Ensemble based data assimilation with global aerosol climate model

Airborne aerosol particles can impact not only air quality but also climate through direct and indirect effects. To estimate these effects, it is required to evaluate temporal and spatial distributions of aerosols. However, uncertainty in the estimate of aerosol distributions remains large. Data assimilation integrates observations and numerical simulations to obtain the optimal solution and reduce the uncertainty. In this study, we have applied ensemble based data assimilation method to global aerosol climate model to improve estimates of aerosol distributions and its radiative forcing. Local Ensemble Transform Kalman Filter (LETKF: Hunt et al., 2007) and global aerosol climate model (SPRINTARS: Takemura et al., 2000, 2005) were used to develop the assimilation system. A one-month experiment with Aerosol Optical Thickness (AOT) measured by MODIS/AQUA was performed for May 2007. Figure 1 shows comparison of MODIS, a priori and a posteriori AOT distributions. Increment between a priori and a posteriori AOT is shown in Fig. 1 (d). The assimilation improves global AOT distribution. In detail, mineral dust is increased over the Middle East, India, East Asia and North Pacific Ocean. Carbonaceous aerosol is increased over Central America, and sulfate aerosol increases over North America. On the other hand, sea salt reduces over the Southern Hemisphere. Root Mean Square Difference (RMSD) between model and MODIS AOT is reduced by 32 %. Assimilated aerosol distribution adjusts direct radiative forcing by natural and anthropogenic aerosols from 0.14 to 0.32 (W/m2) in long wave, from -0.09 to 0.02 (W/m2) in short wave reflecting large increase of mineral dust and carbonaceous. Further analysis and validation of assimilation results will be shown. Monthly mean distribution of aerosol optical thickness (AOT). a) MODIS/AQUA, b) SPRINTARS before assimilation, c) SPRINTASRS assimilated. d) Increment of AOT between assimilated and before assimilation.