Researchers from Nazarbayev University and collaborators have developed and validated a Hybrid Adaptive-Bandwidth Kernel Density Estimation method, or HAKDE, to model wind speed probability distributions more accurately. Using hourly NASA POWER 2023 data from 16 cities in Kazakhstan, the study found that HAKDE performed robustly across diverse wind regimes and supported estimates of expected average power, capacity factor, and annual energy production for wind energy planning.

Key findings

• The study proposes HAKDE, a hybrid approach that combines curvature-based and k-nearest-neighbor bandwidths to better capture wind speed distribution peaks while stabilizing the tails.
• The method was validated using NASA POWER hourly wind data for 16 cities across Kazakhstan.
• HAKDE was not rejected by the reported goodness-of-fit tests in all 16 cities.
• In log-likelihood comparison, HAKDE achieved the highest mean log-likelihood in 9 of the 16 cities.
• The model was integrated with the power curve of the Vestas V150/4200 wind turbine to estimate expected average power, capacity factor, and annual energy production for wind resource assessment and planning.

Why it matters
Accurate wind speed modeling is essential because small changes in wind speed can strongly affect electricity output. By improving how complex wind regimes are described, this research provides a more reliable tool for wind resource assessment and can support planning and investment decisions related to wind energy development in Kazakhstan.