Neural-dynamical hybrid coupled model forecasts of the tropical Pacific sea surface temperatures

Youmin Tang and William Hsieh

A new neural-dynamical hybrid coupled model has been developed for giving seasonal predictions of the tropical Pacific sea surface temperatures. A 6-layer dynamical ocean model of the tropical Pacific is driven by the FSU wind stress; then during the forecasting period (Tang and Hsieh, 2001a), the ocean model is coupled to a nonlinear neural network atmospheric model, which estimates the surface wind stress anomalies from the upper ocean heat content anomalies (Tang et al, 2001).

For better forecast skills, different types of data-- surface wind stress, upper ocean heat content anomaly (HCA) (White, 1995), sea surface temperature (SST) and sea surface height anomaly-- have been assimilated into the coupled model using a 3D-Var data assimilation scheme (Derber and Rosati, 1989). The results show that assimilating HCA (kindly provided by Warren White and Ted Walker at SIO) yields the greatest improvement in the forecast correlation skills (Tang and Hsieh, 2001b). Fig. 1 shows the correlation skills of the predicted SST anomalies (SSTA) in the NINO3 region (5S-5N, 150W-90W) in the equatorial eastern Pacific during 1980-1989 and 1990-1999 using our model with HCA assimilation. The predictions were made at three months intervals (starting on 1 January, 1 April, 1 July and 1 October) and continued until a lead time of 15 months.

Fig.1 Correlation skills of the predicted NINO3 SSTA.

Fig.2 Observed and predicted NINO3 SSTA at lead times of 3, 6, 9 and 12 months.

Fig 3 shows our latest forecasts (initialized using data till the end of February, 2001), indicating that the moderate cool anomalies in the western equatorial Pacific and the moderate warm anomalies in the eastern equatorial Pacific present during late spring, 2001, will gradually fade away to near normal conditions by November, 2001, and remaining near normal till spring, 2002.

Fig.3 Predicted SSTA of the tropical Pacific.

Contour intervals are 0.5 degree Celsius. Positive anomalies above 1 degree are shaded in red, and negative anomalies below -1 degree are in blue. The zero contour is in purple.


  • Derber, J. and A. Rosati, 1989: A global oceanic data assimilation system. J. Phys. Oceanogr., 19, 1333-1347.

  • Tang, Y., and W. W. Hsieh, B. Tang and K. Haines, 2001: A neural network atmospheric model for hybrid coupled modelling. Clim. Dynam., 17, 445-455.

  • Tang, Y. and W.W. Hsieh, 2001a: Hybrid coupled models of the tropical Pacific -- ENSO prediction. Clim. Dynam. (submitted). []

  • Tang, Y. and W. W. Hsieh, 2001b: Impact of data assimilation on ENSO simulations and predictions. J. of Climate (submitted). []

  • White, W.B., 1995: Design of a global observing system for gyre-scale upper ocean temperature variability. Prog. in Oceanogr., 36, 169-217.