The 3DREAMS project: A study of the effect of 3D surface heterogeneity on aerosol retrieval based on synthetic images

To this date, most of the algorithms used to retrieve aerosol properties from multi-angular satellite images use a forward radiative transfer model with a baked-in 1D assumption. Among other things, this means neglecting the effects of surface heterogeneity, such as adjacency (the fact that neighbouring surface reflection properties perturb the diffuse component of incident radiation) and topography (the fact that the surface is not smooth and flat). This introduces bias in the radiative transfer simulation, and thus in the retrieved aerosol properties.

This project aims to investigate the impact of neglecting these heterogeneities on aerosol and surface retrievals from multi-angular satellite observations. For that purpose, a series of benchmarking cases was designed to assess the performance of the GRASP retrieval algorithm (currently state-of-the-art for the processing of EPS-SG/3MI observations) against a known reference created using Eradiate (an accurate 3D radiative transfer model). Benchmarking cases range from simple 1D setups aimed at verifying the alignment of the GRASP forward model and Eradiate, to complex, plausible 3D scenes generated after actual locations on Earth. All assume a 1D atmosphere to focus on the effects of surface heterogeneity.

The complex scenes incorporate topography and land cover information, with varied dominant land cover setups: agricultural, urban, coastal, mountain, in-land water (lake). Locations are situated near key AERONET stations, and the simulated instrument is derived from actual satellite specifications (i.e. geometries, wavelengths) based on PARASOL/POLDER.

In this presentation, we introduce our approach and discuss our conclusions on both the benchmarking approach and the results.