Forecasts of Tropical Pacific SST Using a Comprehensive Coupled Ocean-Atmosphere Dynamical Model

A non-simple coupled ocean-atmosphere model has been developed for use for long-lead climate forecasts in the Coupled Model Project at NOAA's National Centers for Environmental Prediction (Ji et al. 1994a,b). The NCEP Medium Range Forecast (MRF) atmospheric model is used with a dynamic Pacific Basin ocean model originated at the Geophysical Fluid Dynamics Laboratory. The MRF has a reduced spatial resolution and is tuned for more realistic tropical circulation. The ocean thermal field, including SST and subsurface temperature, is initialized using an ocean data assimilation system (Ji et al. 1995). Research has shown that when observed SST fields are prescribed, this coupled model's atmospheric response is fairly reliable in the tropics but considerably less so in the extratropics. The extratropical response is best during the warm or cold phase of ENSO as reflected in the SST. Much attention has in fact been given to the prediction of ENSO itself--the tropical Pacific SST anomaly. Such a forecast is presented here.

In the September and December 1993 issues of this Bulletin, the expected forecast skill of the coupled model version used in 1993 (called CMP6) was shown. A tendency for a horseshoe-shaped spatial pattern of maximum model skill was noted, with highest equatorial skill near the date line and higher skill just north or south of the equator than immediately along it to the east of 165oW. Mean skill for the Nino 3 and Nino 4 regions was shown as a function of season and lead time. The model generally outperformed persistence by a substantial margin. A seasonal dependence in skill was noted, where forecasts were affected by a "spring barrier" as found both in other dynamical as well as statistical predictive models.

Starting with the forecasts presented in the September 1994 issue, the NCEP coupled model was upgraded in several ways, including a refinement of the flux climatology and the installation of a MOS correction for the stress anomalies produced by the atmospheric model. Thus, skills became higher than those cited above; e.g. the high skill horseshoe "thickened" along the equator, extending farther eastward into the western part of the Niño 3 region. Figures 2-1 and 2-2 of the September 1994 issue show a comparison of hindcast skills between the newer (called CMP9) and the previous model versions as a function of lead time for Niño 3 and Niño 4, respectively.

During the last few months, a still newer version of the NCEP model has been developed, called CMP10 (Ji et al. 1996). A major difference between CMP9 and CMP10 is in the heat flux coupling: CMP9 contained a negative feedback procedure for coupling the anomalous net heat flux, while CMP10 uses anomaly coupling for the net heat flux forcing. While mean skills are not as different between CMP9 and CMP10 as they are for CMP6 and CMP9, CMP10 behaves more realistically for high amplitude SST anomalies. CMP9, with its negative feedback mechanism, ran the danger of damping strong ENSO events too much and/or too soon.

The CMP9 and CMP10 coupled model forecasts for the SST anomaly field averaged over Mar-Apr-May 1996 (no lead), Jun-Jul-Aug 1996 (3 months lead) and Sep-Oct-Nov 1996 (6 months lead) are shown in Fig. 1, where the systematic model bias for hindcasts over the 1983-95 period has been subtracted. This forecast is actually the mean of an ensemble of 7 to 11 individual cases, each based on a different one- to two-week-apart initial ocean condition ranging from early January through early March, 1996. The CMP9 and CMP10 forecasts are similar, indicating some weakening and meridional narrowing of the present cool equatorial anomalies. Near dissipation of the cold episode is implied going into winter 1996-97. Fig. 2 shows the Nino 3 forecast in the form of a time series for the three lead times used to form the 3-month forecast averages used in Fig. 1 for both versions of the model.

The observed anomalous SST and subsurface equatorial temperature field for the week centered on February 28 (Fig. 3) include a region of negative subsurface sea temperature anomalies in the central and especially the eastern tropical Pacific Basin. Positive subsurface anomalies appear in the western Pacific, which may presage a return to warm conditions farther east at the surface in 12-18 months (Smith et al. 1995).

Ji, M., A. Kumar and A. Leetmaa, 1994a: A multi-season climate forecast system at the National Meteorological Center. Bull. Am. Meteor. Soc., 75, 569-577.

Ji, M., A. Kumar and A. Leetmaa, 1994b: An experimental coupled forecast system at the National Meteorological Center: Some early results. Tellus, 46A, 398-418.

Ji, M., A. Leetmaa and J. Derber, 1995: An ocean analysis system for seasonal to interannual climate studies. Mon. Wea. Rev., 123, 460-481.

Ji, M., A. Leetmaa and V.E. Kousky, 1996: Coupled model forecasts of ENSO during the 1980s and 1990s at the National Meteorological Center. J. Climate, 9, submitted.

Smith, T.M., A. G. Barnston, M. Ji and M. Chelliah, 1995: The impact of Pacific Ocean subsurface data on operational prediction of tropical Pacific SST at the NCEP. Wea. Forecasting, 10, 708-714.

Figures

Figure 1. NCEP (formerly NMC) coupled model SST anomaly forecast fields for Mar-Apr-May, Jun-Jul-Aug, and Sep-Oct-Nov 1996. Each forecast is an average of three individual month ensemble averages, which in turn are composed of 2 to 4 individual forecasts from 1 to 2-week-apart initial SST conditions. Forecasts from both the CMP9 and CMP10 are shown.

Figure 2. NCEP coupled model SST anomaly forecast time series for Nino 3 for Mar-Apr-May, Jun-Jul-Aug, and Sep-Oct-Nov 1996, as well as intermediate 3-month Mar-Apr-May, Jun-Jul-Aug, and Sep-Oct-Nov 1996, as well as intermediate 3-month periods at 1-month increments. The broken line in each panel represents the SST anomaly forecast (C), and the solid line the observed SST anomaly. The predictions represent the mean of three ensemble mean forecasts, each for one of the 3 most recent months, produced by forecasts from two to three individual 1 to 2-week-apart initial conditions per month.

Figure 3. Equatorial depth-longitude section of ocean temperature anomaly with respect to the 1983-92 mean for the week centered on February 28, 1996.