Probability of Occurrence and Extent of Rainfall-Induced Landslides, Hawke's Bay, New Zealand

Abstract

A probabilistic frequency magnitude model for rainfall induced landsliding has been developed in order to link return periods of rainstorms to the extent of landsliding. The link between rainstorms and landsliding was established through determination of critical water contents for slope failure. The study site is a hillslope in the landslide prone Lake Tutira catchment, Hawke's Bay, New Zealand. Data recorded with a monitoring network at the study site formed the basis for the model. The network recorded water contents, pore water suction, positive pore water pressure and weather data. Methodologically the research aims to link geomorphology, slope hydrology and geotechnical principles if the analysis of stability. Since the landslides under investigation are rainfall triggered, the first and most important task is to determine the role of water in the stability of the slope, and consequently establish a critical water content for slope failure. The high spatial variability of input data for the stability analysis suggests the application of a probabilistic approach and Monte Carlo Simulations are used for this part of the study. The results indicate that the failure probabilities are related to the critical water contents which in turn depend strongly on the slope angle. Therefore, failure probabilities can be expressed as functions of critical water contents which are different for each slope angle class. In order to link failure probability to rainstorm size, the data obtained from the monitoring network are used to establish an empirical model of water content change due to rainstorms, based on the physical understanding of the water movement process. Results suggest that water content changes are related to rainstorm size as well as antecedent moisture conditions and topography. The relationship between water content changes and topography allows the transfer of results from the study site to the larger adjacent Waikopiro catchment, in order to test the model. In a GIS the results from above were combined and the effect of rainstorms of certain return periods (5, 10, 20, 50 and 100 years) on water contents were modelled on a seasonal basis for the Waikopiro catchment. Considering soil properties as constant throughout the Tutira and the Waikopiro catchment, failure probabilities for water contents and slope angle classes established with the Monte Carlo Simulations can be applied to the Waikopiro catchment. The ultimate result provides a method to determine the impact of rainstorms on the probability of landslide occurrence and can be used as a means of assessing the spatial extent of landslides triggered by rainstorms of a certain return period.