A Novel Single-Channel Algorithm for Land Surface Reflectance Retrieval From FengYun-4B AGRI Daytime Mid-Infrared Data

Accurately estimating land surface reflectance (LSR) is critical for separating reflected solar irradiance from the total surface energy in the mid-infrared (MIR) spectral region. Existing approaches for MIR LSR retrieval typically rely on synergistic observations from multiple MIR and thermal infrared channels. However, in the 3–5 μm spectral range, many sensors are equipped with only a single high-transmittance channel for Earth observation. To address this limitation, this study proposes an innovative physically based single-channel retrieval algorithm for LSR, utilizing MIR data acquired by the Advanced Geostationary Radiation Imager (AGRI) onboard China’s next-generation geostationary meteorological satellite Fengyun-4B (FY-4B). Considering different solar zenith angles (SZAs), a simulation dataset was generated using MODTRAN 5.2 radiative transfer model through iterative combinations of reasonable atmospheric and surface parameters, in order to determine the coefficients of the proposed algorithm. In practical retrieval, the algorithm coefficients are only related to the SZA and land surface temperature (LST). A detailed sensitivity analysis found that the errors from atmospheric parameters and instrument noise are almost below the 10-4order of magnitude, and adding a 1 K uncertainty in LST can introduce a maximum LSR retrieval error of 4×10-3. Therefore, the proposed algorithm can be considered to have good robustness. The accuracy of MIR LSR is validated in two ways in this study. First, the LSR is directly validated using a simulation dataset generated by MODTRAN, with a root mean square error (RMSE) of-3and a bias within the range of ±2×10-4. Second, LSR is indirectly validated using the MODIS emissivity product MYD11C1 and AGRI LST product, with an RMSE of-3and a maximum bias of −1.2×10-3. Overall, the proposed method can accurately retrieve LSR from AGRI single-channel MIR data, which helps promote its application in related fields.