Are band-pass variance statistics useful measures of storm track activity? Re-examining storm track variability associated with the NAO using multiple storm track measures

Abstract

Over the past few decades, band-pass filtered (BP) variance statistics have been widely used as indicators of storm track activity. A recent study suggested that these statistics may be subjected to biases related to Doppler shifting of variance into and out of the fixed frequency band, in the process perhaps greatly exaggerating the correlation between BP variance and mean flow variability. In this study, BP statistics are examined together with other storm track measures to assess whether BP variances are useful indicators of the patterns of storm track activity. Storm track variability related to the North Atlantic Oscillation, as well as the five leading EOFs of 300 hPa BP filtered variance statistics, have been examined. Results presented here suggest that BP variance statistics are useful measures of the patterns of storm track activity. Firstly, BP variance statistics largely reflect changes in total unfiltered transient eddy variance as well as spatially filtered statistics. Secondly, different BP statistics all give qualitatively similar patterns. Thirdly, patterns derived from BP statistics are well correlated (spatially) with precipitation anomalies. In addition, BP statistics are straightforward to compute, and easily reproducible. Moreover, BP variance and covariance statistics are closely related to how storm tracks interact with the large scale circulation. However, results shown in this paper support the suggestion that some biases may be caused by the Doppler effect, especially for variations such as the NAO which are tightly tied to changes in the background flow speed. Thus it is argued that BP statistics should be examined together with other measures of storm track activity to obtain a broader perspective on storm track variations.