Objective  In highly urbanized areas, roof greening is a supplement to ground greening. Although the benefits of roof greening in rainwater detention and retention can play an important role in improving water security and realizing sustainable developments for cities, there is rare convincing measurement and evaluation on such benefits from actual roof greening products under various climate features and roof conditions in different areas. Concentrated rainfalls and frequent storms during the plum rain season are the main causes of floods in Shanghai. Being light-weight and low-maintenance, the Sedum lineare roof greening product with a substrate depth of 8 cm is well applicable to a majority of existing buildings in Shanghai. In order to evaluate the product’s actual benefits of rainwater detention and retention, this research measures its rainwater detention and retention effects on typical roofs in response to natural rainfalls throughout the plum rain season.

Methods  Taking a 0° roof and a 26.6° (1∶2) pitched roof as typical roofs in Shanghai, the research installs six modules of the aforesaid product on the roof of Wenyuan Building, College of Architecture and Urban Planning (CAUP), Tongji University, and sets up seven WatchDog 2000 weather stations from the Spectrum Technologies, Inc. to continuously record natural rainfalls and water discharge from these modules during the plum rain season from June to July 2019. Referring to the measured soil moisture data obtained by a ZD-06 soil tester from the ZD Instrument Ltd. during the same period, the research analyzes the detention and retention effectiveness of the modules in groups with respect to both the flat roof and the pitched roof, and both the humid substrate and the semi-humid substrate, respectively for the entire plum rain season and for typical rainfall events. Rainwater retention rate is the index to value the modules’ runoff reduction capability for the entire plum rain season, while rainwater retention rate, discharge delay and peak delay are the indices to value the modules’ runoff reduction, peak flow cutting and discharge delay capabilities for typical rainfall events. Six hours is determined as the minimum rainfall interval to distinguish a single rainfall event, and several pairs of single rainfall events with the same rainfall pattern but different previous precipitation and soil moisture conditions are selected for comparing the modules’ performances.

Results  The plum rain season in 2019 is among the typical ones in Shanghai because of its relatively long period and abundant precipitation. Throughout the whole season, the Sedum lineare modules of the test groups of flat and pitched roof respectively retain 38.9% and 30.6% rainfall volume on average. There is no statistical difference in rainfall retention rate between the two test groups, but the Cohen’s d value still indicates a significant difference. Compared to the modules of the pitched roof group, those of the flat roof group retain an additional precipitation of 29.4 mm on average, which makes for a higher retention rate of 8.3%. The weather station for recording the natural rainfalls records 16 rainfall events during the whole season, including 10 light rains, 3 heavy rains and 3 rainstorms. Among these rainfall events, totally 6 ones (i.e., 2 rainfall events with similar precipitation characteristics but obvious differences in previous precipitation and substrate moisture for each rainfall pattern respectively) are selected and classified into the test groups of humid substrate and semi-humid substrate. The rainwater retention rates for the 6 selected rainfall events vary significantly, i.e., from 11.6% in terms of the test group with humid substrate on the pitched roof under a rainstorm to 100% in terms of the test group with semi-humid substrate on the flat roof under a light rain. The retention rate variation between the Sedum lineare modules among the selected rainfall events conforms to most previous researches, i.e., when either the roof slope, the substrate’s moisture content or the precipitation intensity decreases, the retention rate goes high. At the same time, most test groups present discharge delay. However, regardless of the substrate moisture variation, the difference in discharge delay between the flat roof group and the pitched roof group is very small, with few obvious peak delay being recorded.

Conclusion  The Sedum lineare modules present ideal performance in peak flow cutting, runoff reduction and discharge delay except under rainstorms or heavy rains during continuous rainfalls in Shanghai. It should be an effective land-saving low impact development (LID) measure for Shanghai to extensively apply roof greening to existing buildings, especially those with flat roof. The promotion of roof greening is expected to be an effective measure of reducing flood risks not only for Shanghai, but also for the other cities in the mid-lower reaches of the Yangtze River featuring the plum rain. In the future, Chinese cities should conduct more research on the detention and retention effects of roof greening under natural rainfall events with local climate features, so as to discover, verify and summarize the actual local performance thereof.