Introduction

Various disciplines emphasise the benefits of urban greenspaces concerning the ideals of the good life for everyone, including people’s well-being, community relations, environmental quality, ecological continuity, and property values (Ekkel and de Vries 2017; Heckert 2013; Karuppannan and Sivam 2013). With different sizes, amenities, vegetation and other characteristics, urban greenspaces include various open and public spaces such as forests, vegetated areas along streets, public squares, cemeteries, sports fields, wetlands, and often parks (Boulton et al. 2018; Fan et al. 2017). Meanwhile, research funds and new technologies for monitoring and assessing the earth and census data (Boone et al. 2009; Maliene et al. 2011; Talen 2010) have accelerated new concepts (e.g., green infrastructure, nature-based solutions and environmental justice) (Heynen et al. 2006; Silva et al. 2018; Niță et al. 2018; Rigolon 2016; Sharifi et al. 2021). Current research takes greenspaces as a significant part of socio-spatial systems. It calls for improving greenspaces as part of integrated systems (Cetin 2015) to handle the challenges of urbanisation processes, socio-spatial inequalities, and climate change (Boone et al. 2009; Heckert 2013; Kronenberg et al. 2020; Niță et al. 2018; Rigolon et al. 2019).

However, without enough guidance about how to provide greenspaces, especially for everyone, such promotion of green measures may pressure local governments challenged already by increasing population, land values, building density and social inequalities, and decreasing public land (Boulton et al. 2018; Garcia-Garcia et al. 2020; Heynen et al. 2006; Niță et al. 2018). For a guide, a few research investigates the provision conditions of greenspaces, including the frameworks for decisions and policymaking about planning and managing greenspaces (for reviews, see Boulton et al. 2018; Garcia-Garcia et al. 2020). Meanwhile, the recent pandemic demonstrated that people’s “quick” and on-foot access to greenspaces is still necessary. Particularly children, the elderly, women with young children, and low-income groups (Day 2008; Loukaitou-Sideris and Sideris 2009; Fan et al. 2017) need such daily access to public settings. With their physical and socio-economic capacities, these groups can afford limited walking distances and daily recreational activities near home. They often have disadvantages with the accessibility to greenspaces, which harms environmental justice (Boone et al. 2009; Fan et al. 2017; Guzman et al. 2017; Rigolon and Nemeth 2020; Vaughan et al. 2013). Because planning and design decisions influence spatial inequities, urban plan evaluations focusing on environmental justice issues can improve greenspace accessibility.

When contributing to this research need, this paper considers parks as the daily needed public greens and urban plans as critical but understudied documents about park accessibility nearby homes. It prioritizes two research points: More studies about park accessibility (for a review, see Şenol and Atay Kaya 2021) and plan evaluations (e.g., Cetin 2015) develop at the city and district levels. We need more investigations on the small socio-spatial scales (Haaland and van Den Bosch 2015; Garcia-Garcia et al. 2020) at which the disadvantaged groups experience conditions of inequity daily (Tan and Samsudin 2017). Also, the plans and plan evaluations must recognise that some people can afford shorter walking distances than healthy adults (Stafford and Baldwin 2018). This research favours the needs of disadvantaged groups to get park access nearby their home. It takes the need-based approach with spatial equity (Lucy 1981) as an alternative to conventional planning approaches that allocate greenspaces according to equality-based equity (Talen 2010). Of these conventional ways, particularly, “the standards approach to greenspace planning” (Boulton et al. 2018, p. 84) uses predetermined measurements (i.e., some standards including a target green area per dweller (m2/person) and/or distance from dwellings) to determine the adequacy of greenspace provision. Ultimately, it tends to evolve without considering socio-spatial differences (Boulton et al. 2018; Niță et al. 2018; Sharifi et al. 2021).

As a plan evaluation concerning equitable park accessibility, this paperFootnote 1 asks: How can we assess and revise urban plans to improve park accessibility within walking distances at the neighbourhood scale? About a part of a redevelopment plan at a scale of 1/1000 in Izmir (Türkiye’s third-biggest metropolitan area), this research contributes to the literature by not only evaluating a plan but also developing its assessment compared to some potential for plan revision and the current site.

When operationalizing need-based equity of park accessibility on urban plans (and) at the neighbourhood level, the paper deploys the GIS-based location-allocation (LA) analysis with a multi-criteria assessment methodology. Unlike similar GIS-based applications in the research (e.g., Boone et al. 2009), LA analysis allows this study to use the street network and physical barriers for walking in real settings, and reconfigure and articulate such contextual features according to the limited walking capacities of various groups. Thus, by its context-dependent configuration of the multi-criteria assessment, this study's LA analysis permits re-specifying socio-spatial determinants of walkability and accessibility to urban facilities.

Overall, how this study deploys LA analysis at the neighbourhood scale allows this paper to highlight that plan evaluations are also necessary for urban design research. Also, it helps with reconfiguring contextual variables, including data limitations: Türkiye has no demographic data at spatial scales smaller than the neighbourhood. When handling this data deficiency, we choose the degrees of spatial thresholds (including slope levels, rail lines, and wide roads) that make walking difficult for children and the elderly. According to study results, reconsidering some contextual variables (e.g., the lands designated for public service areas such as schools and cemeteries and the obscured planning terms such as “green space”) and urban design approaches can provide effective tools for urban policies for equitable park accessibility.

The paper continues with a literature review and the study approach. After the site’s current and planned conditions, the data and methods come. The study findings identify uneven local park accessibility in the plan and ways to revise it with improved park accessibility. The discussion and conclusion highlight the study's contributions.

Literature review: urban greenspaces and their planning

When assessing the socio-spatial distribution of benefits of urban greenspaces, the literature takes these spaces with their availability, accessibility, and amenities (Boulton et al. 2018; Kronenberg et al. 2020). Those considering greenspaces as part of environmental amenities and justice issues (Boone et al. 2009; Heynen et al. 2006; Rigolon 2016; Sharifi et al. 2021) identify the distribution of greenspaces with “fair allocation of/access to benefits for all social groups” (Kronenberg et al. 2020, p. 1; Silva et al. 2018). However, urban greenspaces and their benefits are not accessible to everyone (Heynen et al. 2006; Rigolon and Nemeth 2020). Besides other socio-spatial injustices, the inequitable allocation of greenspaces adversely affects the well-being of and sense of justice among those disadvantaged by this distribution (Boone et al. 2009; Sharifi et al. 2021). Such disadvantaged groups include the low-income and non-white dwellers (Boone et al. 2009; Heckert 2013; Talen 2010; Byrne and Wolch 2009), women, children, and the elderly (Forsyth et al. 2019; Loukaitou-Sideris and Sideris 2009) and others with the characteristics marginalized by modern planning and design approaches (Stafford and Baldwin 2018).

Regarding equitable accessibility of greenspaces, the research with the environmental justice lens has two groups (for a review, see Boulton et al. 2018). Composing the majority, the first group runs socio-spatial analyses about various greenspaces as socio-spatial outcomes more at the regional and city (Beler 1997; Ünal et al. 2016) and less at the neighbourhood scales (Garcia-Garcia et al. 2020; Tan and Samsudin 2017; for a review, Şenol and Atay Kaya 2021).The second group investigates the provisioning processes or contextual determinants producing various patterns of greenspaces. It examines management policies shaped by financial and administrative resources, political-economic structures and political leadership, organizational structure and tools of governance, available public lands, socio-spatial segregation patterns, and other case-specific factors (Boulton et al. 2018; Kronenberg et al. 2020; Garcia-Garcia et al. 2020). Urban plans appear a significant part of the provisioning processes of greenspaces.

Urban plan evaluations

Urban plans provide decisions about the allocation and design characteristics of land use, population density, and public expenditures. They adopt governmental statutes, regulations, rules, codes, and policies. In urban plans, public parks are not only a kind of land use with certain features (e.g., geographical location, use, size, form, and amenities), but also a public space and a public service area developed by public resources. Such characteristics make public parks part of public goods whose socio-spatial distribution affects environmental justice (Boone et al. 2009; Heynen et al. 2006; Talen 2010; Wolch et al. 2005). Ultimately, urban plans are a legislative instrument shaping the differential park accessibility and other socio-spatial (in)equities. Their evaluations should rely on new understandings of socio-spatial equities and updated technologies for spatial analyses, such as GIS.

The literature agrees with the necessity and value of plan evaluations to advance plan implementations (Lyles et al. 2016). Overall, the plan evaluations investigate either the plan outcomes or the processes of plan making (Kaza 2019; Laurian et al. 2010) and have differences in their choices of the type, scale, and details of plans (Kaza 2019). Besides the primary outcomes, the subjects of evaluations include specific goals, such as stormwater management, protection of built heritage, landscape (Laurian et al. 2010), ecology (Cetin 2015) and hazard mitigation (Lyles et al. 2016).

Yet the evaluations of park planning are scarce (Gebhardt and Eagles 2014). As a contribution to this small field, this paper evaluates park accessibility as an outcome of plans. Moreover, compared to more research at the city and regional levels (Cetin 2015; Laurian et al. 2010; Lyles et al. 2016), it develops at the neighbourhood scale. It assumes that the plan evaluations at the neighbourhood and small spatial scales can highlight better visions and strategies for grasping the local needs for greenspace planning and design (Garcia-Garcia et al. 2020; Haaland and van Den Bosch 2015; Talen and Anselin 2021; Tan and Samsudin 2017).

Of a few plan evaluations at the neighbourhood scale, some identify that the public investments for greenspaces favour the well-off over low-income neighbourhoods in Chicago (Talen and Anselin 2021) and Los Angeles (Wolch et al. 2005). Also, such investments (for instance) in Atlanta, Chicago, and Philadelphia trigger “environmental gentrification” (Rigolon et al. 2019, p. 2), a process causing an increase in housing prices and displacement of low-income residents. A few others detail how planning decisions about land use and the walking and driving features (e.g., road network density) shape current disparities in public greenspaces, for instance, in Wuhan city (Xing et al. 2018). Both groups deploy GIS to assess plan outcomes in comparison to actual neighbourhoods. Another group uses some GIS-based modelling to propose plan revisions with the assistance of a checklist of some principles (e.g., Gebhardt and Eagles 2014; Rigolon et al. 2019; Silva et al. 2018) or future models of, for instance, traffic networks in Nanjing (e.g., Xie et al. 2019).

Neighbourhood scale and walking conditions

Plan evaluations at the neighbourhood scale take “neighbourhoods” (if available, the census tracts) according to their administrative boundaries. Accordingly, these spatial units become geographies with homogenous conditions for people’s walking and accessibility to greenspaces. In contrast, this paper considers the neighbourhood as the built environment with various geographical (e.g., topographical slopes or a waterway) and physical features (e.g., wide roads or various land uses) that shape people’s walking conditions on streets (Arefi and Naser 2021; Stafford and Baldwin 2018) and to public areas (Karuppannan and Sivam 2013; Li et al. 2015). Furthermore, due to similar socio-spatial and individual factors, one’s neighbourhood can range from their dwelling to a “home area” (i.e., an area of a “short” walk from one’s home) or larger territories (Moser et al. 2002).

This paper operationalizes the neighbourhood scale as the home area defined by walking distance from residential buildings. It does so because, for equitable park accessibility, it prioritizes children, the elderly, women with young children, low-income groups, and others whose everyday life spends more in public spaces nearby homes (Day 2008; Fan et al. 2017; Forsyth et al. 2019; Loukaitou-Sideris and Sideris 2009; Moser et al. 2002). However, we do not investigate dwellers’ perceptions about and experiences with walkability, neighbourhood boundaries, or park accessibility. We assume that these home-bounded groups have limited walking distances due to their physical and socio-economic capacities (Stafford and Baldwin 2018). Thus, their need to get to parks and other public service areas by a short walk should be a priority in park planning and design.

Accordingly, this plan evaluation relies on the understanding of need-based equity (Lucy 1981), a critical alternative to conventional planning practices (Talen 1998, 2010) that rely on the equality-based distribution of public services and facilities (Lucy 1981). A part of the latter, greenspace planning uses “the standards approach” (Boulton et al. 2018, p. 84) with some predetermined quantitative measures (or standards) about a target green area per person in selected geographies and walking distance. These geographies (often the neighbourhood) become the administrative and planning unit where individuals (free from any socio-economic and physical characteristics) can walk to local facilities in 5 to 10 and a maximum of 20 min, or 400–800 m and 1600 m (Goličnik and Thompson 2010).

Recent research (Boulton et al. 2018; Niță et al. 2018; Sharifi et al. 2021; Ståhle 2010; Stafford and Baldwin 2018) criticize the standards approach for not responding to contextual and individual variables for walkability. Some plan evaluations, too, demonstrate how the deployment of this approach created gaps between the plans and the historical development of greenspaces, for instance, in some neighbourhoods of Madrid (Garcia-Garcia et al. 2020) and other cities (Boulton et al. 2018; Sharifi et al. 2021). The critical point is that urban policies relying only on predetermined measurements may not allow locally determined targets for greenspace planning (Haaland and van Den Bosch 2015; Tan and Samsudin 2017). Similarly, with the standards approach, local policies for green infrastructure increase the amount of green coverage, for instance, in 213 Romanian cities but do not include quantitative and local indicators about the multi-benefits of green areas (Niță et al. 2018).

Most research in Türkiye, too, evaluates local greenspace accessibility concerning the standards approach. A few studies with need-based equity at the neighbourhood scale underline the spatial inequities in the cities of Ankara (Beler 1997), Adana (Ünal et al. 2016), and Izmir (Şenol 2019; Şenol and Atay Kaya 2021). Regarding park accessibility in Izmir, Şenol (2019) demonstrates both the differences between rural and urban residential areas and the presence of urban "park-poor areas" with high ratios of children or elderly and small park areas per person.

This study’s call to recognise multiple characteristics of settlements and individuals in greenspace planning is part of urban design and urban studies research. For instance, the neighbourhoods with high population and building density, small lots, fewer private gardens (for a review, see Boulton et al. 2018), or low car ownership (Talen 2010) may require more public greenspaces. Also, the research emphasises age, gender, and income-related social differences as part of the determinants of people’s distinct needs in neighbourhoods, streets, and parks (Arefi and Nasser 2021; Day 2008; Forsyth et al. 2019; Ho et al. 2005; Loukaitou-Sideris and Sideris 2009; Stafford and Baldwin 2018) across cultural contexts (Karuppannan and Sivam 2013; Li et al. 2015). When contributing also to this research, this paper identifies spatial and individual determinants of walkability concerning not the standards for healthy adults but local socio-spatial characteristics.

Study site and methodology

Unlike more studies, this study utilizes GIS-based tools at three research stages: about the contemporary site, the plan, and a revision to improve the planned park accessibility. All stages identify the neighbourhood scale concerning walking distances rather than administrative boundaries. They have multiple contextual variables for walkability rather than any checklist or predetermined measures of the standards approach. The revision develops with a reconsideration of all public service areas. A comparison of all stages evolves concerning park planning with equitable versus equal park accessibility at home areas.

Study site

This study examines a part of a redevelopment plan (dated 2019, at a scale of 1/1000) for the Buca district of Izmir (Fig. 1A). In Türkiye, development plans at the 1/1000 scale follow the Zoning Law (dated 1985, numbered 3194) and detail the spatial arrangement of buildings, roads, and public service areas. Parks are open public spaces managed often by local municipalities. According to the Spatial Planning Regulation (dated 2014), the open and green areas include parks, children's parks and playgrounds, passive greens, recreation areas, fairs, zoos, urban squares, urban forests, botanical parks, forestation areas, cemeteries and so on. Urban plans must assign open and green areas according to predetermined per capita values and a future population at the neighbourhood scale. Park design must provide cultural, social, and recreational needs of individuals and communities and improve the quality of life and healthy environments. However, other than 10 m2/person as a standard for the minimum size of “open and green areas” (consisting only of children’s park, playfield, park, public square, sports field, botanical park, and recreational area), no legal criteria exist about park design and planning.

Fig. 1
figure 1

A Buca District in İzmir; B the study site as the section of the plan and a contemporary overview of its environs in Buca, C its land use, and D population density (person/km2) in 2019

Full size image

In Buca district, this case study focuses on “the park-poor areas,” identified by Şenol (2019) with high ratios of children or elderly and a low amount of park area per person in Izmir’s central city. Izmir metropolitan area has 30 districts with around 4.5 million people. Buca is the most populous district, with 510.695 people (TURKSTAT 2019). But its public green area per person is below the city average (Şenol 2019). Park-poor areas in Buca include parts of the three neighbourhoods (Göksu, Mustafa Kemal, and İnönü) (Fig. 1B) with 64.982 people, an amount of park area per person below one square meter, and a ratio of children (0–13 ages) (24%) higher than the city average (14%) (TURKSTAT, 2019). Of the redevelopment plan of Buca, four sheets cover these park-poor areas. To determine the study site boundaries on these sheets (see Plan Boundary in Fig. 1B), we took the residential lots overlapping with the park-poor areas and surrounded by major physical hindrances to walking (including the railway, water streams, and a highway). Called the Plan onwards, the selected site in 2019 had around 40,000 people in 5400 buildings, usually with three stories. To the north, the population density increases (Fig. 1C and D).

Data and methods

This study deploys GIS to simulate the Plan, a plan revision, and the contemporary site concerning equitable park accessibility. Such socio-spatial simulations allow some flexibility for a context-based plan evaluation. Deployment of GIS provides new opportunities for urban planning and design research assessing the physical and perceived features of the built environment at multiple scales (Maliene et al. 2011; Ståhle 2010; Şenol and Atay Kaya 2021). Overall, a group of researchers measure the accessibility of public service areas concerning the number of people with different socio-economic characteristics in selected spatial units (e.g., Boone et al. 2009; Heckert 2013; Talen 1998). Others assess individual differences in space–time experiences (Kwan 2010; Neutens et al. 2010). Although this study takes a spatial unit for its evaluation, it deploys the location-allocation (LA) analysis (a GIS-based location-based network analysis) (Yeh and Chow 1996). This tool enables us to resolve spatial problems (e.g., park accessibility) as part of a multi-criteria decision process (Malczewski 2006), including the built environment and individual characteristics (Fan et al. 2017; Kwan 2010; Neutens et al. 2010). Figure 2 presents our process for preparing multiple data sets for LA analysis and comparing three stages of evaluation.

Fig. 2
figure 2

Flowchart of the methodology

Full size image

The LA analysis uses a probabilistic algorithm that takes the distance between two locations (called demand and supply points) and each demand point with its closest supply point. In this study, the demand (i) and supply points (j) are the dwellings and parks, respectively. Also, this algorithm calculates alternative (or the best) locations for supply points (z) considering, for instance, a minimum distance between the demand and alternative supply points (Ekkel and de Vries 2017). Thus, for park accessibility, the LA analysis can determine the parks and alternative sites accessible within a minimum walking distance from dwellings.

$$z = \mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{j = 1}^{m} a_{ij} w_{i} d_{ij}$$

where: \({\mathrm{a}}_{\mathrm{ij}}\)= “(1 if demand point ‘i’ is closest to supply centre ‘j’, 0 otherwise)”, \({\mathrm{w}}_{\mathrm{i}}\)= “weight associated with each demand point”, \({\mathrm{d}}_{\mathrm{ij}}\)= “distance between demand point i and supply centre j”

The literature defines the travel cost between supply and demand points (Xi Yi; Xj Yj) based on distance, time, or travel effort. This study takes "distance" as the accessibility criterion. It calculates the distance between each dwelling and park (dij) with the following formula (Yeh and Chow 1996):

$$d_{ij} = \sqrt {\left( {x_{i} - x_{j} } \right)^{2} } + \sqrt {\left( {y_{i} - y_{j} } \right)^{2} }$$

We deploy the LA analysis to measure the park accessibility at the contemporary site, the Plan, and a revision of the planned park accessibility. All three stages consider certain physical and geographical features as the thresholds for walking between dwellings and parks. These thresholds include a continuous street network, a railway, water streams, and wide streets. We consider streets larger than 20 m with less pedestrian safety and narrow ones with high walking activities (Lee and Park 2018). The site and the Plan have one road in 20 m wide. Also, like Rehling et al. (2021) comparing greenspace accessibility by children and adolescents, this study identifies 300 m as the maximum walking distance for park trips rather than a typical 400–800 m defined for healthy adults (Stafford and Baldwin 2018). As the threshold value for the topographical slope levels, it accepts 6% and above. Slope levels of less than 8% and, notably, 6% appear as walkable slopes for people with various socio-economic characteristics, according to cases in Lisbon (Pereira et al. 2021) and Sicily (Ignaccolo et al. 2020).

Data preparation for and comparison of LA analysis

For an assessment with the LA analysis, this research used multiple variables, including the location of and the number of floors and dwellings in each residential building, location and kind of public service areas, street network, rail line, water streams, and slope levels. It produced its data from open digital sources with geographically coordinated data. We gathered the raster file of the selected urban plan from the municipalities. However, because Türkiye has no demographic data at the spatial scales smaller than the neighbourhood, we had no data about the distribution of age groups in the study site. After determining the study site boundaries on the plan (or the Plan), the data preparation started with georeferencing the Plan and digitising the spatial vector data about residential buildings, public service areas, and streets as identical layers in a GIS environment. The following steps (see Fig. 2) detail the data preparation for the LA analysis:

  1. a.

    Determine the locations of residential buildings and public service areas (including parks) as the demand and supply points, respectively (Fig. 3A).

  2. b.

    Assign a weight to each residential building according to its population (calculated by multiplying its number of dwelling units by the average household size in the Buca district) (Fig. 3B, C).

  3. c.

    Identify 300 m as the maximum walking distance via the georeferenced street network from each building to any park and 6% as the highest walkable slope degree. Assign a weight to each street according to its width (EarthExplorer 2020) and slope level (calculated with a digital elevation model, DEM, data of Izmir).

  4. d.

    Identify the spatial thresholds that may harm walkability or increase travel distance (including the streets with a slope level above 6%, a road with 20 m width, a rail line, and water streams) as the line barriers in the analysis (Fig. 3D).

  5. e.

    Add the weighted vector data of residential buildings, public service areas, and spatial barriers to a dataset by using the Network Analyst extension of ArcGIS.

Fig. 3
figure 3

Characteristics of the Plan with the spatial distribution of A land uses, B the number of building stories, C population density (person/km2), and D the identified spatial barriers to walking on the Plan

Full size image

After deploying the LA analysis at each research stage, the study compares the results regarding the site, the plan, and the revision in terms of the number of parks, the amount of both public greenspace and park area per person, and the number and ratio of residential buildings (thus, population) with park access.

Results

Park accessibility on the site

The contemporary site has 5400 buildings with approximately 40,000 dwellers, five parks with a total area of 10,493 m2, and 0.26 m2 as park area per person. However, according to Fig. 4, only 517 residential buildings (with 10% of the population) have park accessibility within 300 m walking distance. With the lowest accessibility level, two parks serve 14 residential buildings.

Fig. 4
figure 4

Park accessibility in the contemporary site

Full size image

Park accessibility on the plan

The Plan proposes approximately 35,000 people in 5218 residential buildings. It suggests around a 10% population decrease with a slight decline in building numbers. The population density in the Plan has a more even distribution than it is (Fig. 1D), whereas the Plan supports two, three, and four-story buildings (Fig. 3B, C). The number of parks increases from five to 16, including two sports areas and three playgrounds. Public service areas (a total of 87) include 16 parks, three health clinics, eight schools, five mosques, five institutional sites, one bazaar area, nine parking lots, 38 “green areas” (as stated by the plan legend), and three forestation areas. The Plan has more places for mixed-uses and commercial activities. The proposed street network is identical to the contemporary network, including a portion of the Izmir Highway to the eastern part and one subway station connected with the local bus system. (Fig. 3A). A rehabilitated streamline runs along the railway line and through the southern part, where the slope level increases by over 6%, and the Plan allocates more “green areas” and parks. The slope level across the Plan varies between 2 and 10% (Fig. 3D).

Figure 5 illustrates the park accessibility for all five existing and 11 planned parks. Nearby commercial activities and a stream, one planned park has no access from dwellings within 300 m walking distance. Five planned parks can serve relatively low numbers of buildings. Thus, half of the planned parks have disadvantaged locations. Out of 5218 residential buildings, only approximately 30% (1609 buildings) have park access. Furthermore, the geographic distribution of buildings without park accessibility (70%) shows some concentrations, which creates the park-void regions extending from the Plan’s mid-section to the northeast.

Fig. 5
figure 5

Park accessibility on the plan

Full size image

Plan revision for an improvement of park accessibility

This stage reconsiders all kinds of public service areas on the Plan as the potential locations to be redesigned as parks or with park-related purposes. Accordingly, 70 of all 87 public service areas are accessible to around 4253 (or 82% of total) residential buildings, introduced in Table 1 and Fig. 6. Significantly, 30 of 38 “green areas” are accessible and serve 2297 (or 44% of all) residential buildings. At this stage, the LA analysis assigns 547 buildings to 15 parks, compared to 1609 in the first stage with only 16 (parks) supply points in Fig. 5. Also, in Table 1, although the number of accessible parks is more than that of accessible mosques and local schools, both groups’ demand capacity (or the number of buildings with park access) is close to each other.

Table 1 Public service areas with accessibility and demand capacity within 300 m
Full size table
Fig. 6
figure 6

Non-park public areas and those with the highest demand capacity (see Table 2)

Full size image

Plans cannot allocate all accessible public areas as neighbourhood parks. To determine the best candidates for new locations with park features at a plan revision, we identified some non-park public places according to their higher demand capacity and geographical location, preferably in park-void areas on the Plan (Fig. 5). Table 2 and Fig. 6 indicate the top five non-park public places with the highest demand capacity, including two mosques with a yard (Location 48, Location 53), two “green areas” (Location 8, Location 50), and a school with a yard (Location 42).

Table 2 Top five non-park public service areas according to their demand capacity
Full size table

Except for the mosque at Location 48, all four locations stand in the park-void regions. As the final step for determining a plan revision, we run an LA analysis with these four locations and 16 parks (all 20 supply points). According to Fig. 7 and Table 3, the designation of four non-park places for park purposes can boost park accessibility from 30 to 59% of residential buildings.

Fig. 7
figure 7

Park accessibility in the plan revision

Full size image
Table 3 Comparison of the conditions for park accessibility on the site, the Plan and the revision
Full size table

Summarized in Table 3, the Plan triples the number of parks and doubles the total park coverage compared to the site’s conditions. Although still low, the park area per person triples. On the Plan, park accessibility inclines from 10 to 30%. However, with a reconsideration of only the planned public service areas, our plan revision increases these values, particularly to 59% for park accessibility. Also, the transformation of yards of a mosque and a school (Location 53 and Location 42) for park-related purposes can increase public greenspaces per person. Furthermore, Fig. 7 shows that the revision decreases the size of the park-void regions in the Plan.

Discussion

This paper contributes to the research on park accessibility as proposed by the urban plans, an understudied research area (Boulton et al. 2018; Gebhardt and Eagles 2014). It argues that as part of the greenspace provisioning, the plan evaluations should respond to the concerns about environmental justice, equitable allocation of greenspaces, and social inclusion (Boulton et al. 2018; Garcia and Garcia 2020; Gradinaru et al. 2023; Talen and Anselin 2021). Thus, plan evaluations making relationships between these concepts difficult to comprehend can contribute to mainstreaming equitable greenspace provisioning in strategic plan-making (Gradinaru et al. 2023).

When operationalising the task of interrelating these concepts to produce solid results for plan-making, we emphasise the need for developing more plan evaluations at the neighbourhood scale. Because, at small spatial scales, various social groups experience daily the socio-spatial conditions of inequity (Haaland and van Den Bosch 2015; Garcia-Garcia et al. 2020; Tan and Samsudin 2017), including park accessibility. Due to limited physical and socio-economic capacities, children, the elderly, low-income groups, and women with children tend to use often the areas nearby homes for daily needs (Day 2008; Fan et al. 2017; Forsyth et al. 2019; Loukaitou-Sideris and Sideris 2009). Also, these home-bounded groups usually afford shorter walking distances than healthy adults (Stafford and Baldwin 2018). Accordingly, when framing the neighbourhood scale, this investigation uses the home area defined by short walking distances from home rather than the official neighbourhood boundaries common in the literature.

Like a few research, this paper evaluates a plan and its potential for revision. These evaluations are developed as simulations supported by GIS-based LA analysis rather than with a checklist (Gebhardt and Eagles 2014; Lyles et al. 2016) or descriptive analysis (Niță et al. 2018; Talen and Anselin 2021; Byrne and Wolch 2009). Significant also for urban design research, these simulations focusing on the home area enable us to recognise certain physical and geographic features of the study site as the local determinants of walkability and measure actual walking distances shaped by these determinants rather than deploying Euclidean distances that ignore local variables. Furthermore, the LA analysis at the home area helps this study handle a contextual limitation (i.e., lack of demographic data about sub-sections of the neighbourhoods in Türkiye) and difficulty of measuring walkability concerning social differences simultaneously: This study reconfigures the degrees of spatial thresholds (including slope levels, rail lines, and wide roads) and walking distance according to their difficulty levels for children and the elderly. By involving the voice of disadvantaged groups about their needs and capacities (Gradinaru et al. 2023), plan evaluations and strategic plans can deploy more reconfigurations with this GIS-based tool and measure park accessibility with social inclusion even in data-limited contexts.

In this paper, by tripling the number of parks (total of 16), the Plan increases the total park area and park area per person but not equitable park accessibility. Within the identified spatial thresholds to walkability, more buildings (70%) lack park accessibility, whereas half of all 11 planned parks have no or low access to buildings within 300 m walking distance. In the neighbourhood, the distribution of parks creates some areas with and without park accessibility, both of which have physical hindrances to walking. Meanwhile, the choice of park locations may relate to the availability of public lands, a factor influencing park provision across multiple contexts (Boulton et al. 2018). Yet the Plan has more public lands (here, public service areas) than those assigned to 16 parks. Considering non-park public lands, our plan revision identified four areas whose development with park features can increase park accessibility from 30 to 59% of residential buildings, particularly in areas without park accessibility. All given, the Plan implementation can lead to underutilising some parks and a sense of social division within the neighbourhood. However, deploying LA analysis, plan evaluations and revisions at the small spatial scales can guide planners and policy-makers (thus, strategic plans) on decreasing uneven park accessibility by managing available public lands.

As another contribution to the literature, the revision stage shows that equitable park accessibility is an issue of also urban design. For instance, the four candidates for the plan revision include areas designated for “green areas,” a mosque and a school. The mosque and school are active today. Traditionally, a single-use zoning approach in plan evaluation may assign one kind of land use to each location, ignoring any potential for designing public facilities’ open spaces for diverse local needs. However, urban design decisions can redevelop these open spaces as flexible with new features and time schedules, improving local coverage and accessibility of public greenspaces. Besides, the use of yards of mosques and schools outside their planned functions by the elderly and children is compatible with the cultural characteristics of Türkiye. In this respect, plan evaluations and revisions at the neighbourhood scale must collaborate with urban design approaches and plans’ cultural context to address transforming such semi-public spaces into flexible public spaces.

As other candidates to be new parks, two “green areas” provide another case for a need for urban planning to collaborate with urban design in the plan evaluations. Overall, the Plan assigns “green areas” to a high number (38 out of 87) of public lands with a high accessibility level (or 44% of buildings). Two “green areas” are smaller (around 150 m2) than the mosques and the school but provide higher accessibility. The question of how to design small areas with high demand capacity (or the number of residential buildings with accessibility) as “good” public spaces (with park features) requires the contribution of urban design.

With a focus on daily life settings, such collaborations between these disciplines are also needed to improve strategic urban plans, regulations, and concepts concerning operationalising equitable accessibility to greenspaces with social inclusion, as presented by this case. For instance, in Türkiye, the Spatial Planning Regulation (dated 2014) does not mention the service and design features of “green areas” considered “passive green areas.” Including cemeteries, state forests and woods, forestation areas, and gardens of historical buildings, “passive green areas” in Türkiye are public greens with low or no public access due to their characteristics related to preservation, aesthetic, economic, and so on. Interestingly, despite this lack of legal description and limited public lands for public facilities, the Plan designates more locations for “green areas.” To observe any motivation driving this outcome, we examined the locations of “green spaces.” The Plan sets “green areas” on hard-to-access public lands, including those with steep slopes and next to major roads, rail lines, and water streams. However, “green areas” in neighbourhood areas without park accessibility lack such locational characteristics. Thus, for better explanations, further research should include the reasoning of “greenspace managers,” an understudied area in various contexts (Boulton et al. 2018, p. 99), about designating “green spaces.” Also, it should include how planners interpret legal and popular notions about greenspaces (Garcia-Garcia et al. 2020; Gradinaru et al. 2023) and, in Türkiye, public access to “green spaces” and “passive green areas.” In other words, research about plan evaluations should simultaneously address plan outcomes and planning processes.

Conclusion

Given the commonality of limited public land for parks and a call for developing urban greens as part of integrated systems in cities with new urbanisation challenges, research about how to achieve equitable park accessibility is necessary to guide and improve urban planning and design practices. Concerned about operationalising the interrelations between environmental justice, equitable allocation of greenspaces, and social inclusion, this research at the neighbourhood level develops a simulation of the selected site, its plan, and the revision with the assistance of LA analysis. Its stages of analysis can guide local governments when revising plans at small spatial scales, which can contribute to mainstreaming equitable greenspace provisioning in strategic plan-making.

This study also calls for changing the obscured legal notions in greenspace planning and urban design and land management research to get involved in the plan evaluations about park accessibility. The analysis illustrates how urban design too can deploy GIS-based LA analysis to re-identify and assess physical and geographical features for walkability according to some disadvantaged groups (e.g., children and the elderly). Such detailed characteristics of the built environment are usually separate from the plan evaluations at the city and the neighbourhood scales. The results of the revision stage show that urban design-based solutions, rather than a single zoning approach, are necessary for reconsidering how to use available public lands to respond to diverse needs for greenspaces, especially in dense urban textures.

Study limitations

As its study site, the paper analysed only a part of a plan and the spatial distribution of public service areas. The whole document and analysis of multiple urban systems are out of the study scope. We could not develop the spatial distribution of demographic data within the neighbourhoods because Türkiye officially has no such data for small spatial scales. We did not consider any attributes of parks’ design, size, maintenance, and safety. Also, because the default setting of the LA analysis allows one criterion, we kept the minimum distance.