UFIM: A Community-Scale Urban Flood Intelligence Framework for Climate-Driven Extreme Rainfall

Urban flooding is increasing worldwide due to the combined effects of climate change driven extreme precipitation and rapid urbanization. Flood impacts within cities exhibit strong spatial heterogeneity, yet most existing urban flood models remain highly complex and computationally demanding, limiting their applicability for targeted risk assessment and early warning in urban governance. In practice, decision-makers increasingly require refined simulations focusing on high-risk and high-impact scenarios, such as underpasses, residential communities, underground garages, metro systems, vulnerable buildings, and urban reservoirs.

To address this gap, we present the Urban Flood Intelligent Model (UFIM), a community-scale urban flood modelling software specifically designed for fine-scale flood simulation and early warning in critical urban environments (https://www.antmap.net/web/ufim-en/). UFIM integrates high-resolution topographic data with a dynamic real-time 1D-2D coupled hydrodynamic framework, explicitly accounting for drainage network and surface interactions and backflow processes. Flexible coupling strategies allow both loosely and tightly coupled configurations, enabling realistic representation of complex urban drainage and surface flow dynamics while maintaining computational efficiency. UFIM supports heterogeneous rainfall inputs, multiple infiltration schemes, diverse outlet boundary conditions, and grid-based surface roughness parameterization. The model is implemented with a user-oriented interface, predefined parameter sets, and advanced visualization tools, lowering the technical barrier for operational use. In addition, UFIM offers cross-platform compatibility (Windows/Linux), rapid deployment via Docker, seamless GIS integration, and AI-assisted diagnostics for model performance evaluation and optimization.

UFIM has been extensively tested across multiple urban scenarios, including residential communities, functional zones, and complex mixed-use areas, under both observed extreme rainfall events and design storms with different return periods. Validation results demonstrate stable long-term simulations and consistently high predictive performance, with inundation detection accuracies exceeding 85% across tested applications.

These results indicate that UFIM provides a robust and practical tool for community-scale flood risk assessment, scenario-based early warning, and resilient urban planning, bridging the gap between advanced hydrodynamic modelling and real-world urban flood governance needs.