An ensemble adjustment Kalman filter study for Argo data  

作　　者：尹训强 乔方利 杨永增 夏长水 

机构地区：[1]The First Institute of Oceanography SOA [2]Key Laboratory of Marine Science and Numerical Modeling (MASNUM)SOA

出　　处：《Chinese Journal of Oceanology and Limnology》2010年第3期626-635,共10页中国海洋湖沼学报（英文版）

基　　金：Supported by the Project of National Basic Research Program of China (No. 2007CB816002);Special Fund for Fundamental Scientific Research (No. 2008G08)

摘　　要：An ensemble adjustment Kalman filter system is developed to assimilate Argo profiles into the Northwest Pacific MASNUM wave-circulation coupled model, which is based on the Princeton Ocean Model (POM). This model was recoded in FORTRAN-90 style, and some new data types were defined to improve the efficiency of system design and execution. This system is arranged for parallel computing by using UNIX shell scripts: it is easier with single models running separately with the required information exchanged through input/output files. Tests are carried out to check the performance of the system: one for checking the ensemble spread and another for the performance of assimilation of the Argo data in 2005. The first experiment shows that the assimilation system performs well. The comparison with the Satellite derived sea surface temperature (SST) shows that modeled SST errors are reduced after assimilation; at the same time, the spatial correlation between the simulated SST anomalies and the satellite data is improved because of Argo assimilation. Furthermore, the temporal evolution/trend of SST becomes much better than those results without data assimilation. The comparison against GTSPP profiles shows that the improvement is not only in the upper layers of ocean, but also in the deeper layers. All these results suggest that this system is potentially capable of reconstructing oceanic data sets that are of high quality and are temporally and spatially continuous.An ensemble adjustment Kalman filter system is developed to assimilate Argo profiles into the Northwest Pacific MASNUM wave-circulation coupled model, which is based on the Princeton Ocean Model （POM）. This model was recoded in FORTRAN-90 style, and some new data types were defined to improve the efficiency of system design and execution. This system is arranged for parallel computing by using UNIX shell scripts： it is easier with single models running separately with the required information exchanged through input/output files. Tests are carried out to check the performance of the system： one for checking the ensemble spread and another for the performance of assimilation of the Argo data in 2005. The first experiment shows that the assimilation system performs well. The comparison with the Satellite derived sea surface temperature （SST） shows that modeled SST errors are reduced after assimilation; at the same time, the spatial correlation between the simulated SST anomalies and the satellite data is improved because of Argo assimilation. Furthermore, the temporal evolution/trend of SST becomes much better than those results without data assimilation. The comparison against GTSPP profiles shows that the improvement is not only in the upper layers of ocean, but also in the deeper layers. All these results suggest that this system is potentially capable of reconstructing oceanic data sets that are of high quality and are temporally and spatially continuous.

关 键 词：ensemble adjustment Kalman filter Argo profile data assimilation 

分 类 号：P731.11[天文地球—海洋科学]