one weather station starting from May 2007 onwards. Time series analysis revealed that the discharge of the catchment will respond within 1 hr after a rainfall event while perched water level may need about 6 hrs. Soil properties necessary and the soil depth were measured at representative locations and spatially interpolated using geostatistical techniques. Depending on the dominant plant species in each land use unit and the soil depth the corresponding root reinforcement applicable was derived from measured root tensile strength, pull out strength, root diameter and root density data using the perpendicular root model. Nine species of plants were tested of which Teak (Tectona grandis) trees offered the highest amounts of net root reinforcement. In order to evaluate the effects of long term and short term land use changes on slope stability in the region, historic (1913 to 2008) land use maps, soil depth maps and root reinforcement maps were derived. In addition the land use maps and corresponding root reinforcement maps of two future scenarios (2016 and 2058) were also derived. These data sets were used to simulate the corresponding slope stability and probability of failure conditions based on the 1985 rainfall time series which included the most extreme daily rainfall that has caused at least one shallow landslide in the study area in the past 57 years. In light of the empirical observations and the modelling results it can be concluded that a continuous rain storm of 4 to 26 hrs, totaling about 106 to 290 mm can cause a steep rise in the perched water table up to critical levels in regolith filled bedrock depressions and the persistence of this level for ~10 hrs may lead to shallow landslides in the study area. It was evident that land use changes which occurred in the early part of 20th century have reduced the root cohesion and altered the soil depth significantly by terracing. The land use changes from the pre-plantation (1913) scenario to present (2008) have resulted in an average increase in the potential area of failure by 43% and the spatio-temporal probability of failure by 49%. Thus, the research showed that in the study area the transition probability of land use change (and consequently the changes in soil depth and root cohesion) outweighs the rainfall quantity in determining the spatio-temporal probability of shallow landslide occurrence. This is in contradiction with the commonly held belief that the temporal probability of shallow landslides can be quantified with only the return probability of landslide-causing extreme rainfall events. Hence, transition probabilities of land use should be assessed and incorporated in regional scale landslide hazard assessments that utilize heuristic and stochastic techniques, especially in such anthropogenically modified terrains