Hazard assessment of debris flows based on a PCA-GRNN model: a case study in Liaoning Province, China

Abstract

In recent years, many studies have investigated the dangerous range of debris flows, but the accuracy with which the dangerous range is predicted is low. A novel hybrid generalized regression neural network (GRNN) model combined with principal component analysis (PCA) is proposed to predict the debris flow hazard range. First, through a detailed analysis of the factors influencing the development of debris flows in the study area, six factors related to the source conditions, dynamic conditions, and accumulation characteristics are selected. Second, PCA is used to analyze the correlations among the factors and reduce the number of dimensions; this method can optimize the parameter selection and improve the prediction accuracy of the model. Finally, the extracted features and the selected parameters for the GRNN model are employed to predict the hazard range. Experimental results show that the PCA-GRNN model boasts a strong nonlinear prediction ability. When 30 small samples of debris flow disaster points are selected, the error is 8.75%, which is less than the error of 11.32% of the traditional large-sample method. Therefore, it is feasible to use the PCA-GRNN method to predict the debris flow hazard range.