基于Landsat数据的2种地表温度空间降尺度算法比较研究

以2003年5月29日福州市LandsmETM＋影像为数据源，对2种地表温度空间降尺度算法——EM算法和HUTS算法进行实验、比较与分析,EM算法是利用高空间分辨率的地表比辐射率对低空间分辨率的亮度温度影像进行调节，从而达到提高热红外影像空间分辨率的目的；HUTS算法则是引入了归一化差异植被指数NDVI和地表反照率d，通过在低空间分辨率拟合二者与地表温度LST之间的关系，然后根据其尺度不变性，将该关系应用到高空间分辨率的影像上，从而达到提高热红外影像空间分辨率的目的．研究结果表明：1）2种算法所得结果影像都能在有效提高空间分辨率的同时较好地保持了原始地表温度影像的空间分布特征，达到了较好的降尺度效果；2）以RMSE为定量评价指标，HUTS算法要略优于EM算法，其中，EM算法的RMSE为1．2494，而HUTS算法仅为0．9869．

A Comparison Study of Two Algorithms for Spatial Downscaling of Land Surface Temperature Retrieved from Landsat Data

A comparison of two algorithms, Emissivity Modulation （EM） and High-resolution Urban Thermal Sharpener （HUTS）, for land surface temperature （LST） image downscaling, was made using a Landsat ETM ＋ image acquired on 29 May, 2003 over a study area of Fuzhou, Fujian, China. The EM method was to enhance the spatial resolution while simultaneously correcting the image derived temperatures for emissivity differences of earth surface materials. The HUTS algorithm was developed for increasing the resolution of thermal infrared （TIR） data to that of visible and near infra-red （NIR） data by fitting relationships among radiometric surface temperature, normalized difference vegetation index （NDVI） and surface albedo （α）. The results showed that both algorithms could effectively enhance the spatial resolution while simultaneously preserving the characteristics and spatial distribution of the original LST image retrieved from the ETM ＋ thermal band. It was also found that the RMSEs of the downscaled LST images from the EM and HUTS algorithms were 1. 249 4 and 0. 986 9, respectively, indicating that the latter algorithm had performed slightly better.