Hurricane season complexity: The case of North-Atlantic tropical cyclones

The forecast of tropical cyclone (TC) seasons remains an elusive subject of study. In order to characterize their complexity, the energy of North Atlantic (NA) TC seasons is studied in this research work. The time evolution of the accumulated cyclone energy probability distribution is analyzed using mobile windows to calculate the statistical parameters: mean, standard deviation, kurtosis and skewness. A nonstationary distribution with clear excess of extreme energetic seasons when compared to a Gaussian distribution was found for the most recent years. The data also show a clear correlation between variability and mean, i.e. more energetic periods are more variable. Finally, the dynamics is analyzed using several techniques such as the lag plot, Hurst exponent and fractal dimension. A stochastic but persistent behavior was found. It is finally concluded that the NA TC season energy seems to be increasing both in magnitude and variability. This helps to elucidate the possible effect of global warming into increasing hurricane hazards.