Identifying Better Indicators of Aerosol Wet Scavenging during Long-Range Transport
MIGUEL HILARIO, Avelino Arellano, Ali Behrangi, Ewan Crosbie, Josh DiGangi, Glenn Diskin, Michael Shook, Luke Ziemba, Armin Sorooshian,
University of Arizona Abstract Number: 105
Working Group: Aerosols, Clouds and Climate
AbstractAs the dominant sink of aerosol particles, wet scavenging greatly influences particle lifetimes and their downstream interactions with clouds, precipitation, and radiative forcing. However, wet scavenging remains highly uncertain in models, hindering model capabilities to accurately predict aerosol effects. To understand the role of transport conditions on aerosol scavenging, previous studies used enhancement (Δ) ratios of black carbon and carbon monoxide (ΔBC:ΔCO) and accumulated precipitation along HYSPLIT trajectories (APT). However, APT inherently neglects in-cloud scavenging, which is more efficient than below-cloud scavenging at removing submicrometer particles. In this work, we aim to identify an improved indicator of wet scavenging that accounts for both below- and in-cloud processes by assessing an array of scavenging-related predictor variables from satellite retrievals (e.g., APT) and reanalysis (e.g., relative humidity (RH), water vapor mixing ratios). We utilize aircraft measurements of ΔBC:ΔCO over the West Pacific region, which hosts a dynamic transport environment rich in aerosol sources and cloud-precipitation systems. To assess each predictor variable, we applied a general first-order exponential decay model to derive relationships between ΔBC:ΔCO and each predictor variable. Bootstrapping and k-fold cross-validation were implemented to evaluate the robustness of each model fit. We find that among the analyzed predictors, the fraction along trajectories where RH exceeds 95% (fRH95) is a robust improvement upon APT from IMERG when predicting observed ΔBC:ΔCO, with predicted-to-observed regression slopes closer to 1 (fRH95: 0.5; APT: 0.1), higher predicted-to-observed correlations (R = 0.8) than APT (0.2 – 0.3), and approximately 30% lower prediction bias than APT. Analysis also isolates in-cloud scavenging (e.g., fRH95 where APT = 0) and considers other model fits (e.g., logarithmic) to characterize aerosol-scavenging relationships. This work demonstrates a method of assessing alternative indicators of wet scavenging that can be applied in a variety of environments to provide more accurate estimates of aerosol scavenging during transport.