Integration of geographic information system-based method in weight assignment and parameter justification for rockfall hazard rating system

Abstract

The purpose of this study is to do a comparison study of integration of heuristic AHP and GWR methods in predicting rockfall hazard. Rockfall Hazard rating System (RHRS) has been an established method to monitor the rockfall hazard leading to development of other system to study rockfall across different countries. The data used for this study involves classified slope datasets in East-West Federal Highway from Gerik to Temenggor, Perak. The study shows that the hazard map produced from AHP yield better predictive model due to the high percentage match of 85 % with the original score from the field. Heuristic method produces better results for rockfall prediction using only 10 parameters. Regression method are proven to be unsuitable for predictive model if the dataset is small with only several parameters. Addition of parameters in a larger study area may improve the predictive score for GWR. GWR lower score than OLS suggest that the occurrence of rockfall may not be heavily influenced by the surrounding factors, thus a general parameter’s coefficient is enough to predict the rockfall hazard.