Energy storage with virtual inertia and virtual droop control has attracted wide attention due to its improved frequency stability with high penetration of renewable energy sources. However, there are significant spatial differences in frequency response. The location and capacity of energy storage are urgent issues to be resolved to support frequency. This study addresses the minimum investment of hybrid energy storage systems for providing sufficient frequency support, including the power capacity, energy capacity, and location of energy storage. A frequency response model is developed taking into account the network structure and frequency spatial distribution characteristics. In addition, a numerical computation method is provided for determining the frequency dynamic indices and calculating the output power of energy storage. Based on a simplified frequency response model, an optimal hybrid energy storage configuration method is proposed to optimize the control parameters, location, and capacity to satisfy the frequency dynamic constraints. This configuration method can exploit the potential of energy storage with different rates in different frequency support stages. To address the nonconvex drawback of this configuration, a numerical calculation method is provided based on the explicit gradient of the frequency and energy storage indices to enhance the computational efficiency. Simulations of a two-area system and the south-east Australian system verify the effectiveness of the proposed hybrid energy storage configuration method.