基于频率比和逻辑回归模型的东北地区火险制图研究

Mapping Fire Occurrence Susceptibility in Northeast China:Comparison of Frequency Ratio and Binary Logistic Regression

以我国东北地区为例,选取了地形、可达性和植被覆盖共7个静态预测因子,分别使用频率比和逻辑回归模型对火险敏感度进行制图表达。使用历史火灾数据和AUC方法对模型进行内部检验和独立检验,结果显示:1)开发的两个模型具有较好的可靠性和区分能力,BLR模型的区分能力更稳健。2)选取的几个静态因子对东北地区火险空间格局起着结构化作用,表明在区域空间尺度上使用静态预测因子对火险建模的可行性。最后,研究了2000-2009年东北地区的火灾事件响应敏感区。

Fire occurrence suseeptibility mapping provides the information on the spatial distribution of fire danger and it is valuable to fire management. Take Northeast China as the study area, seven static fire-influeneing factors involving topography, accessibility and vegetation eover are selected and multi-temporal random sampling methodology is employed to create the training subset,and then the training subset is utilized to develop the frequency ratio（FR） and binary logistic regression（BLR） models which are used to map the fire danger of Northeast China. The cumulative frequency curve of fire danger ranks is drawn and the area under ROC curve（AUC） procedure is employed to validate the reliability and the diserimination eapacity of the developed FR and BLR models respectively. Evaluated by the inner testing and independent validation,good reliability and reasonable discrimination capacity of the two developed models can be conclude＆ When the difference of the reliability is compared between the FR model and the BLR model, it is six of one and half a dozen of the other. However, the discrimination capacity of the BLR model is more robust than that of the FR model. The study suggests that the static predictors employed play a structural role in forming spatial pattern of fire danger in Northeast China. In addition, those regions that are sensitive to fire events from 2000 to 2009 are illuminated.