Understanding the scaling behavior of flood peaks in relation to urban imperviousness and reservoir regulations is essential for effective flood management and mitigation. This study utilizes the MOSART-Urban model, a watershed-scale, semi-distributed hydrological tool specifically designed for the Conterminous United States (CONUS). The model explicitly accounts for Below-ground Urban Stormwater Networks (BUSNs) and reservoirs, enabling a comprehensive investigation of their impacts across selected Mid-Atlantic watersheds. In this framework, BUSNs and street networks are interconnected through street inlets, allowing water to flow both from streets to BUSNs and vice versa during overflow events. Recognizing that BUSN data is often unavailable to the public, we have derived BUSN representations using an innovative algorithm based on graph theory, combined with extensive datasets of street networks, topography, and land use/land cover. Operating at the Hydrologic Unit Code 12 (HUC12) watershed resolution, the model provides a robust spatial framework for urban hydrological analysis. Through a series of numerical experiments, the scaling behavior of flood peaks under various scenarios has been assessed. The analysis of changes associated with catchment size, using the HUC12 drainage network representation, reveals significant impacts on flood peak scaling behavior. The results indicate that both reservoir regulations and stormwater management are essential in reducing peak flows, leading to a complex relationship in the scaling behavior of flood peaks across various catchment sizes. This complexity emphasizes the need for flood risk management strategies to be adaptable and informed by scale.