Analysis Of EAKF Assimilation Experiment Results Of Marine Satellite Data In Climate Model

Climate change has always attracted the attention from international governmentand cooperation. Climate simulation and prediction, which is also the currentworld-wide research focus. Earth System models, as an important tool to predictclimate variability has been rapid development in recent years. However, the currentcoupled climate models have limited predictive ability of global and regional climate.In addition to exploring multi-scale physical processes of ocean, atmosphere andair-sea interaction, adding relatively impact of human life, and other alikesophisticated physical approach to improve the predictability of climate models,making full use of the marine and atmospheric observation to establish a coupled dataassimilation system for climate models, is also an effectively method to improve andenhance the simulations and prediction of climate models. The study of correspondingdata assimilation experiments will promisingly promote the development of climatemodels, and also play an important role in excavating the further value ofobservational data.Based on the First Institute of Oceanography Earth System Model, version1.0(FIO-ESM), a10-member ensemble adjustment Kalman filter (EAKF) was used toassimilate two types of satellite surface data, namely the sea surface temperature (SST)and the sea level anomaly (SLA). Five numerical experiments were established,including a control experiment without data assimilation and four assimilationexperiments. Evaluation and analysis of assimilation effects have been carried out bycomparing the assimilated results with EN3from Met Office Hadley Centreobservations datasets. The results show the EAKF data assimilation system cansignificantly improve the temperature and salinity in maritime component of theFIO-ESM. Four assimilation experiment proved to have different characteristics inspatial and temporal distribution, and the experiments which assimilate SST have abetter performance in simulate tropical warm pool and cold tongue. The comparison of two individual assimilation experiments shows that SST maximum contribute atthe surface while SLA maximum contribute at subsurface after assimilation.Comparison between combined assimilation experiments suggest that the sequence ofassimilating SST and SLA has little influence on the effects of assimilation. Bothindividual and combined assimilation of satellite-derived SLA and SST showsobviously improved temperature and salinity at depths shallower than1000m. And allresults could be concluded than the combined assimilation is optimal in an overallsense. Vertical correlation between the assimilation data and the adjust variables isone of the main reason, which make a different performance in surface and subsurfacelayers between assimilation experiments. The results showed that data assimilationcan also significantly improve the precipitation simulation. For the global annualmean precipitation, the data assimilation can effectively inhibit the spurious doubleinter-tropical convergence zone (ITCZ) phenomenon which is a common problem inall climate models. The precipitation peak in the southern equator is reduced, and as aresult the peak in the southern equator is lower than that in the northern equator. Thispattern is more consistent with the datasets of GPCP and CMAP. The improvement ofprecipitation in low-latitude regions is the largest, followed by the mid-latituderegions, and then in high-latitude regions. The root mean square errors of the monthlymean precipitation also indicated much improvement due to the ocean dataassimilation in FIO-ESM.