Near Real-Time Aerosol Assimilation and Prediction Using JEDI and NOAA’s UFS-Aerosols

A near real-time (NRT) aerosol assimilation and prediction system was recently developed at NOAA/OAR/GSL using NOAA’s future operational global aerosol model and data assimilation system: Unified Forecast System - Aerosols (UFS-Aerosols) and Joint Efforts for Data assimilation Integration (JEDI), respectively. The Visible Infrared Imaging Radiometer Suite (VIIRS) aerosol optical depth (AOD) retrieval products at 550 nm from NOAA/NESDIS are assimilated using the ensemble-variational framework in JEDI to constrain initial conditions in UFS-Aerosols. To account for aerosol emission uncertainty and thus improve the ensemble-based AOD assimilation, a procedure was developed to stochastically perturbed aerosol emissions in UFS-Aerosols in the ensemble forecasts. To assess the longer-term impact of the AOD assimilation and the overall skills of aerosol forecasts against climatology, we produce NRT five-day UFS-Aerosols forecasts. Specifically, the daily biomass burning emission of Global Biomass Burning Emissions Product (GBBEPx) version 4 from NOAA/NESDIS on the current day is projected forward in time through the aforementioned stochastic emission perturbation approach, to provide future biomass burning emissions in the five-day UFS-aerosols forecasts. The NRT experiments using UFS-Aerosols will be evaluated against various aerosol-related observations and analyses/reanalyses, and compared with another NRT AOD assimilation using Global Ensemble Forecast System - Aerosols (GEFS-Aerosols). In addition, the five-day UFS-Aerosols forecasts using the projected GBBEPx emissions will be compared with those using the real-time GBBEPx emissions valid on the future days.