Rapid urbanization and population growth have become rising global concerns. They challenge the existing urban infrastructure and cause several social and environmental issues. One of the most pronounced impacts is the significant increase in the impervious surface in built-up areas. In terms of stormwater management, it causes more flash flooding in terms of increasing frequency and intensity and the pollution of stormwater runoff to receiving water channels. Additionally, the reduction in vegetation cover results in the urban heat island (UHI) effect due to more significant solar heat absorption, the degradation of natural habitat, and loss of biodiversity. As a result, an appropriate solution is required to address the concerning situation.
Among various green infrastructure (GI) practices, green roofs (GR), also known as living roofs, which have recently been introduced, offer a variety of ecosystem services. The temperature and stormwater runoff volume reductions have been widely documented as GR benefits [1
]. Other GR services include enhancing runoff quality, mitigating air and noise pollution, recovering urban ecology, and improving social and economic aspects. GRs are generally divided into two main groups: intensive green roofs (IGR) and extensive green roofs (EGR) with a substrate depth of more than 30 cm and less than 15 cm, respectively [1
]. Each type of GR is suitable for specific purposes and site conditions based on their different advantages. While IGRs support a wide range of plants and prevail over EGRs in terms of ecosystem services, EGRs are a lighter system that can be widely implemented due to their affordability, less maintenance, and easy installation without structural reinforcement [2
]. Semi-intensive green roofs (SIGRs), which have a 15 cm to 30 cm substrate thickness, are a combined GR system that takes advantage of both EGR and IGR [1
Some attempts have been made to integrate GR with other systems. This combined system is described as “hybrid GR” in this paper. One of the noteworthy hybrid GR systems is the photovoltaic GR (PV GR), which was studied by Hui and Chan [7
] and Chemisana and Lamnatou [8
]. Whereas Hui and Chan [7
] found the surface temperature (Ts
) of a PV GR was 5 °C cooler than that of the traditional GR due to the shading effect of the PV panels, a substantial difference of 14 °C between the PV GR and the concrete roof was monitored by Chemisana and Lamnatou [8
]. An improved electricity productivity from the PV panels, which is believed to be due to the cooling effect of the GR, was also detected. Another integrated GR system is the blue GR initially introduced in South Korea. This system has the same design as the conventional GR except for a larger storage layer. The runoff outflow from the blue GR was 0.1 L/s compared to 0.3 L/s from the normal roof in the study of Shafique, et al. [9
]. Additionally, the combination of GR and green wall brings outstanding thermal and energy reductions as compared to stand-alone GR systems [10
]. In spite of the above-mentioned improvements, studies on GRs are insufficient and further research is required before making firm conclusions regarding their use.
Though GRs have been well-studied for decades to quantify the numerous ecosystem services that they provide, the implementation of GRs still remain restricted by barriers and challenges. More specifically, the lack of local GR research, especially in developing countries due to costly GR installation, could make building owners and authorities unaware of GRs’ benefits [1
]. Another noticeable constraint is the safety concern regarding the weight of a GR system. Given that most of the urban area consists of existing buildings, the retrofitting of GRs must be carried out by considering whether any structural reinforcements are required or not. Moreover, there exist many ambiguities and uncertainties about the capabilities of GRs. Nguyen, Muttil, Tariq and Ng [3
] pointed out that published results of GR services were inconsistent in different studies. Those issues need to be resolved by future research that is conducted locally to match with specific climate characteristics. Valuable information from local research is prerequisite to motivate policy makers issuing financial incentives regarding GR application. Addressing all of the discussed problems contributes to the feasibility of the widespread implementation of GR.
Applying simulation tools is a well-known approach to investigate the effectiveness of a GR system before the actual implementation at a building scale or even at a catchment scale. They inform investors and other stakeholders about what gains and losses GRs can generate and then contribute to the decision-making process. Simulation tools are extensively used to study the relationship between GR parameters. They are also able to model building-scale GR behaviors as compared to its actual performance to analyze the model accuracy. In contrast, a relatively smaller number of studies were conducted to assess the effectiveness of GR at large scales [3
]. In this study, the term “large-scale” refers to studies considering the application of GRs at scales that exceed the single-building scale, such as the city-wide scale, municipal scale, or catchment scale. Significant efforts are required to stimulate the thorough adoption of the GR concept as a part of water-sensitive urban design (WSUD).
Existing models are generally distinguished according to different approaches including the empirical-based rainfall–runoff (R–R) relationships and conceptual physics-based numerical models [14
]. Each of them has its own advantages and disadvantages and requires a comprehensive understanding to apply them in particular circumstances and purposes. The principle of the conceptual model is the conceptualization of physical rainfall–runoff processes; hence, each parameter is responsible for components of the physical process. Therefore, conceptual models are suitable for different levels of users primarily due to their simplicity [14
]. However, the limitation of conceptual models is that they need to be properly calibrated to produce accurate results. On the other hand, physics-based models such as HYDRUS are more complicated with a significant number of parameters; thus, they produce outputs at a high level of accuracy. Nevertheless, the complexity of these models leads to several computational constraints and difficulties for non-modeling users [16
]. While they are ideally suited for detailed design, conceptual models are preferably used for conceptual-level planning [18
]. In general, none of the models clearly prevail over others and the vast majority of them must be well calibrated against climate conditions in the area of interest [16
Among several available tools, the Model for Urban Stormwater Improvement Conceptualization (MUSIC) is Australia’s most popular stormwater management tool [18
]. In spite of MUSIC’s extensive use in Australia, the application of this tool for GR research is limited. Table 1
illustrates some recent studies that have used MUSIC, with only two of them simulating GRs. This could be because of MUSIC’s lack of a built-in module for modeling GRs. Some recent studies include those undertaken by Hannah, et al. [19
] and Liebman, et al. [20
], which provide valuable information and a foundation for future studies. MUSIC with built-in Australian meteorological and climate data is suitable to assess impacts of WSUD systems as part of preliminary design at a catchment scale [16
]. MUSIC, a conceptual model, has advantages over complex physics-based models due to its simplicity and low computational requirements, allowing modeling of large-scale GRs and long-term continuous simulations [17
]. Although MUSIC is designed with in-built meteorological data templates for Australian regions, it can also be applied anywhere in the world where appropriate climatic data are available. This includes sub-daily rainfall data (ideally at a 6 min timestep but other timesteps including 5 min, 10 min and hourly are available). The model has been applied in Singapore, Israel, China, Malaysia, and other countries (personal communication, Dale Browne, 2022). Therefore, MUSIC models can be applied internationally with appropriate local climatic data.
Considering the above-discussed gaps, many opportunities exist for future research, which motivates the present study. This research aims to test the performance of green roofs using available industry software (namely MUSICX, which is an upgraded version of MUSIC) to deliver on relevant stormwater management objectives at a campus level. Specifically, MUSICX models are developed to evaluate the effectiveness of installing GRs on all building rooftops at the Footscray Park campus of Victoria University (VU), Melbourne. The performance of large-scale campus-wide implementation of GRs was assessed through the reduction objectives for runoff volume and runoff quality as set out in the Environment Protection Authority (EPA) Victoria guidelines. EGRs are chosen in this study due to numerous well-documented benefits provided by the widespread implementation of such GRs. Although eWater MUSICX can model the stormwater runoff from many types of urban surfaces such as paved roads, roofs, and landscapes, there is no built-in module or package in MUSICX to model GRs. Subsequently, the outcomes of this study will contribute to understanding the impact of GRs in terms of runoff quantity and quality at a catchment scale. The widespread application of GRs would cost a lot in terms of effort and investment; consequently, it requires sufficient technical information from such studies to foresee the potential gains and losses [22
]. Given that MUSICX has some limitations due to its conceptual nature [18
], the selection of this tool is mainly based on the primary research aim of introducing a simple approach to assess the impacts of GRs that can provide accurate results, especially at the initial stage of conceptual design. Moreover, the framework proposed in this paper could be easily included in decision-support tools that can be used by different stages of decision making [23
A summary of application of MUSIC in recent studies.
A summary of application of MUSIC in recent studies.
|Study||Location||Type of WSUD Treatments||Reduction in TSS/TP/TN (%)||Flow Reduction (%)|
|Zhang, Bach, Mathios, Dotto and Deletic ||Brisbane/Melbourne/Perth, Australia||Bio-retention cells, wetlands, and ponds||85/60/45||N/A|
|Ghofrani, et al. ||Tarwin Lower, South Gippsland, Victoria, Australia||Rainwater tanks, bio-retention cells, vegetative swales, and infiltration systems||94.4||16|
|Noh, et al. ||Cameron Highlands, Pahang, Malaysia||Wetlands, bio-retention cells, on-site detention, sediment basin, and gross pollutant traps||65–83/52–78/40–66||N/A|
|Schubert, et al. ||Little Stringybark Creek (LSC) watershed, Melbourne, Victoria, Australia||Rainwater tank, infiltration systems, and bio-retention cells||N/A||60 for storms ≤2 h and 30 for storms >2 h and ≤12 h|
|Montaseri, et al.  ||ACT, Australia||Swales, rainwater tanks, bio-retention cells, infiltration system, and wetlands||80/75/70||N/A|
|Hannah, Wicks, O’Sullivan and de Vries ||Bannockburn, Central Otago, New Zealand ||Green roof||73.9/–12.9/87||62|
|Liebman, Wark and Mackay ||Western Sydney, Australia||Green roof||N/A||22 and 56 for 12.5% and 37.5% GR coverage, respectively |
GRs have been extensively used worldwide in the recent past as compared to other varieties of GI as a potential strategy to address several social and environmental issues. GRs have been significantly studied during the last decade and the results show both its advantages as well as disadvantages.
Due to the existing research gap related to assessing GR performance at a large scale (i.e., scales exceeding the single-building scale), this paper attempts to investigate the hydrological effect of the implementation of GRs at the Footscray Park Campus of Victoria University in Melbourne, Australia. The simulation was carried out using eWater MUSICX, which is a modeling tool that is widely used in Australia. MUSICX possesses the advantages of a conceptual model with built-in Australian climate data and has a huge potential to effectively simulate the hydrological response of GRs at large scales.
Green roofs are still not very popular in Australia and there has been very little modeling of these green infrastructures to understand their benefits in terms of improvement in stormwater quality in the Australian context. This research presents the outcomes of continuous simulation modeling assessment, demonstrating that an existing model (namely MUSICX) can be adapted and used for this application. Thus, this paper aims to assess the performance of GRs using the MUSICX software and provides a simple but effective framework to inform investors and policy makers about the benefits of GRs, which is a prerequisite for widespread implementation of GRs. Additionally, this study attempted to evaluate the impacts of large-scale GRs on runoff quantity and quality on the campus. The simulation results showed a positive performance of GRs, especially with regard to the reduction in stormwater runoff volume. On the other hand, the combined use of GRs and other stormwater treatment devices is required to meet runoff quality objectives according to local stormwater guidelines.
The following is a summary of the key observations and recommendations obtained from this study:
The modeling results show that GR is effective in reducing runoff volume, TSS load, TP load, and TN load. While the largest reductions of roughly 30% are in runoff volume and TN load, the smallest reduction is in TSS load in both studied approaches;
Land use node and bioretention node approaches can be used interchangeably since the difference in MUSIC modeling outputs was found not to be substantial.
The land use node-based method is recommended to be applied when modelers focus on studying runoff quantity due to several simulation settings of the GR substrate’s hydraulic characteristics. On the other hand, the bioretention node-based method is preferable in runoff-quality-related research because of the modifications of plant types and nutritional characteristics of the GR substrate.
In this paper, the importance of model calibration is highlighted. Though no soil testing and flow monitoring data were obtained to calibrate the MUSIC-GR model, they are still part of future work to strengthen the connection between the VU GR design and modeling settings, thereby enhancing the modeling accuracy. On the other hand, concerns about the low accuracy of a model even with properly-calibrated parameters have also been discussed.
The application of GR for the entire VU campus area did not meet runoff quality objectives as set out in the EPA Victoria guidelines. Therefore, it is recommended that a treatment train including GR and other WSUD strategies be implemented to meet several stormwater management objectives.
Irrigation for the GR vegetation contributes to a substantial amount of GR runoff. This paper provides an explicit recommendation to include irrigation into MUSIC to model GRs more accurately.