Remote sensing of atmospheric aerosols has advanced substantially over recent decades, driven by progress in satellite instrumentation and the expansion of ground-based networks such as AERONET. While agreement between satellite aerosol optical depth (AOD) retrievals and AERONET reference data has improved, our analysis highlights a critical yet underrecognized imbalance in global validation frameworks. Specifically, AERONET sites are disproportionately concentrated in urban and vegetated regions, where fine-mode aerosols over dark surfaces favor retrieval accuracy, while drylands, dominated by coarse-mode aerosols over bright surfaces, are underrepresented by nearly a factor of two. This sampling bias introduces a systematic distortion in global validation outcomes. We show that nearly half of the global grid cells with elevated disagreement between MODIS and POLDER AOD products are located in drylands with angström exponent values below 0.75. In these areas, the mean top-of-atmosphere aerosol radiative cooling is weaker by 0.24 watts per square meter than in other regions and has an associated uncertainty of 22% higher. These findings highlight that, for improving estimation of global aerosol effects on climate, there is a need for a more stratified validation framework based on surface type and aerosol regime and an importance of continuing improving ground-based observations over drylands, with some network expansion if possible.
Chen et al. (2026) studied this question.