Risk prediction and regression analysis of debris flow disasters: Taking the debris flow disaster in Weiziping Village, Xi’an as the object

Abstract

[Objective] To study the movement characteristics and key influencing parameters of debris flow, construct a debris flow disaster risk prediction model, and provide data support for assessing the disaster risk of transportation infrastructure caused by debris flows. [Methods] Taking the debris flow disaster in Wei Ziping Village, Xi'an City, Shanxi Province as the research object, a debris flow movement model based on the depth-integrated continuum mechanics theory was established. The friction coefficient and debris flow discharge were selected as key parameters to explore the variation laws of debris flow velocity and mud depth in the accumulation area, and evaluate the risk level of the area where the debris flow exits the gully. Combined with the multiple linear regression, polynomial regression and support vector machine models, the prediction equations for the relationship between debris flow discharge, friction coefficient and flow velocity, and mud depth were established. [Results] There is a significant positive correlation between the flow velocity and the flow rate of debris flow. The flow velocity fluctuates under different flow rate conditions, with the maximum flow velocity ranging from 2.58 m·s^-1 to 8 m·s^-1. The mud depth keeps increasing with the increase of flow rate, and the maximum mud depth range is 0.5 m to 4 m. The friction coefficient is negatively correlated with both the flow velocity and the mud depth of the debris flow. With the increase of the friction coefficient, the maximum flow velocity and the maximum mud depth show an overall downward trend. The risk prediction result show that the risk of debris flow increases significantly with the increase of debris flow. The increase of the friction coefficient can effectively reduce the risk, and the effect is significant at a lower debris flow, while the effect of frictional resistance is not significant at a larger debris flow. The support vector machine model has a better prediction effect on the maximum flow velocity of debris flow, while the polynomial regression model has the best prediction effect on the maximum mud depth in the debris flow accumulation area. [Conclusion] Monitoring the flow of debris flows to predict the flow velocity and the depth of the accumulated sediment in the affected area plays a crucial supporting role in pre-disaster assessment of the risk level of debris flow and post-disaster auxiliary evaluation of the severity of debris flow.

Keywords

debris flow / motion parameters / dynamic characteristics / risk assessment / risk prediction model / influencing factors / support vector machine model / engineering geological disasters