论文部分内容阅读
基于国家海洋局第一海洋研究所发展的地球系统模式(FIO-ESM),设计了集合调整Kalman滤波(EAKF)同化方案并开展了海洋卫星资料同化实验.设计的数值实验包括1组控制实验和4组同化实验.控制实验由一组初始场各不相同的模式组成,将FIO-ESM模式积分1年;同化实验则在积分过程中不断对海洋模式分量进行海洋卫星数据同化.前2组同化实验分别对卫星海面高度异常(SLA)和卫星海面温度(SST)数据进行同化,后2组实验中SLA和SST均加入同化,但两种数据同化顺序不同.为了检验同化过程对气候模式中海洋模拟的影响,将实验结果与再分析数据集EN3进行了对比分析.与海洋模式不同,耦合模式在海面的动量和热量通量是由耦合过程实时计算得出,耦合模式中多变量之间的约束关系更接近实际的物理过程.海洋卫星资料EAKF同化整体显著改善了FIO-ESM中海洋模式分量的模拟结果,尤其在1000 m以浅效果更为显著.不同类型的卫星观测数据的同化效果在深度上有所不同,SST的改善效果在表层附近较大,而SLA则对次表层改善较大,且在深层SLA的改善大于SST.联合SLA和SST同化的实验结果均比单独同化一类数据的效果更佳,但不同顺序对这2种数据进行同化的差异不显著.
Based on the FIO-ESM developed by the First Institute of Oceanography of the State Oceanic Administration, an ensemble-adjusted Kalman filter (EAKF) assimilation scheme and an ocean satellite data assimilation experiment are designed.The numerical experiments include 1 set of control experiments and Four groups of assimilation experiments.The control experiment consists of a group of different patterns of the initial field and integrates the FIO-ESM model for one year, while assimilation experiments continuously assimilate ocean-satellite data from the ocean model components in the integration process.The first two groups of assimilation Experiments assimilate satellite SLA and SST data respectively, and SLA and SST are assimilated in the latter two experiments, but the two data assimilation sequences are different.In order to test the effect of assimilation process on the oceanic climate Simulation, the experimental results were compared with the reanalysis dataset EN3.Unlike the ocean model, the momentum and heat fluxes of the coupled modes at the sea surface were calculated in real time by the coupling process, and the correlation between the multivariate The relationship between the constraints is closer to the actual physical process.East satellite data EAKF assimilation significantly improves the simulation results of the ocean mode components in the FIO-ESM, especially at 1000 m, the shallow effect is more significant.The assimilation effect of different types of satellite observation data is different in depth, the improvement effect of SST is larger near the surface layer, while the SLA improves the sub-surface layer more greatly, and the improvement of deep SLA is greater than SST. The experimental results of the assimilation of SLA and SST are better than the assimilation of a class of data alone, but there is no significant difference in assimilating the two kinds of data in different orders.