A Multiphysics Method for Increasing Model Background Error in the Aerosol Data Assimilation with an Ensemble Kalman Filter

Abstract

This study addresses the issue of insufficient ensemble spread in the Ensemble Kalman Filter (EnKF) for aerosol data assimilation, which limits the adequate reflection of observational data. To address this, new methods are proposed to increase ensemble spread by incorporating multiple parameterizations for planetary boundary layer (PBL) and cloud microphysics in each ensemble run, representing model uncertainty in physical parameterizations. Variations in wind speed, rather than temperature or relative humidity, effectively induce more perturbations in the PBL, improving aerosol simulation uncertainty and successfully transferring the data assimilation impact to the upper atmosphere. This approach significantly increases model background errors, thereby improving surface PM2.5 analysis quality and forecasting skills for PM2.5 by up to 24 hours. The results highlight the importance of considering uncertainties in meteorological states, particularly in PBL schemes, for advancing PM2.5 data assimilation and forecasting performance.