Trends in vertical wind velocity variability reveal cloud microphysical feedback

By controlling supersaturation vertical air motion influences how aerosols activate into cloud droplets and ice crystals. This effect is difficult to represent accurately in atmospheric models as they cannot typically resolve the sub-kilometer scale component of wind motion, however it can be addressed by machine learning. Here we apply a generative technique combining storm-resolving simulations, observational and climate reanalysis data, to predict the spatial standard deviation in vertical wind velocity, σ_W. This analysis reveals significant trends in σ_W, reaching up to 1%yr^-1 in low and mid-level oceanic regions, indicating enhanced atmospheric turbulence. We attribute these trends to global shifts in water vapor, temperature and convection, suggesting a feedback connection between enhanced warming and turbulence, which in turn has a microphysical effect through the activation of more cloud hydrometeors. Focusing on low level clouds this effect has had a radiative impact of about -0.10 ± 0.04Wm^-2 since the year 1900, slightly mitigating greenhouse warming.