Robustness of serial clustering of extra-tropical cyclones to the choice of tracking method
This is the author accepted manuscript. The final version is available from Co-Action Publishing via the DOI in this record.
Cyclone clusters are a frequent synoptic feature in the Euro-Atlantic area. Recent studies have shown that serial clustering of cyclones generally occurs on both flanks and downstream regions of the North Atlantic storm track, while cyclones tend to occur more regulary on the western side of the North Atlantic basin near Newfoundland. This study explores the sensitivity of serial clustering to the choice of cyclone tracking method using cyclone track data from 15 methods derived from ERA-Interim data (1979–2010). Clustering is estimated by the dispersion (ratio of variance to mean) of winter [December – February (DJF)] cyclone passages near each grid point over the Euro-Atlantic area. The mean number of cyclone counts and their variance are compared between methods, revealing considerable differences, particularly for the latter. Results show that all different tracking methods qualitatively capture similar large-scale spatial patterns of underdispersion and overdispersion over the study region. The quantitative differences can primarily be attributed to the differences in the variance of cyclone counts between the methods. Nevertheless, overdispersion is statistically significant for almost all methods over parts of the eastern North Atlantic and Western Europe, and is therefore considered as a robust feature. The influence of the North Atlantic Oscillation (NAO) on cyclone clustering displays a similar pattern for all tracking methods, with one maximum near Iceland and another between the Azores and Iberia. The differences in variance between methods are not related with different sensitivities to the NAO, which can account to over 50% of the clustering in some regions. We conclude that the general features of underdispersion and overdispersion of extratropical cyclones over the North Atlantic and Western Europe are robust to the choice of tracking method. The same is true for the influence of the NAO on cyclone dispersion.
The authors would like to thank Swiss Re for sponsoring the Intercomparison of Mid-Latitude Storm Diagnostics (IMILAST) project. T.E. and D.B.S. were supported by NERC project CREDIBLE. We thank the European Center of Medium Range Weather Forecast (ECMWF, www.ecmwf.int) for the ERA-Interim reanalysis, and the Climate Prediction Center from the National Oceanic and Atmospheric Administration for the NAO index (http://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/nao.shtml). We thank all the members of the IMILAST project (www.proclim.ch/imilast) for making the cyclone tracks available. We also thank Kevin Hodges for making the cyclone tracks available for this study and for discussions. The present manuscript profited from discussion with various members of the IMILAST project group. We are thankful to the two anonymous reviewers for their valuable comments and suggestions on a previous version of this manuscript.
Vol. 68, Art. No. 32204