Landsat-8数据地表温度反演单通道修正算法

Single-channel Correction Algorithm to Invert Land Surface Temperature with Landsat-8 Satellite Data

Jiménez-Mu1oz提出的陆地表面温度(land surface temperature,LST)反演单通道(single-channel,SC)算法因其需要的大气参数少而被广泛应用,但大气水汽含量较高时,SC算法精度会受到影响。该文针对不同的大气模式建立适应于Landsat-8数据且在大气水汽含量为0~4.5g/cm^3时也有较高精度的单通道修正(singlechannel correction,SCC)算法。在1976美国标准大气、热带大气、中纬度冬季和中纬度夏季4种大气模式条件下,SCC算法平均误差/平均绝对误差分别为0.04/0.55K、0.07/0.25K、0.02/0.51K、-0.08/0.24K。通过参数敏感性分析发现,LST反演误差与地表比辐射率、大气水汽含量呈线性相关。地表比辐射率每改变0.01,SCC算法反演得到的LST在不同大气模式下大约改变0.46K、0.49K、0.46K、0.47K。大气水汽含量每改变0.1g/cm^3,SCC算法反演得到的LST在不同大气模式下大约改变0.41K、0.10K、0.06K、0.06K、0.13K。

Single-channel （SC） algorithm to invert land surface temperature developed by Jimenez-Munoz is widely used because the algorithm just needs a small quantity of atmospheric parameters. Nevertheless, the accuracy of SC algorithm will be affected when the atmospheric water vapor content is high. The paper designs a single-channel correction （SCC） algorithm which can not only adapt to Landsat-8 data, but also have a relatively high accuracy when the atmospheric water vapor content is from 0 g/cm3 to 4.5 g/cm3 for different atmospheric models. Under the conditions of four different atmospheric models,namely 1976 US Standard Model Atmosphere,Tropical Model Atmosphere, Mid-Latitude Winter Model Atmosphere and Mid-Latitude Summer Model Atmosphere,the average errors/average absolute errors of SCC algorithm under each model are 0.04/0.55 K, 0. 07/0.25 K,0. 02/0. 51 K,-0.08/0.24 K respectively. Through the sensitivity analysis of parameters in the SCC algorithm, we can find that linear correlation exists between LST inversion errors and land surface emissivity, atmospheric water vapor content. When land surface emissivity changes by every 0. 01 ,LST obtained from the inversion of SCC algorithm may roughly change by 0.46 K, 0.49 K, 0.46 K, 0. 47 K under the different atmospheric models. When atmospheric water vapor content changes by every 0.1 （g/cm3） ,LST got from the inversion of SCC algorithm may roughly change by 0.41 K,0.10 K,0.06 K, 0.06 K,0.13 K under the different models.