Runoff modelling using QSWAT+ to assess the contribution of piped network stormflow in the Haaganpuro semi-urban catchment

Water-related problems have always represented a tough challenge for human beings throughout history. Their wide collection of issues, ranging from water quality degradation to stormwater management in urban catchments, reflects the controversial power of this fundamental resource, vital for human life and in some cases destructive if not well controlled. Especially, in a context where the climate fluctuations and the development of urban areas are increasing their impacts on both streamflow and water quality, water management solutions are acquiring much more importance day by day. In particular, the rising in the frequency of extreme rainfall events justifies the need for an improvement in the actual stormwater management systems of the cities. In this perspective, hydrological modelling assumes a relevant role since it constitutes a powerful instrument in understanding how rivers and channels act in a watershed and in simulating runoff processes, a crucial part for planning stormwater management solutions. This study focuses on the implementation of the Soil and Water Assessment Tool (SWAT) which is a continuous time, semi-distributed small watershed to river basin-scale model to simulate streamflow from the Haaganpuro catchment, an urban brook located in the city of Helsinki (Finland), in order to separate how large a share of runoff originates from areas drained with a stormwater piping system as compared to more natural park areas. The catchment delineation was performed with the support of the QGIS software, while the input data (including weather data) were uploaded through the SWAT+ Editor to allow flow simulation. The model was calibrated for the months between May-August 2017 and validated for September-November of the same year using observed daily streamflow imported on the SWAT+ Toolbox, a software mainly implemented to perform sensitivity analysis, calibration and more. The most sensitive parameters were identified through a sensitivity analysis which showed that the ones related to HRUs (Hydrologic Response Units) and soil were the most significant, calibrated manually to establish their variation range and then automatically to find their optimal value. Simulated daily surface runoff was estimated applying the Curve Number method while model performance was assessed through the Nash-Sutcliffe Efficiency (NSE) and the percent bias (PBIAS) which respectively give insights on how the model is able to resemble measured values and how much they are under or overestimated. At first, the calibration focused only on a specific section of the Haanganpuro catchment named Lansi-Pakila, an urban neighbourhood in the North-East of the basin whose calibrated parameters are assumed to be representative of all the other built-up areas. Then, the analysis was shifted to the more natural zones with the purpose of identifying some reasonable calibration values to be compared with the urban ones. In this way, two sets of parameters were derived, one for urban areas and one for forested/park areas. These sets were then used to modify the input files of the SWAT+ Editor in order to run the final runoff simulation for the decade 2012-2022 (the study period of this thesis) and split the contribution of the two different domains to streamflow. The results were finally compared to other studies findings, understanding in this way their reasonability and enhancing the relevance of the SWAT model for water resources management and controls.