利用AMSR-E遥感数据反演地表温度——以2008年广东省春季寒害为例

摘要

This paper uses AMSR-E/Aqua L2A global swath spatially-resampled Brightness Temperatures (TB) data between Jan. 25,and Feb. 5,2008 to calculate three vegetation indices:Microwave Polarization Difference Index (MPDI) .Normalized Difference Vegetation Index (NDVI) and Ratio Vegetation Index (RVI) in Guangdong province of China. The land surfaces with different vegetation cover conditions are classified into five types:barren land,grassland.shrub.coniferous forest and broa-dleaved forest. Using the step regression analysis method,we establish a multi-bands regression model between AMSR-E microwave brightness temperatures and the land surface temperatures observed from meteorological observation stations for each land surface vegetation cover type. Results show that the retrieval accuracy of the multi-band regression model can reach under 3. 0℃. Eighty-six percents of which can reach under 2. 5℃. Hence the LST retrieval method can be used as a technical support to the loss-evaluating of cold disaster in Guangdong,Southern China.%基于2008年1月25日至2008年2月5日期间的AMSR-E/Aqua L2A微波亮度温度数据,以广东省为研究对象,依据微波极化差异指数(MPDI)、归一化植被指数(NDVI)和比率植被指数(RVI)等3种植被指数,将广东省地表植被覆盖情况分为裸地、草地、灌木林、针叶林和阔叶林等5种类型,利用逐步回归分析方法,建立了基于不同植被覆盖类型的微波亮度温度与地面气象温度多元回归模型.同步地面气象温度数据验证表明,本文建立的基于地表植被覆盖分类的多波段地表温度反演模型,地表温度反演精度基本可达到3.0℃,其中有大约86％的地区地表温度反演精度可以控制在2.5℃以内,为广东省作物寒害预测提供了微波遥感技术支持.