光谱学与光谱分析, 2016, 36 (8): 2414, 网络出版: 2016-12-23
利用FTIR和MODIS数据估算塔克拉玛干沙漠宽波段地表比辐射率
Estimating Surface Broadband Emissivity of the Taklimakan Desert with FTIR and MODIS Data
摘要
分析并提供了一个利用MODIS窄波段数据, 估算地表宽波段(8~14 μm)比辐射率的最优估算方程, 并根据该方程获得了塔克拉玛干沙漠地区地表比辐射率特征分布情况。 首先, 沿塔克拉玛干沙漠的两条南北穿越公路, 使用傅里叶变换热红外光谱仪(FTIR), 选取20个观测点, 获取实测的地表宽波段比辐射率。 其次, 利用MODIS温度产品MOD11A1和MOD11C1热红外区域第29, 31和32波段和MOD09A1近红外区域第7波段数据, 建立待定系数的地表宽波段比辐射率多元线性回归估算方程。 通过FTIR的观测值和MODIS数据确定该估算方程的系数, 并进行误差分析。 研究发现, 使用FTIR观测值, 由MODIS第29, 31和32波段数据的线性回归方程, 可以产生高精度的地表宽波段比辐射率。 加入MODIS第7波段后, 新的线性回归估算方程的精度更高, 均方根误差RMSE为0.004 5, 平均偏差Bias为0.000 1。 与文献中的其他六种估算方程横向对比, RMSE和Bias分别比其他六种估算方程低1和2个数量级。 最后, 利用该估算方程获得了研究区的地表比辐射率分布图, 结果显示, 沙漠中心区域的值为0.880~0.910, 平均值为0.906; 有稀疏植被区域的值为0.910~0.940; 靠近沙漠边缘的绿洲的值为0.950~0.980。
Abstract
Surface broadband emissivity in the thermal infrared region is an important parameteras for the studies of the surface energy balance. This paper analyzed and offered an equation to estimate the surface broadband emissivity for the spectral domains 8~14 μm against the MODIS data, and then, the distribution characteristic of surface emissivity for Taklimakan Desert was obtained with this equation. Firstly, along two highways crossing the Taklimakan Desert, twenty sample sites were selected and their spectral of broadband emissivity were observed with Fourier Transform Infrared spectrometer (FTIR). Secondly, using the Moderate Resolution Imaging Spectrometer (MODIS) land surface temperature and emissivity product MOD11A1 and MOD11C1, derived emissivities in three thermal infrared channels 29 (8.4~8.7 μm), 31 (10.78~11.28 μm) and 32 (11.77~12.27 μm) and MODIS surface reflectance products MOD09A1, derived reflectance in near-infrared channel 7 (2.105~2.155 μm), developing an empirical regression equation to convert these spectral emissivities and reflectance to a broadband emissivity. The FTIR data were used to determine the coefficients of the regression equation, another part of FTIR data were used to investigate the accuracy of equation. It was found that the equation consist of MODIS channels 29, 31 and 32 has more accuracy; furthermore, the accuracy is improved when channel 7 data was added in the regression equation. The root mean square error (RMSE) and Bias were 0.004 5 and 0.000 1, respectively. Comparing to other six equations originated from literatures, which also estimate the surface broadband emissivity from narrowband emissivities. The RMSE and Bias of our equation are lower one order and two orders of magnitude than other six equations, respectively. Lastly, our equation is applied in the Taklimakan Desert area to build a distribution image of emissivity based on MODIS data. It demonstrates that the emissivity of Taklimakan Desert is in the range of 0.880~0.910 over the central regions, the averaged value is 0.906; The emissivity is in the range of 0.910~0.940 where the areas covered by spare vegetation; The emissivity is in range of 0.950~0.980 where the regions near to the oasis.
李火青, 吴新萍, 买买提艾力·买买提依明, 霍文, 杨兴华, 杨帆, 何清, 刘永强. 利用FTIR和MODIS数据估算塔克拉玛干沙漠宽波段地表比辐射率[J]. 光谱学与光谱分析, 2016, 36(8): 2414. LI Huo-qing, WU Xin-ping, Ali Mamtimin, HUO Wen, YANG Xing-hua, YANG Fan, HE Qing, LIU Yong-qiang. Estimating Surface Broadband Emissivity of the Taklimakan Desert with FTIR and MODIS Data[J]. Spectroscopy and Spectral Analysis, 2016, 36(8): 2414.