Characterization and classification of offshore wind profiles via a shape-based cluster analysis framework
Zhenru Shu, Pak Wai Chan, Xuhui He · Physics of Fluids · 2025
Understanding vertical wind profile variability over marine environments is fundamental to advancing offshore wind resource assessment and the aerodynamic design of wind turbines. This study proposes a shape-based clustering framework to characterize and classify offshore wind profiles using long-term Doppler Lidar observations collected at an offshore platform near Hong Kong. The analysis categorizes wind profiles according to their morphological features across a range of wind speeds, seasonal conditions, and atmospheric stability regimes. In general, the results reveal two dominant profile classes. The primary class displays a near-monotonic increase in wind speed with height, consistent with shear-driven boundary layers. The secondary class, by contrast, exhibits more complex structures, including monotonic decreases, low-level wind speed maxima, and multi-layered inflections. These findings underscore the limitations of conventional extrapolation models, such as power-law and logarithmic profiles, which assume stationary, homogeneous conditions and fail to capture the dynamic variability of the marine atmospheric boundary layer. In contrast, the data-driven approach proposed in this study retains physical interpretability, improves regime detection, and supports stability-aware modeling of offshore wind fields. Overall, the study highlights the need for incorporating non-ideal, shape-dependent classifications into wind engineering practice and contributes toward a more physically representative understanding of offshore wind dynamics.