Although pioneering RUA projects exist, most suffer from a lack of promotion, specific laws, legal procedures, and urban codes (Cerón-Palma et al., 2012; Sanyé-Mengual et al., 2013). Studies in Barcelona, Berlin, and Bologna developed a preliminary classification of such barriers (Cerón-Palma et al., 2012; Specht et al., 2015, 2016a; Sanyé-Mengual et al., 2016; Specht and Sanyé-Mengual, 2017). Social obstacles include limited acceptance by users, the conceptualization and perception of UA, by many stakeholders, that it is not “true” agriculture, and the urgent need for training qualified technical personnel (Sanyé-Mengual et al., 2016; Specht et al., 2016a; Specht and Sanyé-Mengual, 2017). Social and health risks have also been repeatedly identified in several surveys on citizen perceptions (Sanyé-Mengual et al., 2016, 2018b; Specht and Sanyé-Mengual, 2017). Additionally, the possible environmental impacts associated with materials used for the construction of RTG facilities require careful consideration (Cerón-Palma et al., 2012). The low level of income generated by RUA products and difficulties in developing a viable business model were found to be the principal economic concerns (Palmer et al., 2016; Specht and Sanyé-Mengual, 2017). Technological and architectural barriers included the visual/aesthetic impact (especially within historical centers), structural load limitations in buildings, building height limits according to the building size, and the overall building envelope (Cerón-Palma et al., 2012; Sanyé-Mengual et al., 2016; Specht et al., 2016a). Legal challenges range from the lack of RUA regulations in current urban building codes and the difficulties in managing food safety protocols and certification schemes within small-scale farms. While the few studies that have been conducted have identified the barriers, opportunities, and risks associated with urban agriculture, there is a gap in identifying specific policies and key factors that can contribute to or limit urban agriculture on rooftops.

Four cities from Europe, Barcelona, Berlin, Bologna, and Paris were chosen as case studies, given that they recently hosted some highly innovative RUA projects aimed at social inclusion, technological development and research. Among them, policies for supporting RUA have been implemented only in Paris (Paris City Council, 2018), whereas in other cities, existing regulations do not specifically target these kinds of projects (Cerón-Palma et al., 2012; Orsini et al., 2014; Freisinger et al., 2015). Table 2 shows a summary of the main characteristics of the case studies. Barcelona is a compact Mediterranean city (Rueda, 2007; Parés et al., 2013). It has 1.6 million inhabitants in 101 km² and features a population density of 16,420 inhabitants/km² (Statistical Institure of Catalonia, 2020), being among the densest and most compact municipalities in Europe (Barcelona City Council, 2018a). The lack of land has led to exploring new ways to promote horticulture in the city, such as the RTG located on the ICTA-ICP building with a focus on research for technology innovation (Fertilecity, 2018). The city of Berlin has 3.7 million inhabitants (Berlin Business Location Center, 2019) who live over a surface of 892 km² (OECD, 2010) with a population density of 4,147 inhabitants/km² (Environmental Atlas Berlin, 2018). Today, among existing RUA projects, two rooftop gardens have a particularly social focus, whereas an RTG for applied research in botany and plant biology can be found at the Humboldt University building. Bologna is the main city of the Emilia Romagna Region, situated in northcentral Italy, and with a population of 394,463 inhabitants in 140.7 km², resulting in a population density of 2,802 inhabitants/km² (ISTAT, 2010, 2021). Paris has 2.2 million inhabitants living on a surface area of 105 km². This results in one of the highest urban densities in the world, reaching values in inner Paris of 20,755 inhabitants per km² (INSEE, 2017). The City Council was recently very active in promoting projects targeting biodiversity preservation, greening, UA and food initiatives (Delgado, 2018). In 2016 and 2017, the first and second editions of the Parisculteurs program were launched, creating social inclusion and research spaces (Collé et al., 2018).

The second stage of the research consisted of the identification and classification of the UA experts to be involved. This included UA project developers (e.g., architects, agronomists), public administrators (with responsibilities in assigning municipal licenses and developing urban planning strategies), academics and researchers, and citizens involved in UA initiatives. Furthermore, relevant stakeholders from the cities of Barcelona, Berlin, Bologna, and Paris were identified and invited to define the state of the art of RUA in their cities.

To collect data from stakeholders, a workshop (Cerón-Palma et al., 2012) was conducted at the ICTA-ICP building (located in Barcelona) in September 2017. During the workshop session, interventions were recorded by the workshop organizers. The five stages developed in the workshop for data collection are described below.

Six main categories were identified, namely, urban, environmental, social, legal, technological/architectural, and economic barriers and opportunities (Cerón-Palma et al., 2012; Sanyé-Mengual et al., 2016; Specht and Sanyé-Mengual, 2017; Nadal et al., 2018).

Experts involved in UA, including project developers, public administration, academics, and citizen initiatives from Barcelona, Berlin, Bologna, and Paris, were identified and invited to participate in the survey.

Data collection was carried out from November to December 2017. The survey was designed to evaluate stakeholder perceptions through Likert scales that provided a range of responses to a series of statements. Five categories of responses were included (Croasmun and Ostrom, 2011), ranging from 5 to 1. The survey was structured into six sections: (1) survey description and context, (2) stakeholder information, (3) key issues for integrating RUA, (4) factors that hinder or promote RUA, (5) public policies to promote RUA, and (6) barriers to integrating RUA. Participants indicated their degree of agreement with a specific statement regarding the environmental, urban, social, legal, technological, architectural, and economic dimensions. Survey sections are further described in Supplementary Material 1.

A quantitative analysis of the survey results was performed, enabling us to define local and global trends in the responses and overall perceptions of the stakeholders.

Figure 2 shows a summary of the results (see complete data in Supplementary Material 2, Supplementary Table 1). Increased biodiversity, generation of green spaces, educational, research and social purposes, environmental CO₂ reduction, building energy optimization, new business generation, and new technology development were identified as potential key factors for integrating RUA. Urban planning, building laws, tax reduction, subsides, educational policies, and local policies, e.g., the Paris Climate Action Plan, Parisculteurs, and Plan Local d'Urbanisme de Paris (Paris Local Urban Plan) from Paris and the Primer Concurs de Cobertes Verdes (First Green Roof Contest) from Barcelona, were identified as policies that potentially “promote” and are related to RUA. Potential barriers identified included legal gaps, lack of a specific legal framework, building codes, administrative processes, restrictions on food sales, urban codes, health risks, historical building codes, rooftop accessibility difficulties, building designs, building structural features (overloading), high costs of infrastructure, climatic conditions, residents opposed to agricultural roofs on their buildings, lack of economic benefits, cost of water, firemen codes, food-free distribution, economic crisis, and a lack of interest by society.

Thirty stakeholders responded to the survey. Figure 3 shows the distribution of participants and their field of expertise regarding UA, made up of five participants from Barcelona, seven from Bologna, six from Berlin, and twelve from Paris. Fifty percent of the respondents were academics and researchers, 40% of stakeholders were project developers, and 10% were public administrators.

Figure 4 summarizes the key factors identified by more than 50% of participants (see all information in Supplementary Figures). Two factors—educational and environmental—were unanimously perceived as “relevant” by participants from Barcelona and Berlin. Educational factors refer to the integration of RUA as a tool for developing educational activities. Environmental factors include functions such as increasing biodiversity, generating green areas, reducing CO₂, and mitigating urban heat islands. Research from a multidisciplinary approach, including agriculture, environmental sciences, urban planning, architecture and social sciences, technological innovation related to new forms of UA, food production within city limits, and social functions are key factors perceived as “relevant” for integrating RUA.

Figure 5 summarizes the contextual factors and global and local trends that hinder or promote RUA that were identified by more than 50% of the stakeholders (see all information in Supplementary Figures). Globally, pollution was the only factor identified as a condition that “hinders” RUA. Those factors that “promote” RUA with the highest agreement (66%) among stakeholders were cultural values and social rural-urban links. Local trends showed five context factors perceived as “promoting” RUA: productive spaces, cultural values, social cohesion, social rural-urban links, and the high cost of urban land. The cost of water was perceived as a “hindering” factor to a similar degree both in Barcelona (75%) and Bologna (71%). There was some disagreement on the pollution factor; participants from Bologna (71%) identified it as a “promoting” factor, while participants from Barcelona and more than half from Paris (71%) identified it as a “hindering” factor.

Figure 6 presents a summary of policies selected by more than 50% of participants, both globally and by city (see all information in Supplementary Figures). Five policies were identified in the target cities as “promoting” RUA. The sustainability policy obtained greater agreement (79%) among participants from all cities, followed by the Milan Food Policy Pact (62%). According to local trends, six policies were identified as “promoting” RUA. One was found to be common in all cities: policies targeting sustainability. The remaining five policies targeting financial incentives, the development of new technologies, education, and food production, as well as the Milan urban food policy pact and the Parisculteurs program, were identified as “promoting” RUA. Policies related to food trade were considered to “hinder” RUA by all the participants from Barcelona and, to a lesser extent, by the participants from Paris. Policies related to urban planning were perceived by participants from Barcelona and Paris as “hindering” RUA and by those from Bologna as a “promoting” factor, while those from Berlin were “neutral.”

Figure 7 summarizes the barriers, and the frequency of their presence, to integrating RUA, both globally and by city. Only those barriers identified by more than 50% of participants are shown (see all information in Supplementary Figures). The following section presents the barriers, by category.

Constraints for integrating RUA in historical buildings and the limits on the height of buildings according to building codes were identified in this study as barriers to RUA development but had not been reported before (Cerón-Palma et al., 2012; Specht et al., 2015; Sanyé-Mengual et al., 2016), perhaps due to the recent growth of RUA experiences. Regarding the limits on the height of buildings, since the implementation of the Parisculteurs program (Paris), the city council has changed urban regulations to allow farming on rooftops even when the building exceeds height limits (Brin et al., 2016). Nevertheless, according to stakeholders' perceptions, this barrier still applies. In Barcelona, RTGs cannot be built on some rooftops due to height/volume restrictions (Metropolitan Area of Barcelona, 2018). Among the constraints for RUA integration, it was mentioned that building codes did not consider RUAs and the need for building structure reinforcement. Indeed, building overloading and the need for reinforcement (Cerón-Palma et al., 2012; Sanyé-Mengual et al., 2016) are still major barriers. Currently, many buildings may not have a suitable structure or load-bearing capacity for RUA (Toporova, 2018), which may also have economic repercussions due to the cost of building reinforcement and the need for professionals to develop and execute such projects.

Building designs to operate RUA and the difficulties of roof access were mentioned as factors that “constrain” RUA development. Cerón-Palma et al. (2012) also identified the complexity of adapting or renovating existing buildings; in this sense, it is essential to identify how users would access the roof spaces considering safety norms. In general, the technical adaptations necessary to operate RUA can lead to extra costs and limit the economic feasibility of projects. Competition for integrating other systems and/or functions on roofs is still present, as reported in previous works (Cerón-Palma et al., 2012; Sanyé-Mengual et al., 2016). However, current practice demonstrates that the integration of multiple systems/functionalities can also take place in parallel. In Barcelona, the RTG Lab Fertilecity integrates a rainwater harvesting system (RWHS) for crop irrigation, reaching 100% water self-sufficiency (Sanyé-Mengual et al., 2014). Five RUA projects recently built integrate RWHS and photovoltaic (PV) systems (Barcelona City Council, 2017). Thus, RUA, RWHS, and PV systems can coexist, providing significant benefits (Benis et al., 2018; Toboso-Chavero et al., 2018; Corcelli et al., 2019).

Regarding economic categories, stakeholders perceived that the high cost of infrastructure is a major barrier; previous studies also reported this constraint (Cerón-Palma et al., 2012; Sanyé-Mengual et al., 2016). Such barriers can be addressed with the support of financial policies and incentives, such as those already implemented in Barcelona and Paris. However, the initial investment goes beyond the financial cost, since maintenance of this kind of infrastructure is also expensive and constitutes an additional barrier during the operation stage (Zambrano-Prado et al., 2021). Therefore, for RUA that are successful and do not turn into short-lived projects, maintenance costs beyond the initial costs must be studied and considered. Food sales policies are related to urban land zoning ordinances, and together with the perception of high-cost infrastructure, can lead investors with commercial interests to easily lose interest. If there is no specific legislation for the trade of products grown within the city, it is difficult to integrate large-scale RUA projects. Fletcher et al. (2012) recognized restrictions by municipalities on sales of locally grown products in cities. To address this barrier, some cities have made policy changes, especially in North America. In 2012, the Berkeley Planning Commission adopted the definition of “Non-Processed Edibles,” which allowed the production of different kinds of food products within urban areas and their sale in residential districts (Fletcher et al., 2012).

As in previous barrier categories, no common social barrier was found in the target cities. Compared to other barriers, social aspects were associated with fewer constraints. The survey revealed that exclusive access to food growing on rooftops, exclusive access to developing RUA, a lack of interest by society in RUA projects and limited acceptance by residents of RUA on their building are not frequent. However, during the workshop, stakeholders manifested their concerns about these social barriers. A risk that large companies may transform RUA into an exclusively profit-oriented (Specht et al., 2016a) initiative and, thus, aggravating social disparities in accessing systems and products (Sanyé-Mengual et al., 2016) have been identified by stakeholders in the previous literature. These risks could also be drivers of green gentrification in neighborhoods. Currently, however, this risk does not seem to be a major concern among stakeholders.

Urban planning codes that do not contemplate urban agricultural land use are still barriers. Castillo et al. (2013) identified barriers related to zoning codes, such as a lack of clear ordinances that are friendly to agriculture. In Singapore, urban planners included rooftop farms in the definition of urban green spaces and diversified the classification of agricultural land use, allowing this activity in urban areas (Diehl et al., 2020). Additionally, cities in the U.S., such as New York and Chicago, were included (The New York City Council, 2010; Urban Sustainability Exchange, 2011; City of Chicago, 2020). Of the cities involved in this study, in Barcelona, the General Metropolitan Plan does not allow agricultural activities inside the city, effectively making the commercialization of food produced in the city illegal. In the case of Paris, programs to encourage UA have been launched, which may allow agricultural activities in the city, while in Bologna, the workshop findings indicate that agricultural activities are not allowed in the city.