利用增强回归树分析中国野火空间分布格局的影响因素  

Using boosted regression trees to analyze the factors affecting the spatial distribution pattern of wildfire in China

作　　者：焦琳琳[1,2,3] 常禹[1] 申丹[1,3] 胡远满[1] 李春林[1] 马俊[1,3] 

机构地区：[1]中国科学院沈阳应用生态研究所森林与土壤生态国家重点实验室,沈阳110016 [2]中国科学院地理科学与资源研究所,北京100101 [3]中国科学院大学,北京100049

出　　处：《生态学杂志》2015年第8期2288-2296,共9页Chinese Journal of Ecology

基　　金：国家自然科学基金项目(31470516;41201185和41271201);中国科学院战略性先导科技专项(XDA05050201)资助

摘　　要：确定野火空间分布区格局的影响因素,对野火的预测、火险区划具有重要的意义,同时也为制定合理的野火管理措施提供科学依据。以中国2006—2011年野火发生面积和次数作为因变量,选取了5个影响因素,利用增强回归树(boosted regression tree,BRT)方法分析了影响野火空间分布格局的主要因素。结果表明:影响中国野火过火面积空间分布的因素从大到小依次是海拔(36.92%)、年平均气温(27.85%)、年平均降雨量(13.17%)、人口密度(13.00%)和植被类型(9.07%),总体来说,气候因子和海拔的差异决定了研究区过火面积的空间分布格局;影响中国野火火烧次数空间分布的因素从大到小依次是人口密度(27.44%)、海拔(25.97%)、植被类型(22.84%)、年平均气温(18.98%)和年平均降雨量(4.77%),可以看出,人类活动、地形和可燃物共同影响火烧次数的空间分布格局。建议在制定全国火险区划时,除了气候和植被因素外,还应该重点考虑地形,特别是海拔的影响;并且加强野外用火管理,开展可燃物的处理以减少野火的发生。Determining factors that affect the spatial distribution pattern of wildfires has signifi- cant implications for wildfire prediction and fire risk zonation, and could also provide scientific basis for making rational wildfire management decisions as well. We chose five factors including annual average temperature, mean annual precipitation, elevation, vegetation type and popula- tion density, and utilized boosted regression tree （BRT） method to analyze the main factors that influence the spatial distribution pattern of burned area and the number of fires from 2006 to 2011. Results showed that the factor most affecting the spatial distribution of burned area accor- ding to its relative contribution was elevation （36.92%） , followed by annual average temperature （27.85%）, mean annual precipitation （13.17%）, population density （13.00%） and vegetation type （9.07%）. In general, climate and elevation determined the spatial distribution pattern of burned area. The factor most affecting the spatial distribution of the number of fires according to its contribution was population density （27.44%）, followed by elevation （25.97%）, vegetation types （22.84%）, annual average temperature （18.98%） and mean annual precipitation （4.77%） according to their relative contributions. Human activities together with topography andfuels determined the spatial distribution pattern of the number of fires. We suggested that, in ad- dition to climate and vegetation factors, topographic factors especially elevation should be inclu- ded when making national wildfire risk zonation; and that the management of field fire use should be strengthened and fuel treatment be performed to reduce the occurrence of wildfires.

关 键 词：增强回归树分析 野火 影响因素 

分 类 号：X932[环境科学与工程—安全科学]