Yao Qian, Li Zhengqiang, Xu Wenbin, Wang Siheng, Xu Hua, Zhao Liang, Zhang Hao, Ji Zhe
Opt Express. 2024 Nov 4;32(23):42091-42111. doi: 10.1364/OE.541016.
The hybrid nature of the mid-infrared (MIR) spectrum complicates the separation of reflected solar irradiance from total energy. Consequently, existing studies rarely use MIR satellite data alone for retrieving land surface temperature (LST) and land surface emissivity (LSE). In this study, we developed What we believe to be a novel physics-based approach to retrieve LSE and LST using MIR channel data from the MEdium Resolution Spectral Imager II (MERSI-II) onboard China's new-generation polar-orbiting meteorological satellite Fengyun-3D (FY-3D). MERSI-II includes two MIR channels (channels 20 and 21) with a spatial resolution of 1 km, suitable for applying the split-window (SW) algorithm. First, considering the unequal but linearly related land surface bidirectional reflectivity (LSR) in channels 20 and 21, we propose an improved nonlinear SW algorithm. This algorithm, combined with the radiative transfer equation (RTE), accurately retrieves LSR from MIR data. Second, using a kernel-driven bidirectional reflectance distribution function (BRDF) model, the RossThick-LiSparse-R model, we estimate hemispherical directional reflectance from the time series of LSRs (10 days) and subsequently retrieve LSE based on Kirchhoff's law. Atmospheric correction is performed using ERA-5 atmospheric reanalysis data with the radiative transfer (RT) code (MODTRAN 5.2). Finally, LST is retrieved using the RTE in the MIR spectral region. The retrieved LSR was compared with those fitted using the BRDF model, yielding a root mean square error (RMSE) < 0.006 and a bias < 0.003. Cross-validation using the MODIS LSE and LST products (MYD11C1) as a reference showed that the RMSE of the retrieved LSE over 10 days was < 0.027 with a bias < 0.023. For the retrieved LST, the RMSE was < 1.8 K with a bias < 0.7 K. Overall, the proposed method demonstrates potential for retrieving global LSE and LST from MERSI-II MIR data, contributing to advancements in related applications.
中红外（MIR）光谱的混合特性使得从总能量中分离出反射太阳辐照度变得复杂。因此，现有研究很少单独使用MIR卫星数据来反演陆地表面温度（LST）和陆地表面发射率（LSE）。在本研究中，我们开发了一种我们认为新颖的基于物理的方法，利用中国新一代极轨气象卫星风云三号D（FY - 3D）上的中分辨率光谱成像仪II（MERSI - II）的MIR通道数据来反演LSE和LST。MERSI - II包括两个空间分辨率为1 km的MIR通道（通道20和21），适用于应用分裂窗口（SW）算法。首先，考虑到通道20和21中陆地表面双向反射率（LSR）不相等但呈线性相关，我们提出了一种改进的非线性SW算法。该算法与辐射传输方程（RTE）相结合，可从MIR数据中准确反演LSR。其次，使用核驱动双向反射分布函数（BRDF）模型，即RossThick - LiSparse - R模型，我们根据LSR的时间序列（10天）估计半球方向反射率，随后根据基尔霍夫定律反演LSE。使用ERA - 5大气再分析数据和辐射传输（RT）代码（MODTRAN 5.2）进行大气校正。最后，在MIR光谱区域使用RTE反演LST。将反演得到的LSR与使用BRDF模型拟合得到的LSR进行比较，均方根误差（RMSE）< 0.006，偏差< 0.003。以MODIS LSE和LST产品（MYD11C1）作为参考进行交叉验证表明，反演得到的LSE在10天内的RMSE < 0.027，偏差< 0.023。对于反演得到的LST，RMSE < 1.8 K，偏差< 0.7 K。总体而言，所提出的方法展示了从MERSI - II MIR数据中反演全球LSE和LST的潜力，有助于相关应用的发展。
