干旱区精河流域地表温度的模型反演研究

以精河流域绿洲为研究区，使用Landsat ETM⁺数据，采用单窗算法和普适性单通道算法对研究区地表温度进行反演，并将这两种算法的反演结果与研究区MODIS温度产品(MODIS LST)进行比较。结果表明：(1) 单窗算法和普适性单通道算法反演的结果总体趋势比较接近，研究区整体的平均温度相差约2k；(2) 采用改进型土壤调整植被指数(MSAVI)代替归一植被指数(NDVI)计算地表比辐射率可有效提高反演精度，并且同等条件下单窗算法的反演精度高于普适性单通道算法，两种算法的反演结果与MODIS LST的相关系数分别是0.9255和0.8651；(3) 在城镇区域，普适性单通道算法反演结果与MSAVI的相关性高于单窗算法，相关系数为0.8136，说明普适性单通道算法更适合干旱区大范围城镇地表温度的反演研究。

Study on retrieval model of land surface temperature in Jinghe watershed in arid region

Based on Landsat ETM⁺ image data in Jinghe watershed oasis, both methods of mono-window algorithm and single-channel algorithm were adopted to retrieve land surface temperature (LST) in the study area, and the comparison between the retrieved results and MODIS temperature products (MODIS LST) was made. The results showed: (1) The retrieved results from these two algorithms were similar each other in overall trend, and the mean temperature difference of the whole study area was about 2k. (2) The retrieval accuracy could be improved effectively by using modified soil-adjusted vegetation index (MSAVI) instead of normalized differential vegetation index (NDVI) to compute land surface emissivity, and the retrieval accuracy of single-window algorithm was higher than that of single-channel algorithm. The correlation coefficients between the retrieved data of these two algorithms and MODIS LST were 0.925 and 0.8651, respectively. (3) In the urban areas, the correlation coefficient between the retrieved data of single-channel algorithm and MSAVI was 0.8136, being higher than that of mono-window algorithm. Therefore, the method of single-channel algorithm was more suitable for the retrieval research of LST in urban areas in large scale.