A global aerosol model based on the analysis of 30-year ground measurements from AERONET (AEROEX model): Implications for satellite-derived aerosol retrievals.

The AErosol RObotic NETwork (AERONET), established in 1993, now spreads across 900 global sites, has about three decades of ground-based aerosol measurements. An aerosol model characterizes the physical and optical properties of atmospheric particles used in satellite and ground-based retrievals and climate simulations. Earlier aerosol models, developed using limited data (∼10-12 years, ∼250 sites), could not capture recent environmental shifts and associated changes in aerosol emissions driven by industrialization, land use changes, intensified wildfires, and dust storms. These changes necessitate a comprehensive re-evaluation of global aerosol models for enhanced satellite-derived aerosol optical depth (AOD), their properties retrieval, and climate studies. This study introduces new AERONET-extended (AEROEX) models, derived from the machine learning based cluster analysis of over 202,000 AERONET measurements (1993-2023) using Gaussian Mixture Models. The measurements are classified into four fine-mode (Fine-1 to Fine-4) and two coarse-mode (Coarse-1, and Coarse-2) types of aerosols, by integrating particle linear depolarization ratio, fine mode fraction (FMF), single scattering albedo, and asymmetry parameters to capture regional and seasonal variations across nearly all continents, except Antarctica. Compared to earlier models, AEROEX incorporates particle non-sphericity and evolving atmospheric conditions, offering a more globally consistent yet regionally representative aerosol classification. Based on these, newly derived aerosol models, seasonal global aerosol maps at 1° × 1° resolution were generated using a Random Forest classifier with expert-guided refinement. These maps capture spatial and seasonal variability of dominant aerosol types and are suitable for use in satellite aerosol retrieval algorithms. Validation using independent AERONET data from 2024 (excluded from model development) shows substantial improvement in satellite-based AOD and FMF retrievals. Spectral comparison against AERONET "sun" observations at global sites confirms AEROEX's ability to represent ambient aerosol conditions across seasons and source regions. Overall, AEROEX provides an improved framework for satellite aerosol retrievals using single-view spectroradiometers and climate impact studies.