This chapter investigates the critical role of urban open spaces (UOS) in fulfilling essential urban functions, highlighting their importance for environmental sustainability, social interaction, and urban connectivity. The study challenges the conventional wisdom that denser neighbourhoods have poorer access to high-quality open spaces, proposing instead that compact urban environments may offer better access to such spaces. The research employs a comprehensive methodology that assesses the quantity, quality, and accessibility of UOS, integrating various datasets and GIS network analyses. The pilot study in Stavanger, Norway, focuses on four areas within the municipality, comparing densely populated neighbourhoods with more sprawled, peripheral areas. The findings suggest that residents in denser neighbourhoods, such as Storhaug, enjoy better access to both the quantity and quality of urban open spaces compared to those in less dense areas like Hundvåg. This chapter not only contributes to the ongoing debate on urban density and open space provision but also offers a robust framework for future studies and urban planning practices aimed at enhancing urban sustainability and quality of life.
AI Generated
This summary of the content was generated with the help of AI.
Abstract
The sufficient provision of urban open spaces (UOS) is fundamental for sustaining good quality of life in cities. This is an aspect long established within urban planning. Indeed, the balance between built masses and open spaces has been crucial for the discipline’s development and innovative practices in spatial design. This paper presents a pilot study correlating the concentrations of built form and population with 10-minute walk access to UOS, including both grey and green areas. We realise this in three steps. First, we verify the availability of all UOS for flexible utilisation by the public. Second, all UOS are evaluated through a qualitative score matrix. And third, we assess the amount of UOS that is accessible from each residential building within the 10-minute isochrone through network analysis. The research results challenge the notion that citizens in denser neighbourhoods have poorer provisions of good quality UOS. On the contrary, the examined case studies illustrate that inhabitants within denser areas have physical access to a greater quantity of UOS. This fact becomes even more significant when one considers the individual open spaces’ elements, amenities and connectivity. The realised analysis indicates that if one aims to effectively assess the provisions of UOS within cities, a more complex and diverse picture must be drawn in the evaluation process. Only then the impact of planning strategies such as urban densification can be fairly evaluated with respect to the pressure the process puts on the available green and grey UOS.
6.1 Introduction
Open space in cities is as essential as built-up space for fulfilling urban functions. Thus, the sufficient provision of urban open spaces (UOS) is considered crucial in spatial planning, serving vital functions that can be grouped into three large categories. First, UOS are critical for sustaining a healthy and safe environment in cities by guaranteeing sufficient ventilation and natural lighting in buildings and urban residents’ access to green areas for recreation and comfort. Second, UOS facilitate a meaningful environment enriching the human experience of living in society by providing spaces for social interaction and possibilities to (re)shape collective identities. And third, UOS are fundamental for accessibility and connectivity, enabling a well-functioning urban environment for a prosperous life.
Even in very compact urban environments, such as in some preindustrial areas, between 50 and 60% of the ground floor area is open space. This proportion has grown enormously with the introduction of modern planning since the 1850s. Indeed, the quantity, the quality and the connectivity of open spaces were fundamental elements in the modernisation of the city since the second part of the nineteenth century. The Haussmann plan for Paris and the Plan Cerdá for Barcelona are prime examples of this approach to reshape and expand existing cities from a rational logic of accessibility and connectivity to and through open spaces. For example, the open space per block in the Plan Cerdá is between 60 and 80% (Pallares-Barbera et al. 2011; Santasusagna Riu et al. 2021).
Advertisement
During the first half of the twentieth century, the question seemed to be how to increase open space in urban areas by taking advantage of new technological advances. The advances in transport technology facilitated access to vast peri-urban areas making land for urban development cheaper. The development of new construction techniques for high-rise buildings enabled an increased density while decreasing building footprints. Frank Lloyd Wright’s Broadacre City is a prime example of the first, and Le Corbusier’s Ville Radieuse of the second. In both cases, the amount of open space increased enormously, reaching over 90%. Broadacre City and the Ville Radieuse were both unrealised plans. However, they inspired new trends in urban design that, among many other things, multiplied the area of open spaces in cities. With the increase in quantity came a decrease in quality, mainly triggered by the need for green buffer areas to separate roads from buildings and the loose connection between buildings and open space in new neighbourhoods.
6.2 Research Problem
Today, the problem with open spaces in cities, particularly in affluent countries, is not scarcity but wastefully spent. This problem is especially omnipresent in most areas developed after the 1950s. Squandered open space is synonymous with poor urban quality. The question then is how to requalify or better use these spaces, including the possibility of using some of them for new buildings, better articulated with their adjacent open spaces to enhance urban quality through densification. This relevant question has been explored in the work of Ståhle (Ståhle 2010). In this work, we use a different approach. Evaluating these spaces’ quantity, quality and accessibility is a fundamental step towards a densification strategy to integrally improve the city's sustainability. This study aims to contribute to this endeavour by testing the following hypothesis.
Inhabitants of more compact neighbourhoods in Stavanger enjoy better access to high-quality open spaces than inhabitants in sprawled peripheral areas.
6.3 Methodology
The principal objective of this study is to propose a methodology that can explore the provision of open spaces. The method considers distribution, accessibility, quantities (areal aggregation) and qualities of open spaces. Moreover, it relates these measurements with urban form quantifications and local inhabitants’ concentration. The proposed approach is structured on the following operations:
(1)
Verifying the possibilities for utilising the individual urban open spaces based on their physical access, ownership and functional regulation.
(2)
Applying a qualitative score matrix for individual open spaces based on their characteristics.
(3)
Performing network analysis within the selected areas of interest where individual residential buildings are utilised as a travel origin.
To demonstrate the possibilities of the research method, we employ it in a pilot study focussing on four areas of interest (aggregation of grunnkrets1 units) located in two districts within the municipality of Stavanger, i.e. Storhaug and Hundvåg. The following paragraphs describe in further detail the specific steps of the proposed methodology.
Advertisement
6.3.1 Utilised Datasets
The study integrates datasets from various sources, whose description is presented in Table 6.1. Most of them are collected in geo-referred format, but certain adjustments, integration and modifications are required to serve the empirical explorations. The three most significant data processing operations, part of the preparatory step, are:
Table 6.1
Employed datasets in the research. Hyperlinks to access the datasets via public platforms are added where applicable (in blue)
Road network dataset comprising all driveable roads that are longer than 50 m or part of a network and pedestrian and cycle paths and cycle paths represented as road link geometry. The dataset is exported from the National Road Data Bank (NVDB)
.shp format (SOSI 4.5)
Statens Kartverket (Norwegian Mapping and Cadastre Authority)
Statens Vegvesen (Norwegian Public Roads Administration)
Detailed building information, including building types, roof superstructures, descriptive building lines (e.g. ridgeline), and building attachments (e.g. verandas)
.shp format (FGDB 10.0, SOSI 4.5)
Statens Kartverket (Norwegian Mapping and Cadastre Authority) via Geovekst
2020
Bygning Matrikkel
Cadastral building register dataset with information per individual building, including gross floor area (GFA), gross floor area used for housing (ResGFA), gross floor area for other purposes (OthGFA), number of units (residential—dwellings or commercial), and height (H), expressed in number of floors
.xlsx tabular format
Statens Kartverket (Norwegian Mapping and Cadastral Authority)
June 2020
Performing spatial interpolation of various public land use datasets (FKB-AR5, FKB-Naturinfo, FKB-Vann), Eiendom Matrikkel (cadastral properties register data) and Stavanger Anleggsregister (open spaces register) to structure the basis for the performed assessment of the open spaces.
Setting up the pedestrian accessibility model by processing the transport networks dataset (Elveg 2.0) and integrating the regional traffic lights signals (Signalanlegg Rogaland).
Integration of FKB-Bygning (a constructions dataset), Bygning Matrikkel (cadastral building register) and census data on grunnkrets level to set the basis for calculating spatial densities and demographic distribution within Stavanger municipality.
6.3.2 Urban Open Space Assessment
In broader terms, urban open space (UOS) is defined as any urban ground space, regardless of public accessibility, that is not roofed by an architectural structure (Stanley et al. 2012). In the context of Stavanger municipality, such spaces include, among others, streets, parking lots, sidewalks, pathways, parks, urban gardens, courtyards, waterfront promenades, cemeteries, green areas (covered with grass, trees, shrubs, or other vegetation), schoolyards, playgrounds, squares, plazas and vacant lots. However, this diversity of spaces serves urban life in different ways and therefore requires other valuations. For this purpose, we propose two basic valuation criteria: availability and meaning.
Availability: Urban open spaces offer different degrees of availability. For example, not all are publicly accessible; some are publicly accessible but have functions limiting public appropriation, some are used in multiple ways and others are only available for a limited group of people.
Quality: People value urban open spaces differently based on their position within the urban system, the environmental and human-made elements they incorporate, their functions and the socio-cultural meaning they convey.
Availability and quality assessment of individual UOS are incorporated in the analysis through interpolating different data sources (FKB-AR5, FKB-AR5-Naturinfo, FKB-Vann, Eiendoms Matrikkel, Stavanger Anleggsregister) and the additional verification through satellite images and Google Maps. The following paragraphs reveal further details regarding how we methodologically specify these operations.
Availability Assessment of Urban Open Spaces—AVi
The availability of urban open space is mainly determined by its accessibility and readiness for use by ordinary people. The assessment, therefore, consists primarily of classifying the spaces according to their degree of availability. In this regard, we made three categories: unavailability, private availability and availability. Unavailable open spaces (uaUOS) are areas that, despite the proximity, do not have conditions for everyday people’s uses. Such areas are typically inaccessible to the public. However, they provide some specific services, for example, agricultural lands, military sites, roads with a speed of over 30 km/h, port yards and other industrial open areas. The second demarcation concerns the private open spaces (prUOS), including gardens, patios and courtyards, depending on building typologies. Private open spaces are available for appropriation only for the users of the building(s) they belong. The rest are considered as available urban open spaces (aUOS). The classification of open spaces in one of these three categories is the basis for estimating the index of urban open space availability per inhabitant, i.e. AVi.
where AVi is the Availability index of urban open spaces, y represents the residential building for which the estimation is performed, CA represents the catchment area around y (10-minute walking isochrone), aUOS is the aggregation of the available urban open space for various uses within CA, GBF is the gross building footprint within CA, uaAOS is the aggregation of unavailable for flexible use urban open space within CA and POP is the total number of occupants within y.
Quality Assessment of Urban Open Spaces—AQi
Quality assessment is executed through an evaluation matrix relating to the characteristics of individual spaces. This matrix adopts the conceptual and theoretical considerations presented in the previous sections of this paper. Based on this, we define three aspects and six indicators upon which all UOS are assessed. Table 6.2 and Fig. 6.1 depict how this assessment matrix is structured and applied to evaluate individual UOS quality values (Q). Since this study is developed from the perspective of spatial planning and urban design, it is worth underlining that the quality assessment is shaped with an emphasis on usability.
Table 6.2
Description of the different aspects/indicators used to estimate individual UOS values
#
Qualitative aspects and indicators
Weight
[A1]
Environmental quality
- evaluating whether the space provides basic environmental comfort for users
1.00
[1.1] Air ventilation and natural sunlight
/0.50
[1.2] Integration of natural elements within the space
/0.50
[1.2.1] Presence of greenery and vegetation within the space
//0.125
[1.2.2] Presence of trees within the space
//0.125
[1.2.3] Presence of water elements (creeks, rivers, ponds, lakes, sea, fountains, etc.)
//0.125
[1.2.4] Scenic views
//0.125
[A2]
Social meaning and collective identity
- evaluating the conditions to facilitate social interactions, meaning and identity
1.00
[2.2] Activities within the space
/0.50
[2.1.1] Conditions for regular activities (daily/weekly), including social gatherings, sports, recreational activities, etc.
//0.25
[2.2.2] Conditions for occasional economic and organised cultural activities (monthly or rarer)
//0.25
[2.2] Amenities, socio-economic and cultural uses of the buildings surrounding the space
[A3]
Supporting mobility and accessibility
evaluating the basic conditions for urban connectivity and accessibility
1.00
[3.1] Serving spatial connectivity and urban mobility networks
/0.50
[3.2] Spaces of significant activity because of their position in the urban system (highly integrated and connected spaces attracting pedestrians)
/0.50
Aggregated UOS qualitative value (Q)
3.00
Fig. 6.1
Visual diagram illustrating the aspects and indicators applied to evaluate individual UOS qualitative value (Q). Source The graphic is designed and created by the authors
An important demarcation to be considered here is that as part of estimating the quality assessment of individual urban open spaces (AQi), the unavailable spaces (uaUOS) are incorporated since they also provide at least environmental qualities to the respective urban areas. Thus, we assess the green uaUOS, such as agricultural lands and unavailable private spaces (prUOS) within the area of aggregation, with a score of 0.75 according to the matrix, since they secure air ventilation, natural sunlight, greenery and trees. Similarly, the grey uaUOS, including vehicular roads with a speed limit above 30 km/h and industrial and harbour sites, are assigned a score of 0.50 due to the air ventilation and natural sunlight they provide. Consequently, the AQi is calculated as follows:
where AQi is the Aggregated Quality index of urban open spaces, y represents the residential building for which the estimation is performed, wUOS is the aggregation of the urban open spaces (UOS) within the respective catchment area (CA) around y (10-minute walking isochrone) while considering their individual qualities (Q), S represents the areal coverage of an individual UOS which is within CA, and POP is the total number of occupants within y.
6.3.3 Estimating Urban Open Space Quality Index (UOSi)
GIS Network Analysis
The empirical estimations presented in this study are generated through GIS network analyses. We use the technique to calculate service areas around each point of interest (POI), i.e. the residential buildings within the selected grunnkrets units for this study. To estimate the available provisions of UOS for each POI, we employ the threshold of 10-minute travel by foot, following the work of Øksenholt et al. (2016). Furthermore, we estimate the time isochrones around each POI by utilising the average walking speed for healthy adults of 4.86 km/h along the pedestrian network (Montufar et al. 2007) and incorporating time barriers, i.e. waiting times at the traffic light signals within the municipality of Stavanger. This GIS processing operation is visually depicted in Fig. 6.2.
Fig. 6.2
A highlight of the urban elements used to process the estimation of the Availability index and Aggregated Quality index, incl. grunnkrets, serving as units of analytical aggregation (A), residential buildings, serving as POIs and pedestrian network to generate service area polygons
As a result, we can estimate an individual service area polygon around each POI, referring to a 10-minute walking time isochrone. Thus, we can analyse the provision of UOS per occupant for each residential building through its respective service polygon. Consequently, we can also logically calculate the average UOS provisions per inhabitant for each grunnkrets unit by aggregating the estimated scores per occupant for each residential building and considering the latter’s total number. We repeat the process concerning the Availability index (AVi) and the Aggregated Quality index (AQi). The difference between the two estimations is visualised in Fig. 6.3.
Fig. 6.3
Representation of the index estimations for an individual residential building (POI) concerning AVi (left), estimating the available provision of UOS and AQi (right), estimating the provision of UOS, considering the qualities they provide
where UOSi is either the Availability index (AVi) or the Aggregated Quality index (AQi) of urban open spaces, A represents the area of aggregation (grunnkrets units of interest), y represents the residential buildings located within A and n is the total number of residential buildings within A.
6.4 Pilot Study Case Selection
For this pilot study, we explore two districts within the municipality of Stavanger, i.e. Storhaug and Hundvåg. Both areas accommodate large populations in the context of Stavanger and represent prominent districts of the city. However, they differ in terms of urban morphologies and density developments. Storhaug is an early twentieth-century neighbourhood near the city centre, where urban villas, many of them divided into apartments, are the main architectural typologies. More recently, a large part of the district has been densified by the introduction of multi-family housing. The area has a higher concentration of inhabitants and built density than the rest of the municipality. On the contrary, the residents of Hundvåg traditionally occupy single-family houses. Still, due to the area’s relative proximity to the city centre of Stavanger, newer typologies have been adopted after the construction of bybrua (the vehicular and pedestrian connection with the city). These new expansions have been realised as greenfield developments. To a large extent, they are characterised as mixtures of individual row houses and low-height multi-family apartment blocks, spatially arranged as spatial blocks next to agricultural lands. Based on this description, we can identify specific typologies which can be explored as individual study areas within these two municipal districts.
To ensure that the specific areas to analyse are representative in terms of compactness, we consider two indicators in the selection process: ground space index (GSI) and population density (POPd), referring to the concentration of buildings and residents, respectively. Based on the values of these two indicators, we identify the representative ranges for each of the districts (i.e. the mean value for the whole area ± a quarter of the standard deviation) and specific study case areas fitting within these ranges (see Figs. 6.4 and 6.5). In addition, Table 6.3 illustrates the details regarding this selection and includes additional parameters of the case studies, such as areal coverage and approximate distance from the central core of Stavanger.
Fig. 6.4
Selected areas for examination within the district of Storhaug. Basemap image: Norwegian Mapping Authority (Kartverket), obtained via Norgeskart.no [accessed March 2024]
Selected areas for examination within the district of Hundvåg. Basemap image: Norwegian Mapping Authority (Kartverket), obtained via Norgeskart.no [accessed March 2024]
Details regarding the selected grunnkrets (aggregation) units to be examined, the districts where they are located and the municipality of Stavanger
Study areas of interest (aggregated number of grunnkrets)
Built density (GSI—ground space index)
Population density (inhabitants per hectare)
Area (ha)
Apprx. distance from city centre (km)
Apprx. walking time from city centre (minutes)
Storhaug (26*)
0.239
(SD/4 = 0.029)
51.3
(SD/4 = 7.2)
325.8
–
–
Study Area 1 (2)
Johannes, Lervik 2
0.268
54.5
21.2
1.2
15
Study Area 2 (2)
Storhaug 3, Paradis 2
0.210
52.5
15.1
1.3
17
Hundvåg (25)
0.117
(SD/4 = 0.017)
22.7
(SD/4 = 3.8)
655.6
–
–
Study Area 3 (1)
Kuneset
0.110
22.9
14.7
6.3
73
Study Area 4 (4)
Austbø 3, Austbø 4, Austbø 5, Skeie 5
0.100
24.2
36.3
5.2
68
Municipality of Stavanger (247)
0.038
5.6
25 596.3
-
-
6.5 Empirical Results and Interpretation
The GIS-based operations performed allow us to extract useful empirical outcomes to test the initial hypothesis. Table 6.4 depicts these results and includes the aggregated scores for each study area concerning the provision of UOS per inhabitant within 10 minutes of walking regarding their availability (AVi) and assessed qualities (AQi). Furthermore, as part of the analytical process, we extract the aggregated values of the mean UOS qualitative scores (Q) per m2 for the aUOS and the average prUOS per inhabitant for each study area since these variables seem to be also relevant to the research discussion. These pieces of extracted empirical information outline a few avenues for further interpretations, which are intertwined with the presentation of results in the following paragraphs.
Table 6.4
Provision of UOS per inhabitant within the study areas of interest, estimated through AVi and AQi and concerning both public and private open spaces, their availability and quality
Study areas of interest (aggregated number of grunnkrets)
AVi
AQi
aUOS quality/m2mean value
prUOS per person (m2)
Storhaug (26*)
Study Area 1 (2)
Johannes, Lervik 2
48.3
89.838
1.86
26.50
Study Area 2 (2)
Storhaug 3, Paradis 2
46.0
85.100
1.85
90.79
Hundvåg (25)
Study Area 3 (1)
Kuneset
27.6
40.572
1.47
77.38
Study Area 4 (4)
Austbø 3, Austbø 4, Austbø 5, Skeie 5
40.5
62.370
1.54
56.82
First, we can look at the results regarding the availability assessment of urban open spaces (AVi), an index that refers to the value of the available urban open space area (in hectares) per inhabitant within a 10-minute walking isochrone. On average, the residents of the Study Area 1 (Johannes and Lervik 2) have the highest UOS available, 48.3 ha, within a 10-minute walk. The scores for Study Area 2 (Storhaug 3 and Paradis 2), Study Area 3 (Kuneset) and Study Area 4 (Austbø 3–5 and Skeie 5) are, respectively, 46.0, 27.6 and 40.5. These estimations already reveal an interesting finding. Indeed, the residents of Study Area 3, which is characterised as the examined case with the lowest POPd (22.9 inh/ha), have the poorest availability of UOS to utilise for various purposes in their vicinity. On the contrary, in the cases of Study Areas 1 and 2 within Storhaug, where the POPd is more than double, residents have a significantly higher availability of UOS.
Second, we examine the results concerning the quality assessment of urban open spaces (AQi), estimated as explained in the previous section. The index refers to the value of the total UOS area (in hectares) per inhabitant within the 10-minute walking isochrone after the quality assessment for each space is performed. According to AQi, we can outline that Storhaug score’s higher values, respectively, 89.838 for Study Area 1 and 85.100 for Study Area 2. On the other hand, the estimated outcomes for Study Areas 3 and 4 are 40.572 and 62.370. However, it is worth underlining that there is a greater proportional difference between the scores of AVi and AQi for the cases in Storhaug compared to those in Hundvåg. This results from the generally higher values that individual aUOS scores concerning the qualitative assessment matrix presented above (see Fig. 6.1 and Table 6.2). As we can see in Table 6.4, the mean of the UOS qualitative value (Q) per m2 is the highest in the case of Study Area 1, i.e. 1.86, followed by 1.85 (Study Area 2), 1.54 (Study Area 4) and 1.47 (Study Area 3).
Third, although not examined as a focal point of this pilot study, it is also interesting to pay attention to the findings depicting the provision of private UOS per inhabitant for the examined areas. The estimated values reveal certain results worth highlighting. On the one hand, the residents of Study Area 2 have higher provisions of prUOS (90.79 m2), followed by those inhabiting Study Areas 3 (77.38 m2), 4 (56.82 m2) and 1 (26.50 m2). From these figures, we can underline that the provisions of prUOS within Study Area 2 in Storhaug are much higher than the ones in the study areas in Hundvåg (3 and 4) despite being a significantly denser area concerning both built form and population.
6.6 Reflection and Conclusion
This research elaborates on a method to evaluate the provisions of urban open spaces (UOS) in detail. The approach considers distribution, accessibility, quantities (areal aggregation) and qualities of open spaces in relation to urban form and local population densities. Furthermore, this pilot study also explores the hypothesis of whether inhabitants of more compact neighbourhoods in Stavanger enjoy better access to high-quality open spaces than inhabitants in sprawled peripheral areas.
As a highlight of the empirical outcomes presented in the previous section, we can argue that the executed study can confirm this research hypothesis. This is evident for the concentration of residents; according to AVi and AQi estimations within the examined cases, the areas with higher population density enjoy the better provision of urban open spaces. The correlations between built density, expressed in gross space index (GSI), and AVi/AQi illustrate a similar picture for the analysed sample. Residents in Storhaug areas, characterised by higher GSI, have greater availability and higher quality urban open spaces within 10-minute walking than the study areas in Hundvåg.
However, the limited size of the sample used in this study does not allow generalisations. The validation of this approach requires larger models, including entire cities or a more significant number of samples from different cities. The method can also be improved by a more detailed analysis of building typologies and private/semi-private UOS, such as balconies and rooftops. Another limitation of this methodology is the substantial amount of local knowledge required for its implementation. The flexible uses of specific UOS for the organisation of seasonal events, closing of vehicular traffic along streets for a period throughout the year, and similar activities reflect on the qualitative assessment of individual UOS. Combining publicly accessible datasets and tools such as Google Maps is valuable. Still, knowledge of how urban open spaces are used over time is fundamental. Indeed, exploring these shortcomings provides avenues for further research directions and can eventually reveal additional details regarding the method's applicability, potential and limitations.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
The grunnkrets unit is the smallest statistical unit in Norway, and they vary in size and population. In the case of the municipality of Stavanger, there are 247 units located within 9 districts (Bydeler). The grunnkrets with higher population and built density tend to have smaller areas than those in sparsely populated zones.
Montufar J, Arango J, Porter M, Nakagawa S (2007) Pedestrians’ normal walking speed and speed when crossing a street transportation. Res Rec: J Transp Res Board 2002:90–97. https://doi.org/10.3141/2002-12CrossRef
Santasusagna Riu A, Tort Donada J, Vadri Fortuny MA, Carril PV (2021) Estimating public space metrics from nineteenth-century urban cartography: Barcelona’s Cerdà Plan of urban expansion. Environ Plan B: Urban Anal City Sci 48(9):2640–2655. https://doi.org/10.1177/2399808320987562CrossRef
