火烧迹地遥感指数与地表温度空间差异性

以2003年根河和金河林业局森林火灾火烧迹地作为研究区,以2000年、2004年的landsat影像为数据源,运用辐射传输方程反演地表温度,提取影像12种常见的遥感指数。分析火烧强度与地表温度的关系、不同遥感指数对地表温度的解释效果以及不同尺度下遥感指数与地表温度的空间差异性,建立遥感指数和地表温度的最小二乘拟合模型(OLS)和地理加权回归模型(GWR)。结果表明:(1)火烧迹地的火烧强度对地表温度具有明显的解释效应,重度火烧具有较强的热聚集效应;(2)火烧迹地地表温度的解释因素主要为下垫面状况的因素,从12种宽带波段遥感指数对地表温度的解释模型中,发现差分归一化火烧指数(IDNBR)、归一化不透水面指数(INDIS)、植被归一化水体指数(IVNDW)、归一化植被指数(INDV)、绿度总和指数(ISG)、改进型归一化水体指数(IMNDW)等6种指数对地表温度的解释效应更好;(3)不同尺度下,地表温度与遥感指数具有空间差异性,不同指数空间相关性随尺度增大而减弱。(4)在解释地表温度模型选择上,局部回归GWR模型比全局回归OLS模型的回归结果好。

Spatial Difference of Remote Sensing Index and Land Surface Temperature in Burned Area

Taking 2003 "Genhe-Jinhe" forest fire burned area as research area, with LANDSAT 5 and LANDSAT 7 images as data sources, by the radiation transmission equation inversion of LST, 12 common indices were extracted. The relationship between fire intensity and LST, the interpretation result of different indices on surface temperature, the spatial difference between index and LST at different scales, the Ordinary Least Squares model(OLS) and Geographically Weighted Regression model(GWR) based on index-LST were established and compared. The fire severity in the burning place has obvious explanatory effect on the LST, and the severity burned areas has a strong heat accumulation. The explanatory factors of the LST of the burning area are mainly the underlying surface factors. Through Exploratory Regression, 12 kinds of wideband band remote sensing indices are explored to explain the optimal model of LST, and DNBR, NDISI, VNDWI, NDVI, SGI and MNDWI have the best interpretation effect on LST. There are spatial differences between the LST and the remote sensing indices at different scales, and the spatial correlation of different indices decreases with the increase of scale. The local regression GWR model has more ideal regression analysis results than the global regression OLS model because of the comprehensive consideration of spatial differences of different factors.