基于Logistic回归分析的穆棱市地质灾害风险分析

Risk Assessment of Geological Hazards Based on Logistic Regressive Analyse for Muling County in Heilongjiang Province

通过分析穆棱市自然地理环境、区域地质条件、人类活动等与地质灾害的空间分布关系,得到植被覆盖度、坡度、土地利用类型、土壤类型、降水和地貌岩性等作为地质灾害影响因素,将影响因子引入二元Logistic回归分析,构建出穆棱市地质灾害风险概率模型,基于此模型开展了穆棱市地质灾害的敏感性区划,并进行了地质灾害防治分区。通过随机点验证和ROC曲线的检验发现,该模型的评价精度达到70%以上,模型的预测结果可行有效,具有较高的准确度,适用于穆棱市潜在地质灾害的预测,可作为穆棱市地质灾害防治的参考依据之一。

This paper analyzes the space distribution relation-ship between regional geographical conditions, natural geographical environment, human activities and the geological hazard. It is found that vegetation coverage degree, slope, soil types, geological lithology, precipitation and the land use types all have effect on the geological hazard, we introduced these impact factors into the Logistic regression model, build the occurrence probability model of geological hazard, get the sensitivity region classification. And the overall accuracy of this model is over 70%, the result is acceptable and suitable for the potential geological hazards prediction and has a high accuracy. The model have great significance for the control of geological hazard for Muling County.