Spatiotemporal analysis and nonlinear statistics of hourly wind speed variability over tropical Cuba

Wind energy is a rapidly expanding renewable resource that plays a pivotal role in the global transition from fossil fuels to sustainable and low-carbon energy systems. However, its potential application requires understanding its intricate dynamics.. The objectives of the present work are to i) evaluate empirically the hypothesis of nonlinearity, determinism and chaotic behavior of wind speed time series and ii) investigate statistical relationships between different nonlinear parameters. Hourly wind speed time series consisting of 87648 data points were collected from eleven meteorological stations in Granma Province, Cuba, from January/2014 to December/2023. We calculated the Lévy-stable index (α) for the original and pre-whitened time series. The nonlinearity parameter, determinism test (Λ), global Lyapunov exponent (m), Hurst exponent (H) and multiscale entropy (MSE) were also computed for each series series using, respectively, the time reversibility, delay, detrended fluctuation analysis and multiscale entropy methods. These parameters were compared with those computed from 40 surrogate time series generated at each station. We found that the original wind speed time series could be classified as Gaussian noises with Lévy index α=2.00, lower deterministic component ∧=0.318±0.023, time reversibility Z-score < 2.021 and approximately constant MSE. The decorrelated data fitted sub-Gaussian distributions with the α exponent in the range 1.227 α 1.704. A multiple regression analysis found a significant empirical link between H, m, the exponent of the MSE function () and Λ with the correlation coefficient of r=0.972. These findings could be useful for the siting, operation and optimization of wind energy-based technologies.