Forecasting ENSO events: A neural network-extended EOF approach
- Univ. of British Columbia, Vancouver (Canada)
The authors constructed neural network models to forecast the sea surface temperature anomalies (SSTA) for three regions: Nino 4. Nino 3.5, and Nino 3, representing the western-central, the central, and the eastern-central parts of the equatorial Pacific Ocean, respectively. The inputs were the extended empirical orthogonal functions (EEOF) of the sea level pressure (SLP) field that covered the tropical Indian and Pacific Oceans and evolved for a duration of 1 yr. The EEOFs greatly reduced the size of the neural networks from those of the authors` earlier papers using EOFs. The Nino 4 region appeared to be the best forecasted region, with useful skills up to a year lead time for the 1982-93 forecast period. By network pruning analysis and spectral analysis, four important inputs were identified: modes 1, 2, and 6 of the SLP EEOFs and the SSTA persistence. Mode 1 characterized the low-frequency oscillation (LFO, with 4-5-yr period), and was seen as the typical ENSO signal, while mode 2, with a period of 2-5 yr, characterized the quasi-biennial oscillation (QBO) plus the LFO. Mode 6 was dominated by decadal and interdecadal variations. Thus, forecasting ENSO required information from the QBO, and the decadal-interdecadal oscillations. The nonlinearity of the networks tended to increase with lead time and to become stronger for the eastern regions of the equatorial Pacific Ocean. 35 refs., 14 figs., 4 tabs.
- OSTI ID:
- 576797
- Journal Information:
- Journal of Climate, Vol. 11, Issue 1; Other Information: PBD: Jan 1998
- Country of Publication:
- United States
- Language:
- English
Similar Records
Interannual and interdecadal oscillation patterns in sea level
Seasonal climate summary. [Global climates summary for June-August 1992]