Evaluating the stormwater reduction of a green roof under different rainfall events and antecedent water contents with a modified hydrological model

This study simplifies the recession process, includes the hourly evapotranspiration, and modifies the surface layer of the green-roof hydrological model based on the Green-Ampt (G-A) method and Curve Number (CN) method. The existing G-A and CN methods are usually applied for event-total runoff. The results show that the proposed model improves the predicting accuracy of the outflow, and the model performs well for rainfall events with different intensities, duration, and hyetograph. the CN method performs as prominent as the G-A method but is more efficient. Based on observation, the retained volume and peak reduction are close to 50 % on average with a 1-to-2-hour peak delay. The results of the three-year simulation show that the retained volume and peak reduction reach 50 % for 75 % of the events when the antecedent water content is 30 %. Almost all rainfall is retained for events with rainfall amounts smaller than 20 mm when the antecedent water content is smaller than 30 %. On the other hand, 50 % (60 %) rainfall (peak) is retained (reduced) for events larger than 20 mm when the antecedent water content is smaller than 20 %. With the model verified using various rainfall events, the study generalizes an equation to estimate the rainwater retained by green roofs. Moreover, the stormwater-detention performance of a green roof in general is concluded quantitatively under different antecedent water content to assist the decision-making of irrigation concerning the goal of stormwater detention.