The use of a calculus-based cyclone identification method for generating storm statistics

Abstract

Maps of 12 hr sea-level pressure (SLP) from the former National Meteotrological Center (NMC) and 24 hr SLP maps from the European Centre for Medium-range Weather Forecasts (ECMWF) 40 yr re-analysis (ERA40) were used to identify extratropical cyclones in the North Atlantic region. A calculus-based cyclone identification (CCI) method is introduced and evaluated, where a multiple regression against a truncated series of sinusoids was used to obtain a Fourier approximation of the north—south and east—west SLP profiles, providing a basis for analytical expressions of the derivatives. Local SLP minima were found from the zero-crossing points of the first-order derivatives for the SLP gradients where the second-order derivatives were greater than zero.

Evaluation of cyclone counts indicates a good correspondence with storm track maps and independent monthly large-scale SLP anomalies. The results derived from ERA40 also revealed that the central storm pressure sometimes could be extremely deep in the re-analysis product, and it is not clear whether such outliers are truly representative of the actual events. The position and the depth of the cyclones were subjects for a study of long-term trends in cyclone number for various regions around the North Atlantic. Noting that the re-analyses may contain time-dependent biases due to changes in the observing practises, a tentative positive linear trend, statistically significant at the 10% level, was found in the number of intense storms over the Nordic countries over the period 1955–1994 in both the NMC and the ERA40 data. However, there was no significant trend in the western parts of the North Atlantic where trend analysis derived from NMC and ERA40 yielded different results. The choice of data set had a stronger influence on the results than choices such as the number of harmonics to include or spatial resolution of interpolation.