In power systems with a high penetration of renewable energy, the strong volatility of renewable power output significantly increases the demand for system frequency regulation capacity. To address this challenge, this paper proposes a data-driven statistical modeling method for characterizing renewable power fluctuations, along with a quantitative evaluation framework for frequency regulation capacity requirements considering renewable variability. Firstly, based on historical operation data, the output fluctuation characteristics of wind farms or photovoltaic (PV) plants are analyzed across different seasons and intra-day periods. A two-dimensional “month-hour” statistical model for the power fluctuations of individual renewable plants is established. Subsequently, a method for evaluating the total renewable power fluctuation level at the power grid scale is proposed, which effectively accounts for the spatiotemporal complementary effects of multiple renewable sites. By integrating load forecast results, the proposed approach enables quantitative estimation of frequency regulation capacity requirements considering the combined volatility of both generation and demand sides. Finally, a real-world case study of a provincial power grid demonstrates the effectiveness and practicality of the proposed method in renewable fluctuation characterization and frequency regulation capacity assessment.