Susceptibility Assessments and Validations of Debris-Flow Events in Meizoseismal Areas: Case Study in China's Longxi River Watershed

The 2008 Wenchuan earthquake triggered tens of thousands of landslides, resulting in rich loose materials for the 8.13 debris flows. Longchi Town in southwest China's Sichuan Province was one of the worst hit areas both by the Wenchuan earthquake and 8.13 debris flows. The loose source materials generated by coseismic landslides in the Longxi River watershed of the upper Min River are widely distributed and cover an area of 7.97 km2. The volume has reached as high as 13,364.99x104 m3. After catastrophic debris flows occurred in Longchi Town on August 13, 2010, there were still large amounts of loose debris remaining. This presented major potential risks for additional debris-flow formations in the meizoseismal region. In this paper, an index of entropy models and logistical regression models was implemented for the purpose of identifying the areas that were potentially susceptible to debris-flow formations based on the same catchment units. In order to reflect the impacts of the Wenchuan earthquake on the subsequent debris-flow events, the volumes of the source debris accumulations were calculated in each catchment unit. These were then incorporated into the models as controlling factors. The performances of the models were evaluated using the receiver operating characteristic curve (ROC) technique. Then, the quality ratings of the susceptibility maps were compared and verified using the results of the field investigations. The spatial mismatches of the index of entropy (IE) and logistic regression (LR) models with the observed best predictive abilities were 16.7% and 8.3%, respectively, were revealed in the susceptibility maps. The LR model showed better performance results in the division of the susceptibility zones, and also in the reliability and robustness of the assessment results. The two models had both incorporated a spectrum of the leading factors for the debris-flow occurrences, including the lithology; distances from the coseismic surface fault-ruptures; densities of the loose source material; monthly average precipitation rates; and slope gradient rates, which are necessary for debris-flow formations in strong seismic zones. The debris-flow prone areas were found to be mainly distributed within 4 km of the Beichuan-Yingxiu fault. Furthermore, the verification results showed that more than 90% of the actual debris flows were distributed in the very high and high susceptibility zones. Sufficient attention should be paid to debris-flow prevention and early warning procedures in very high and high susceptibility zones. The results of this study provided practical implications for developing debris-flow disaster management and hazard mitigation strategies in similar earthquake-prone environments.