Code and data for Constraining Aerosol–Cloud Interactions via the Droplet–Aerosol Trends Discrepancy

This dataset and associated code are provided in support of the Science Advances paper “Constraining Aerosol–Cloud Interactions via the Droplet–Aerosol Trends Discrepancy” (Wang et al., 2026) and fulfill the Data and Materials Availability requirements of the publication. Abstract: Aerosol–cloud interactions strongly affect Earth’s climate, yet their radiative forcing (RFaci) remains poorly constrained. Satellite estimates typically rely on column-integral aerosol proxies that may not capture cloud-base cloud condensation nuclei (CCN) that directly govern droplet formation. Using multi-source observations from 2003–2020, we demonstrate that cloud droplet concentrations declined even as column-integral aerosol increased over the Southern Hemisphere, contrary to theoretical expectations. By explicitly decomposing droplet sources and sinks, we trace this paradox to systematically elevated cloud bases, which limit CCN supply, and to enhanced precipitation, which accelerates cloud droplet removal. Including these processes constrains to –1.36 [-1.70, -1.12] W m⁻². Previous global assessments using column-integral proxies are therefore biased by –35% to +26%, reaching +42% across the Southern Hemisphere. These results reconcile long-standing droplet–aerosol discrepancies and provide a physically grounded basis for refining climate sensitivity estimates and guiding model development.