Assessment of wind energy potential using hybrid adaptive bandwidth kernel density technique: A case study of major cities in Kazakhstan

Abstract

To accurately assess the wind energy potential in Kazakhstan, this study proposes and validates an advanced wind speed distribution estimation method, Hybrid Adaptive-Bandwidth Kernel Density Estimation (HAKDE), for modeling wind speed probability distributions (WSPDs). This method is designed to overcome the limitations of traditional models in describing complex wind characteristics. HAKDE is a hybrid model that combines a curvature-based technique to represent the peak with k-nearest neighbors (k-NN) to stabilize the distribution's tail. This combination gives HAKDE high flexibility while ensuring a balance between detail and smoothness. Based on hourly wind data from NASA POWER across 16 cities, the proposed model is implemented and evaluated using goodness-of-fit tests (CvM, K–S, A–D, χ2) and a likelihood-based comparison. The results show that HAKDE is not rejected by any of these tests in all 16 cities. According to the log-likelihood comparison, HAKDE attains the highest mean log-likelihood in 9 of the 16 cities, indicating robust and consistent performance across diverse wind regimes. In terms of practical applications, the HAKDE distribution is integrated with the power curve of the Vestas V150/4200 wind turbine to calculate more accurate estimates of expected average power (EAP), capacity factor (CF), and annual energy production (AEP) for assessment and planning. As a result, HAKDE demonstrates flexible, reliable performance as a wind resource assessment tool, playing an important role in supporting planning and investment decisions in the wind energy field. © 2025 The Authors