基于FY-33D/MERSI-Ⅱ和Terra/MODIS数据的森林火点提取算法——以我国西南地区为例  

Forest fire spot extraction algorithm based on FY-3D/MERSI-Ⅱand Terra/MODIS data--Taking southwest China as an example

作　　者：黄欢 马晓红 杜昊 张义钊 张洪铭 邓帆 HUANG Huan;MA Xiaohong;DU Hao;ZHANG Yizhao;ZHANG Hongming;DENG Fan(Power Science Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 550002,China;Guizhou Xingyi Power Supply Bureau,Guizhou Power Grid Co.,Ltd.,Xingyi 562400,China;Guizhou New Meteorological Technology Co.,Ltd.,Guiyang 550002,China)

机构地区：[1]贵州电网有限责任公司电力科学研究院,贵州贵阳550002 [2]贵州电网有限责任公司兴义供电局,贵州兴义562400 [3]贵州新气象科技有限责任公司,贵州贵阳550002

出　　处：《安全与环境工程》2024年第5期163-169,198,共8页Safety and Environmental Engineering

基　　金：贵州省科技支撑计划项目(黔科合支撑[2020]4Y054号);南网科技项目(GZKJXM20200655);中国气象局卫星专项(FY-3(03)-AS-12.13)。

摘　　要：中国西南地区森林覆盖率高,土地利用类型多样,人类活动频繁,森林火灾风险较高,防火任务十分艰巨。遥感技术具有覆盖范围广和周期性监测等优势,被广泛应用于森林火灾监测。气象卫星搭载了对火灾敏感的红外谱段传感器且时间分辨率较高,已成为森林火灾监测的重要遥感数据源。物理阈值法是遥感数据火点提取中比较成熟的算法,但其精度受所选阈值的影响较大,导致不同区域的火点提取精度存在差异。以中国西南地区的重庆市为例,提出了一种采用FY-3D/MERSI-Ⅱ和Terra/MODIS极轨卫星数据作为数据源进行森林火点监测的算法。首先,通过分析前一年MOD14监测产品的历史数据,确定该地区的初定火点阈值;随后,消除云层、水体和太阳耀斑的干扰;最后,采用上下文背景阈值法进行火点提取。对2022年8月重庆市山火的分析表明,该算法的火点提取平均准确率超过了90%。研究结果表明,根据区域特点调整阈值可以提高该地区火点监测的准确率。The southwest China has a high forest coverage,complex land use,and frequent human activities,resulting in a high risk of forest fires and a heavy burden on fire prevention tasks.Remote sensing technology,with its capability for extensive and periodic observations,is widely used in monitoring forest fires.Meteorological satellites equipped with infrared sensors sensitive to fires and high temporal resolution have become important remote sensing data sources for forest fire monitoring.The physical threshold method is a relatively mature algorithm for remote fire point extraction,but its accuracy is significantly influenced by the chosen threshold,leading to variations on fire point extraction accuracy in different regions.In this study,the FY-3D/MERSI-Ⅱand Terra/MODIS polar-orbiting satellites were used as data sources for forest fire point monitoring in Chongqing City of southwestern China.Through an analysis of historical data from the previous year's MOD14 monitoring product,regional background thresholds were determined.The context background threshold method was then applied for fire point extraction,with the removal of cloud,water,and solar glare interference.An analysis of Chongqing City forest fire in August 2022 using this method yielded an average fire point extraction accuracy exceeding 90%.The research results indicate that adjusting the threshold according to regional characteristics can improve the accuracy of fire point monitoring in this area.

关 键 词：森林火灾监测 FY-3D/MERSI-Ⅱ Terra/MOD14 火点提取 上下文背景阈值法 阈值选取 

分 类 号：X4[环境科学与工程—灾害防治] TP79[自动化与计算机技术—检测技术与自动化装置]