You are on page 1of 14

Urban Forestry & Urban Greening 80 (2023) 127827

Contents lists available at ScienceDirect

Urban Forestry & Urban Greening


journal homepage: www.elsevier.com/locate/ufug

Original article

Assessment of spatial equity of urban park distribution from the perspective


of supply-demand interactions☆
Jingyuan Zhang a, *, Puay Yok Tan b, **
a
School of Architecture, Harbin Institute of Technology (Shenzhen), 518055 Shenzhen, China
b
Department of Architecture, 4 Architecture Drive, National University of Singapore, 117566, Singapore

A R T I C L E I N F O A B S T R A C T

Keywords: Urban parks have been widely recognized for the critical functions they provide in enhancing health and
Spatial equity wellbeing. However, parks are known to be inadequately provided and non-uniformly distributed in cities,
Supply-Demand interactions leading to an increasing interest to examine spatial equity of park distribution. Such studies have mostly focused
Most often used parks
on the distributional attributes of park supply in determining spatial equity, while the role of residents’ demand
Perceptional demand
is usually ignored or not adequately assessed. In addition, people’s preferences for park use were normally
Multi-scale analysis
ignored and nearest parks were assumed as supplier of park services. Moreover, even though equity assessment is
scale-dependent, few studies adopt a multi-scale assessment. We report on a study to address these gaps. A multi-
scale Supply-Demand Equity Index (SDEI) was developed to assess spatial equity of park distribution from the
perspective of supply and demand interactions, and data were collected at a fine scale of neighborhood level.
Residents’ accessibility to their most often used parks and nearest parks were used to assess and compare supply
level; while residents’ perceptions obtained from surveys and population density were used to assess and
compare demand level. Results showed that supply measured by accessibility to most often used park more
accurately reflects supply, and demand assessment on the basis of perception could reflect variations among
different social groups and reveal otherwise hidden patterns of spatial inequity. It was also shown that SDEI
could identify areas where mismatch between supply and demand occurs. We demonstrated in a single study that
different definitions and approaches of quantifying spatial equity can lead to distinct conclusions, therein
highlighting that the complex nature of equity requires nuanced approaches and providing a broader view of
equity assessment. In addition, the implications of this study for urban park planning and management were
highlighted.

1. Introduction (Barbosa et al., 2007; Nicholls, 2001; Oh and Jeong, 2007; Tan and
Samsudin, 2017; Zhou and Kim, 2013). In these studies, equity of park
Parks have long been accepted as significant public spaces in urban distribution is studied in two main ways: first, assessment of park
history, serving as central points of recreation and social interaction, accessibility and provision use indicators such as distance or time for
promoting health and well-being, as well as providing a range of envi­ reaching parks, per capita park area, park service area, etc., in relation to
ronmental benefits (Byrne and Wolch, 2009; Kim and Jin, 2018; Ward amount received by different socioeconomic, demographic, or racial
Thompson, 2011; J. R. Wolch et al., 2014). For the community it serves, groups (Dai, 2011; Giles-Corti and Donovan, 2002; Sister et al., 2010;
the social and ecological benefits of a park system are largely influenced Zhou and Kim, 2013); second, assessment of these indicators in relation
by the amount and distribution of parks (Bedimo-Rung et al., 2005; J. R. to the amount present within different regions or subdivisions within an
Wolch et al., 2014). Increasingly, studies across a range of cities area of interest, from a spatial viewpoint (Neutens et al., 2010; Oh and
worldwide highlight that distribution of parks are often suboptimal Jeong, 2007; Tan and Samsudin, 2017). The commonality between these


This research was founded by the National Natural Science Foundation of China (No. 52108048), Shenzhen Excellent Technology Creative Talents Cultivation
Foundation (No. RCBS20210609104539077), and Shenzhen Imported High-level Talents Research Foundation (No. FB11409008).
* Correspondence to: School of Architecture, Harbin Institute of Technology (Shenzhen), China.
** Correspondence to: Department of Architecture, National University of Singapore, Singapore.
E-mail addresses: jingyuanz@u.nus.edu (J. Zhang), puay.yok.tan@nus.edu.sg (P.Y. Tan).

https://doi.org/10.1016/j.ufug.2022.127827
Received 19 August 2022; Received in revised form 27 November 2022; Accepted 30 December 2022
Available online 31 December 2022
1618-8667/© 2022 Elsevier GmbH. All rights reserved.
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

types of studies is that they rely on disparities in park services among either population density or socioeconomic characteristics. Therefore,
either different social groups or spatial units to assess the extent of residents’ actual demand level measured at a finer scale based on per­
equity. ceptions should be incorporated into equity analysis, and doing so ne­
We suggest that such assessments provide a useful, but incomplete cessitates conceptual and methodological approaches for demand to be
view of equity assessment. The primary limitation is that adequacy of quantified. This also would advance the assessment of spatial equity,
parks provides only one perspective—the “supply” of parks. Equity since residents’ perceptions, demand and satisfactions have become
however, is a concept that embeds additional components over and important indicators for the level of reasonableness of park distribution
beyond the supply of parks. Equity refers to the quality of being equal or (Larkin et al., 2021; Y. Lin et al., 2021; Lotfi and Koohsari, 2009).
fair in a situation or distribution, which would affect individual well- Apart from the lack of methods for measurement of demand, there
being (Alonso, 1968; Deutsch, 1975). Spatial equity of parks implies a are also additional challenges associated with spatial equity assessment.
social or political consensus about the fairness or justice of the spatial First, a commonly held assumption is that people would use nearest
distribution of costs and benefits of parks (Dear, 1978; Truelove, 1993). parks irrespective of individual preferences, see (Barbosa et al., 2007; B.
Equity does not mean equal shares of public goods for all, but rather the B. Lin et al., 2014; Nicholls, 2001; Oh and Jeong, 2007; Sister et al.,
supply of goods in variable proportions to different people with different 2010). However, residents’ most often used parks are not necessarily the
demand (Bennett, 1980), i.e., whether the distribution of parks is closest ones, which may not be favored and used by residents. In addi­
effective and optimal for satisfying residents’ demand (Nicholls, 2001). tion, the distinction between nearest and most often used parks is usu­
In practice, accessibility to parks has been widely used as an indicator of ally not considered in park equity studies, and the adoption of nearest
equity (Nicholls, 2001; Oh and Jeong, 2007; Talen, 1998; Xiao et al., park as recreational destinations would result in oversimplification of
2017). Accessibility to parks refers to the ease with which a park can be park supply level, which would further undermine the accuracy of eq­
reached (C. Liu et al., 2010; Pirie, 1979). Equity on the other hand as uity analysis. Second, spatial scale is also a key issue that needs to be
suggested, addresses whether there exists disparity between supply and considered. The critical role of scale in influencing analysis results is
demand (Lee and Hong, 2013). The one-dimensional focus on supply well-recognized within the discourse and practice of spatial equity
disparity therefore does not provide a complete view of equity. With the analysis (Omer, 2006). However, most studies have been limited to the
development of theories and practices in human-oriented urban plan­ use of large-scale aggregated data, such as census tract or planning area
ning, the importance of “demand” in influencing the assessment of level data (Omer, 2006; Zhou and Kim, 2013), a point also emphasized
urban park equity has gradually become considered and incorporated in by Tan and Samsudin (2017). Data at smaller spatial scales such as at the
recent years (Z. Liu et al., 2021; Shao and Luo, 2022). Therefore, we neighborhood level is more limited, and the specific ways in which eq­
suggest that the assessment of spatial equity should focus on the balance uity manifests in relation to scales of analysis are still relatively un­
between accessibility to parks, which represents “supply” of parks, and known. In addition, while the spatial quantification of whether supply
the extent to which these are actually needed by residents whom they meets demand at a fine scale is increasingly recognized as an important
are meant to serve, i.e., the “demand” for parks. The spatial match of topic in the area of urban ecosystem services assessment, approaches
supply and demand of park resources is then a critical indicator in the and empirical assessment of this area for urban parks as an important
assessment of adequacy and reasonableness of park distribution. source of cultural services are scarce. Therefore, a method for equity
Although clearly important, demand is much less addressed in assessment that could be applied to multi-scale levels would contribute
studies on spatial equity of parks. By definition, demand refers to peo­ to the illustration of equity in a fuller picture.
ple’s behavioral intentions and motivations, and indicates how much an The aim of this paper is to assess the application of a framework for
individual is willing to try, and how much effort a person is willing to equity analysis of urban parks, that considers both park supply and
exert, for the aim of performing a specific behavior (Ajzen, 1991). demand for parks in concert using fine-scale data. The objectives of this
However, demand for urban parks has yet been fully explored and study are to: (1) develop and apply a framework and index for equity
adequately measured in literature (Peña et al., 2015). The majority of analysis that integrates park supply and residents’ park use demand
previous studies have either ignored the measurement of demand, or simultaneously, using data from fine-scale neighborhood level, and
only relied on desktop measurements. In the latter, demand is usually assess the application of the equity index across multiple scales; (2)
assessed using proxies such as population, population density (Lee and compare differences in supply and demand assessment arising from the
Hong, 2013; B. Liu et al., 2022), ratio of population to park service area use of accessibility to most often used parks compared to nearest parks
(Sister et al., 2010), age group coefficient (Y. Lin et al., 2021), combi­ as park supply, as well as residents’ perceptional demand compared to
nations of population and other indicators (Cortinovis and Geneletti, population density as park use demand; (3) assess the spatial pattern of
2020; Y. Lin et al., 2021), etc. In these studies, the implicit assumption supply, demand, and resulting equity level of urban parks in four lo­
for using population as a proxy is that different groups of residents have calities in Singapore, and discuss its implications for planning and
a uniform demand (B. Liu et al., 2022; Shao and Luo, 2022), and the management.
level of demand of a certain region is mainly determined by the quantity
or density of residents. The incorporation of social group coefficient is an 2. Methodology
improvement and considers group stratification. Even in recent litera­
ture, demand was still coarsely assessed by using population density (B. 2.1. Overview of key components and framework
Liu et al., 2022). However, population size or social group coefficient as
a single indicator of demand is one-dimensional which is useful in An overview of combination of assessing park supply and demand in
assessing demand for ecosystem services, but has limitations in equity analysis is shown in Fig. 1, which in essence, addresses equity as
revealing differences in demand among population and in providing the matching level of residents’ demand for using parks and the supply
useful information on residents’ actual willingness for park use. In of parks. We conducted this assessment at two scales: subzone and
addition, these traditional proxies of demand are usually based on planning area, and data were collected at the household level. Subzone
aggregated geographical units, and fine-scale data from smaller units and planning area are areal units used by Singapore Urban Redevelop­
like neighborhood level were seldom used, which inevitably under­ ment Authority. Subzone is the smallest areal unit, which is usually
mined the accuracy of demand assessment. Demand in the realm of centered on a neighborhood center or activity node. Each planning area
recreational services provided by urban parks, in essence is a variable is composed of several subzones, and typically represents a residential
that is associated with perceptional data and cannot be simply assumed town, with a population of around 150,000 (Urban Redevelopment
to be uniform across the population or certain social groups; hence a Authority, 2014).
direct assessment of demand is more accurate than proxies derived from In our study supply is defined as the recreational potential provided

2
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 1. Conceptual framework of spatial equity assessment integrating supply and demand.

by parks, and was measured by accessibility from respondents’ home to spatial equity assessment on park distribution is less studied in
their most often used parks and nearest parks respectively, for high-density Asian cities, with the majority of past studies in U.S. or
comparative analysis. In accessibility assessment, researchers have al­ European cities (Crompton and Wicks, 1988; Neutens et al., 2010;
ways tried to incorporate the positive influence of quality and the Nicholls, 2001; Talen and Anselin, 1998; Wen et al., 2013; J. Wolch
negative influence of distance into the measurement of accessibility, et al., 2005; X. Zhang et al., 2011; Zhou and Kim, 2013). As a widely held
such as gravity potential (Luo and Wang, 2003) and 2-step floating model of compact city (Tan et al., 2013), the approaches used and results
catchment area method (Shao and Luo, 2022). These two methods were of this study could hold useful lessons for rapid urbanizing cities in East
not used in this study since the influence of park area on accessibility and Southeast Asia.
would be exaggerated in fine-scale analysis, and even hinder the dif­ In this study, household level data were collected through door-to-
ferences in distance. In addition, our study design requires the com­ door survey in four planning areas: Choa Chu Kang, Seng Kang, Bukit
parison between accessibility to nearest and most often used park, while Batok, and Pasir Ris. The selection of study sites was based on case study
2-step floating catchment area method inevitably calculates accessibility logic and the analysis of cases adopts a logical inference approach
to several surrounding parks. Data on respondents’ most often used (Berg-Schlosser et al., 2009; Ragin and Becker, 1992; Small, 2009). An
parks were obtained through household surveys, while nearest parks important principle for selecting multiple cases is that each subsequent
were selected based on network distance, using ArcGIS 10.2 (ESRI, case is expected to replicate the prior one under two different ap­
Redlands, CA, United States). As for the specific measurement of proaches. The first approach is literal replication, through which a
accessibility, network distance was adopted since it most closely ap­ similar case is selected to compare whether similar mechanism and re­
proximates the actual walking distance from residents’ home to park, sults are present; the second approach is theoretical replication, in
with a clear connotation of accessibility, and could be easily used for which a different case is chosen to investigate if the presumed con­
comparative analysis among different study sites. trasting mechanism and results would be found (Small, 2009). There­
Demand is defined as residents’ behavioral intentions to use parks for fore, Choa Chu Kang and Seng Kang both of which have relatively low
recreational services provided by urban parks, and it shows an in­ park provision (park area ratio is 10.0% and 9.1% respectively) and high
dividual’s attitudinal commitment or desire towards using parks. In population density (663 and 465 capita/ha respectively) were selected,
other words, demand signifies how much residents need to use or and per capita park area is 1.5 and 2.0 square meters respectively. In
dependent on parks as a source of recreation. Demand was measured in contrast, Bukit Batok and Pasir Ris were selected as they have relatively
two approaches in this study for comparative analysis: calculating high park provision (park area ratio is 21.9% and 24.1% respectively)
population density and obtaining residents’ perceptional demand level and low population density (356 and 316 capita/ha respectively), and
for park use through questionnaire in a Likert scale, ranging from 1 (very per capita park area is 6.2 and 7.6 square meters respectively.
low demand) to 5 (high demand). The grading assessment was adopted Within each study site, stratified random sampling was used to
as it is a widely adopted measurement that can provide an assessment of ensure that sampled residents reflect a range of socio-demographic and
demand. For instance, Zhang and Tan (2019) and Wang et al., (2015) socio-economic groups, and thus be representative of the whole popu­
used a Likert scale (1− 5) to measure demand for parks; Peña et al. lation of study sites. Target respondents were adults aged above 21 years
(2015) and H. Zhang et al., (2021) used questionnaire survey to collect old, which is the official adult age in Singapore. One eligible respondent
demand data on recreational services. In order to make it clear for re­ was randomly selected from each sampled household, using the “last
spondents to understand what demand is, we added explanations for the birthday method” (person whose birthday is closest to the day of sur­
question of park use demand, and set a uniform standard to make it vey). If no one in the selected household agreed to take the survey, a
comparable for different responds (details in Appendix I of Supple­ replacement household two units away from the original one was used.
mentary Material). Perceptional demand obtained from each respondent
at household scale was aggregated at subzone and planning area scale 2.2.2. Household survey and spatial data collection
for subsequent analysis. Household level data were collected using door-to-door question­
naire survey between February and June in 2016 (approved by Insti­
2.2. Case study design and data collection tutional Review Board, No. A-15–234). Information on residents’ socio-
demographic and socio-economic attributes, home locations, name of
2.2.1. Site selection and sampling process most often used parks, park use demand levels, etc. were collected, and
Singapore was selected as the study site for several reasons. First, the questionnaire used was provided in Supplementary Material. We
Singapore as a “Garden City” has dedicated significant resources to park sampled 600 households and collected 597 valid questionnaires,
supply, planning, and design (Tan et al., 2013), but there is no localized comprising 167 from Choa Chu Kang, 145 from Seng Kang, 142 from
study on whether the level of park supply match with demand. Second, Bukit Batok, and 143 from Pasir Ris (Fig. 2). The reliability assessed

3
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 2. Spatial distribution of respondents in four study sites.

using Cronbach’s alpha showed that internal consistency was satisfied Supply-Demand Equity Indexes (SDEI) were developed. Subzone scale
(α = 0.85 >0.70). The proportions of surveyed samples from different equity assessment aggregates the congruence or mismatch between park
dwelling types (HDB, condominium, and landed property), racial groups supply and demand, thus could be used to identify under or ‘over’
(Chinese, Malay, Indian, and Others), and gender types (male, female) supplied subzones. Functions (1− 3) are used to derive SDEI at the
were largely proportional to the composition of the whole population in subzone level. Specifically, park supply proxied by accessibility to
each study site, suggesting that our samples represent the population nearest park and park demand proxied by population density were used
well (Table 1). to calculate SDEII (Function 1); accessibility to most often used park in
For spatial data, the boundaries of planning areas and subzones, combination with population density were used to calculate SDEIII
residential land use, geographic data on urban parks and park connec­ (Function 2); accessibility to most often used park in combination with
tors were obtained from Urban Redevelopment Authority, while build­ perceptional demand were used to calculate SDEIIII (Function 3). A value
ing footprints and road network data were from Singapore Land of SDEI close to 1 indicates a higher level of spatial equity, otherwise
Authority. The specific locations of park entrances of all parks in study (SDEI close to 0 or larger than 1) a lower level of spatial equity is
sites were mapped during fieldwork. Road network data were refined by indicated, i.e., mismatch between supply and demand was identified.
adding detailed footpaths and small service roads, and were used for the To aggregate demand data in subzone and planning area, demand
calculation of network distance. proxied by population density (i.e. Djs and Dis) was directly used to
represent the overall level of demand; demand proxied by perceptional
2.3. Formulation of supply-demand equity index levels (i.e. Djp and Dip) was aggregated by the summation of demand
value of all the samples in subzone (1 to k) and planning area (1 to n).
On the basis of our proposed conceptual framework, multi-scale Supply data were also aggregated accordingly, using the average value

Table 1
The proportions of dwelling types, racial groups, and gender types in samples and whole population of each study site.
Study site Comparison of proportion Dwelling type (%) Racial group (%) Gender type (%)

HDB Condominium Landed properties Chinese Malays Indians Others Male Female

Choa Chu Kang Sample 88.0 10.8 1.2 68.9 14.4 10.8 6.0 46.1 53.9
Population 88.3 10.4 1.3 71.4 16.8 9.1 2.7 50.1 49.9
Seng Kang Sample 88.3 9.7 2.1 75.2 11.7 10.3 2.8 51.0 49.0
Population 92.2 7.2 0.6 79.0 10.2 8.3 2.5 49.5 50.5
Bukit Batok Sample 73.9 22.5 3.5 69.0 10.6 14.1 6.3 45.8 54.2
Population 71.4 24.7 3.9 72.3 14.2 10.6 2.9 49.5 50.5
Pasir Ris Sample 72.0 23.1 4.9 63.6 16.8 14.0 5.6 54.5 45.5
Population 71.5 24.1 4.4 67.3 20.3 8.3 4.1 49.7 50.3

4
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

of accessibility (Function (1− 2)) and the summation of accessibility ⃒


⃒ ∑k


(Function 3). Thereafter, we used division operation to compare the ⃒
⃒ A ⃒
Djs × Al ⃒⃒
jm
1 ∑m ⃒ j=1
differences between supply and demand in each subzone, and used the SDEI pII = × ⃒ ∑n − ⃒ ∈ [0, 1] (6)
2 l=1 ⃒ A Dis × Am ⃒
level of mismatches between supply and demand in planning area as a ⃒ i=1 im ⃒
⃒ ⃒
baseline. The main reason accounting for the difference in the units of
supply and demand is that they essentially capture different aspects of ⃒ ⃒
⃒ ∑
k ∑k ⃒
the park system, i.e., supply reflects the physical capacity of recreational ⃒ Ajm Djp ⃒⃒

1 ∑m ⃒ j=1
resources, which should be assessed by amount or distance; demand SDEI pIII = × ⃒∑n
j=1
− ∑n

⃒ ∈ [0, 1] (7)
reflects social capacity of recreational resources expressed by users, 2 l=1 ⃒ A
⃒ i=1 im

i=1 Dip ⃒

which thus has to be assessed by scale levels or social group composi­ ⃒ ⃒


tions. Therefore, the combined consideration of supply and demand
addresses both the physical and social aspects of park system. Although where l refters to lth subzone; m is the total number of subzones in the
the units for assessing supply and demand are different, they both could whole planning area; Al is the area of lth subzone; Am is the area of the
be used to illustrate whether the level of supply and demand is high, whole planning area.
medium, or low, so this does not undermine the reasonableness of their
interactions and the calculation of SDEI. In addition, this difference in 3. Results
units was offset by the division operation, resulting in a mathematical
ratio that is used to represent the equity level. Similarly, Tardieu and 3.1. Comparison of different approaches to measuring park supply
Tuffery (2019) used geometric mean and B. Liu et al. (2022) used di­
vision to compute the combined effect of supply and demand. In previous studies, supply was normally measured by accessibility to
/ nearest parks, even though residents may not necessarily use the nearest
∑k
park. Our results lend credence to this observation. As shown in Fig. 3,
Ajn k/∑ Ain /n
n

j=1 there were more than 85% of residents can reach nearest parks within
SDEI I = i=1
(1) 400 m′ Euclidean distance range in four study sites, and 100% of resi­
Djs Dis
dents live within 800 m to nearest parks, indicating a high level of park

k / supply. However, when Euclidean distance was replaced with network

n
Ajm k/ Aim /n distance, these percentages were reduced to around 60–70% and 96%,
(2) respectively. More notably, if accessibility was measured by network
j=1 i=1
SDEI II =
Djs Dis
distance to most often used parks, the proportions of accessible residents
decreased further to 30–50% and 50–75%, respectively, indicating
∑ medium to low supply level. In addition, regarding the transportation
k
∑n
Ajm / Aim
SDEI III
j=1
= k i=1
∑n (3) means for accessing most often used park, 75.2% of the respondents
∑ i=1 Dip chose walking, while using private car, MRT & Bus, and cycling have
Djp
j=1 almost equal shares, i.e., 7.6%, 7.6%, and 7.3%, respectively. In
contrast, jogging and others are the least favorite transportation means
where Ajn and Ajm are the accessibility level (i.e. reciprocal of network for reaching parks (1.9% and 0.4%). This finding shows that people
distance, to transform distance to the level of accessibility) of jth sample prefer to walking to their most often used parks.
in subzone l to nearest park and most often used park, respectively; k is To further illustrate the importance of incorporating park users’
the total number of samples in subzone l; Djs is population density of preferences, the differences in residents’ destination choices between
subzone l; Djp is perceptional demand of jth sample in subzone l; Ain, Aim, nearest park and most often used park were assessed (Table 2), using
Dis, Dip, and n have the same meaning with aforementioned pairwise study site Bukit Batok as an example, and the distribution of parks in
items but the calculation region is changed to the whole planning area. Bukit Batok are shown in Fig. 4. There were general mismatches be­
Equity analysis at planning area scale was also performed to address tween the percentage of respondents living adjacent to a specific park
whether each subzone within this area enjoys the same equity level. The (A) and the percentage of respondents using it most often (B). It is worth
concept of residential segregation between racial groups was extended noting that for Beauty Garden Park and Park #251, the differences were
to assess the spatial disparity between supply and demand. As shown in 24% and 10.6%, respectively. The common attribute for these two parks
Function (4), Duncan and Duncan (1955) proposed a dissimilarity index is that they are both tiny neighborhood parks (<1 ha) with limited
D to assess the degree of residential segregation between Black and public spaces and facilities in a high-density residential area. More
White residents in a certain region. We modified this index, substituting importantly, there is a large attractive regional park Bukit Batok Town
population of different racial groups with supply and demand and Park near these two parks (<1 km), which is naturally preferred and
deriving three sets of SDEI at planning area scale, i.e. SDEIpI, SDEIpII, and chosen by the majority of residents as their most often park. This finding
SDEIpIII, ranging from 0 to 1, with 0 representing ideal equity, while 1 suggests that although there are relatively high population residing near
indicates extreme inequity (Functions 5–7). Higher values of SDEIp these two parks, only a few choose to use them regularly, and thus
indicate lower equity level and should provoke planning concerns. signifies the importance of considering residents’ preferences on parks.
⃒ ⃒ In contrast, for Bukit Batok Nature Park and Bukit Batok Town Park, the
1 ∑n ⃒⃒bi wi ⃒⃒
D= × ⃒ − ∈ [0, 1] (4) discrepancy is − 35.2% and − 5.6%, respectively, which indicates that
2 i=1 B W⃒
there were far more residents attracted by these large parks apart from
where bi and wi represent population of Black and White residents in ith users from surrounding neighborhoods. The main reason accounted for
area; B and W are the total population of these two racial groups in the this notable difference is that these two parks are both attractive
whole area. regional parks. Our results showed that the choice of destination dataset
⃒ ⃒ would have an impact on the measurement of park supply, and acces­
⃒ ∑

k ⃒
⃒ sibility to most often used parks would reflect supply level in a more
⃒ Ajn
1 ∑m ⃒ j=1 Djs × Al ⃒⃒ accurate manner.
SDEI pI = × ⃒∑n
l=1 ⃒
− ⃒ ∈ [0, 1] (5)
2 A Dis × Am ⃒
⃒ i=1 in ⃒
⃒ ⃒

5
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 3. Comparison of park supply assessment based on different approaches.

Table 2
Differences in destination choice between nearest park and most often used park (using Bukit Batok as a sample).
Park Name Park Area Percentage of respondents living adjacent to it Percentage of respondents using it most often Percentage of difference (%)
(m2) (%) (A) (%) (B) (A-B)

Bukit Batok Central Park 23,686.4 6.3 8.5 -2.2


Bukit Batok Playground 26,182 9.9 2.8 7.1
Park #180 21,589 7.7 3.5 4.2
Bukit Batok Town Park 413,765 15.5 21.1 -5.6
Jalan Dernawan 1935.29 0.7 0 0.7
Playground
Lorong Kemunchup Park 1396.83 7.0 0 7
Chu Yen Playground 2856.57 0.7 0 0.7
Hillview Crescent 4187.42 2.8 0 2.8
Playground
Bukit Batok Nature Park 352,891 6.3 41.5 -35.2
Jalan Remaja Interim Park 954.25 0.7 0 0.7
Burgundy Crescent 2707.76 1.4 0 1.4
Playground
Pavilion Playground 4342.06 0.7 0 0.7
Phoenix Heights 3772.95 2.1 2.1 0
Playground
Park #251 7016.10 12.7 2.1 10.6
Beauty Garden Park 3483.75 25.4 1.4 24

3.2. Comparison of park demand across population and condominiums, one possible reason is that there are normally few
comprehensive regional parks located near landed properties. In addi­
Demand proxied by population density assumes that people have tion to sociodemographic attributes, two attitude and behavior related
uniform demand level across the population(Shao and Luo, 2022), and groups were also included in comparison. Specifically, respondents who
the differences of demand among geographical units are solely deter­ chose current house for being closer to parks reported higher park use
mined by the aggregation of population. However, our results showed demand, and respondents who used park last week also reported higher
that there exist group variations among residents regarding perceptional demand level. In contrast, perceptional demand did not show significant
demand. What is interesting about the cross-group comparison is that differences among socioeconomic groups like gender, education level,
perceptional demand levels were statistically and significantly different working status, individual and household monthly income, indicating
in population with different ethnic attributes (p = 0.004), family that socioeconomic status did not constrain residents’ inner demand for
structure (p = 0.005), type of dwelling (p = 0.035), reasons for moving using parks. It is also worth noting that although residents from different
(p = 0.000), and whether used park last week (p = 0.000) (Kruskal-­ age groups have statistically similar demand, people aged 25–34 showed
Wallis H Test was used for variables with multi-groups; Mann-Whitney U lowest average demand compared with other groups.
Test was used for variables with only two groups; the details about
groups were provided in Supplementary Material). Specifically, for
3.3. Spatial pattern of park supply and demand
ethnic attributes, pairwise comparisons showed that Indians express
significantly higher demand than Chinese. This finding confirms our
From the comparative analysis conducted above, supply assessed by
hypothesis that people with different ethnic characteristics would have
network distance to most often used parks is more accurate and closer to
different demand levels, which is also in agreement with the findings of
the actual level of park service attributed to residents. Similarly, demand
previous studies (Dwyer, 1994; Payne et al., 2002). Families with chil­
measured by perceptional demand could better reflect demand varia­
dren under 21 years old (official adult age in Singapore) expressed
tions among the population in a more precise manner than population
stronger demand than families without children, which is consistent
density (the specific data for supply and demand measured through
with the commonly held assumption that children are generally more
different approaches as well as location of subzones were provided in
eager to use parks. It is worth noting that residents from landed prop­
Appendix IV in Supplementary Material). In order to better illustrate the
erties expressed higher demand for using parks than others from HDB
spatial variations of park supply and demand measured through

6
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 4. Park distribution in study site Bukit Batok.

aforementioned approaches, the spatial patterns of them in four study SDEI value. SDEI close to 0 indicates a lower level of spatial equity and a
sites are shown in Figs. 5 and 6, respectively. Less spatial variations series of yellow to brown color was used to show that demand exceeds
among subzones were observed when supply was calculated by acces­ supply, i.e., undersupplied areas (low supply-high demand). SDEI that is
sibility to nearest parks, and residents can easily reach a park within larger than 1 indicates that supply exceeds demand, i.e., oversupplied
800 m′ walking distance (Fig. 5). However, when the destinations were areas (high supply-low demand), with green and dark green used to
replaced with most often used parks, there is at least one subzone has illustrate this oversupplied situation. SDEII, SDEIII, and SDEIIII led to
low supply level (network distance > 800 m), and the proportion of different spatial patterns due to different definitions and methods for
subzones with high supply level was largely decreased. This supports our assessing equity. Spatial equity revealed by SDEIII generally is more
earlier analysis that accessibility to nearest parks would overestimate diverse than by SDEII, while SDEIIII derived mixed-level of spatial
supply and mask spatial variations. variation. These inconsistencies reinforced our proposition that fine-
The spatial distribution of park demand showed different patterns scale data on park supply and demand used in SDEIII and SDEIIII
when it was measured in different approaches. It is observed that several would reveal hidden patterns, while general proxies like accessibility to
subzones with high population density exhibited low perceptional park nearest parks and population density used in SDEII tends to show
use demand, and vice versa (Fig. 6). For instance, the subzone located at oversimplified spatial equity assessment. The underlying reason
the eastern part of Seng Kang was identified with high population accounted for the mixed-level equity revealed by SDEIIII might be that
density but low demand level. It is counterintuitive but if we analyze in the disparity level of demand among subzones is different across plan­
conjunction with supply, we may infer that residents’ low demand might ning areas, which would be further discussed in the following section.
be influenced by the low supply level of this area (averagely 950 m). For the planning area scale, SDEIp was calculated for each study site
This finding supports our proposition that aggregation of population based on Function (5− 7). The major difference between SDEIp and SDEI
would not necessarily be transformed to equal portions of demand for is that SDEIp captures the general spatial equity level from a synthesized
using parks. It also indicates that demand might be influenced by other perspective, and transfers the local disparities as shown in Fig. 7 to an
factors. We also found that people residing in subzones close to regional overall assessment. The strength of SDEIp is that it is able to demonstrate
parks, such as in the western subzones of Seng Kang and Pasir Ris, tend the overall equity level of a region intuitively, hence it is especially
to have medium to high perceptional demand, which might be rein­ useful when comparative analysis is required. The value of SDEIp was
forced by their interactions with parks. calculated for four study sites (Table 3), ranging from 0 to 1, with
0 representing ideal equity, while 1 indicates extreme inequity. In
comparing the data in columns of Table 3, we showed that for the same
3.4. Spatial equity assessment at the multiple scales planning area, the equity level revealed by SDEIpI and SDEIpII is
different, and SDEIpIII indicated mixed-level of spatial equity. Another
At the subzone scale, SDEI was calculated for each subzone in four observation is that SDEIpI measured by traditional indicators would
study sites using different approaches to measuring park supply and overestimate equity level. For instance, the equity level of Choa Chu
demand (Function (1− 3)), as with the ones used in Section 3.3. The Kang actually is the lowest as indicated by SDEIpII and SDEIpIII, but it
spatial variation of equity value of subzones was illustrated in Fig. 7 (the ranked as the second highest shown by SDEIpI. As for the comparison of
specific data for spatial equity measured through different approaches as data in rows, it is shown that Pasir Ris generally has the highest level of
well as location of subzones were provided in Supplementary Material). spatial equity, Choa Chu Kang ranks as the lowest, Seng Kang and Bukit
A value of SDEI that is close to 1 indicates a higher level of spatial equity, Batok showed mid-level spatial equity. We showed that although Pasir
and light green was used to illustrate the spatial match of high supply Ris and Bukit Batok both have high park supply level, equity levels are
and high demand within the range of (0.8, 1.2], while light yellow in­ different. This similarly applied to the low park supply level group of
dicates low supply accompanied by low demand with the same range of

7
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 5. Spatial pattern of park supply proxied by accessibility to nearest park (upper) and most often used park (lower).

8
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 6. Spatial pattern of park demand proxied by population (upper) and perceptional demand (lower).

9
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Fig. 7. Spatial pattern of equity measured by SDEII, SDEIII, and SDEIIII.

Choa Chu Kang and Seng Kang. Overall, these results showed that 4. Discussion
supply-demand interactions presented a different assessment of park
equity than supply, or demand alone, affirming the use of SDEIp as a 4.1. A more accurate approach to measuring park supply and demand
quantitative tool to assess overall equity level at this scale.
A key contribution of this study is that it demonstrates conceptually
and empirically that the integration of park supply and demand vari­
ables provided a more nuanced picture of park equity. On the supply
side, residents’ accessibility to the most often used parks avoided the
assumption that people would use the nearest parks, which tends to

10
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Table 3
Value of SDEIp in three approaches for four study sites (planning area scale).
Study sites SDEIpI SDEIpII SDEIpIII Average supply (accessibility Average supply (most Average demand (population Average demand (perceptional
to nearest park; m) often used park; m) density; 1000 person/m2) demand; Likert scale)

Choa Chu 0.149 0.328 0.244 308.82 564.29 32.916 3.33


Kang
Seng Kang 0.296 0.184 0.062 349.41 720.67 27.251 3.22
Bukit 0.203 0.167 0.225 330.71 894.28 20.189 3.26
Batok
Pasir Ris 0.083 0.123 0.186 309.62 781.88 22.088 3.32

overestimate actual park supply and equity level. The observed To further compare the different approaches to measuring park
mismatch between nearest park and most often used park (Fig. 3) im­ supply and demand, we conducted comparative analyses using three
plies that different approaches to measuring supply would lead to types of SDEI, and the results further highlighted the influence of mea­
different even contrasting understandings of park supply. In addition, surement indicators and the importance of incorporating this more
the comparison between differences in destination choice between nuanced approach. Specifically, our comparative analysis showed that
nearest park and most often used park (Table 2) helps us to understand when supply was assessed by accessibility to most often used parks and
residents’ decision-making process. Specifically, based on data from demand was assessed by perceptional demand, some new information
questionnaire survey and interview, it is implied that distance is only regarding spatial variation is shown, which would contribute to the
one of the influential factors of park usage, while other factors also exert investigation of actual situation and underlying reasons of inequity.
an effect, like park area, aesthetic value, number and quality of facilities,
convenience for parking, etc. In addition, from the perspective of 4.2. Park equity assessment at multiple scales based on fine-scale data
random utility theory, the preferences of an individual among available
alternatives can be described by a utility function, and the increase of Within the discourse and practice of spatial equity analysis,
utility is gradually slow with the increase of input. Therefore, a rational geographic analysis scale is an indispensable component (Omer, 2006).
individual would choose the one with maximum utility, no matter the Most studies have been limited to data aggregated over a large spatial
total number of alternatives is large or very limited (Termansen et al., scale, which inevitably makes results prone to aggregation error
2013). This also explained the reasonableness of using most often used (Apparicio et al., 2008). Our study tried to overcome this problem by
parks as the supply source, since these parks were believed to generate using household-level data on accessibility and demand: the point of
the maximum utility of recreational experiences, and were thus origin is the location of respondent’s home, making the collection of
preferred and chosen by residents. data on more nuanced supply and demand possible. Another major
On the demand side, the conventional use of proxies of demand like strength of fine-scale data is that it partly overcomes the limits posed by
population implicitly assumes that different segments of the population analytical scale. Specifically, park supply does not benefit in-situ users
have equal demand for parks, which is unlikely to be universally true. only, as they also benefit users dwelling in other units and who are
Indeed, it is more commonly reported that different segments of the willing to access to these parks. Therefore, collecting data on supply and
population, as influenced by age, ethnicity, and even education level demand based on self-reported data at household scale can provide a
manifest different needs for parks (Y. Lin et al., 2021; Payne et al., matching unit and more accurate data source for subsequent
2002). In addition, assuming equal demand at aggregated levels can lead supply-demand matching analysis. Studies also highlighted that larger
to ecological fallacy, which normally arises when inferences and con­ spatial scale tends to result in a more equitable measurement result,
clusions are made about individuals based on analyses of the group regardless of the measurement used, leading to the suggestion that
performance to which those individuals belong. Differentiated demand different scales should be adopted in spatial equity assessment (Tan and
among population will thus contribute to a more nuanced examination Samsudin, 2017; Truelove, 1993). Our multi-scale analyses provide a
of equity (Omer, 2006; Talen, 1998). We applied perceptional demand more comprehensive understanding of spatial equity.
data and showed that there exist group variations. For instance, our
results suggest that residents’ reasons for moving and whether used park 4.3. Interpreting the relationship between park provision, supply, demand
last week are good indicators of park use demand, as these two factors and equity
actually reflected people’s attitudes towards parks and actual park use
behavior, which have been shown to hold significant and positive cor­ Park provision has long been used as an indicator of the spread of
relations with park use demand in a prior study (Zhang and Tan, 2019). park benefits and the effectiveness of park planning in practice, and the
In addition, we found that income did not differentiate people’s demand equal share of park provision has also been widely used as an indicator
level, and this is partly attributed to the fact that almost all parks are free of equity (C. Liu et al., 2010; Nicholls, 2001). However, our results
of charge in Singapore. Not surprisingly, people aged 25–34 showed showed that study sites with higher park provision do not necessarily
lowest average demand, which may be partly attributed to the fact that have higher equity level (Table 3), which indicates that provision alone
people from this age group would mostly be young working adults and as one aspect of supply cannot fully explain equity. For instance, we
tend to spend more time at work. On the whole, the result of demand observed that Pasir Ris and Bukit Batok both have higher park provision
variations indicates that perceptional park use demand is more relevant and lower population density, but the equity level in Pasir Ris is
with personal preferences, ethnic characteristics, considerations for consistently higher. This difference was accounted by the lower park
children, and housing types, rather than with socioeconomic status. This supply of Bukit Batok assessed by accessibility to most often used parks,
finding expanded the current literature on demand variations among and this comparison helps to highlight and locate the strengths of park
population, as previous studies normally focus on the influence of supply other than provision as an indicator of equity. However, the
population aggregation or speculate the group variations from litera­ heterogeneity of park supply level is also not adequate to explain equity,
ture, lack of the support of empirical data (Cortinovis and Geneletti, and should be analyzed in conjunction with demand. A major reason is
2020; Z. Liu et al., 2021; Shao and Luo, 2022). The variation of demand that the equal inputs of services seldom result in output equality, due to
across population also confirms our hypothesis that perceptional de­ the variation in individual’s demand and abilities. For instance, it has
mand would capture different attitudes among population, and is more been suggested that equal inputs to unequally endowed residents may
accurate than population density as a proxy of demand. result in severely unequal outputs (Lineberry, 1977). One example is

11
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

that the average supply level of Choa Chu Kang measured by network regeneration, to enhance the equity of park services. For instance, the
distance to most often used parks was the highest (Fig. 3), but its equity subzone located at the western part of Pasir Ris was identified with
levels shown by SDEIpII and SDEIpIII were the lowest. This finding in­ undersupply (Fig. 7), and there is a clear mismatch between supply and
dicates that although on average residents in Choa Chu Kang have easier demand (Figs. 5 and 6). In addition, two types of matching (high
access to parks, disparities of supply among residents are larger than supply-high demand and low supply-low demand) are also worth noting.
other study sites. This disparity is also accounted by the unequal demand The first type would contribute to the optimized utilization of park re­
level expressed by residents living in each subzone. It can be seen from sources, and generate positive circulation between supply and demand.
Figs. 2 and 6 that the majority of parks are concentrated in the central In contrast, areas with low supply-low demand should be prioritized in
area, while residents from peripheral areas expressed high levels of the regeneration of park resource although their equity levels are high,
demand. In contrast, Seng Kang had lowest park supply (Fig. 3), but its since we found that residents’ low demand might influenced and rein­
equity level was not the lowest in general. This is mainly due to the fact forced by the low supply of parks in a certain area. On a larger scale,
the park distribution in Seng Kang is dispersive and evenly, and the planning area level provides an overall equity level assessment and al­
matching degree between supply and demand in each subzone is high. lows for an explicit comparison across study sites, also could be used as
This finding also indicates another issue that deserves further explora­ an assessment tool to facilitate equitable park planning at the citywide
tion, is that high equity level also could be derived by evenly but low scale. For instance, planning area Choa Chu Kang was identified with the
levels of supply. Therefore, statistical equity alone as an indicator of lowest equity level and should be prioritized in future planning to
park distribution cannot adequately explain the reasonableness of park reduce regional inequity.
distribution, and this is the reason why we emphasized that we should Second, our results are helpful for the discussion of trade-off between
explore and map the spatial pattern of supply and demand separately providing larger regional parks and smaller neighborhood parks. Park
first and then in concert to derive equity, and the results derived in each size is critical in the sense that a large regional park may provide more
stage could be used as guidance of planning intervention. We thus functions and is more attractive, and elicit more planned activities
suggest that practitioners should have a clear understanding of the (Boone et al., 2009). Smaller neighborhood parks are also essential since
connotation of park provision, supply, demand and equity levels as well they can be easily accessed. In our study, it was observed that diversity
as their interrelationships before developing strategies. On the whole, of park types in a certain area, including both regional and neighbor­
this comparative analysis has shown that high park provision does not hood parks, would contribute to the ease of reaching parks and
equate with high supply and equity level. In other words, although improvement of equity level, as seen in the central part of Choa Chu
residents live in a planning area with abundant park resources, there still Kang and Bukit Batok. In contrast, if an area only has neighborhood
could exists high spatial disparity of supply and demand within that parks, people would travel longer to reach more attractive parks which
region, resulting in certain groups of residents receive deprived park they prefer to use, as seen in the northern part of Choa Chu Kang. This
resources. finding provides significant insights for park planners, that the quality
and attractiveness of a park play a substantial role in influencing resi­
4.4. Implications of SDEI for park planning and management and dents’ choice of most often used parks and thus improve equity level.
limitations SDEI in turn, can be used for devising planning and community in­
terventions to alleviate localities with inequitable park provision.
The concept of supply-demand interactions remains underused in Third, apart from supporting practical decisions in park planning, we
supporting practical decisions in urban park planning and management. further validate the usefulness of SDEI by using it to facilitate park
The generation of practical decisions is largely based on physical attri­ management. On one hand, based on comparison between the spatial
butes of parks, ignoring individuals’ preferences and demands for these heterogeneity of supply measured by accessibility to nearest and most
attributes. As a result, the main metric used for park system planning is often used parks, it is indicated that if the management goal is to
accessibility to nearest parks, assuming that every resident located maximize the recreational value, park managers should focus on
within the planned distance from park is a potential user. However, a improving facilities, services and events that are attractive for most of
number of studies have shown that this assumption is far from being the residents nearby, so as to transfer nearest park to most often used
verified (Tardieu and Tuffery, 2019; J. Zhang and Tan, 2019). In this parks and minimize this observed mismatch. On the other hand, based
study, we developed an index to spatially explicitly show the spatial on the results of heterogeneity of demand across population, it is high­
heterogeneity of supply based on accessibility to most often used park, lighted that if the management focus is to enhance the recreational
demand based on residents’ expressed preferences, and resulting equity. values for certain social groups, the demand of these groups should be
This approach is innovative in the sense that supply assessment is nor­ carefully examined. For instance, we found that families with children
mally based on an unverified assumption of potential users, and demand expressed stronger demand for using parks, and Mäntymaa et al., (2022)
assessment is usually inaccurate and specified with only the density or found that low-income groups would benefit more than others from the
sociodemographic characteristics of individuals. reconstruction and management of parks.
First, SDEI could be used as a decision-support tool for urban park Nevertheless, as a concept and tool, SDEI has limitations. A major
planning, and may shed light upon the identification of areas worth limitation is that obtaining perceptional demand is time consuming, and
prioritized attention, so as to generate optimized suggestions for inter­ therefore, a resource intensive exercise that will be challenging to
vention at a fine scale. As several scholars have pointed out, planning for conduct at a larger scale and over frequent periods. However, with the
public amenities need to be conducted at smaller spatial scales in order development of new techniques for data collection and processing,
to assess distributional inadequacies (Tan and Samsudin, 2017). Spe­ including perception data collected from social media (Song et al.,
cifically, subzone level equity assessment explicitly shows the spatial 2020), perception data generated based on machine learning of re­
heterogeneity of supply and demand, and the mapping of spondents’ ratings (F. Zhang et al., 2018), perception data deduced from
supply-demand imbalance explicitly shows two types of mismatching street view images (Ye et al., 2019), etc., there would emerge promising
(oversupplied (low demand-high supply) and undersupplied (high ways to overcome this challenge in the future. Another limitation of
demand-low supply)) and two types of matching (high supply-high de­ SDEI lies in the lack of time dimension in supply and demand assess­
mand and low supply-low demand) that are worth noting. For over­ ment, since residents’ most often used parks and demand may vary in
supplied areas, decision makers could consider how to optimize the everyday experiences, weekends, and holidays. In our study, people’s
attractiveness of abundant park resources and benefit a larger group of demand in the general situation was asked to generate an overall value.
population apart from local residents. While for undersupplied areas, Therefore, we would like to highlight that in future studies, it would be
priorities should be given to them in future park development and insightful to assess perceptional demand in a more refined manner,

12
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

taking the time dimension into consideration. CRediT authorship contribution statement
Despite these limitations, we suggest that our study is useful to
provide a method for operationalizing the challenges of integrating Zhang Jingyuan: Conceptualization, Methodology, Data curation,
supply and demand in spatial equity assessment. Demand, in particular, Investigation, Formal analysis, Writing – original draft, Writing – review
cannot be easily represented by more conventional data such as popu­ & editing, Visualization, Funding acquisition. Tan Puay Yok: Concep­
lation within a catchment of interest, as it refers to people’s behavioral tualization, Methodology, Data curation, Investigation, Writing – review
intensions to use parks (Zhang and Tan, 2019). We demonstrated the use & editing, Supervision.
of perceptional demand is reasonable and helpful in the spatial visual­
ization of supply-demand mismatch. In addition, perceptional park use
Declaration of Competing Interest
demand is more relevant with personal preferences, ethnic character­
istics, considerations for children, and housing types, rather than with
The authors declare that they have no known competing financial
socioeconomic status. This difference might also be extrapolated to the
interests or personal relationships that could have appeared to influence
population at the whole city level, and could be used to derive social
the work reported in this paper.
group coefficients for easier calculation of demand in future studies. In
addition, SDEI has the potential to be further refined and adapted to
other cities. Appendix A. Supporting information

5. Conclusion Supplementary data associated with this article can be found in the
online version at doi:10.1016/j.ufug.2022.127827.
Assessment of spatial equity should emphasize the degree of spatial
match of accessibility to park resources and residents’ demand level. References
While this is a logical inference from the conceptual underpinnings of
Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50
equity, simultaneous assessment of supply and demand has been inad­
(2), 179–211.
equately addressed in the majority of spatial equity studies. We attribute Alonso, W., 1968. Equity and its Relation to Efficiency in Urbanization. Institute of Urban
this to the lack of suitable methods for assessment and challenges in data & Regional Development, University of California,.
Apparicio, P., Abdelmajid, M., Riva, M., Shearmur, R., 2008. Comparing alternative
collection on demand. We developed an integrated framework and a
approaches to measuring the geographical accessibility of urban health services:
series of SDEI to assess equity at multiple scales, as replicable method­ Distance types and aggregation-error issues. Int. J. Health Geogr. 7 (1), 7.
ology to support spatial assessment and management of urban parks. We Barbosa, O., Tratalos, J.A., Armsworth, P.R., Davies, R.G., Fuller, R.A., Johnson, P.,
further validated its usefulness by applying SDEI in four study sites in Gaston, K.J., 2007. Who benefits from access to green space? A case study from
Sheffield, UK. Landsc. Urban Plan. 83 (2), 187–195.
Singapore. On the whole, our study demonstrated a method to oper­ Bedimo-Rung, A.L., Mowen, A.J., Cohen, D.A., 2005. The significance of parks to
ationalize more accurate assessment of supply and demand, contributing physical activity and public health: a conceptual model. Am. J. Prev. Med. 28 (2),
to existing literature on assessment of park equity. 159–168.
Bennett, R.J., 1980. The Geography of Public Finance. Routledge,.
Our study also demonstrated three other areas of improvement. First, Berg-Schlosser, D., De Meur, G., Rihoux, B., Ragin, C.C., 2009. Qualitative comparative
through the comparative mapping of park supply and demand pattern analysis (QCA) as an approach. Config. Comp. Methods.: Qual. Comp. Anal. (QCA)
using different approaches, we concluded that residents’ assessment on Relat. Tech. 1–18.
Boone, C.G., Buckley, G.L., Grove, J.M., Sister, C., 2009. Parks and people: An
the basis of their preferences and perceptions may serve as a more ac­ environmental justice inquiry in Baltimore, Maryland. Ann. Assoc. Am. Geogr. 99
curate and reasonable proxy for supply and demand of urban parks. (4), 767–787.
Second, through the multi-scale analysis of spatial equity assessed using Byrne, J., Wolch, J., 2009. Nature, race, and parks: past research and future directions for
geographic research. Prog. Hum. Geogr.
different types of SDEI, a more comprehensive understanding of spatial
Cortinovis, C., Geneletti, D., 2020. A performance-based planning approach integrating
equity in a locality was provided. We particularly showed the usefulness supply and demand of urban ecosystem services. Landsc. Urban Plan. 201, 103842.
of data collection at the household level, which can be aggregated at Crompton, J.L., Wicks, B.E., 1988. Implementing a preferred equity model for the
delivery of leisure services in the US context. Leis. Stud. 7 (3), 287–304.
larger scales. In addition, SDEI could be easily replicable in other cities
Dai, D., 2011. Racial/ethnic and socioeconomic disparities in urban green space
and regions, at multiple scales. Third, our proposed framework also has accessibility: Where to intervene? Landsc. Urban Plan. 102 (4), 234–244.
the potential to be widely adopted as a comprehensive approach to Dear, M., 1978. Planning for mental health care: a reconsideration of public facility
support decision making, and could be used for communication by re­ location theory. Int. Reg. Sci. Rev. 3 (2), 93–111.
Deutsch, M., 1975. Equity, equality, and need: What determines which value will be used
searchers and practitioners. Our results highlighted that higher park as the basis of distributive justice? J. Soc. Issues 31 (3), 137–149.
provision do not necessarily result in higher spatial equity level, indi­ Duncan, O.D., Duncan, B., 1955. A methodological analysis of segregation indexes. Am.
cating that more efforts should be made on improving equity level apart Sociol. Rev. 210–217.
Dwyer, J.F., 1994. Customer Diversity and the Future Demand for Outdoor Recreation.
from traditional focus on park provision. Spatially, areas with a low park DIANE Publishing,.
supply but high demand can be identified and targeted for intervention. Giles-Corti, B., Donovan, R.J., 2002. Socioeconomic status differences in recreational
Urban planners can also consider the baseline level of demand in plan­ physical activity levels and real and perceived access to a supportive physical
environment. Prev. Med. 35 (6), 601–611.
ning for parks. It is not uncommon, for instance, that parks located at the Kim, D., Jin, J., 2018. Does happiness data say urban parks are worth it? Landsc. Urban
fringe of cities away from core residential areas where the demand is Plan. 178, 1–11.
low, may then become less accessible and less often used than those Larkin, A., Gu, X., Chen, L., Hystad, P., 2021. Predicting perceptions of the built
environment using GIS, satellite and street view image approaches. Landsc. Urban
closer to residential areas where the demand is the high. Incorporating
Plan. 216, 104257.
demand into park planning systematically can help to generate Lee, G., Hong, I., 2013. Measuring spatial accessibility in the context of spatial disparity
reasonable match between supply and demand for decision making. A between demand and supply of urban park service. Landsc. Urban Plan. 119, 85–90.
Lin, B.B., Fuller, R.A., Bush, R., Gaston, K.J., Shanahan, D.F., 2014. Opportunity or
better understanding of park distribution and equity assessment are
orientation? Who uses urban parks and why. PloS One 9 (1).
worthy topics for further studies that can influence the well-being of a Lin, Y., Chen, X., Huang, L., Zhu, C., Shahtahmassebi, A., Zhang, J., Gan, M., 2021. Fine-
large number of urban dwellers. We believe that the methods and scale mapping of urban ecosystem service demand in a metropolitan context: A
assessment described in our study can lead to a more nuanced picture of population-income-environmental perspective. Sci. Total Environ. 781, 146784.
Lineberry, R.L., 1977. Equality and Urban Policy: The Distribution of Municipal Public
equity in the endeavor to plan for and provide parks in a more equitable Services. Sage Publications, Inc,.
manner. Liu, B., Tian, Y., Guo, M., Tran, D., Alwah, A.A.Q., Xu, D., 2022. Evaluating the disparity
between supply and demand of park green space using a multi-dimensional spatial
equity evaluation framework. Cities 121, 103484.
Liu, C., Li, X., Han, D., 2010. Accessibility analysis of urban parks: methods and key
issues. Acta Ecol. Sin. 30 (19), 5381–5390.

13
J. Zhang and P.Y. Tan Urban Forestry & Urban Greening 80 (2023) 127827

Liu, Z., Huang, Q., Yang, H., 2021. Supply-demand spatial patterns of park cultural Talen, E., Anselin, L., 1998. Assessing spatial equity: an evaluation of measures of
services in megalopolis area of Shenzhen, China. Ecol. Indic. 121, 107066. accessibility to public playgrounds. Environ. Plan. A 30 (4), 595–613.
Lotfi, S., Koohsari, M.J., 2009. Analyzing Accessibility Dimension of Urban Quality of Tan, P.Y., Samsudin, R., 2017. Effects of spatial scale on assessment of spatial equity of
Life: Where Urban Designers Face Duality Between Subjective and Objective Reading urban park provision. Landsc. Urban Plan. 158, 139–154.
of Place. Soc. Indic. Res. 94 (3), 417–435. Tan, P.Y., Wang, J., Sia, A., 2013. Perspectives on five decades of the urban greening of
Luo, W., Wang, F., 2003. Measures of spatial accessibility to health care in a GIS Singapore. Cities 32, 24–32.
environment: synthesis and a case study in the Chicago region. Environ. Plan. B: Tardieu, L., Tuffery, L., 2019. From supply to demand factors: What are the determinants
Plan. Des. 30 (6), 865–884. of attractiveness for outdoor recreation? Ecol. Econ. 161, 163–175.
Mäntymaa, E., Jokinen, M., Louhi, P., Juutinen, A., 2022. Visitors’ heterogeneous Termansen, M., McClean, C.J., Jensen, F.S., 2013. Modelling and mapping spatial
preferences for urban park management: The case of a city park in Oulu, Finland. heterogeneity in forest recreation services. Ecol. Econ. 92, 48–57.
Urban For. Urban Green. 77, 127751. Truelove, M., 1993. Measurement of spatial equity. Environ. Plan. C. 11.
Neutens, T., Schwanen, T., Witlox, F., De Maeyer, P., 2010. Equity of urban service Urban Redevelopment Authority, 2014, Master Plan 2014.
delivery: a comparison of different accessibility measures. Environ. Plan. A 42 (7), Wang, D., Brown, G., Liu, Y., Mateo-Babiano, I., 2015. A comparison of perceived and
1613. geographic access to predict urban park use. Cities 42, 85–96.
Nicholls, S., 2001. Measuring the accessibility and equity of public parks: a case study Ward Thompson, C., 2011. Linking landscape and health: The recurring theme. Landsc.
using GIS. Manag. Leis. 6 (4), 201–219. Urban Plan. 99 (3–4), 187–195.
Oh, K., Jeong, S., 2007. Assessing the spatial distribution of urban parks using GIS. Wen, M., Zhang, X., Harris, C.D., Holt, J.B., Croft, J.B., 2013. Spatial disparities in the
Landsc. Urban Plan. 82 (1), 25–32. distribution of parks and green spaces in the USA. Ann. Behav. Med. 45 (1), 18–27.
Omer, I., 2006. Evaluating accessibility using house-level data: A spatial equity Wolch, J., Wilson, J.P., Fehrenbach, J., 2005. Parks and park funding in Los Angeles: An
perspective. Comput., Environ. Urban Syst. 30 (3), 254–274. equity-mapping analysis. Urban Geogr. 26 (1), 4–35.
Payne, L.L., Mowen, A.J., Orsega-Smith, E., 2002. An examination of park preferences Wolch, J.R., Byrne, J., Newell, J.P., 2014. Urban green space, public health, and
and behaviors among urban residents: the role of residential location, race, and age. environmental justice: The challenge of making cities ‘just green enough’. Landsc.
Leis. Sci. 24 (2), 181–198. Urban Plan. 125, 234–244.
Peña, L., Casado-Arzuaga, I., Onaindia, M., 2015. Mapping recreation supply and Xiao, Y., Wang, Z., Li, Z., Tang, Z., 2017. An assessment of urban park access in
demand using an ecological and a social evaluation approach. Ecosyst. Serv. 13, Shanghai–Implications for the social equity in urban China. Landsc. Urban Plan. 157,
108–118. 383–393.
Pirie, G.H., 1979. Measuring accessibility: a review and proposal. Environ. Plan. A 11 (3), Ye, Y., Richards, D., Lu, Y., Song, X., Zhuang, Y., Zeng, W., Zhong, T., 2019. Measuring
299–312. daily accessed street greenery: A human-scale approach for informing better urban
Ragin, C.C., Becker, H.S., 1992. What is a case?: exploring the foundations of social planning practices. Landsc. Urban Plan. 191, 103434.
inquiry. Cambridge university press,. Zhang, F., Zhou, B., Liu, L., Liu, Y., Fung, H.H., Lin, H., Ratti, C., 2018. Measuring human
Shao, Y., Luo, W., 2022. Supply-demand adjusted two-steps floating catchment area perceptions of a large-scale urban region using machine learning. Landsc. Urban
(SDA-2SFCA) model for measuring spatial access to health care. Soc. Sci. Med. 296, Plan. 180, 148–160.
114727. Zhang, H., Zhang, Z., Dong, G., Yu, Z., Liu, K., 2021. Identifying the supply-demand
Sister, C., Wolch, J., Wilson, J., 2010. Got green? Addressing environmental justice in mismatches of ecorecreation services to optimize sustainable land use management:
park provision. GeoJournal 75 (3), 229–248. A case study in the Fenghe River watershed, China. Ecol. Indic. 133, 108424.
Small, M.L., 2009. How many cases do I need?’On science and the logic of case selection Zhang, J., Tan, P.Y., 2019. Demand for parks and perceived accessibility as key
in field-based research. Ethnography 10 (1), 5–38. determinants of urban park use behavior. Urban For. Urban Green. 44, 126420.
Song, X.P., Richards, D.R., He, P., Tan, P.Y., 2020. Does geo-located social media reflect Zhang, X., Lu, H., Holt, J.B., 2011. Modeling spatial accessibility to parks: a national
the visit frequency of urban parks? A city-wide analysis using the count and content study. Int. J. Health Geogr. 10 (1), 31.
of photographs. Landsc. Urban Plan. 203, 103908. Zhou, X., Kim, J., 2013. Social disparities in tree canopy and park accessibility: A case
Talen, E., 1998. Visualizing fairness: Equity maps for planners. J. Am. Plan. Assoc. 64 (1), study of six cities in Illinois using GIS and remote sensing. Urban For. Urban Green.
22–38. 12 (1), 88–97.

14

You might also like