Author(s): Virginia Stovin; Simon De-Ville; Zhangjie Peng

Linked Author(s): Virginia Stovin, Simon De-Ville

Keywords: Green roof, Retention, Detention, Performance metrics, Climate

Abstract: Green roofs represent one of the hydrologically simplest Sustainable Drainage Systems (SuDS) components. Nonetheless, significant efforts have been invested over the last 10 years or more to collect data on their hydrological performance and to develop/validate modelling tools to accurately represent them in drainage design tools. Observational data sets, including the six year rainfall-runoff record obtained at Sheffield for nine different test bed configurations, provide valuable information at multiple levels. Even with quite short observational periods, it is possible to obtain approximate data on ‘typical’ performance and to run side-by-side comparisons of alternative configurations. However, these observations cannot be used to inform design, where extreme (non-routine) events and statistically representative time-series are required. Nor are the findings transferable to locations with different climatic characteristics. Robust modelling tools are needed for these engineering design and stormwater management objectives, and observational data sets have a key role to play in informing the development of appropriate modelling tools. The hydrological functioning of a green roof can be separated into the retention (or initial losses) component and the detention (runoff delay) component. In a green roof the recovery of retention capacity is driven by the process of evapotranspiration (ET), and it is now widely accepted that appropriate modelling of ET will lead to good estimates of retention losses. The models need to incorporate a Soil Moisture Extraction Function to account for the reduction in ET that occurs when moisture is restricted, and crop factors may be required to characterise different vegetation. Application of this retention model allows us to explore how the green roof’s hydrological performance varies depending on local climate. We will present a contour map of the expected annual retention across the UK as an illustrative example. Compared with retention, modelling of runoff detention is more complex and less well advanced. Approaches range from black-box models (reservoir routing or machine learning based) that are empirically fitted to observational data to complex physically-based models that separate the green roof into its multiple layers. Whilst the black-box approaches lack transferability, the more physically-based models may not necessarily offer a better solution because of uncertainties (and natural variability) in key input parameters. A brief overview of these alternative approaches will be provided in the presentation. There is strong argument for using continuous simulation rather than design storms to design and model SuDS components; otherwise important information about drying process and functionality in response to day-to-day events (important for river quality) is lost. In the final part of the presentation we will share our initial thoughts on which metrics should be derived from the continuous simulation results to properly characterise SuDS performance in response to both routine and extreme rainfall events.

DOI: https://doi.org/10.3850/IAHR-39WC252171192022SS1301

Year: 2022