Nonlinear canonical correlation analysis of the tropical Pacific wind stress and sea surface temperature

Abstract

Nonlinear canonical correlation analysis (NLCCA) formulated by a neural network approach was applied to the monthly surface wind stress (WS) and sea surface temperature (SST) in the tropical Pacific. The strength of the nonlinearity varies with the lead/lag time between WS and SST. Compared to the CCA modes, the NLCCA modes explain more variance of the two sets of variables and have higher canonical correlations, particularly, at longer lead/lag times. Unlike the CCA, the NLCCA modes are capable of capturing the asymmetry in the spatial patterns between El Nino and La Nina episodes in both SST and WS fields. With the WS lagging and then leading the SST, the roles of the predictor field and the lagging response field were interchanged, the spatial asymmetry was found to be considerably stronger in the response field than in the predictor field. Hindcasts for SST (using WS as predictor) show that the NLCCA model is generally slightly better than the CCA model in terms of the correlation skills and the root mean square error, mainly in the eastern equatorial regions (e.g. Nino12 and Nino3). Hindcasts for WS (using SST as predictor) show that despite the relatively rapid decrease of skills with the lead time (compared to the skills of SST prediction from WS), the NLCCA is slightly better than the CCA model, with greater improvement over the western equatorial Pacific.