Typhoon complexity: Northwest Pacific tropical cyclone season complex systems analysis

Abstract

The complex nature of tropical cyclones (TCs) has been recognized in a vast literature yet only few works perform complex systems diagnostics to understand their dynamics. This is especially important in order to study the effects of global warming on TC hazards. Here, such analysis is performed from a data-driven perspective using statistical and nonlinear dynamics diagnostics to the annual Accumulated Cyclonic Energy (ACE) data over the most active basin, the Northwest Pacific, from the years 1950 to 2021. The best quality data period, from 1984 to 2021, is also considered for a separate analysis in order to test the possible differences due to the data acquisition process. The following results are obtained: (i) The use of mobile windows shows a lack of trend. (ii) The closeness to a normal probability distribution indicates unpredictability, as confirmed by the return map and the autocorrelation function. As an explanation for such unpredictability, the ordering and ranking analyses reveal the presence of several processes governing the dynamics with fractal forcing dominating the larger ACE values. (iii) The Hurst exponent analysis shows a slight persistence for less than 3 years and a very slight antipersistence for longer periods of time suggesting the presence of negative feedback. (iv) Finally, the TC dynamical system dimensionality is studied. A discontinuity is inferred from a Katz fractal dimension value of 2.8, suggesting the system dynamics to be embedded by at least three independent variables, in agreement with statistical models for the TC season.