Quantifying Aerosol Effects on Convective Updraft Velocities in the Amazon using Causal Inference

Correlations between aerosol number concentrations and convective updraft intensities have been found mostly in model simulations in previous studies. One of the proposed mechanisms is that aerosol loading invigorates updrafts through their impacts on cloud microphysics. However, it remains uncertain to what extent and through which pathways the invigoration might occur. In this study, a causal inference approach is employed to quantify the impacts of aerosol number concentrations on convective updraft properties using unique observations of convective vertical velocities during the Green Ocean Amazon (GoAmazon 2014/15) experiment. Isolated convective cells passing over the GoAmazon site are tracked using a Lagrangian tracking method. From this tracking data, the lifecycle stage of the cells is determined such that aerosol impacts on developing and mature convection can be investigated separately from dissipating convection. To run the causal inference model, vertical velocity retrievals for tracked cells are collocated in space and time with surface measurements of aerosol concentrations measured at different supersaturations and size ranges. To isolate the impacts of aerosols on updraft velocities from effects of other meteorological factors, we account for confounders such as convective available potential energy and environmental lapse rate. Since aerosol number concentrations in the Amazon exhibit a strong seasonality, results are also separated between the relatively clean wet season and the polluted dry season. A statistically significant reduction of updraft velocities of 1.6 m s^-1 is causally inferred for conditions with comparatively high fine aerosol concentrations during the wet season.