基于分组分裂窗算法的MTSAT-1R地表温度反演

Retrieval of Land Surface Temperature Using a Classified Split-window Algorithm from MTSAT-1R Data

以黑河流域上游和中游为研究区,针对MTSAT-1R卫星数据,运用MODTRAN 4.0及晴空状态下的TIGR大气廓线数据,发展了根据地表比辐射率、大气水汽含量、传感器观测角度分组模拟的分裂窗算法,进行地表温度反演。分析了传感器噪声、地表比辐射率和大气水汽含量3个参数对该算法的影响,并结合模拟数据、地面观测数据及MODIS地表温度产品,对反演结果进行分析评价。结果表明:当传感器垂直观测或大气水汽含量小于2.5g/cm2时,反演精度在1K以内;反演结果与地面观测数据对比差异较小,在阿柔站RMSE为3.7 K(日)/1.4 K(夜),在盈科站RMSE为2.4K(日)/2.0K(夜);与MODIS地表温度产品比较,空间分布呈现出一致性。总之,分组分裂窗算法能较好地用于MTSAT-1R卫星数据进行地表温度反演。

A Classified Split-Window（C-SW）algorithm is developed to retrieve Land Surface Temperature（LST）from the thermal infrared data observed by the Multifunctional Transport Satellites-1R（MTSAT-1R）using atmospheric radiative transfer model MODTRAN 4.0and Thermodynamic Initial Guess Retrieval（TIGR）clear-sky atmospheric profile data.The coefficients of the C-SW algorithm are divided into several groups according to different ranges of the three parameters,i.e.the atmospheric water vapor content,the land surface emissivity and the satellite zenith viewing angle.The method is applied to the upstream and midstream region of the Heihe river basin.The errors of the LST retrieval caused by the uncertainties of instrument noises,land surface emissivity and atmospheric water vapor content are analyzed.Finally,the retrieved LST from MTSAT-1Ris compared with the simulation data,the ground observations and MODIS LST products over the whole study area and at the experimental sites.The results indicate that the accuracy of LST retrieval is less than 1Kwhen the sensor viewing zenith angle is close to nadir or atmospheric water vapor is less than 2.5g/cm2.The comparison between the estimated LST and the in-situ measurements show that the root mean squared errors（RMSE）are 3.7K（daytime）/1.4K（nighttime）at Arou site and 2.4K（daytime）/2.0K（nighttime）at Yingke site,respectively.Moreover,the comparison with MODIS products shows consistent spatial pattern over the study area.As a conclusion,the proposed classified split-window algorithm can be successfully applied to the LST retrievals from MTSAT-1Rdata over the study area.