Abstract: With the continuous development of new power systems featuring “dual high” characteristics and the increasing scarcity of dispatchable thermal power resources, the issue of reserve allocation caused by the forecasting errors of renewable energy has gradually become a focal point. Meanwhile, the enriched adjustable resources within the source-load-storage system provide a wider range of approaches for reserve provision. To address the challenge of coordinated reserve optimization at different time scales in the source-load-storage system, we first utilize the complementary ensemble empirical mode decomposition (CEEMD) method to decompose the forecast errors of net load at multiple time scales. To balance the trade-off between risk and reserve allocation, we employ the conditional value-at-risk (CVaR) to characterize the system risk and propose a hierarchical reserve coordination optimization method that considers the risk of wind curtailment and load shedding. Based on this, considering the response characteristics of multi-type source-load-storage systems, we establish a coordinated and complementary optimization scheduling model considering hierarchical reserve allocation. The objectives of this model are to minimize the reserve allocation costs, maximize generation benefits, and mitigate potential risks. Case study results demonstrate that the proposed hierarchical reserve optimization method effectively strikes a balance between system risk and reserve allocation, thus improving system reliability and achieving reliable power supply and enhanced renewable energy integration.