Sensitivity of urban flood simulations to stormwater infrastructure and soil infiltration

Abstract

Flooding from intense rainfall is a major hazard in many urban areas. One of the major challenges in urban flood simulations is lack of data about the location and properties of stormwater infrastructure and land cover. In this paper, we investigate the sensitivity of urban flood simulations to inputs of stormwater infrastructure and soil characteristics. We use a 2D hydrodynamic model (MIKE URBAN) to simulate flood events on the University of Alabama (UA) campus in Tuscaloosa U.S.A. Infiltration rate, soil moisture, and soil texture were measured in the field. Soil texture was found to be homogeneous across campus (sandy loam) but with a high degree of spatial and temporal variation in infiltration rate and soil moisture. Comparison between different storm event return periods (10, 25, 50, and 100 years) shows that the same flooding hotspots are persistent but with considerable variation in water depth and flood extent. To investigate the sensitivity of the simulations to stormwater infrastructure, simulations without the stormwater infrastructure input were conducted. The results show that stormwater infrastructure decreases flooding volume of RP 10, 25, 50, and 100 by factors of 20, 14, 12, and 8, respectively. This shows that urban flood simulation is highly sensitive to the inclusion of stormwater infrastructure, though with decreasing relative impact for larger events. To investigate the sensitivity of the simulations to soil characteristics, four land-cover simulations (actual, uniform, entirely pervious, entirely impervious) were compared. The results show flood simulation predictions are sensitive to both the value and spatial explicitness of the soil input data. We discuss challenges in urban flood simulation and their potential solutions in the context of emerging frameworks for national and global hyper-resolution flood forecasting and analysis.