基于多光谱遥感图像的森林过火面积监测研究  

Research on Forest Burnt Area Monitoring Based on Multispectral Remote Sensing Images

作　　者：方志平[1] 周会娥[2] 刘爽[2] 王志婷 FANG Zhi-ping;ZHOU Hui-e;LIU Shuang;WANG Zhi-ting(Yunnan Forestry Investigation and Planning Institute,Kunming 650000,Yunnan,P.R.China;Ecological Branch of Yunnan Forestry Investigation and Planning Institute,Kunming 650000,Yunnan,P.R.China)

机构地区：[1]云南省林业调查规划院,云南昆明650000 [2]云南省林业调查规划院生态分院,云南昆明650000

出　　处：《森林防火》2025年第1期85-87,共3页JOURNAL OF WILDLAND FIRE SCIENCE

基　　金：云南省人才培养专项(80201402)。

摘　　要：森林火灾严重威胁森林生态系统,传统监测手段难以满足当下精准高效要求。多光谱遥感图像技术能捕捉不同波段森林地表特征信息,有助于区分过火区域与未过火区域。基于此,本研究聚焦森林过火面积监测,通过精准提取火灾信息,剖析多光谱遥感图像对过火面积监测效果。结果表明:当火灾进入稳定期(18~24 h),多光谱遥感图像可以对整个过火区域进行最后的确认和分析,过火面积识别平均准确率为93%,较人工监测增加7.46%。因此多光谱遥感图像能够快速获取目标区域信息,可以迅速捕捉到火灾区域的光谱特征变化,及时监测林区过火面积变化。Forest fires are a serious threat to forest ecosystems,and traditional monitoring means can hardly meet the current requirements of precision and efficiency,while multispectral remote sensing image technology has the ability to capture the forest surface characteristics in different bands,which is helpful to distinguish the burned area from the unburned area.Based on this,in this study,the monitoring of forest overfire area was focused on,and the monitoring effect of multispectral remote sensing images on overfire area by accurately extracting fire information was analyzed.The experimental results showed that when the fire entered the stable period(18~24 h),the multispectral remote sensing image can make the final confirmation and analysis of the whole overfire area,and the average accuracy of overfire area identification was 93%,which was an increase of 7.46%compared with manual monitoring.Therefore,multispectral remote sensing images could quickly obtain information about the target area,and could rapidly capture changes in the spectral characteristics of the fire area and monitor changes in the overfire area of the forest in a timely manner.

关 键 词：多光谱遥感 森林防火面积 监测 平均准确率 火灾区域 

分 类 号：S762.3[农业科学—森林保护学]