A stochastic hydro-geomorphological model for shallow landsliding due to rainstorm

A prediction model of shallow landsliding is proposed. It considers not only the deterministic aspects containing slope stability, saturated throughflow and a soil (regolith) depth development, but also the stochastic aspects of intensity and duration of rainfall. It turns out that the probability of saturated throughflow, which is the direct trigger mechanism to shallow landsliding, can be expressed by a log-normal distribution. The short term probability of landsliding is defined as the excess probability that the depth of saturated throughflow surpasses the critical value. The average recurrence interval T-av of landsliding can be calculated as the expected value. This model was applied to a test field where a lot of shallow landsliding occurred at a heavy rainstorm in 1988. Then a DEM of 5-m grid interval was utilized to calculate T-av at every grid point. Consequently, it was found that the percentage of the landslide grid number to the total grid number for every T-av rank increases when T-av decreases. Therefore, it is confirmed that T-av is an index of the susceptibility to shallow landsliding and the distribution map of T-av can be regarded as a kind of hazard map. The spatial distribution of T-av reveals its significant dependence on the topography. (C) 1999 Published by Elsevier Science B.V. All rights reserved.