Long-term variability of storm tracks characteristics

Abstract

The variability of storm track characteristics is studied in terms of polar lows in the North Atlantic and extratropical cyclones in the Northern Hemisphere by detection and tracking algorithms. Two different cyclone-tracking algorithms to detect North Atlantic polar lows, which are very intense mesoscale cyclones, are compared. Both approaches include spatial filtering, detection, tracking and constraints specific to polar lows. Comparisons between these two methods show that different filters lead to different numbers and locations of tracks. The discrete cosine transform is more precise in scale separation than the digital filter and the results of this study suggest that it is more suited for the bandpass filtering of mean sea level pressure fields. The detection and tracking parts also influence the numbers of tracks although less critically. The winter storm activity over the North Hemisphere during the last one thousand years within a global climate simulation was also investigated by tracking and clustering analysis. The numbers of extratropical winter storms exhibit notable inter-annual variability, but only little variability on centennial time scales. The respective storm track clusters also show only small changes between the different centuries. Storm track clusters with longest lifetime and largest deepening rates are found over the North Atlantic and Pacific oceans, with negative correlations between the numbers of members of neighboring oceanic clusters. A linear relationship was found between the numbers of members per storm track clusters over the North Pacific or North Atlantic Ocean and specific atmospheric circulation patterns using a canonical correlation analysis.