Predictability of atmospheric minor constituents validated by a high-performance data assimilation system

利用先进的数理统计方法—集合卡尔曼滤波，开发了一套大气微量成分数据同化系统。数据同化是一种将模型仿真结果与观测结果相融合的信息技术。利用该资料同化系统计算了平流层臭氧和对流层气溶胶浓度的分析。我们使用这些分析作为初始条件，对臭氧层和粉尘气溶胶进行了数值后播模拟。因此，臭氧/粉尘数小时/数天的可预测性明显提高。

We developed a data assimilation system for atmospheric minor constituents using the Ensemble Kalman filter that is an advanced mathematical statistic scheme. Data assimilation is an informational technology, which merges model simulation results and observation results. The analyses of stratospheric ozone and tropospheric aerosol concentrations were calculated by this data assimilation system. We performed numerical hindcast simulations of the ozone layer and dust aerosol using those analyses as initial conditions. Consequently, the ozone/dust predictability in several hours/several days was evidently improved.