Hydrological performance of green roofs

Due to an increase in impermeable hard surfaces, urbanization has led to the deterioration of urban watercourses and increased the quantity of stormwater runoff. It may be argued that the current norm of impermeable roofs represents a wasted opportunity. Green roofs have the potential to replace some of the hydrological characteristics of natural catchments that are normally lost as a consequence of urbanization and the removal of vegetation. The overall aim of this study was to develop a generic green roof rainfall runoff response model capable of predicting the temporal variations within any configuration of green roof in response to an arbitrary rainfall input. It was recognized that the preliminary investigations has led to the identification of a subset of processes/parameters for a green roof which warranted more detailed investigation. In this case the substrate moisture holding capacity and the losses due to evapotranspiration were identified as key controlling variables to be identified. To simulate the function of stormwater drainage, a direct observation of the system's behaviour is required. Hence, an established `typical' small scale green roof (1.0 in x 3.0 m) on the roof of Sheffield University has been monitored with the intention to relate both retention and detention with fundamental, measurable, physical properties of the system. A continuous long time-series of data, in the period of 29 months, from the test rig was analysed and interpreted. Laboratory analyses on physical properties and evaporation of the substrates were undertaken and relationships between measureable physical properties and model parameter values were identified. The empirical (requiring site-specific calibration using monitored data) conceptual model now has been developed into a physically-based model. Although the model still needs to be refined, independent physically-based methods have been identified for defining two key parameters (evapotranspiration (ET) and the maximum moisture-holding capacity (WC,,, a,, )). ET can be estimated using a modified form of Thornthwaite's equation, and WC.., may be determined by physical laboratory assessment of the substrate. The proposed hydrological model has been shown to reproduce monitored data, both during a storm event, and over a longer continuous simulation period