Forecasting the equatorial Pacific sea surface temperatures by neural network models

Abstract.

We used neural network models to seasonally forecast the tropical Pacific sea surface temperature anomalies (SSTA) in the Niño 3.4 region (6 ^°S–6 ^°N, 120 ^°W–170 ^°W). The inputs to the neural networks (i.e., the predictors) were the first seven wind stress empirical orthogonal function (EOF) modes of the tropical Pacific (20 ^°S–20 ^°N, 120 ^°E–70 ^°W) for four seasons and the Niño 3.4 SSTA itself for the final season. The period of 1952–1981 was used for training the neural network models, and the period 1982–1992 for forecast validation. At 6-month lead time, neural networks attained forecast skills comparable to the other El Niño-Southern Oscillation (ENSO) models. Our results suggested that neural network models were viable for ENSO forecasting even at longer lead times of 9 to 12 months. We hypothesized that at these longer leads, the underlying relationship between the wind stress and Niño 3.4 SSTA became increasingly nonlinear. The neural network results were interpreted in light of current theories, e.g., the role of the “off-equatorial” Rossby waves in triggering the onset of an ENSO event and the delayed-oscillator theory in the development and termination of an ENSO event.