High-Resolution Parameterization of a Hydrological Model in Areas with Varying Urban Density

Simulation and evaluation of green infrastructure (GI) measures, such as green roofs or pervious pavers, necessitate high spatial and temporal resolutions of input data. The studied urban watersheds need to be delineated into small-scale subcatchments that represent single roofs, sidewalks, street segments, etc. While such a time-consuming approach is suitable for watersheds of the size of a few building blocks, it is not feasible for larger urban areas. Regionalizing results from high-resolution research catchments, which support detailed GI-simulation, to surrounding areas provides one avenue for assessing larger regions at the requested detail. Three research catchments located in Lahti, Finland, representing central (TP and AP) and sub-urban (KP) areas of the city and monitored from 2008 to 2010, were selected to develop high-resolution model parameterizations (using SWMM 5.0) in an attempt to simulate GI impacts for the entire Lahti urban area.

This presentation presents the development, calibration, and validation of a high-resolution hydrological model in the three research catchments and discusses key factors controlling the model performance. The developed models were calibrated using the multi-objective genetic optimization algorithm NSGAII (one-minute rainfall recording frequency) and validated against events with one-minute and ten-minute recording frequency of rainfall. Calibration parameters were individually selected for each catchment based on a sensitivity analysis.

The variation in model performance statistics for the three catchments indicates what the controlling factors for the performance of a high resolution model are. The significance of the distance of the rainfall measurement station from the study site is shown for modeling urban runoff at a high temporal resolution. Less impervious catchments (KP in this study) imply a less homogenous surface structure and involve larger uncertainties in catchment delineation and surface water routing. These uncertainties particularly concern the interface between constructed and natural areas and the interaction between areas directly connected to the drainage network and areas connected by overland flow.