Empirical relationships for the estimation of debris flow runout distances on depositional fans in the Wenchuan earthquake zone

Abstract

The Wenchuan area has become highly susceptible to landslides and debris flows since the earthquake occurred on 2008. A detailed debris flow hazard assessment is necessary to provide information for future risk management. Debris flow runout on a depositional fan is an important factor for assessing debris flow hazard, and its estimate becomes essential for the planning of the mitigation works that must be built. We therefore developed two simple empirical relationships based both on univariate and multivariate approaches, which provide the runout distance using the data for 134 channelized debris flow events in 134 catchments in the Wenchuan earthquake zone. We generated a correlation matrix of the debris flow runout distances and the relevant variables. Because of the high correlations, the independent variable debris flow volume (VD) was selected for the univariate approach. Using a multicollinearity analysis and a stepwise regression technique, the catchment (basin) internal relief (H), and VD were used to develop a multivariate runout relationship. The coefficients of the variables were obtained using 80% of the dataset (training dataset); the remaining 20% was used for test of estimated performance by comparing the computed runout distances with observed values. The validation demonstrated that the proposed relationships are suitable for estimating the runout distances of debris flows on depositional fans in the Wenchuan earthquake zone. The univariate runout relationship has the advantage of being simple, whereas the multivariate runout relationship provides higher accuracy. Additionally we rearranged and reformulated existing relationships using the same training dataset and compared the results with those from the relationships here proposed. The estimations provided by our relationships were the closest to the observed values. The presented approaches may be applied to estimate debris flow runout distances in other areas after they are retrained using local datasets.