Data-driven modeling of synthetic wind speed signals with adaptive temporal resolution

As wind energy systems continue to expand in size and complexity, ensuring their structural reliability under stochastic wind loading has become increasingly critical. Accurate wind speed modeling across multiple temporal scales is essential for reliability assessment, extreme load estimation, and risk-based design. However, a central challenge in data-driven wind speed and storm hazard analysis is the limited temporal resolution of available wind measurements. In practice, wind speed data are often provided as 10 min or hourly averages. While such coarse discretizations are sufficient for estimating aggregate power production, they fail to capture the high-frequency variability that governs extreme loading, structural response, and short-term system dynamics. This paper introduces a novel signal-processing based framework for synthesizing high-resolution wind speed time series from coarse measurements while maintaining statistical and spectral consistency. The methodology operates in the frequency domain, where spectral components are extrapolated and modeled to reflect physically plausible high-frequency behavior while also enforcing a strict energy conservation and preserving extreme-value characteristics. The resulting synthetic wind speed exhibits essential statistical characteristics, including mean values, variance, maximum amplitudes, and distributional properties, this enables high-fidelity stochastic simulations at resolutions suitable for advanced engineering and environmental analyses. The results indicate that the proposed methodology is capable of generating high-resolution, data-informed wind speed realizations with consistent statistical and spectral behavior. The framework provides a robust and computationally efficient tool for wind modeling.