Forecasts from September 1995

Published in the Experimental Long-Lead Forecast Bulletin, September 1995

The El Niño forecasts are part of an ongoing research effort at COLA, and do not represent official forecasts. These forecasts should not be used as the basis of any commercial, policy, or other decisions. They are strictly research tools for advancing the understanding of the ocean-atmosphere system.

In order to better sample the annual cycle we have doubled the hindcast ensemble from the original 30 cases to 60 cases. The figure below shows the time evolution of the NINO3 SSTA for all 60 hindcasts and the verifying SSTA analysis. The systematic error in the predicted SSTA has been removed and the observed NINO3 SSTA is indicated by the bold solid line. For the most part, the predictions track the observed anomalies for the first 12 months although there are some notable exceptions and, as expected, the hindcasts have the greatest degree of fidelity with the analysis during the major warm and cold events. Considering all 60 cases, the NINO3 correlation (not shown) remains above 0.6 for lead times up to 12 months.

Time evolution of the NINO3 SSTA for all 60 hindcasts and the observed SSTA. The thin curves denote the predicted hindcast and the thick curve is the observed SSTA.

The figure below shows the NINO3 time series of the predicted SSTA for three forecasts initialized on January 1, 1995, April 1, 1995 and September 1, 1995, respectively. Each forecast is for 18 months although only the last 10 months of the January 1, 1994 forecast and the last 13 months of the April 1, 1995 forecast are shown. Both the January 1995 and April 1995 forecasts predict near normal temperature in September-October-November 1995 and moderately cold temperature in the boreal winter and spring of 1995-96. In contrast, the forecast initialized in September 1995 has near normal temperature throughout the boreal winter and spring of 1995-96. The April 1995 forecast indicates a warming trend in mid-to-late 1996, whereas the September 1996 forecast has a cooling trend in late 1996.

Time evolution of the NINO3 SSTA forecast. The dotted curve is the last 10 months of the forecast initialized in January 1995, the dashed curve is the last 13 months of the forecast initialized in April 1995 and the solid curve corresponds to the forecast initialized in September 1995.

The horizontal structure of the predicted (September 1, 1995 initial condition) SSTA for the winter of 1995-96 and the spring of 1996 is shown below. In the boreal winter season, there are weak cold anomalies in the eastern Pacific and weak warm anomalies in the central and western Pacific, but, in general, the model predicts normal conditions for the tropical Pacific. This is in contrast to the forecast initialized in April which indicates 0.3-0.9 cold anomalies throughout most of the basin between 5S-5N for the boreal winter of 1995-96. In the spring season, the temperature rises slightly in the east to produce near normal conditions throughout the basin.

Top panel shows December-January-February 1995-96 mean SST anomaly for the forecast initialized in September 1995 forecast. Bottom panel show the March-April-May 1996 SSTA for the forecast initialized in September 1995. The contour interval is 0.30K and the zero contour is suppressed.