新型Landsat8卫星影像的反射率和地表温度反演

Landsat 8卫星自2013年2月发射以来,其影像的定标参数经过了不断调整和完善,针对Landsat 8开发的各种算法也相继问世.本文采用最新的参数、算法和引入COST算法建立的大气校正模型,对Landsat 8多光谱和热红外波段进行了处理,反演出它们的反射率和地表温度,并与同日的Landsat 7数据和实测地表温度数据进行了对比.结果表明,现有Landsat 8多光谱数据的定标参数和大气顶部反射率反演算法已有很高的精度,本文引入COST算法建立的Landsat 8大气校正模型也与Landsat 7的COST模型所获得的结果几乎相同,相关系数可高达0.99.但是现有针对Landsat 8提出的地表温度反演算法仍不理想,已提出的劈窗算法误差都较大.鉴于TIRS 11热红外波段的定标参数仍不理想,因此在现阶段建议采用单通道算法单独反演TIRS 10波段来求算地表温度,但要注意根据大气水汽含量的情况选用正确的大气参数计算公式.

Retrieval of the reflectance and land surface temperature of the newly-launched Landsat 8 satellite

The quality of Landsat 8 sensor data has been improved after a series of corrections and data reprocessing since its successful launch on February 11, 2013. These include the correction of all calibration parameters, the improvement of the radiance conversion coefficients for the all Operational Land Imager (OLI) sensor bands, the refinement of the OLI detector linearization, the radiometric offset correction for the two Thermal Infrared Sensor (TIRS) bands, the slight improvement to the geolocation of the TIRS data, and the reprocessing of all Landsat 8 data held in the USGS archives. In addition, several algorithms specially developed for Landsat 8 data have also been proposed over the pass year. This paper aims to assess the accuracy of the retrieved the reflectance of the OLI sensor and the land surface temperature (LST) of the TIRS sensor of the new satellite with those of the well-calibrated Landsat 7 ETM+ sensor and the ground-measured LST. This study retrieved the top of the atmosphere (TOA) reflectance of the OLI multispectral bands and the LST of the TIRS thermal bands with the most recent calibration parameters, algorithms and USGS-processed Landsat 8 image data. In addition, the Chavez's COST model has been introduced to correct the atmospheric effects on the OLI multispectral bands. To examine the performance of the calibration parameters and the developed algorithms, the retrieved TOA reflectance of each multispectral band and the computed LST of the thermal infrared bands have been compared with that of the corresponding band of the synchronized, well-calibrated Landsat 7 data and the in situ LST, respectively.#br#The results show that the current Landsat 8 calibration parameters of multispectral bands can achieve high accuracy for the retrieval of TOA reflectance. The proposed COST-based atmospheric correction algorithm can also have a nearly identical performance when compared with the Landsat 7 COST model's result. Nevertheless, two recently-proposed split window algorithms for computing the LST from Landsat 8 thermal infrared bands did not perform well, as they offered a large difference between the algorithm-modeled LST and the ground-measured LST. Given the scaling parameters of the TIRS thermal infrared band 11 is still unstable, as announced by the Landsat 8 project team, it is recommended that at this stage users might use the single channel (SC) algorithm of Jiménez-Muñoz and Sobrino to retrieve the LST from Landsat 8 thermal band 10 (like working on Landsat TM/ETM+ band 6) rather than attempt a split-window algorithm using both TIRS bands 10 and 11. However, care should be taken in the selection of correct atmosphere parameters for the SC-based LST computing, especially when a very high atmospheric water vapor condition occurs.