Enhancing surface albedo and aerosol retrieval from MTG-I/FCI by accounting for Earth's sphericity and light polarization effects using a simple near-real-time-compatible approach

Geostationary satellites allow continuous monitoring of the Earth including land surfaces and aerosols, which can now benefit from the advanced measuring capabilities of the new Meteosat Third Generation-Imager and its Flexible Combined Imager (FCI) on board. FCI offers many opportunities to improve the near real time (NRT) clear-sky retrieval of shortwave surface albedo as it is operationally conducted in the EUMETSAT Land Surface Analysis Satellite Application Facility (LSA-SAF) project (Juncu et al. 2022). For instance, FCI's new VIS04 measuring channel centered at 444 nm was found to enable better aerosol characterization (Georgeot et al. 2025) compared to what can be achieved with the previous generation GEO multi-spectral imager SEVIRI, which is essential to achieve high-quality estimation of surface properties. In addition, FCI's high spatio-temporal resolution (10-minute full-disk scan frequency and 1 km in most visible and near-infrared channels) enables enhanced surface monitoring overall. Currently, the atmospheric correction scheme used to retrieve surface albedo from FCI in the LSA-SAF is being improved to exploit all the relevant data provided by FCI while meeting the time constraints of near-real-time processing.

To achieve this goal, we use fast radiative transfer (RT) codes that make assumptions for the sake of computational constraints. This includes the plane-parallel and scalar approximations, which respectively neglect the Earth's sphericity and light polarization effects. Based on accurate top-of-atmosphere (TOA) reflectance simulations from the SMART-G Monte-Carlo RT code, we assess the errors resulting from these two simplifications in the case of FCI data processing. First, the plane-parallel approximation is found to impact significantly 36\% of FCI observations over the year, including errors larger than 10\% in some cases (e.g., at the beginning and end of each day) (preprint by Klein et al. 2025). Second, neglecting light polarization is found to lead to errors up to 6 \% in TOA reflectance simulations, especially in short visible wavelengths. Based on this study, we propose a simple approach that compensates fast RT simulations for the errors coming from these two assumptions by using pre-calculated look-up-tables of accurate Rayleigh reflectance accounting for Earth's sphericity and light polarization. According to our results, this simple approach leads to a significant error reduction overall, especially in FCI's VIS04 channel where error is divided by 4.

In addition to presenting the results above, we will discuss the upcoming integration of this simple approach in the atmospheric correction scheme of the algorithm iAERUS-GEO, which jointly retrieves aerosol optical depth and surface albedo from FCI (Ceamanos et al. 2023). Finally, we will present a first assessment of the benefits offered by this method when used to process real FCI data corresponding to relevant case studies.