The variability and intermittence of wind power will cause the large imbalance power that demands more expensive ancillary service. Energy storage, fast response but costly, is a viable solution to suppress the fluctuation of wind power. However, the determination of energy storage is a great challenge given the load demand and wind power uncertainties This paper proposes to use discrete Fourier transform (DFT) to decompose the imbalance power into different time-varying components, i.e., intra-week, intra-day, intra-hour and real-time. Therefore, the imbalance power to be compensated by energy storage can be quantified. By compensating the fast-changing imbalance power (the slowly changing power is provided by the conventional generators), energy storage can be optimized to accommodate integration of high penetration wind power. The simulation results on the 2030 Western Electricity Coordinating Council (WECC) system demonstrate effectiveness and efficiency of this approach.