Long monthly European temperature series and the North Atlantic Oscillation

Abstract

In this paper the relationship between the surface air temperatures in 28 European cities and towns and the North Atlantic Oscillation (NAO) are modelled using the Vector Seasonal Shifting Mean and Covariance Autoregressive model, extended to contain exogenous variables. The model also incorporates season-specific spatial correlations that are functions of latitudinal, longitudinal, and elevation differences of the various locations. The empirical results, based on long monthly time series, agree with previous ones in the literature in that the NAO is found to have its strongest effect on temperatures during winter months. The transition from the winter to the summer is not monotonic, however. The strength of the error correlations of the model between locations is inversely related to the distance between the locations, with a slower decay in the east–west than north–south direction. Altitude differences also matter but only during the winter half of the year.