Uncertainty Estimates of the EOF-Derived North Atlantic Oscillation

Abstract

ifferent approaches to obtaining uncertainty estimates of the North Atlantic Oscillation (NAO) are explored. The resulting estimates are used to enhance the understanding of spatial variability of the NAO over different time periods. Among the parametric and nonparametric approaches investigated in this study, the bootstrap is nonparametric and not confined to the assumption of normally distributed data. It gives physically plausible uncertainty estimates. The NAO uncertainty estimates depend on sample sizes with greater sampling variability as sample size is smaller. The NAO uncertainty varies with time but common features include that the most uncertain values are centered between the centers of action of the NAO and are asymmetric in the zonal direction (more uncertainty in the eastward direction or downstream). The bootstrap can also be used to provide direct measures of uncertainty regarding the location of the NAO action centers. The uncertainty of the location of the NAO action centers not only helps assess the shift in the NAO but also provides evidence of more than two action centers. The methods reported on here could in principle be applied to any EOF-derived climate pattern.