Rapid growth continues at unprecedented rates in downtown Toronto with a population increase of 16.14% between 2016 and 2021 (Statistics Canada, 2022). The population between the downtown core and distant suburbs is also rising but growing at a slower rate compared to previous years (Statistics Canada, 2022). Increased urbanization has been attributed to multiple driving factors that include a general lack of land availability, demographic changes, economic growth, and an overall improvement in transport infrastructure (Chokhachian et al., 2020). In Toronto, different zoning permissions across the city have a considerable impact on neighbourhood density and growth rates, as zoning regulations can dictate building type, use, location, and form. A study observed that more permissive zone types tend to result in denser neighbourhoods (Toronto City Planning, 2021).

Furthermore, there are environmental concerns associated with increased building activity that may compromise the standard of living for city residents in the long term (Chatzipoulka et al., 2016). Residents in dense neighbourhoods typically face issues related to indoor and outdoor thermal comfort due to the urban heat island effect (Aghamolaei et al., 2020). As the emphasis on liveable cities continues to grow, studies investigating the impact of density on neighbourhood sustainability at various scales, typologies, and climates become more important. One way to measure neighbourhood sustainability is with environmental performance.

In the past decade, the demand for on-site renewable energy production to enhance building resiliency and meet rigorous energy performance standards has led to numerous studies investigating the correlation between urban geometry and solar potential as an environmental performance measure. Solar potential is the irradiance intensity on building surfaces and can essentially translate into the potential for electricity production through photovoltaic systems or solar thermal systems (Amado and Poggi, 2014). A combination of high-performance buildings and on-site renewable energy supports the City of Toronto’s ambitious goal to achieve Net-Zero Emission by 2040 (City of Toronto, 2021a). Several studies have confirmed that building geometry, orientation, and adjacent obstructions can influence solar access to the roof, facade, and ground (Amado and Poggi, 2014; Chatzipoulka et al., 2016; Mohajeri et al., 2016; Lobaccaro et al., 2019). The solar potential for roofs is less affected than facades by neighbourhood density, which can be attributed to the inter-building effect, another term for the overshadowing of neighbourhood surfaces caused by tall buildings built close to one another (Wang et al., 2021). Buildings located in the north and west parts of downtown Toronto experience less solar access due to the majority of high-rise buildings being located in the south and east parts of the city (R. Zhu et al., 2020b).

Recent studies have also presented evidence regarding the impact of urban morphology on building energy demand, thermal energy generation, daylighting, indoor thermal comfort, and outdoor thermal comfort (Chokhachian et al., 2020; Pan and Du, 2021; Wang et al., 2021; Giostra et al., 2022). Using a generative approach to produce different parametric models of urban forms, Chokhachian et al. (2020) explored the relationship between density metrics representing these forms and outdoor solar access. In this study, outdoor solar access was regarded as an environmental performance measure that is signified by the ratio of the pedestrian area that gets at least 50% of direct sunlight on the summer and winter solstices. The density of urban forms were represented through parameters such as building block size, canyon width (the inter-building distance), number of floors, and floor area ratio (FAR). The study found that a high canyon width improved outdoor solar access, and this was specifically significant in summer. Another notable finding of this study was that a high urban density characterized by a high FAR value (for example, FAR = 7) can have an important bearing on outdoor solar access during the winter causing it to drop by about 96.5%. In a similar study, Giostra et al. (2022) examined how different urban forms characterized by three different building block typologies, namely, the tower, courtyard, and bar blocks impact solar access, whilst exploring geometrical indicators such as building form and layout, notably building width, length, height, depth, inter-building distance, size of inner courtyards, orientation and the number of buildings. The study also found that towers have the highest solar potential, followed by the bars and courtyard block typologies. Orientation has an insignificant impact on the solar potential, as vertical surfaces receive more diffuse and reflected radiation than horizontal surfaces. Using a different methodological approach compared to the aforementioned studies, Pan and Du (2022) investigated the relationship of urban form density metrics with outdoor solar access, which was represented by horizontal illuminance levels (HIL) and site illuminance uniformity (SUo), by measuring these metrics onsite using digital equipment. The study found that Sky View Factor (SVF), Mean Building Height (MBH), and vertical uniformity are positively correlated with HIL and SUI, while Building Coverage Ratio (BCR) and Floor Area Ratio (FAR) are negatively correlated with these metrics. They also found that Mean Nearest Neighbour Distance (MNN), which represents building dispersion within a neighbourhood block, is positively correlated with HIL and SUI, such that a 1% increase in building dispersion can increase HIL by 0.07 lux and SUI by 0.01lux. Wang et al. (2021)’s review study found that through the interbuilding effect, a metric that qualitatively represents reduced solar access on building surfaces (roofs and facades) owing to the proximity of buildings to one another, the power generation output of roof-integrated active solar technologies can diminish by 50.73%.

Many similar studies analysed the impact of neighbourhood morphology on the rooftop and façade solar potential in two ways. The first method employs a simplified quasi-generic model consisting of a single building typology organized in a uniform layout. Parametric simulations commonly use the courtyard, tower, and slab typologies (Natanian et al., 2019; Czachura et al., 2022; Giostra et al., 2022). The best-performing typologies to optimize energy demand, solar potential, and daylighting are low-rise courtyard and slab buildings (Giostra et al., 2022). Natanian et al. (2021) noted that the courtyard typology is preferable in terms of energy balance. The hypothetical models are ideal for typology performance comparisons and sensitivity analyses. The main drawback to this method is the oversimplification of building geometries, creating a scenario that does not reflect the complex conditions of the built environment (Shi et al., 2021).

The second method simulates existing neighbourhoods that comply with local building codes and zoning regulations. Chatzipoulka et al. (2016) conducted a solar analysis for 24 urban sites with varying levels of density located in London, England. The research found that comparing sites with different layouts, but similar densities changed the irradiance by 11% for facades and 32% for ground surfaces. Sarralde et al. (2015) also examined various metrics and indicators to assess the solar potential of buildings' roofs and façades. Their analysis considered the increase in solar irradiation over a base-case scenario and compared different scenarios for optimizing solar irradiation, whereby potential increases ranged from 0.44% to 8.78%. A similar study confirmed that compact neighbourhoods received a lower solar irradiance compared to less dense neighbourhoods (Mohajeri et al., 2019). A case study approach provides designers and urban planners with a guideline for the real-world integration of solar technology in urban centers. From a Toronto context, the authors of this paper have also investigated how modifying roof morphologies in an existing suburban neighbourhood led to higher rooftop solar potential and solar power generation (Hasan et al., 2021). However, this study, in particular, examines existing neighbourhoods forming part of Toronto’s organic urban sprawl, whereby they manifest varying morphological configurations and use classifications, making the second typological method entailing the assessment of existing neighbourhoods’ solar potential the most relevant methodological approach to undertake.

Several studies have utilized various density metrics to examine the relationship between existing urban forms and solar energy potential, energy consumption, and environmental quality. A seminal study in this domain is by Mohajeri et al. (2016) who investigated the impact of urban compactness on solar energy potential using site coverage (SC), plot ratio (PR), volume-area ratio, building density (BD), nearest neighbour ratio (R_n), and entropy (S) with solar access. They found that a higher SC reduced solar access, while a higher entropy increased it. Chatzipoulka et al. (2016) studied the relationship between urban geometry and solar availability using SC, mean building height, the standard deviation of building height (Sd_height), the standard deviation of building footprint (Sd_area), directionality, complexity (CX), compacity (CP), number of buildings (NoB), and mean outdoor distance. The results showed that complexity, Sd_area, and directionality were the most influential metrics for the solar performance of building facades. Waibel et al. (2017) explored the impact of building geometry on energy consumption and solar energy potential using building footprint ratio (BFR), building height-to-width ratio (BHW), and floor area ratio (FAR). BFR and FAR are also known as SC and PR, respectively. They identified optimal building geometries that maximize solar potential and minimize energy consumption. Chokhachian et al. (2020) adopted a generative modeling approach entailing variation of parameters representing urban geometry. They investigated how daylighting, urban heat island, and solar access varied based on changes in BD, FAR, open space ratio (OSR), and building-to-street width ratio (BHSW). Zhu D. et al. (2020) investigated ten different cities to understand the effect of urban morphology on solar potential. They observed a general correlation where neighbourhoods with a higher density of tall buildings and significant variation in building height offer greater solar potential. The density metrics used include BD, Sd_height, average neighbourhood building height (ANBH), average building perimeter (AP), and envelope-to-roof surface area ratio. Boccalatte et al. (2022) proposed a general GIS-based methodology that explored the impact of building height, building distance, building footprint, total floor area, mean building volume (MBV), volume-to-facade ratio, FAR, and the number of neighbourhoods on the rooftop overall shading rate at a city-scale. The density metrics that were common to most of these studies were BD, SC, and PR. Table 1 summarises the density metrics obtained from the review of the literature, commonly used terminologies to represent them, and the corresponding formulas.

The City of Toronto’s Zoning Map indicates several use classifications and sub-classification, which are mainly categorized as Residential, Commercial, Employment, and Institutional (City of Toronto, 2013a). The residential category is subdivided into low-rise (typically houses) and high and mid-rise buildings referred to as Apartment. The Commercial category is further subdivided into Local, Residential, and Residential–Employment. For the Institutional category, the sub-divisions were General, Hospital, Education (typically colleges and universities), School, and Place of Worship. In the Employment category, the sub-categories included Light Industrial, Industrial, Heavy Industrial, and Industrial Office. A distinction between the Employment Industrial Office and Commercial categories is that they represent different types of building functional uses. A building classified as an ‘Industrial Office’ can support both office and industrial functions, such as manufacturing or warehousing, whereas the ‘Commercial - Residential Employment’ use classification refers to a building that can only support office-related functions, such as administrative activities or related services (in case of the employment sub-classification) (City of Toronto, 2022). The zoning map had additional categories such as Open Space and Utility and Transportation, which are areas without building development, and hence were out of the scope of this study. Given that large-scale integration of solar technologies on existing low-rise residential buildings is feasible on a case-to-case basis, but in actuality depends upon a multitude of factors unique to an individual building’s rooftop configuration, orientation, and proximity to shading structures, low-rise residential buildings were also omitted from the study analysis (Hasan et al., 2021). Similarly, the Place of Worship sub-category was also not considered, as the integration of technologies on the building envelope may compromise the protected heritage status of the buildings and disrupt its ability to blend in architecturally with the surrounding neighbourhood (Munari and Roecker, 2015). A summary of the land-use classifications used in this study for neighbourhood selection has been identified in Table 2.

The City of Toronto has numerous neighbourhood blocks that match the land use classifications investigated in the study. To select blocks that could be potential case studies, visual examinations were made using the City of Toronto Zoning Map and Google Maps. The following constraints were applied to ensure consistency in the selection approach.

1 Neighbourhood blocks that were close to buildings or blocks that represented different land-use classifications were filtered out, to ensure homogeneity in neighbourhood form and density.

For example, it was ensured that ‘E’ was close to a similar type of neighbourhood and was not potentially overshadowed by a high-rise building in the vicinity, as this is essentially an anomaly and does not reflect the density of such archetypical neighbourhoods. The neighbourhoods identified for further analysis in this study are indicated in Figure 4.

As can be seen from Figure 4, most of the neighbourhoods and surrounding areas are uniform in terms of use classifications. The selected neighbourhoods for the Residential zone are near ravines and creeks. High-rise buildings with identical heights are grouped to create the R and RAC zones. Predominantly low-rise apartments can be seen in RA. Located near the Toronto core, CR and CRE are similar high-rise residential neighbourhoods with commercial at-grade. Common characteristics include limited street trees and narrow street widths. The CL site includes low-rise buildings located in a typical suburban area. Single-family detached houses, parking lots, and wide streets surround the site. The Institutional neighbourhoods are composed of low-rise buildings. IE and IS are surrounded by mostly residential buildings, while IG and IH are healthcare centres that feature extensive parking lots and adjacencies to substantial green spaces. The neighbourhoods classified as Employment are near parks and similar in density to the healthcare centers. Large parking lots and roads keep the buildings spaced far apart. It can be observed that the neighbourhoods become more compact with closer proximity to downtown Toronto. Additionally, use classifications such as CRE or CR, despite being densely clustered, may also higher prevalence of buildings that are configured non-uniformly i.e., having a great variation of height. On the other hand, other use classifications particularly those from E or I are more dispersed and are located in neighbourhoods that are fragmented having predominantly buildings of the same height.

For the distinct Toronto neighbourhood archetypes, 20 different density metrics that are holistic representations, albeit in varying capacities, of the spatial distribution and geometrical characteristics of the built volume; the extent of spatial randomness between neighbourhood buildings and compactness were computed in this study. These neighbourhood archetypes were differentiated based on the land-use classifications mainly categorized as Institutional, Industrial, Residential, Employment, and Commercial. Within each classification category, there were further sub-classifications that bore intra-neighbourhood morphological variations too. Table 4 summarizes the values for the density metrics for the 14 neighbourhood archetypes. The relatively higher values for PD, SC, R_n, NoB, and BHSW (plot density, site coverage, nearest neighbour ratio, number of buildings, and building height to street width ratio, respectively) and lower values for OSR (open space ratio) are indicative of dense and compact neighbourhood archetypes, and those that stand out are, namely, CRE, CR, and IS, whereby this corroborates the visual observations of the sites as they are located in Toronto downtown. Additionally, the same neighbourhood archetypes also have higher values for CX and AVBH (complexity and average building height), which would be high-level indicators of the level of inter-building overshadowing in the respective neighbourhood. For relatively less dense neighbourhoods, namely, IEO, ELI, EHI, CL, and RA, Sd_height, has been lower in the latter four. In some neighbourhood archetypes, there is a high probability of finding buildings with very different geometrical characteristics, which is signified as S. This is another way of representing “organic” or great uncertainty in the geometrical characteristics of buildings in a neighbourhood. For example, in the E and, IE, the distribution of building heights is non-uniform. Additionally, E, CR, and CRE also have the most variation in terms of the footprint of buildings.

Figure 7 and Figure 8 delineate the average of the annual solar irradiance on all building roofs and facades in each of the 14 neighbourhood archetypes. The average roof solar potential of all the neighbourhoods hovers around 1,259 kWh/m², and about 64% of the neighbourhoods in the study sample have an above-average roof irradiance. Neighbourhoods that have the highest and lowest rooftop solar irradiance are CL (1,341 kWh/m²) and CR (1,101 kWh/m²) respectively. The denser neighbourhoods, particularly the CR, CRE and IS (all present in the Toronto downtown core) have below-average rooftop potential. These neighbourhoods also have greater randomness, in terms of the magnitude of height difference between adjacent buildings whereby the average Sd_height = 43.4 m, and this is indicative of the overshadowing of building rooftops due to other buildings in proximity (see Figure 9). In contrast to the solar radiation intensities of the neighbourhood roofs in the study sample, a greater level of variation in the facade solar irradiance is observed, whereby as expected the typical dense neighbourhoods have a lower than average solar potential. However, the anomaly also worth identifying is the lower solar potential for the lesser dense archetypes, which could plausibly be explained by the S_area, which represents the non-uniform distribution of facade sizes and their orientation along specific directions. This also indicates the potential for overshadowing.

The results on the variation of solar potential on building rooftops and facades across the different neighbourhod use-classifications is especially important from the standpoint of identifying the applicability of retrofits involving the integration of solar energy technologies, such as photovoltaics and solar thermal technologies. Therefore, for urban planners, architects, and energy modelers, this is an invaluable insight into which neighbourhoods and building surfaces would be a priority in being recipients of active solar technology retrofits. The density metrics discussed earlier provide a quantified assessment of neighbourhood characteristics and this correlates with the solar potential of certain neighbourhood-use classifications, whereby an evaluation of these metrics would determine the feasibility of potential sustainability measures in existing Toronto neighbourhoods.

Scatter graphs of the annual average solar potential of the roofs and facades of the sample neighbourhood archetypes versus the respective density metrics were plotted. A least square regression best-fit line was drawn to investigate the influence of each metric on the solar potential of the neighbourhoods, whereby the value of R² represented the strength of the relationship and the level of variance owing to the metric (see Figures 10, 11). Figure 12 plots the R² values for each neighbourhood archetype indicating the metrics having the highest influence on the roof and façade solar potential. For the roof solar potential, it is evident that the metrics bearing the most influence include Sd_height, PD, Cx, R_n, and NoB (standard deviation of neighbourhood building height, plot density, complexity, nearest neighbourhood ratio, and the number of neighbourhood buildings respectively). The R² values for these metrics are 0.90, 0.68, 0.63, 0.60, and 0.58, respectively. As seen from past neighbourhood solar potential studies, these density metrics together present a quantitative representation of the level of compactness and proximity to tall buildings (Chatzipoulka et al., 2016; Mohajeri et al., 2016). All these metrics are negatively correlated with the roof solar potential. In contrast, density metrics that were seen to have a moderate effect (0.5 < R² < 0.25) on the roof solar potential were AVBH, OSR, RESA, and BHSW (average building height, open space ratio, roof-to-envelope surface area ratio, building height to street width ratio, respectively).

In contrast to the trends seen in the scatter graphs investigating the relationship between density metrics and neighbourhood roof solar potential, the plots for the facade solar potential did not present a clear indication of the extent of influence of each metric. In this case, the annual average facade solar potential of each neighbourhood archetype was greatly dispersed, and the absence of clustering led to relatively lower R² values. Nevertheless, the density metric that has the highest influence on the neighbourhood facade solar potential are OSR, Cx, and RESA (open space ratio, complexity, and roof envelope to surface area), which are 0.48, 0.41, and 0.36. The listed density metrics are moderately correlated to the irradiance values. Both the CX and RESA are a factor of the facade area, hence a larger facade area of adjacent buildings may potentially cause significant overshadowing on facade surfaces. Similarly, as OSR is positively correlated with solar irradiance, spaces between buildings such that the inter-building distance is higher would lead to facade surfaces having better solar access. The non-uniform distribution of facade solar radiation intensity could be attributed primarily to their orientation and the size of surfaces (area) orientated in a particular direction, which varies even in similar neighbourhood use classifications.

From an urban planning and neighbourhood design standpoint, the findings of this study are an important and much-needed precedent as introducing novel knowledge on a plausible methodological framework in the assessment of the state of solar accessibility and retrofit viability in Toronto neighbourhoods. The research can also help guide planning policy by providing evidence-based recommendations for increasing the solar energy generation potential of other cities in Canada. However, in light of the study’s results, it is also important to recognize the limitations of the study in terms of the methodological approach opted for.

• Small dataset

While a relatively small dataset may be pointed as a limitation of this study, it is important to emphasize that, from the standpoint of the results on the relationship between density metrics and facade solar potential, this could have been augmented to determine a more meaningful relationship between density metrics and the solar irradiance levels. It is also apparent that roof solar potential, barring some exceptions, is typically uniform across the neighbourhood use classifications whereby the study did not consider the shading from trees and other ancillary structures that could potentially have further skewed the results.

• Façade solar potential

As the objective of this study was to investigate neighbourhood-level density and morphological impacts on solar potential, the architectural features of buildings were not accounted for in the simulation analysis. However, it is important to acknowledge the implications of simplifying building facades on the study results. The exclusion of balconies and overhangs from facades of residential high-rises inevitably impacted the estimated overall annual facade solar potential, likely making the simulated values an overestimation of the irradiance in reality. The effects are greater on south-facing facades where balconies cause localized shading, whereas north-facing balconies have less significant impacts. This is especially significant for residential classifications (Commercial-Residential Employment, Commercial-Residential, Residential, Residential-Apartment, and Residential-Apartment Commercial) where balconies typically exist. In contrast, for non-residential categories (Employment and Institutional) lacking balconies, the solar potential computed is closer to that in reality.

• Case studies as a representation of different land-use sub-classifications

While the results of this study provide useful insights into which neighbourhood use classifications and building surfaces are suitable for implementing active solar technology retrofits, it is also important to note that the selected case study neighbourhoods may not be fully representative of each land-use subclassification. Given the small sample sizes analysed for each classification, the results cannot necessarily be generalised to all neighbourhood blocks categorised under those use types. Further analysis across a broader selection of neighbourhoods would be needed to determine retrofit priorities at a city-wide scale.

• Correlation between density metrics

One potential reason for the low correlation (R² values) found between density metrics and the façade solar potential is that this study considered the average solar potential across all building facade orientations. By not differentiating based on orientation, key variables like obstruction angles were not accounted for (Catita et al., 2017; Desthieux et al., 2018; Sepúlveda et al., 2023). Thus, the averaging approach used may have skewed the relationship between neighbourhood density and facade solar potential.

• Future climatic context

The weather file (CWEC 2016) used in the study’s solar analysis provides a reasonable representation of typical meteorological conditions in Toronto based on historical weather data. However, under future climate scenarios, higher mean temperatures resulting in increased cloud cover and precipitation may contribute to reduced solar irradiance on buildings in Toronto (Li et al., 2018).