|dc.description.abstract||The combined effect of urbanization and climate change, which results in an increased level of imperviousness, is currently causing stress to urban drainage systems in urban areas unable to preserve the hydrological cycle of natural catchments. A considerable portion of urban impervious fractions is comprised of rooftops, which have increasingly been retrofitted with green roofs. Vegetated rooftops may contribute to reducing the impact of climate change and urbanization by retaining and detaining runoff, thereby reducing overall runoff volume and peak runoff rates. However, they do not always fulfil the requirements for stormwater detention. Moreover, in climates with limited evapotranspiration, a non-vegetated configuration may be a more favourable option for stormwater management.
A field station for testing different rooftops was established in 2017 on the coast of Trondheim, Norway. Two different solutions comprised of vegetated (green) and non-vegetated (grey) configurations were evaluated with respect to their hydrological performance between years 2017 and 2019. The first solution (Solution 1) consisted of an extensive green roof with a 30- mm thick layer of Sedum mats, a non-vegetated grey roof with a 200-mm thick layer of expanded clay, and a reference black roof. The second solution (Solution 2) offered a detentionbased green roof with a 100-mm thick layer of expanded clay, a non-vegetated grey roof with a 100-mm thick layer of expanded clay covered by paving stones and the reference black roof.
The extensive and detention-based green roofs provided stormwater retention between 24- 37 % of the total long-term continuous runoff in the warm period, experiencing a normalized daily retention of 0.86 mm/day for Solution 1 and 1.05 mm/day for Solution 2. Regarding detention performance, the grey roof outperformed the extensive green roof for the ten largest recorded events for Solution 1. However, underlaying the extensive green roof with expanded clay for Solution 2 resulted in the improvement of the detention performance in addition to providing reasonable stormwater retention. With respect to Solution 2, while the detention-based green roof performed similarly to the grey roof in terms of peak delay, it outperformed the grey roof in volume reduction (retention) by more than four times. Nevertheless, the event-based retention and detention performances of the individual roof may vary since they are a result of the local climate as well as events interpretation.
The detention performance of the detention-based green roof, consisting of Sedum mats underlaid with expanded clay, was tested for current and future climate conditions under extreme precipitation. Events having a return period of 20 years, including a climate factor of 1.4 corresponding to different locations in Trondheim, Oslo and Bergen, were created using an artificial rainfall generator. It was found that the peak delay and attenuation may not provide an ideal evaluation of the roof primarily due to the temporal resolution and specificity of individual rainfall patterns. While peak delay exponentially decreased from 31 minutes to 2 minutes, the centroid delay exponentially decreased from almost 500 to 10 minutes with increasing initial water content. The time of concentration for the black roof was within 5 minutes independent of rainfall intensity, whereas in the case of the detention-based green roof, it was extrapolated between 30 and 90 minutes, being strongly dependent on the rainfall intensity ranging between 0.8 to 2.5 mm/min. The roof performance decreased due to increased initial water content and was most sensitive to longer rainfalls with lower intensity than short, intense rainfalls.
The Environmental Protection Agency (EPA) Storm Water Management Model (SWMM) 5.1.012 with Low Impact Development (LID) Controls was used to evaluate the model performance of the roofs from Solution 1 on the building scale before the calibrated roof models were applied on a catchment scale. High-resolution data from previously monitored roofs and physical parameters within the individual LID layers (soil and drainage mat) were selected for model calibration. The calibrated parameters of the green and grey roofs were compared to several other roofs from another study, including roofs located in Oslo, Bergen and Sandnes, with different types of construction, geometry, and climates. The SWMM model provided a significantly positive match between observed and simulated runoff using a Nash-Sutcliffe model efficiency (NSME) between 0.56 and 0.96 and a volume error between -4 to 29 % for all configurations, including different locations in Bergen, Oslo, Sandnes and Trondheim. The calibrated roof models used during the winter period showed poor performance levels for longterm simulation runs. These were expressed with an NSME of 0.56 (green roof) and 0.37 (grey roof) and a volume error of 30% (green roof) and 11% (grey roof). Even though optimal parameter sets were proposed for each configuration, the model parameters obtained at one site were only partly transferable due to the issue of model equifinality. This could be linked to the choice of a one-year training period and the fact that model transferability is greater when moving from wetter to drier periods.
The PCSWMM model version 7.2.2780, which is a user interface for EPA SWMM 5, was used for simulating runoff from an urban catchment at Risvollan in the city of Trondheim. The calibrated green or grey roof models for Solution 1 were used in the catchment instead of conventional rooftops, and their resulting detention impact on the catchment outlet was evaluated according to nine scenarios. There was a focus on peak runoff reduction in terms of the runoff exceedance of three thresholds (100, 50 and 25 L/s) at the catchment outlet. It was found that as little as 11 % of the roof area can substantially reduce maximum flows, reduce the number of events, including the volume per event as well as the duration of the events themselves. Concerning the peak flows, the reduction performance increases in line with an increase of the area for potential retrofitting by green or grey roofs and with a decrease of the catchment slope. The implementation of grey roofs outperformed the extensive green roofs in all areas: event duration, number of events and volume of exceedance per event.
The green roofs showed a greater variation in their measured moisture content than the grey roofs did due to the transpiration process. However, the relatively high moisture levels in the expanded clay did not affect the detention performance under normal daily conditions. A reduction of detention performance occurred when running consecutive extreme events over the course of one day. In this case, the moisture levels increased with respect to events with a 20-year return period for future climate (2071-2100) and 200-year return period for the current climate. While the extensive green roof was mostly efficient for retention, the detention-based green roof underlaid with expanded clay provided an additional detention effect. Thus, the roof implementation on vulnerable points will mainly occur when using a configuration containing expanded clay.
Considering its initial saturation point and chosen type, rooftop retrofitting as a form of source control will, in most cases, contribute to a change in the runoff characteristics of conventional roofs. This thesis provides information for decision-makers and urban stormwater planners on how to target and focus on implementing rooftop solutions as stormwater measures. The hydrological performance of the detention-based green roof under extreme precipitation confirms and strengthens the applicability of the retrofitted rooftop solutions in an urban environment under both current and future climate conditions.||en_US