THE IMPACT OF INITIAL FORCED WIND ON THE PREDICTABILITY OF THE ZEBIAK-CANE COUPLED OCEAN-ATMOSPHERE MODEL

With simultaneous observed sea surface temperature anomaly (SSTA), the difference betweenNCEP/NCAR 925hPa reanalysis wind stress anomaly (NCEPWSA) and FSU wind stress anomaly (FSUWSA) isanalyzed, and the prediction abilities of Zebiak-Cane coupled ocean-atmosphere model (ZC coupled model) withNCEPWSA and FSUWSA serving respectively as initialization wind are compared. The results are as follows.The distribution feature of NCEPWSA matches better with that of the observed SSTA than counterpart ofFSUWSA both in 1980s and in 1990s; The ZC ocean model has a better skill under the forcing of NCEPWSAthan that of FSUWSA, especially in 1990s. Meanwhile, the forecast abilities of the ZC coupled model in 1990sas well as in 1980s have been improved employing NCEPWSA as initialization wind instead of FSUWSA.Particularly, it succeeded in predicting 1997/1998 El Nio 6 to 8 months ahead; further analysis shows that onthe antecedent and onset stages of the 1997/1998 El Nio event, the horizontal cold and warm distributioncharacteristics of the simulated SSTA from ZC ocean model, with NCEPWSA forcing compared to FSUWSAforcing, match better with counterparts of the corresponding observed SSTA, whereby providing betterpredication initialization conditions for ZC coupled model, which, in turn, is favorable to improve the forecastability of the coupled model.

With simultaneous observed sea surface temperature anomaly （SSTA）, the difference between NCEP/NCAR 925hPa reanalysis wind stress anomaly （NCEPWSA） and FSU wind stress anomaly （FSUWSA） is analyzed, and the prediction abilities of Zebiak-Cane coupled ocean-atmosphere model （ZC coupled model） with NCEPWSA and FSUWSA serving respectively as initialization wind are compared. The results are as follows. The distribution feature of NCEPWSA matches better with that of the observed SSTA than counterpart of FSUWSA both in 1980s and in 1990s; The ZC ocean model has a better skill under the forcing of NCEPWSA than that of FSUWSA, especially in 1990s. Meanwhile, the forecast abilities of the ZC coupled model in 1990s as well as in 1980s have been improved employing NCEPWSA as initialization wind instead of FSUWSA. Particularly, it succeeded in predicting 1997/1998 E1 Nino 6 to 8 months ahead, further analysis shows that on the antecedent and onset stages of the 1997/1998 E1 Nino event, the horizontal cold and warm distribution characteristics of the simulated SSTA from ZC ocean model, with NCEPWSA forcing compared to FSUWSA forcing, match better with counterparts of the corresponding observed SSTA, whereby providing better predication initialization conditions for ZC coupled model, which, in turn, is favorable to improve the forecast ability of the coupled model.