Rainfall-based criteria for assessing slump rate of mountainous highway slopes: A case study of slopes along Highway 18 in Alishan, Taiwan

A data-driven approach is considered in the present to investigate the relationship between the precipitation (rainfall) and the slump rate of highway slopes along Highway 18 in Alishan, Taiwan. The basis for the analysis presented herein is a set of 103 slope failure records collected from the study area, where each record consists of entries for daily rainfall, effective rainfall and other topographical and geological factors. To begin with, Artificial Evolution Neural Network (AENN), a data-driven approach, is first used to learn from past slope failure records. The results show that the developed AENN can accurately predict the occurrence of slope failure with a success rate of 99%. Then, this AENN model is employed to study the slump rate of slopes in the study area, and the computational results show that the slump rate can be accurately assessed based on the precipitation data that consists of daily rainfall and effective rainfall. A rainfall-based criterion is further developed as an early warning indicator for accelerated slump rate of slopes in the study area.

Research Highlights
► The Artificial Evolution Neural Network was performed to assess the slope failures along the Alishan Highway.
► The slump rate can be accurately assessed based on the daily rainfall and effective rainfall.
► A rainfall-based criterion is further developed as an early warning indicator for accelerated slump rate of slopes in the study area.