Study On The Assessment Of Fire Risk And Zoning In Gansu Province

Fire which burns a lot of vegetation resources every year is a global natural disaster,not only caused huge economic losses,but also seriously damaged the natural environment and ecological balance of human survival.The ecological environment of Gansu province is fragile and the vegetation resources are relatively poor.Thus,the scientific fire assessment and warning are beneficial to the protection of ecological environment in Gansu Province.The application of 3S technique,namely,Remote Sense(RS),GIS(Geographic Information System),GPS(Global Position System),can supply informative,fast,high reliability,intuitive dynamic information and technical analysis tools for achieving the time span of fire risk analysis and research division.The remote sensing data,fire statistics data,meteorological data from 2000 to 2014 were used to build fire risk model,by means of ArcGIS,ENVI,SPSS and other software,which was used to divided into fire danger in Gansu Province.The result provides reference for the scientific implementation of fire prevention measures.The following conclusions are drawn through the study:(1)With frequent fire and serious loss in Gansu Province,the total of 37102 fires directly resulted in economic losses of about 4.18×10~8 million during 1996 to 2012.From the perspective of time,the number of fire in 1996~2000 was lower than the average level and rose slowly;however,from 2001 to 2006,the number of fire was higher than the average level;nevertheless,during 2007to 2011 the fire decreased rapidly.Coming to 2012,the loss of fire was the most serious from 1996to 2012.According to the basic situation of forest fire in Gansu Province from 1997 to 2013,forest fires occurred 293 times and caused forest area of 995.9 hectares.Other losses converting into cash are about 8.6×10~7 RMB.Particularly in 2003,the number of forest fires,the fire area,affected forest area and other losses were the highest in nearly 20 years.(2)The area of fire considerably varies in temporal and distribution.The study found that:from the perspective of space,the burned area mainly distributed in the east of Zhangye City,the south of the Jinchang City,the east of Qingyang City,the west of Pingliang City,the south of Tianshui City and south of Gannan Autonomous Prefecture and other places.From time distribution:the number of burned area,up to 1836,was the largest in 2003 among nearly 20 years;the number of burned area,reached 4629,was the largest in autumn among 4 seasons;from the month view,the number of burned area concentrated on September,October,and November.(3)Different factors of fire risk influenced diversely to frequency of fire point.According to the relationship between frequency of fire point and climate factors,vegetation factors,elevation factors,human factors,fire point concentrated on the area of average annual precipitation of 200mm~600mm,average annual temperature of 4~6?,elevation of 1000~3000 m,slope<20 degrees,south aspect and high vegetation coverage,especially cultivated vegetation,grassland,broad-leaved forest.Referring to the relationship between fire frequency and years of per capital GDP,average population density,distance from road distance and distance to residential area which is a logarithmic negative correlation.Fire points ofen locate in the concentration of GDP<3000 RMB per capital,the average population density<2000,distance away from the road<10000 meters,distance from the residents of these areas<10000 meters.(4)The BLR(Binary Logistic Regression)and GWLR(Geogrophical Weighted Logistic Regression)fire risk models were established by using the remote sensing data of Gansu Province during 2000~2014.The area under the ROC curve of GWLR was 0.830,which was significantly higher than BLR's 0.733.The AIC and AICc values of BLR were 3811.486 and 3811.576,respectively,while the AIC and AICc values of GWLR were 2700.963 and 2708.852.It can be seen that the ability of GWLR distinguishing between fire and non-fire point is higher than that of BLR,and the simulation result is better.Distribution of fire risk probability was got by using GWLR model,then the fire risk assessment was delveloped.(5)The spatial distribution of fire risk in Gansu is obviously different.High and extremely high fire risk zones,the area of fire probability>0.6,are mainly distributed in Gannan,Longnan,southern Tianshui,southern Linxia,in the east of Qingyang,Qilian mountain range and oasis area of the Hexi Corridor,Gansu Province.Meanwhile,modetate fire danger area is mainly distributed in the periphery of the high,high fire danger area.Extremely low and low fire district,area of26.71×10~4 km~2,mainly distributed in this rigion of less population,poor ecological environment,dificient resources and low vegetation cover.Establishing 10 fire impact factor partition and proposing the differentiated fire prevention recommendations.