Effect of data and model resolution on urban flood modeling

Abstract:

Hydrologic and hydrodynamic models are useful tools to study urban hydrology and predict urban flooding. A key factor to increase accuracy of these models is the availability of high-resolution terrain and infrastructural data that allow characterizing heterogeneity of urban catchments. Unfortunately, cities do not often have or can release this infrastructure information; and terrain data products such as Light Detection and Ranging (LiDAR) are sporadically available at different spatial resolution across the cities. In this study, we quantify how the accuracy of urban hydrologic-hydrodynamic models varies as a function of data availability and resolution. For this aim, we apply the USEPA’s Storm Water Management Model (SWMM) in an urban catchment in the city of Phoenix, AZ where we have collected detailed infrastructure data, high resolution LiDAR data, and remotely sensed water depths during recent flood events. We test different model configurations by (i) varying the level of terrain aggregation (from 0.3 m to 10 m); (ii) assuming different levels of availability of stormwater infrastructure data (e.g., complete, near complete and missing - components and attributes); and (iii) using two routing methods with different complexity and data requirements. To evaluate the different cases, we compare simulated flood volumes, maximum water depths at flooded locations, and flow rate at the outlet during a set of events when flooding was reported by the city. Our work expands previous studies that analyzed the influence of terrain information on urban hydrologic models and gives cities and county agencies insights on how the availability of different datasets could dramatically improve urban flood forecasting.