以1980—2009年伊春地区林火发生数据为基础,综合考虑高程、植被类型、基础设施和气象条件等因素,运用随机森林算法对伊春地区的林火发生影响因子及其影响力进行分析。结果显示:当日平均相对湿度小于30%时,火灾发生的概率最大;日最高地表温度在15℃的范围内对林火发生的影响最小,大于15℃之后对林火发生的影响逐渐增大。伊春林火主要集中在海拔0~400 m及距离居民区15~30 km的范围内。相对湿度对林火发生的影响最大,地表温度对林火发生的影响次之,距居民区距离对林火发生的影响最小。此外,随机森林算法对伊春地区森林火灾具有较高的预测精度,达80%。林火发生概率插值图和火险区划图显示,伊春地区林火发生主要集中在中部、东南和最北部地区,西南、东北地区也存在高火险地区。%We used random forest algorithm and took into account the elevation , vegetation types , infrastructure and meteorologi-cal factors to analyze the driving factors of forest fire and their effect powers based on the forest fire data set in the period of 1980-2009 .The fire occurrence possibility reached highest when the daily average relative humidity was lower than 30%.The daily maximum ground surface temperature significant influenced fire occurrence at around 15℃, and the higher temperature increased the influence power on fire occurrence .The fires in Yichun more likely occurred at altitude of 0-400 m and the regions which was 15-30 km away from settlements .The relative humidity had the most important influence on fire occurrence , followed by ground surface temperature , and the distance to settlement had relatively less influence .Ran-dom forest had a high predictive ability on fire occurrence , reaching to 80%correctness .By maps of likelihood of fire oc-currence and fire risk division , forest fire occurred mainly in the central region , southeastern and northern Yichun .South-west and northeast region of Yichun also had fire prone areas .
展开▼