Improvement of failure time prediction of rainfall-induced landslides while reducing the fluctuation by the relationship between velocity and acceleration

Abstract

Background: Measurement of surface displacement (SD) plays a major role in predicting the failure time of landslides triggered by rainfall and is the basis for predicting the failure time of a slope Fukuzono (In proc. IVth International Conference and Field Workshops on Landslides, Tokyo, 1985). To study this concept, small-scale indoor slope models of sandy materials were tested under simulated rainfall with constant discharge to observe the behaviour of SD. Five cases of experiments were conducted by changing the conditions of the model, such as rainfall intensity, slope inclination and void ratio. Then, the relationship between the velocity and acceleration of the SD was examined for each case. Accordingly, failure time was predicted using two different methods and considering the behaviour of the slope just before failure. The first method was the original Fukuzon’s inverse-velocity (INV) method, which uses the inverse number of velocities and the increment of velocities at two different times. The second method was a new approach based on the velocity and acceleration (VAA) method and was carried out using the logarithmic acceleration and logarithmic velocity values for linear regression analysis. Results: The linear relationship between velocity and acceleration on a logarithmic scale could be established under different stress conditions. The relationship between velocity and acceleration was identical under different stress conditions within the small range of experimental conditions and variations in the small-scale indoor model slope. However, in the relationship obtained from the natural slope experiment, Futtsu revealed that the relationship varied slightly from the indoor small-scale model slope’s results. The analysis results showed that the VAA method could acquire a more accurate failure time than the INV method. Conclusions: The linear relationship between the velocity and acceleration of the SD shows that an early warning system for rainfall-induced landslides can be well established based on the relationship between the velocity and the acceleration of SD. The most accurate failure time prediction by the VAA method was due to reducing the scattering of the velocity values generated by the noise of the measured SD data using the least-squares method. In contrast, the individual velocity values greatly affected the INV method.