Dust Aerosol Analysis and Prediction with Lidar Observations and Ensemble Kalman Filter

We have developed a state-of-the-art data assimilation system for a global aerosol model with a four dimensional Ensemble Kalman Filter (4D-EnKF) in which Lidar observations, i.e., attenuated backscattering coefficient, depolarization ratio, and extinction coefficient, were successfully assimilated. The concentrations of dust, sulfate, and seasalt aerosols as well as the dust surface emission intensity were treated as control variables in this data assimilation system. The Lidar observations were obtained from the Level 1B dataset of the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) or the dataset of the East Asian ground-based Lidar network operated by the National Institute for Environmental Studies of Japan (NIES). With the use of these Lidar observations and 4D-EnKF system, aerosol data assimilation and prediction experiments were globally performed in the spring (March - May) of 2007. In this paper, we especially focus on the analysis and prediction of Asian dust which is a seasonal meteorological phenomenon sporadically affecting East Asian countries during the springtime. The analysis and prediction results derived from satellite and ground-based observations were compared with each other, and validated by independent observations: 1) aerosol optical depth measured by the Moderate Resolution Imaging Spectro-radiometer (MODIS) over East Asia, and 2) weather reports on aeolian dust events in East Asia derived from the World Meteorological Organization (WMO) Surface Synoptic Observations (SYNOP). Forecast scores were estimated by phenomenal discrimination (i.e. hit or not) using the SYNOP weather reports and a threshold of modeled dust surface concentration, for example, 100 micrograms/m3. Detailed four-dimensional structures of dust outflows from source regions, such as Taklimakan or Gobi desert, to the Pacific Ocean over the Korean Peninsula or the Japanese Archipelago were well reproduced by this data assimilation system. The intensity of dust emission at each grid point was also adjusted as a consequence of the inversion analysis of the four dimensional data assimilation. The short-range dust prediction was generally improved by using the results of the data assimilation analysis as initial conditions. These results are valuable for the comprehensive analysis of aerosol behavior as well as aerosol forecasting.