One of the main challenges in urban hydrology is data scarcity, particularly for below-ground components, such as storm drain networks, at the regional or larger scales. To tackle this challenge, we devise a novel framework for estimating Urban Stormwater Networks (USNs) based on street network, topographic, and land use/land cover datasets and drawing on graph theory concepts. This framework can systematically derive USNs at various urban watersheds with different levels of drainage capacity. We demonstrate the framework's applicability by validating the derived USNs at eight urban watersheds located in the Baltimore, Houston, Los Angeles and Seattle metropolitan areas in the US where existing USN data is available. By using a spatial proximity metric we show that the framework can reproduce the real USN quite well, capturing over 70% of real stormdrain pipes in terms of total length. This framework relies on ubiquitous geospatial data. Therefore, it can be applied to the regional or continental scale. Furthermore, this framework facilitates employing the dual drainage urban system concept, i.e., street conveyance and storm drain network, for large-scale urban hydrologic modeling.