Modelling detention performance of green roofs in cold climates

Abstract

Green roofs (GRs) have become a popular sustainable drainage system (SuDS) technology in urban areas. As many countries and regions experience political encouragement and substitution schemes in implementing GRs, there is a need for reliant models that can support designing purposes.

The stormwater management model s (SWMM) Low Impact Development Green Roof (LID-GR) control is used to simulate the hydrological detention performance of two GRs, GR1 and GR2, with different drainage properties located in Oslo, Norway. This study uses event-based data to replicate GR runoff. Accordingly, four event-models were calibrated using the Shuffled Complex Evolution algorithm with the Nash-Sutcliffe criteria (NSE) as the objective function. Eight events were used for model validation. In addition, sensitivity to the model parameters was analysed in relation to model input (precipitation characteristics and roof initial saturation levels). Simulation results revealed that SWMM s LID module can capture response of the GRs even though the adequacy varies among events.

Parameter sensitivity analysis exhibited significant correlation between conductivity slope and maximum precipitation intensity. During calibration two GR1 (0.55 and 0.72) and three GR2 (0.73, 0.88 and 0.51) event-models yielded NSE>0.5. However, only parameter sets of two GR2 event-models yielded NSE>0.5 when applied to the validation events.

The study shows potential of SWMM as a design tool if supplemented with a calibration algorithm. However, calibration is required and some adjustments to the LID-GR module should be made in the future. This concerns particularly the module s drainage layer.