An analysis of the variability of the North Atlantic Oscillation in the time and the frequency domains

Abstract

Abstract 10.1002/1097-0088(20001130)20:14<1675::AID-JOC564>3.3.CO;2-3 An analysis of the variability of the North Atlantic Oscillation (NAO) since the beginning of the 19th century has been carried out using monthly pressure series from Gibraltar, Lisbon, The Azores and Iceland. In the first part, the combinations of stations that best monitor the NAO at different time intervals (monthly, seasonal and winter-annual) are analysed based on a signal/noise approach. The stability through time of the relationship between the pressures at the different stations is also studied. Based on this analysis, proxy indices of the NAO are obtained and compared with the NAO index resulting from a principal component analysis (PCA) of gridded sea level pressure (SLP) data. In the second part, the cross-spectral characteristics of the data of the northern and southern stations are studied in order to determine for which ranges of periods the pressure undergoes simultaneous coherent and out-of-phase variations. Based on the cross-spectral analysis, several filters are proposed to be applied to the pressure series prior to calculating the NAO index, in order to improve the reliability of the NAO index as an indicator of the NAO. Results show that for a monthly or seasonal index, The Azores must be selected as the southern station, but the strongest relationship is found for winter using Gibraltar as the southern station. The cross-spectral analysis shows, firstly, that the periods that are mainly responsible for coherent and out-of-phase variations are 6 months, 1 year, 2 years and 8 years (the most important). Secondly, several ranges of periods, mainly from 3 to 4 years and less than 6 months, are shown not to contribute to coherent and out-of-phase relationships between stations. Removing these ranges of periods from the pressure series prior to calculating the indices notably improves the reliability of the indices as indicators of the NAO state. The comparison of the filtered and unfiltered indices shows that extreme values of the NAO result from coherent out-of-phase pressure variations between stations. Copyright © 2000 Royal Meteorological Society