利用ASTER资料估算黑河中游沙漠和绿洲地区夏季地表温度

Estimation of Summer Land Surface Temperature over Desert and Oasis Area in Middle Reaches of Heihe River Using ASTER Data

利用美国宇航局地球观测系统EOS-Terra卫星所搭载高级空基热发射反射辐射计（ASTER）5个热红外通道遥感资料，并结合三种地表温度反演算法和可见近红外信息，估算2003年夏季黑河中游地区沙漠和绿洲景观下地表温度空间分布。分析发现三种方法估算的地表温度比较接近，且反演值与地面观测值较为一致，能够作为陆面过程研究输入数据。其中，估算的水体温度主要介于19．0～21．0℃，绿洲内农田温度分布于27．O～29．0℃，荒漠戈壁地表温度分布于40．O～60．0℃；在绿洲内，由植被覆盖度参数化比辐射率方法（Pv方法）所估算地表温度值最低，Alpha导出比辐射率方法（ADE方法）估算值最高，由温度比辐射率分离方法（TES方法）估算值介于二者之问；荒漠戈壁区域Pv方法地表温度估算值最高，ADE方法估算值最低，TES方法估算值仍然介于前两者之间；绿洲和荒漠戈壁均具有较大地表温度空间变率。利用有关比辐射率光谱库地物观测数据拟合出一个经验公式以获取地物宽通道比辐射率。分析计算表明，地物目标在较大地表环境温度变化范围内，利用地物宽通道比辐射率计算地表长波辐射最大绝对误差不超过8．0W·m^-2。

In this paper, a study on land surface temperature（LST） and land surface emissivity estimation over the desert and oasis area in Heihe river basin in the summer of 2003 is conducted by using Advanced Spaceborne Thermal Emission Reflectance Radiometer（ASTER） data, based on Temperature/Emissivity Separation method（TES）, Alpha Derived Emissivity method（ADE） and Emissivity Parameterization method from vegetation coverage （Pv）. The analysis suggests that the estimations of the three methods are relatively close and their retrieval results are consistent with the ground observations. According to the retrieved resuits, the water body surface temperature mainly ranges from 19.0℃ to 21.0℃ and the cropland surface temperature mainly ranges from 27.0℃ to 29.0℃, and the desert and Gobi surface temperatures range from 40.0℃ to 60.0℃. It is found that the Pv parameterization method provides lower LST over oasis region and the ADE method provides higher LST, whereas the result is inverse over desert and Gobi region. In these two cases, the result provided by the TES method is between the one estimated by the Pv parameterization method and the ADE method. It is also found that LST appears relatively high spatial variation both over the oasis and the desert regions. A fitted formula based on emissivity spectra library is established in order to get the broad channel LSE. Using this formula, a maximal 8.0 W.m^-2 error was caused in calculating long- wave radiation at a large environmental LST range.