Developing a framework for urban flood modeling in Data-poor regions

Abstract

Forecasting floods is an effective non-engineering measure for flood control. However, for urbanized regions, the lack of drainage data and complex underlying surface conditions make flood forecasting extremely difficult. This paper proposes a framework for urban flood modeling in data-poor regions to address these issues. The drainage capacity of the pipe system was represented by the constant infiltration rate, which was calibrated by coupling satellite soil moisture product and measured flood data to better reflect real drainage conditions. Subsequently, given the complex underlying surface conditions, this framework divides the land use/cover into several dominant urban land use types, which were assigned unique runoff generation and flow confluence parameters in urbanized regions. The framework also takes the impact of dense buildings on water flow into consideration. A building treatment method for structured grids was proposed, which can consider the anisotropy of buildings and scale of the grid. The results showed that it is feasible to estimate drainage capacity using the constant infiltration approach, and the process of soil infiltration is crucial for urban flood forecasting. Compared with the highest resolution publicly available land use/cover dataset, the urban land use types extracted from high-resolution remote sensing images can reveal the runoff generation and flow confluence mechanism more accurately in urbanized regions. This study verified the necessity of considering the impact of buildings in urban flood modeling. The building treatment method of this framework can identify synergistic effects of local terrain and building boundaries on water flow. The results provide a reference for effective flood forecasting in urban regions with poor data.