Home | Climate Info System -> IRI Forecasts -> SPEF forecasts

Probabilistic Predictions of ENSO Indices Using Statistical Models

Forecasts of monthly NIÑO1&2, NIÑO3, NIÑO3.4, and NIÑO4 sea-surface temperature anomalies with lead-times of up to 11 months are produced using predictive discriminant analysis, canonical variate analysis, four forms of generalized linear models, and multiple linear regression with forecast probabilities derived by integration of the prediction intervals. Full details of the models are provided by Mason and Mimmack (2002). The forecasts presented here represent an average of the forecast probabilities from these statistical models. The first four unrotated principal components of gridded monthly sea-surface temperatures over the tropical Pacific (25ºN—25ºS, 110ºE—70ºW), and the respective NIÑO index are used as the only predictors. The NIÑO index is regressed out of the sea-surface temperatures before the principal components are calculated to ensure that the index remains orthogonal to the components. Skilful forecasts of tropical Pacific sea-surface temperature anomalies can be developed relatively simply using only prior temperatures in the region as predictors (Barnston and Ropelewski 1992; Penland and Sardeshmukh 1995; Latif et al. 1998), and so the relatively simple sets of predictors used here can provide forecasts with high levels of skill.

Careful assessments of the operational levels of forecast skill have been made by using a retroactive forecast procedure: the models were trained over an initial 30-year training period, and were then used to produce 22 years of retroactive forecasts. Five categories of anomalies were defined ranging from "cold", through "cool", "normal", "warm", to "hot". Probabilities for each of the categories over the 22-year retroactive period January 1981 to December 2002 were calculated. The training period was initially set as 30 years (1951—80), and retroactive predictions for the following five years were then made using the optimal model. After this 5-year period the models were retrained over the period 1951—85, possibly selecting different variables and a different number of retained variables, and predictions for 1986—90 were made. This procedure was repeated until the set of 22 years of retroactive predictions had been made. At each stage, the definitions of the five categories were reset to ensure that the categories remained equi-probable a priori. While the categories are defined as equi-probable over the training periods, this is not necessarily the cases for the verifications over the independent period. For 1981—85, the verifications were categorized on the basis of the 1951—80 training period; the verifications for 1986—90 were categorized on the basis of the 1951—85 training period, etc.. For the forecasts presented here, the training period was from 1951 to the end of the latest year available, and anomalies are defined with reference to this same period.

The combined forecast is calculated by averaging the forecast probabilities from the various models. No attempt is made to weight the probabilities from the different models by a measure of model skill, since ranked model performance is sensitive to the precise skill measure used, and can be conditional upon the actual outcome. Good reliability is demonstrated for forecasts of all categories except "normal" (Fig. 1). Ranked probability skill scores (RPSS's) were calculated for each month separately, using the combined forecast probabilities, and comparing the forecasts to a strategy of climatology. The scores for six months are shown in Fig. 2, where they are compared to the skill of persistence and damped persistence forecasts. The seasonal dependence of skill is clearly apparent for both the model and the (damped) persistence forecasts. For example, for forecasts of NIÑO3.4 from August there is positive skill out to March of the following year, although persistence outscores the model at short lead-times.

Fig. Reliability diagram for retroactive combined forecasts at increasing lead-times of "cold" (dark blue line), "cool" (light blue line), "normal" (green line), "warm" (orange line), and "hot" (red line) conditions for the 22-year period Jan 1981—Dec 2002. Forecasts at all lead-times and for all months are pooled. The histograms indicate the frequency of forecasts with probabilities in the ranges 0.00—0.05, 0.05—0.15, 0.15—0.25, ..., 0.95—1.00. The y-axes range to 1700. The top histogram is for "hot" conditions, the second top for "warm" conditions, and the rest as indicated.

Fig. 2. Ranked probability skill scores for retroactive combined forecasts at increasing lead-times of monthly Niño-3.4 sea-surface temperature anomaly categories for the 22-year period Jan 1981—Dec 2002. The skill scores are calculated with reference to a strategy of forecasting climatology. The red bars represent the scores for the models, and the blue gray bars are for forecasts of persisted anomaly categories damped toward climatology.

REFERENCES

Barnston, A. G., and C. F. Ropelewski, 1992: Prediction of ENSO using canonical correlation analysis. J. Climate, 5, 1316—1345.

Latif, M., D. L. T. Anderson, T. P. Barnett, M. A. Cane, R. Kleeman, A. Leetmaa, J. O'Brien, A. Rosati, and E. Schneider, 1998: A review of the predictability and prediction of ENSO. J. Geophys. Res., 103, 14 375—14 393.

Mason, S. J., and G. M. Mimmack, 2002: Comparison of some statistical methods of probabilistic forecasting of ENSO. J. Climate, 15, 8—29.

Penland, C., and P. D. Sardeshmukh, 1995: The optimal growth of tropical sea-surface temperature anomalies. J. Climate, 8, 1999—2024.