Optimizing Data Assimilation Strategy for a Global Aerosol Model with a Multi-Sensor Constellation

Operational models of atmospheric aerosols are used in a variety of civilian and military applications. Most of these models now use satellite observations to initialize the aerosol forecasts, using a variety of different sensors and data assimilation approaches. Integration of new aerosol data sources into data assimilation systems involves significant effort to ensure the quality of the data for assimilation. It is also necessary to ensure the consistency of the assumptions used in generating the satellite products and model priors that are combined in the assimilation cycle. There is now a constellation of at least 8 sensors capable of some form of aerosol retrieval with data available in near-real time for forecast model initialization. The effectiveness of additional sensors in an aerosol data assimilation system is strongly affected by the methods used to select observations for assimilation, and the technical details of the data assimilation system. This contribution discusses the aerosol data sources which have been tested for operational use in the Navy Aerosol Analysis and Prediction System (NAAPS). Different methods of combining data from different sensors are examined and their effect on data quantity, quality, and consistency is demonstrated. Additional experiments are shown using a simple variational assimilation system with different time windows and documenting how this affects the quality of the resulting analysis. These results illuminate the relative role of data quality and quantity in global aerosol analysis, and indicate the pathways for development of data assimilation systems best suited for the future constellation of aerosol observing satellites.