Theoretical research on the relationship between return period of rainfall and shallow landslides

A hydrogeomorphological model for shallow landslide prediction that considers both the stochastic character of rainfall intensity and duration and the deterministic aspects controlling slope stability is presented. The probability of shallow subsurface flow (i.e. the process triggering shallow landsliding) is expressed by a lognormal distribution. The short-term probability of landsliding is defined as the probability that the saturated depth exceeds a critical value. The average recurrence interval of landsliding T,v can be defined as the expected value of the recurrence interval of landsliding. This model was applied to a field site in Japan where extensive shallow landsliding occurred as a consequence of a heavy rainstorm in 1988. A digital elevation model was utilized to calculate T-av for every grid cell. it was found that T-av is a useful index to correlate to the susceptibility to shallow landsliding. A time series of landsliding was then simulated using this model. A series of values of return periods of rainfall (and, consequently, of saturated throughflow) were randomly generated for 10000 years (one rainstorm event per year). It was found that the spatial extent of landsliding (expressed as a percentage of scar area) associated with each rainstorm event was strongly influenced not only by rainfall, but also by the historical sequence of landsliding. The effective return period of rainfall for shallow landsliding was then estimated using the same model. It was found that rainfall events with return periods <500 years and <1000 years trigger about 50% and 65% respectively of the total number of landslides over the long term. Copyright (C) 2004 John Wiley Sons, Ltd.