Abstract: The international ARGO Project can provide more than 100000 profiles (from 0-2000 meters depth) of seawater temperature and salinity every year, which would be of great benefit to the development of ocean and atmosphere sciences. It is very helpful to deepen the comprehension of the ocean processes, to reveal the mechanism of ocean-atmosphere interaction, and to provide the more real model initial field for long-range weather and shortterm climate forecasting so as to improve the model forecasting ability. Presently, it is one of the hot issues in the both science fields of atmosphere and ocean, that utilizing these data to study and to apply these data in practical operation. In this study, the ARGO buoy data were int roduced to the NCC-GODAS assimilation system. It is viewed by the results that the sea water temperatures assimilated with ARGO data agree well with the observations in three oceans not only on the magnitude of temperature but also on the distributing forms. It is also shown clearly that it can better describe the observed sea water temperature pattern in both winter and summer seasons as well as the characters of the seasonal variation and interannual anomalies. Additionally, in this study, the Zebiak-Cane (1987) ocean model has been improved by establishing a new subsurface temperature parameterize scheme for equatorial western Pacific Ocean, based on the ARGO data. It is indicated that it overcomes the limitation that ZC model can't simulate the surface and subsurface temperature in equatorial western Pacific Ocean. With the new parameterization, the simulated subsurface temperature anomalies are improved but the sea surface temperature anom alies are also better simulated. It is also revealed that both the simulated interannual characters and the variability amplitudes of surface and subsurface temperature are similar with the observations.