关键词: MaxEnt, 华蓥山, 林地, 浅表层滑坡, 风险, 植被

Abstract:

Objective: The aim of this study was to determine the relative contribution rate of environmental variables to forestland superficial landslide risk, to clarify the key vegetation factors affecting the superficial landslide risk and their disaster reduction range, and to reveal the coupling effect of vegetation and non-vegetation factors on the superficial landslide risk. This research can provide important theoretical support for landslide risk assessment and disaster mitigation decision. Method: Taking the forest land in Huaying Mountains as the research object,17 environmental variables were selected, and the MaxEnt model was used to determine the relative contribution rate of environmental variables to the prediction of superficial landslide risk of the forest land, the response changes of superficial landslide risk to various factors were analyzed based on the existence of vegetation factors. Result: 1) The ROC test results of the model accuracy showed that the simulation accuracy of the model was 0.887 without considering the vegetation factor, which reached a very accurate level of accuracy, while in the case of considering the vegetation factor, the simulation accuracy of the model was 0.915, which reached an extremely accurate level of accuracy, and the accuracy was improved by 3.1%. 2) The cumulative contribution rate of engineering geological rock group, forest volume, distance to fault, terrain relief, elevation, green-red vegetation index, plane curvature, and stand type to the prediction of superficial landslide risk reached 80%. Vegetation factors played an important role in the superficial landslide risk prediction, mainly reflected in forest volume, vegetation coverage, and stand type. 3) The existence of vegetation factors changed the response of superficial landslide risk to five variables including plane curvature, slope aspect, elevation variation coefficient, slope variable, and profile curvature. Vegetation factors weakened the superficial landslide risk caused by plane curvature, elevation variation coefficient, and profile curvature, and the reduction rates were 4.9%, 5.9%, and 8.1%, respectively. Vegetation factors aggravated the superficial landslide risk caused by slope variability, with an aggravating rate of 10.9%, and it had both positive and negative effects on the superficial landslide risk generated by the slope aspect, with the aggravation and reduction rates of 12.8% and 6.4%, respectively. Conclusion: Our results demonstrated that the MaxEnt model had high simulation accuracy for the simulation of superficial landslide risk of forest land, and it expressed the response of the superficial landslide risk to various influencing factors intuitively. When using this model to predict the superficial landslide risk of forest land, in addition to the conventional influencing factors such as geology, topography, landform, soil, etc., the vegetation factor was also a key environmental variable in the model simulation, which had an important contribution to the simulation accuracy. The existence of vegetation factors generally did not change the response trend of the superficial landslide risk to other influencing factors, but would have an important impact on the superficial landslide risk caused by the extreme values of some non-vegetation factors, showing a coupling effect, which might aggravate or weaken the superficial landslide risk.

Key words: MaxEnt, Huaying Mountains, forestland, superficial landslide, risk, vegetation