Evaluating infiltration models for urban flood assessment: a case study of Pamplona, Spain

Hydrology science provides different methods to estimate infiltration: physically based, semi-empiric, and empiric models. Several researchers have compared the adequacy of these models to different conditions, such as soil types and land uses. Most of them were applied to agriculture and natural/forest lands. One of the main conclusions provided by such studies is that land cover is the most relevant factor, even more than soil texture. As a result, infiltration estimates in urban areas are subject to significant uncertainty. Some of the most employed models in infiltration evaluation are Green-Ampt (GA), Horton, Kostiakov and Philip. GA and Horton methods are particularly prevalent in applications involving urban areas. In addition, some studies showed that combining infiltration methods with flow redistribution models could improve the infiltration analysis performance.

In addition, in the last few years, 2D hydraulic models have incorporated improved functionalities to simulate urban floods generated by short and high-intensity storm events in small catchments. Such models usually include both GA and curve number (CN) infiltration equations for characterizing infiltration processes, obtaining runoff amounts and simulating flooding water depths, velocities and extents.

In this study, we assessed the hydrological response of urban basins in flood events in terms of the selected infiltration equation. We employed two 2D hydraulic models, HEC-RAS 2D and IBER, to analyze the hydrological response of an urban catchment in the city of Pamplona (Spain). We compared the results of two infiltration equations, GA and CN, considering different rain patterns. Firstly, we used twenty-one synthetic hyetographs ranged from 2-yr to 200-yr return period and three hyetograph peak positions (centered, left-skewed and right-skewed). Secondly, we also considered the precipitation fields with a time step of 10 minutes for two real storms. The results show that the GA model provides larger flooding extents than the CN model. In addition, the IBER model simulates larger flooding extents than the HEC-RAS 2D model. Differences appear to be more significant for smaller total rainfall depth storms.