| Forest fire is an important factor in forest ecosystem,which plays a key role in maintaining biodiversity and ecosystem structure.In addition,forest fire also damages forest resources and human life and property safety.The driving factors of forest fires are very complex,and there may be spatial differences between different ecosystems.Therefore,understanding the driving factors of forest fires is very important for forest fire prediction and management.Fujian and Daxing’anling regions are located in subtropical evergreen broad-leaved forest and northern cold temperate coniferous forest respectively in China’s three major forest regions.The average burning area of subtropical evergreen broad-leaved forest is low,but the burning frequency is high,while that of northern cold temperate coniferous forest is high and the burning frequency is low.As coniferous forest in cold temperate zone and evergreen broad-leaved forest in subtropical zone are two typical forest fire occurrence ecosystems,Daxing ’anling and Fujian are selected as research sites for forest fire occurrence and driving factors analysis in this study.Fujian Province is a typical subtropical region in China,and Fujian is also a high incidence area of forest fires.From 2000 to 2016,a total of 6164 forest fires occurred in Fujian Province,with an average annual forest fire area of 7328.97hm2.Daxing ’anling is located in the cold temperate zone of northern China and is also a high incidence area of forest fires.According to statistics,706 fires occurred in Heilongjiang from 2004 to 2014,of which 240 occurred in Daxing ’anling,accounting for 52.41% of the total forest fires in the province.Based on Arc GIS,GWR4,SPSS and other statistical analysis software,this paper uses geographic weighted logistic regression model to analyze the main influencing factors of forest fire occurrence in Fujian and Daxinganling regions from 2001 to 2016,determines the driving factors,distinguishes the differences of the driving factors of forest fire occurrence in two typical ecosystems of subtropical evergreen broad-leaved forest and northern cold temperate coniferous forest in China,establishes the forest fire prediction model and classifies the fire danger grades,and provides scientific support for forest fire prevention and management in the southeast and northern forest regions of China.In this paper,Ripley’s K function and GWR4 software are used to analyze the forest fire occurrence data in Fujian and Daxing ’anling region in time and space.According to the spatial aggregation and distribution of forest fires,the forest fire data in Fujian and Daxing’anling region are divided into four periods(2001-2004,2005-2008,2009-2012 and 2013-2016).Based on satellite fire point data,using Arc GIS10.2 and other software,the weather factors(daily temperature range,daily surface temperature range,accumulated precipitation during fire prevention period,accumulated precipitation in the previous quarter of fire prevention period,The average temperature in the fire prevention period,the average temperature in the previous quarter of the fire prevention period,the annual accumulated precipitation,the annual average temperature,the annual average relative humidity,the daily accumulated precipitation,the daily relative humidity,the daily average wind speed),terrain factors(altitude,slope),vegetation data(vegetation coverage),social infrastructure factors(population density,GDP per capita,distance from residential area,distance from railway,distance from highway)are extracted for spatial information,and combined with random points,the driving factors of forest fires in Fujian and Daxing’anling regions are analyzed by geographic weighted Logistic regression model.Geographically weighted logistic regression model was used to analyze the driving factors of forest fires in Fujian and Daxing’ anling regions.The results show that the important factors affecting the occurrence of forest fires in Fujian are:(1)Daily temperature difference(74.5% in the significant area)and vegetation combustible factors(44.8%)are the most important factors affecting the occurrence of forest fires in Fujian.(2)Daily temperature range from 2005 to 2008(81.9%).(3)Daily temperature range(96.5%)and vegetation combustible factors(36.8%)from 2009 to 2012.(4)Daily temperature range from 2013 to2016(93.7%).Among them,the important factors affecting the occurrence of forest fires in Daxinganling are:(1)accumulated precipitation(48.3%),slope(47.1%),vegetation combustible factors(47.3%)and distance from railway(27.2%)during the fire prevention period from 2001 to 2004.(2)Daily average temperature from 2005 to 2008(87.8%),accumulated precipitation during fire prevention period(87.9%),slope(74.2%),vegetation combustible factor(65%)and distance from railway(27.2%).(3)Cumulative precipitation(49.3%),slope(64.9%)and GDP per capita(43.8%)in the previous quarter of the fire prevention period from 2009 to2012.(4)Cumulative precipitation(59.6%),slope(57.1%)and population density(58.1%)during the fire prevention period from 2013 to 2016.The fitting results based on the geographically weighted logistic regression model show that the Daxing ’anling region is comprehensively influenced by meteorological factors and topographic factors as well as socio-economic factors.Fujian region is highly influenced by meteorological factors,while terrain and socio-economic factors have little influence.Compared with the traditional Logistic regression model,the geo-weighted logistic regression model has higher prediction accuracy for forest fires in Fujian and Daxing’ anling regions.the geo-weighted logistic regression model can provide more comprehensive spatial information in explaining the impact of environmental factors on forest fires.when fitting the relationship between forest fires and environmental factors,it takes more spatial factors into account and has better expression ability. |
PDF Full Text Request