摘要
近年来随着全球气候变化加剧,世界各地森林火灾频发,不仅破坏了生态系统,还使各国人民的生命财产和生存环境遭受威胁。Sentinel-2数据红边波段和短波红外波段可以用于火烧迹地提取研究,本文选取火灾发生前和火灾发生后的Sentinel-2A/B时序数据,采用归一化植被指数(normalized difference vegetation index,NDVI)、修正的NDVI指数(NDVI_(red-edge1)、NDVI_(red-edge2)、NDVI_(red-edge3)、NDVI_(SWIR))和归一化燃烧指数(normalized burn ratio,NBR)分别提取美国加利福尼亚州天堂镇火烧迹地信息,分析不同方法提取火烧迹地信息的潜力与存在的问题,与官方公布的过火面积所转换的像元数比对,以探索Sentinel-2A/B数据的3个红边波段和2个短波红外波段提取火烧迹地的可行性。研究结果表明:(1)对火烧迹地提取,改进的NDVI_(red-edge1)、NDVI_(red-edge2)、NDVI_(SWIR)的精度略优于NDVI,其精度分别为92.85%、89.08%、90.14%。而改进后的NDVI_(red-edge3)精度为83.5%,低于NDVI;(2)基于短波红外波段的NBR及NDVI_(SWIR),精度分别达到了93.83%、90.14%。NBR以及基于NDVI改进后的NDVI_(red-edge1)和NDVI_(SWIR)可用于火烧迹地提取。
In recent years,as the global climate change is aggravating,forest fire occurs frequently all over the world.It not only leads to great damage to the ecosystem,but also endangers the lives and property of people.Research on the red-edged band and short-wave infrared band of Sentinel-2 data used to extract fire burns,this paper selected the Sentinel-2A/B time sequence data before and after the occurrence of fire and used the NDVI(normalized difference vegetation index)method and its improved ND⁃VIred-edge1,NDVI_(red-edge2),NDVI_(red-edge3),NDVI_(SWIR) and NBR(normalized burn ratio)to extract the burning information in town of Paradise California,USA.The capability of different methods for extracting the burning information and the related issues were analyzed.The pixel numbers of fire area converted from the official record were compared with those extracted from three red-edged bands and two short-wave infrared bands of Sentinel-2A/B data to explore their feasibility for extracting the burned sites.The study shows that:(1)for the retrieved fire area,the precision of the improved NDVI_(red-edge1),NDVI_(red-edge2),NDVI_(SWIR) was slightly better than that of NDVI,with values of 92.85%,89.08%and 90.14%,respectively,while the improved NDVI_(red-edge3) accuracy was only 83.5%,lower than that of NDVI;(2)the retrieval precision of NBR and NDVI_(SWIR) based on the short-wave infrared wavelengths reached 93.83%and 90.14%,respectively,indicating that the NBR,NDVI_(red-edge1) and NDVI_(SWIR) can be used to extract the burned areas.
作者
卢涛
张军
胡文英
LU Tao;ZHANG Jun;HU Wenying(College of Tourism and Geography,Yunnan Normal University,Kunming,650091,China;School of Earth Sciences,Yunnan University,Kunming 650500,China)
出处
《三峡生态环境监测》
2021年第2期52-60,共9页
Ecology and Environmental Monitoring of Three Gorges
基金
云南大学研究生创新人才培养项目(C176230200)
云南省高校科技创新团队支持计划。
