Dynamic analysis of meteorological time series in Hong Kong: A nonlinear perspective

Many meteorological systems are chaotic in nature, which inevitably limits its predictability. Accurate prediction of meteorological variables depends primarily on properly diagnosing the complex underlying dynamics. Nonlinear dynamic analysis has shown to be particularly useful for such purpose. In this study, the concept of recurrence analysis was extended to investigate and characterize the underlying dynamics of meteorological time series (i.e., wind speed, temperature, pressure and relative humidity) based on daily observation in Hong Kong during a period from 1998 to 2018. The existence of chaos was clearly identified based on the phase space reconstruction diagram and the recurrence plot. It was shown that the underlying dynamics associated with wind speed appears to indicate higher level of complexity as compared to those of pressure and temperature, which, in consequence, may lead to more irregular time-dependent behaviour and lower predictability. Moreover, season-to-seasonal variability in the dynamics of meteorological time series was evident, particularly for wind speed and temperature. Overall, this study shows that the recurrence analysis can be well applied as a useful diagnostic tool to investigate the dynamics of meteorological systems, which is expected to provide a new avenue regarding the modelling and prediction of the behaviour of meteorological parameters.