Environ Sci Pollut Res Int. 2022 Aug 24. doi: 10.1007/s11356-022-22649-x. Online ahead of print.
ABSTRACT
Environmental geological disasters pose a significant threat to human life, property and environmental safety. It is necessary to conduct targeted governance in key prevention and control areas based on reasonable susceptibility assessment. Using the debris flow disaster in Xiuyan County as an example, this study compares and analyzes prone prediction models such as the frequency ratio (FR), decision tree (DT) and random forest (FR) models and evaluates the cost of prevention and control and the protection of life and property. The research results show that the FR, DT and RF models have good performance. The ROC test, disaster point density statistics and cross-validation results show that the RF model has the best performance. The study area was mainly less and mildly prone areas. The highly prone areas are mainly distributed in the northeast and southwest of the study area. It is the key area of disaster prevention and control. Elevation, rainfall intensity and population density have the largest influence on the susceptibility to debris flows. Based on the RF model, the disaster points in the highly prone areas account for 54% of the disaster points of the whole area, and the project treatment cost of the disaster points is 0.78 million yuan per single gully, which protects 56% of the lives and property in the study area, which is better than the DT and FR models. The RF model not only has good prediction performance in terms of susceptibility. It can realize the targeted management of disasters, achieve the targeted investment of governance costs and the effective protection of life and property and serve the sustainable development of the regional environment and economy with greater value.
PMID:36001261 | DOI:10.1007/s11356-022-22649-x
