302
Views
113
CrossRef citations to date
0
Altmetric
Original Articles

The influence of the North Atlantic Oscillation on the regional temperature variability in Sweden: spatial and temporal variations

&
Pages 505-516 | Received 26 Oct 1998, Accepted 22 Feb 1999, Published online: 15 Dec 2016
 

Abstract

A statistical analysis of the seasonal and interannual variations in the regional temperatureanomalies of Sweden during 1861–1994 is performed. The study uses homogenized monthlytemperatures averaged over 6 regions to minimize the non-climatic and local-scale climaticeffects. It is found that the temperature variability shows a clear regional and seasonal dependency.The topography, the influence of the sea and the synoptic climatology may have determinedthe dependency. The anomaly is related to variations in the North Atlantic Oscillation (NAO) expressed by an index (NAOI) and the extent to which the temperature anomaly canbe explained by the NAO is investigated. The results show that the NAO has an importanteffect on the regional Swedish temperature on the monthly and interannual scales. The relationshipbetween the temperature and NAOI over the period 1985–1994 are strong, implying that the NAOI may be a suitable candidate for a statistical downscaling model of the regional temperature.However, correlation analysis over different 31-year periods shows that the strength of the associationvaries with time and region. The further north the weaker the association. On the other hand, the temporal variations of themoving correlations for the 6 regions are similar. Part of the temporalvariations may be explained by the averaged strength of NAO during different 31-year periods.This is especially evident for southern Sweden. At last, the coherency spectrums between thetemperature anomalies and the NAO index is determined, which enables an examination of theassociation over the frequency domain. The result supports the idea that theNAOhas an importanteffect on the Swedish temperature, though the strength of the association varied with time. Theseresults have implications for statistical downscaling.