基于Landsat 8影像的不同燃烧指数在农田秸秆焚烧区域识别中的应用

Quantification of crop residue burned areas based on burning indices using Landsat 8 image

农田秸秆焚烧造成秸秆资源的巨大浪费和大气环境的污染,利用热红外波段识别火点的方法可以实时、快速地获取焚烧情况,但是不能提供秸秆焚烧的面积、范围等详细的空间分布情况,也不能确定焚烧的严重程度.利用已燃烧与未燃烧区域的光谱差异选择某些波段构建燃烧指数的方法在森林火灾研究中得到了广泛应用,但是这些指数在农田秸秆焚烧中的潜在应用没有被研究.本文基于松嫩平原地区的两景Landsat 8卫星影像,采用归一化燃烧率(NBR)、引入热红外波段的归一化燃烧率(NBRT)、燃烧面积指数(BAI)3种燃烧指数对农田秸秆焚烧和未焚烧区域进行提取,并与秸秆覆盖度进行相关分析.结果表明:NBR、NBRT和BAI指数对焚烧和未焚烧区域的分类精度分别为91.9%、92.3%、87.8%,NBR、NBRT与覆盖度呈线性相关,R2分别为0.73、0.64,BAI与覆盖度呈幂指数相关关系,R2为0.68.燃烧指数方法可以在农田秸秆焚烧中得到很好的应用,可定量评估秸秆燃烧程度,为大气环境评价提供技术与数据支持.

Crop residue burning leads to atmospheric pollution and is an enormous waste of crop residue resource. Crop residue burning can be monitored timely in large regions as the fire points can be recognized through remotely sensed image via thermal infrared bands. However,the area,the detailed distribution pattern and especially the severity of the burning areas cannot be derived only by the thermal remote sensing approach. The burning index,which was calculated with two or more spectral bands at where the burned and unburned areas have distinct spectral characteristics,is widely used in the forest fire investigation. However its potential application for crop residue burning evaluation has not been explored. With two Landsat 8 images that cover a part of the Songnen Plain,three burning indices,i.e.,the normalized burned ratio（ NBR）,the normalized burned ratio incorporating the thermal band（ NBRT）,and the burned area index（ BAI）,were used to classify the crop residue burned and unburned areas. The overall classification accuracies were 91.9%,92.3%,and 87.8%,respectively. The correlation analysis between the indices and the crop residue coverage indicated that the NBR and NBRT were positively correlated with the crop residue coverage（ R^2= 0.73 and 0.64,respectively） with linear regression models,while the BAI was exponentially correlated with the crop residue coverage（ R^2= 0.68）. The results indicated that the use of burning indices in crop residue burning monitoring could quantify crop residue burning severity and provide valuable data for evaluating atmospheric pollution.