• Journal of the Korean earth science society
• /
• v.26 no.7
• /
• pp.706-713
• /
• 2005

Subsurface oceanic data (Z20; Depth of $20^{\circ}C$ isotherm and WWV; Warm Water Volume) from the tropical Pacific Ocean from 1980 to 2004 were utilized to examine upper ocean variations in relation to E1 Nino. Time series analysis using EOF, composite, and cross-correlation methods indicated that there are significant time delays between subsurface oceanic parameters and the Nino3.4 SST. It implied that Z20 and WWV would be more reliable predictors of El Nino events. Based on analyzed results, we also constructed neural network model to predict the Nino3.4 SST from 1996 to 2004. The forecasting skills for the model using WWV were statistically higher than that using the trade wind except for short range forecasting less than 3 months. This model greatly predicted SST than any other previous statistical model, especially at lead times of 5 to 8 months.