Modelagem de telhados verdes extensivos modulares utilizando o storm water management model (SWMM)

Abstract

Green roofs have the ability to retain rainwater partially or entirely, thus reducing peak flows and surface runoff. The retention is provided by the set of green roof layers, including vegetation, substrate, and drainage elements. In this study, a detailed analysis of precipitation, runoff, and monitored moisture data was conducted in relation to green roofs. The LID Bioretention Cell hydrological module of the Storm Water Management Model (SWMM) was used to calibrate the hydrological parameters of five green roofs, which were identical except for properties related to porosity, field capacity, and wilting point. The characteristics of the substrate layer materials were previously determined and used for model parameterization, with only the parameters related to water flow in the substrate (K, Ψ, and the Gradient of the conductivity curve) being calibrated, and, if necessary, the initial moisture of the substrates was adjusted. Thirty-one precipitation events that generated runoff were selected for calibration, along with ten events for model validation. The model's efficiency was evaluated by comparing it with observed data using Volume Error (EV) and Nash-Sutcliffe efficiency (NSE). The results showed that the runoff volume is directly correlated with the precipitation volume, with SC4 presenting the highest average runoff volume. However, the Kruskal-Wallis statistical test indicated no significant difference in monitored runoff volume among the different substrates. During the model analysis, significant variability in water flow parameters between analyzed events was observed. Regarding sensitivity, the parameter K initially showed greater sensitivity compared to the others, with the highest sensitivity in a range of 2 to 120 mm/h. Additionally, very low K values resulted in surface outflow in the model. After calibrating the parameters and adjusting the monitored moisture of the green roofs, the model achieved an excellent reproduction of the runoff volumes from the substrates, with an average NSE of 0.98 and an average EV of 3.51%. During validation, it was also necessary to adjust the monitored moisture to improve the indicators, highlighting the importance of moisture for the model's accuracy, evidenced by an NSE of 1 and an EV of 0.0. During validation, the indicators showed inferior results in low precipitation events. It is recommended that future studies include the evapotranspiration component and continuous simulation for a better evaluation of the hydrological performance of green roofs.