Development of the global data assimilation system for aerosols - Tuning of JEDI-based forecast error covariance

Global aerosol data assimilation is being developed as a component of the Unified Forecast System (UFS) at the National Centers for Environmental Prediction (NCEP). This makes use of the Joint Effort for Data assimilation Integration (JEDI) framework, which is a model independent data assimilation framework under current development. The aerosol forecast system will assimilate aerosol optical depth retrievals from VIIRS N20/N21 instruments. The initial system will be 3d-Var, and the current aerosol assimilation efforts are focused on tuning of the static forecast error covariance. JEDI employs the B-Matrix on an Unstructured Mesh Package (BUMP) for this purpose, and we are conducting 24 and 48 hour lagged forecast experiments to produce new covariance estimates that include seasonal variations. We present the initial results of the validation of this seasonal covariance tuning using VIIRS O-F statistics, along with comparisons with Aeronet observations and MERRA-2 reanalyses.