Forecast of Tropical Pacific SST Using a Simple Coupled Ocean-Atmosphere Dynamical Model

contributed by Stephen E. Zebiak, Mark A. Cane and Dake Chen

Lamont-Doherty Earth Observatory of Columbia University, Palisades, New York

Since the middle to late 1980s, forecasts of the NINO3 SST anomaly have been regularly made at Lamont-Doherty Earth Observatory (LDEO) of Columbia University using a simple coupled ocean-atmosphere dynamical model (Cane et al., 1986; Cane and Zebiak, 1987; Zebiak and Cane, 1987). This represented the beginning of a strong movement toward physical approaches to the diagnosis and prediction of climate and its short-term fluctuations. Here we present a few details of this model's current forecasts of SST for the tropical Pacific basin. In addition to the standard LDEO model (LDEO1), forecasts using a new version of the model with improved initialization (LDEO3) will also be presented.

Figure 1 shows 6, 9 and 12 month lead SST anomaly forecasts by LDEO1 for the tropical Pacific basin, verifying in April, July and October 1999, respectively. The forecasts indicate a moderate cooling of the eastern and central equatorial Pacific in spring of 1999, followed by a gradual weakening of the cooling toward summer and fall. A closer look at the forecast integrations for the NINO3 region in particular is provided in Figure 2 and 3. Figure 2 shows the time series of NINO3 SST anomaly for 6 different lead times, while Figure 3 shows 6 individual forecasts beginning from 1-month-apart initial conditions from May to October 1998 along with the ensemble mean. All runs indicate a cooling trend over the next several months and negative SST anomalies through mid-1999.

The new version of the model, LDEO3, follows the same coupled initialization procedure of its predecessor LDEO2 (Chen et al., 1995) except that it assimilates sea level data in addition to winds (Chen et al., 1998). This version retains the good overall skill of LDEO2 for the 1970s and 1980s while performing much better for the 1990s, especially for the recent El Niño event. The improvement is attributed to the extreme effectiveness of the sea level data in the vicinity of the equator in correcting the model ocean state and preconditioning it for prediction. LDEO3 represents the first step of our effort at LDEO to develop a systematic procedure that assimilates multiple data sets into the Lamont forecast system.

Forecasts made by LDEO3 are shown in Figures 4, 5 and 6. Note that the 6-month ensemble averaging used in LDEO1 (Figures 1 and 2) is no longer needed here because consecutive forecasts are rather consistent with one another. LDEO3 obviously did a better job in predicting the 1997/98 El Niño as compared to LDEO1. Its current forecasts indicate that SST anomaly will remain negative in the eastern and central equatorial Pacific over the next few seasons.

References

Cane, M. A., S. E. Zebiak and S. C. Dolan, 1986: Experimental forecasts of El Niño, Nature, 321, 827-832.

Cane, M. A., and S. E. Zebiak, 1987: Prediction of El Niño events using a physical model, In Atmospheric and Oceanic Variability, H. Cattle, Ed., Royal Meteorological Society Press, 153-182.

Chen, D., S. E. Zebiak, A. J. Busalacchi and M. A. Cane, 1995: An improved procedure for El Niño forecasting: implications for predictability. Science, 269, 1699-1702.

Chen, D., M. A. Cane, and S. E. Zebiak, 1998: The impact of sea level data assimilation on the Lamont model prediction of the 1997/98 El Niño, Geophys. Res. Lett., 25, 2837-2840.

Zebiak, S. E. and M. A. Cane, 1987: A model El Niño-Southern Oscillation. Mon.Wea. Rev., 115, 2262-2278.

Figure 1: LDEO1 predicted SSTA at 6, 9 and 12 month lead times. Predictions are based on ensemble mean individual forecasts initialized from six consecutive months ending in October 1998. Forecasts for each lead time (independent of start month) have been corrected for systematic biases using a singular value decomposition analysis based on the years 1972-92.

Figure 2: LDEO1 forecasts of NINO3 SSTA for 6 different lead times. Each forecast is actually the mean of forecasts from 6 consecutive months, adjusted to have the same mean and standard deviation as the observed. Note that 0 month lead can differ from the observed because SST data are not used in initialization. For each lead time, forecast values are indicated by x's and observed values by a solid line. Error bars represent root-mean-square error, based on the years 1972-1987.

Figure 3: LDEO1 forecasts of NINO3 SSTA initialized one month apart from May to October 1998 (dashed curves). The thin solid line is the ensemble mean of the 6 individual forecasts. The thick bold line is the observed NINO3 SSTA.

Figure 4: Same as Fig. 1 except for LDEO3 without ensemble averaging.

Figure 5: Same as Fig. 2 except for LDEO3 without ensemble averaging.

Figure 6: Same as Fig. 3 except for LDEO3.