摘要
森林火灾是生态系统面临的重大自然灾害之一,利用遥感技术快速提取火烧迹地的信息对灾后管理和评估至关重要。文章以宁夏回族自治区固原市林场火灾为例,利用环境减灾二号A/B卫星影像开展森林火灾火烧迹地范围的识别和提取并进行精度验证。采用支持向量机分类方法和基于归一化植被指数的决策树分类方法,结合灾前和灾后2个时相的影像,分析了火灾后火烧迹地的面积和范围。结果表明:归一化植被指数(NDVI)在火烧迹地范围提取中优于支持向量机(SVM)分类法,识别精度达到97.5%。环境减灾二号A/B卫星具有较高的空间分辨率,可以较好地反映植被燃烧后形成的火烧迹地,有效满足了森林草原火灾灾后开展定量评估的需求,为森林火灾监测应用提供有力支撑。
Forest fire is one of the major natural disasters facing the ecosystem.Using remote sensing technology to quickly extract the information of the burned area is very important for post disaster management and severity assessment.In this study,the forest fire occurred in Guyuan City,Ningxia Hui Autonomous Region was chosen as an example,using HJ-2A/B satellites image to identify and extract the forest fire burned area.Based on two-temporal imagery before and after forest fire,the area and extent of burned land were analyzed by using Support Vector Machine(SVM)classification method and decision tree classification method combined with Normalized Difference Vegetation Index(NDVI).The results showed that NDVI performs better than SVM classification in extracting burned traces range,and the overall accuracy was up to 97.5%.HJ-2A/B satellites have high spatial resolution and can efficiently reflect the burned traces after vegetation combustion,which effectively meets the needs of quantitative assessment after forest and grassland fire disaster,and therefore provides strong support for the application of forest fire monitoring.
作者
金晟玥
王平
刘明博
贾松霖
苏琮
JIN Shengyue;WANG Ping;LIU Mingbo;JIA Songlin;SU Cong(National Disaster Reduction Center of China, MEM, Beijing 100124, China;DFH Satellite Co., Ltd., Beijing 100094, China)
出处
《航天器工程》
CSCD
北大核心
2022年第3期141-146,共6页
Spacecraft Engineering
基金
高分观测重大专项(03-Y30F03-9001-20/22)
国家重点研发计划(2021YFB3901205)。
关键词
环境减灾二号A/B卫星
森林火灾
归一化植被指数
支持向量机
HJ-2A/B satellites
forest fire
Normalized Difference Vegetation Index(NDVI)
Support Vector Machine(SVM)
作者简介
金晟玥,女,硕士,研究实习员,研究方向为卫星遥感技术在应急管理中的应用,Email:syjin_ndrcc@163.com。
