The Relative Importance Analysis Of Debris Flow Factors Based On Hierarchical Partitioning And Dominance Analysis

The Bailong River Basin is one of four most serious geological hazard areas in China. It is located in the interchange among the Qinghai-Tibet Plateau, the Loess Plateau and the Sichuan Basin. From the tectonic view, it is located in the eastern boundary of Cenozoic India-Asia collision zone with heavy neotectonic and frequent earthquake. In this area, the geological hazards occur frequently, which threatens the local people’s lives and property seriously. The debris flow is the result of multiple factors whose effects are different. It is significant to study on the causes of the debris flow and find the factors that play a role on the debris flow, which becomes a hotspot.Based on the historical materials and field survey of debris flow in the BaiLong River Basin, several hazard factors were selected according to the results of previous studies and experts’ advices. The logistic regression model of geological disasters was established by using the GIS spatial analysis and mathematical statistics. Eventually, the contribution of each factor was sorted according to the relative importance of hazard factors, which was quantitatively analyzed by dominance analysis and hierarchical partitioning. The results are as follows:1. The valley unit properties were extracted according to historical materials and field survey. Then the spatial database of debris flow’s factors was built based on Geodatabase, which laid the foundation for study on the impact factors of debris Flow.2. According to the spatial distribution, formation factors of the debris flows, the single factor was under the correlation and sensitivity analysis; combined with the expert advices, the factors of debris flows were selected, i.e., fault distance, proportion of unstable lithology, loose deposit index, relative height of main valley, slope of main valley, cutting density, density of landslide, NDVI, proportion of land use for human activities,10-minute average precipitation.3. The factors of debris flow were selected using stepwise regression, and then the logistic regression model between debris flow hazard and factors was established. The results were verified using the Receiver Operating Characteristic curve (ROC) and hazard risk zoning map.4. The contribution of each factor was sorted according to the relative importance of hazard factors, which was quantitatively analyzed by dominance analysis and hierarchical partitioning. The result was:relative height of main valley> slope of main valley> density of fault distance>density of landslide>10-minute average precipitation> loose deposit index> proportion of land use for human activities. The results indicated that the topographical and geological factors played a greater role in the formation of debris flow.5. The above results were taken as the risk weight of the debris flow factors, and the risk for each valley was calculated by the debris flow hazard index. Two hazard zoning map of the study area were manufactured to verify the results of two method.