Green Urban Drainage Infrastructure : Hydrology and Modelling of Grass Swales

Abstract: The management of urban runoff has evolved along with the advancement of understanding of runoff environmental impacts. Besides the impacts on water quality in the receiving waters, the impacts on the urban hydrologic regime include reduced infiltration by the sealing of pervious land, reduced evapotranspiration by removal of vegetation, and the resulting increase of stormwater runoff peaks and volumes causing flooding, and ultimately degradation of receiving waters. In such considerations, urban stormwater management benefits from the implementation of Green Infrastructure which includes decentralized vegetative controls that capture and infiltrates rain where it falls and thus reduces and improves stormwater runoff. An example of small scale elements of Green Infrastructure are traditional grass swales. Through shallow depressions with mild side slopes grass swales collect and infiltrate stormwater from parking lots and roads, while runoff flows are attenuated and further conveyed depending on the hydraulic loading. Grass swales usually operate reliably and their maintenance needs are well understood. Their hydrological performance is, beside their dimensions and the contributing area, determined mainly by hydraulic and soil-related hydrological parameters that change with the intensity of the storm. Yet, because swales discharge to downstream drainage elements, either to the conventional sewer system or to other stormwater management facilities, the knowledge of the underlying inter-related processes and influential factors that govern the hydraulic and hydrological performance of grass swales is required.Against this background, this thesis is devoted to such questions as (i) what are the differences in the hydraulic and hydrological performance of the studied swales, (ii) how do soil characteristics, including the antecedent soil moisture, influence the swale water balance for various hydraulic loadings; and (iii) how can the related hydrological processes be simulated in high-resolution and reliably predicted using a grid-based, distributed model. For this purpose, full-scale studies were performed in three 30-m grass swale sections in Luleå, Northern Sweden, by collecting hydraulic and hydrological data based on routine storm events mimicking block-rainfall storm events of 2 months and 3 years recurrence. The resulting runoff and soil moisture data were used to calculate the swale water balance, to derive event hydrographs and to obtain calibration and validation data for model simulations. The experimental results showed that the relative swale flow volume reduction decreased with an increasing soil moisture and indicated the transition in dominating swale functions: at low initial SWC, runoff was highly attenuated (up to 74%), but for high SWC, the conveyance function dominated (with attenuation as low as 17%). Runoff flow peaks were reduced, proportionally to the volume reductions. Swale outflow hydrograph lag times varied between 5 to 15 minutes and decreased with increasing soil moisture. The swale wetness affected runoff formation, attenuation and subsequent outlet discharge and, for the short-duration events tested, only the top soil layer contributed to these findings. In the three swales tested, soils, initial soil water content, saturated hydraulic conductivity and topography varied spatially significantly. Double-ring infiltrometer measurements resulted in values of 1.78, 4.04 and 9.41 cm/hr (n=9) in the three swales tested and deviated from estimates from averages of spatially integrated infiltration rates. However, with regard to spatial variability, only the topography, described as irregularities in the swale bottom slopes affected the swale runoff dissipation and conveyance in the early phase of the events. Together with estimates of the water stored in the top soil layer, 4-32% of runoff volumes from the mimicked 2-month storm were temporarily stored. The distributed model Mike SHE was found capable of simulating swale drainage processes, when properly calibrated. Close agreement (NSE>0.8) was found not only for the measured and simulated swale outlet hydrographs, but also for the changes of the soil moisture in the top soil layer, which shows rapid increase up to the saturated soil water content, but minor or no progression in depths of 0.2 m. The model output was little sensitive to the initial soil water content, especially for low inflow which resulted in larger residuals in simulated runoff peak flows and volumes. As in field measurements, spatial variability of the initial soil water content had no effect on the swale outflow, but the accuracy of the topographical representation. The thesis findings include several implications regarding effects of the assessed parameters in the application of the model for swale flow simulation and eventually the design of grass swales.