Evaluation of geohazard susceptibility based on information value model and information value-logistic regression model: A case study of the central mountainous area of Hainan Island

As the basis of geohazard risk evaluation, the geohazard susceptibility evaluation can objectively and accurately reflect the probability of geological hazard occurrence by using quantitative mathematical statistics. This article takes Wuzhishan city, where occurs the most geohazards in Hainan Island, as an example. Factors including structure, rock formations, slopes, topographic undulations, altitude variation coefficients, normalized differential vegetation index (NDVI), rainfall, river systems, roads, and curvatures were selected as evaluation indicators and applied in both information value model and information value-logistic regression model. In the end, by comparing and analyzing the accuracy and adaptability of both models, the article ends with the conclusion that the high-prone areas are mainly distributed along roads and rivers in the mountainous areas, and the extremely low-prone areas are mainly located in the areas where have no rivers, valleys and human activities. In addition, the results also revealed that the prediction accuracy meets the requirements of susceptibility evaluation owing to the high AUC (area under the curve) values occupying 0.897 and 0.896 respectively in both models. Evaluation factors such as rainfall, elevation variation coefficient and highway play a remarkable role on the development of geohazards. Furthermore, it is indicated by experiments that the information value-logistic regression model has better prediction accuracy than the other. The research results provide a scientific and effective discrimination method and a prediction approach for geohazard risk evaluation in this area.