基于MODIS数据和GWLR的甘肃省火灾风险模型与区划

Model and Zoning of Fire Risk in Gansu Province based on GWLR and MODIS Imagery

利用甘肃省2000~2014年MCD45A1数据产品,建立了火灾风险评估的GWLR和BLR模型,从省域和较长时间尺度上进行火险风险的模拟,并进行火险区划。研究表明:GWLR比BLR模型的AICc值更小,可靠性、区分性更强,更适于火灾建模。甘肃省内火点多集中于坡度低于25°,海拔介于1 000~3 000m,坡向介于0°~90°,年均温3℃~11℃,降水量200~400mm,距离公路更近的区域,以及NDVI>0.6或NDVI<0.3,0.2<GVMI<0.4的区域。将甘肃省划分为极高火险区、高火险区、中火险区、低火险区4类火险区。气温、地表温度、NDVI、距离和高程对于甘肃省火灾的发生具有空间上的影响。通过运算形成火灾影响因素空间分布图,可为防火措施的科学实施提供参考。

The remote sensing data sets, MODIS, of Gansu Province and the fire influencing factors from 2000 to 2014 was used to build the GWLR and Logistic model.Fire risk zoning study was conducted in lager temporal scale and provincial spatial scale.Result indicate that fire points are more concentrated in the slope of less than 25°,an altitude ranging between 1 000-3 000 m,slope between 0°-90°,the average annual temperature in 3℃-11℃ ,precipitation 200-400 mm,NDVI〉0.6 or NDVI〈0.3,0.2GVMI〈0.4,and near by the road .GWLR is more stable and reliable based on the smaller AICc of GWLR.Fire area of Gansu was divided into low, moderate, high, and extremely high fire risk zones.Temperature （QW）, land surface temperature （LST）, Normalized Difference Vegetation Index （NDVI）, distance to the nearest path （DL） and altitude（GC） have influenced the fire distribution in Gansu.The fire prevention measures should be take advantage of the distribution map of fire influencing factors.