Is the North Atlantic oscillation a random walk? A comment with further results

Time series on the North Atlantic oscillation (NAO) have been subject to considerable analysis in recent years, with the consensus emerging that the data are not characterized by a random walk process. However, no consensus has yet emerged concerning the short run correlation structure of the data. This comment explores the time series properties of the NAO in more detail, using testing and modelling procedures that have found wide application in areas such as economics and finance. A structural time series model provides a comprehensive explanation of the index's rich dynamics, there being a slowly changing level component and a cyclical component having a period of approximately 7.5 years. These are dominated, however, by an irregular component, whose presence makes accurate forecasting of the index problematic. Copyright (C) 2004 Royal Meteorological Society.