Professional Documents
Culture Documents
GIS Planificación Verde
GIS Planificación Verde
Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind
a r t i c l e i n f o a b s t r a c t
Article history: The issue of accessibility to urban greenspaces is raising as one of the most debated in sustainable urban
Received 11 July 2013 planning, especially in topics such as environmental justice and health inequalities. This is mainly due
Received in revised form 7 October 2013 to the growing attention that is recognised today to health and well-being benefits from greenspaces.
Accepted 6 November 2013
Different people interpret accessibility based on their individual needs and priorities, but it is generally
acknowledged that access to greenspaces may be particularly beneficial for children, lower socioeconomic
Keywords:
groups and for people with other mental/psychological illness. However, if accessibility is the measure of
Accessibility
the ease of reaching valued destinations, clarifying its definition is an important pre-requisite for further
Greenspaces
Indicators
analysis aimed at supporting urban planning choices on greenspaces.
GIS The following paper presents a set of accessibility indicators aimed at quantifying different measures
Urban planning of accessibility to existing open spaces for the city of Catania, south Italy, an urban context characterised
by a general lack of greenspaces and high density of urban settlements.
Proposed indicators are divided into two main categories: simple distance indicators (SIs) and proxim-
ity indicators (PIs). The first accounts for the number of people or users that can have access to a particular
open space, while the second weights these people or users with the distance from their location to the
open spaces. Indicators are calculated using different thresholds of Euclidean and network distances.
Results show different scenarios in terms of rank of greenspaces accessibility, strongly influenced by
chosen distance metric (Euclidean vs network) and thus emphasise a careful use of these indicators as
planning support tools. Some practical implications of measuring accessibility for urban planning can be
highlighted: for instance, specific land uses might be chosen for highly accessible open spaces, especially
those characterised by a high proximity to residential settlements. Examples include allotment gardens,
playgrounds and other informal green areas.
© 2013 Elsevier Ltd. All rights reserved.
1470-160X/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ecolind.2013.11.011
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
2 D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx
or service provision, and describes an integrated system of facil- within a fixed distance (or time) from the origin points of users, or
ities/services and users from the user viewpoint (Halden et al., vice versa. These indicators are simply based on a distance relation
2003). Moreover, in this study, the accessibility has an intrinsic between users and the service under examination, and are gen-
spatial feature that is related to the possibility for citizens to reach erally easy to calculate, without the need for further analyses or
a particular place and is different to other non-spatial character of data collecting. Examples are the number of services (i.e. in this
accessible places such as regime property. According to this feature, paper, GSs) within a fixed distance of residential areas, the number
accessibility can be modelled by GIS spatial indicators. of people located within a fixed distance, the minimum distance to
Approaches to the evaluation of accessibility can be differ- the closest services or the average distance from all services. This
ent according to the particular aims of the assessments, available approach ignores the real spatial distribution of the population that
data, time and resources. Since there is often a trade-off between can use a certain service and the relative distance of the population
the availability of appropriate datasets or software and the to the service.
aims/objectives of a particular project, different approaches have The second approach is to calculate ‘proximity’ measures based
been used to assess accessibility to GSs (Higgs et al., 2012). on the evaluation of the number of users or members of the popu-
For urban policies aimed at enhancing the overall accessibility to lation with reference to their specified distance/time from a certain
GSs and related provided ecosystem services, it is important to find service (i.e. GS). In this way, accessibility to a particular service is
appropriate tools and indicators able to support choices towards weighted with its effective distance from the users.
the creation of new GSs. Particularly, the choice of which area to Both approaches require some geographic information in order
be addressed as a GS should be based on the criterion (among oth- to calculate the indicators.
ers) of the maximisation of its accessibility in order to allow the Three fundamental variables to be considered in the modelling
highest number of people to have easier access to the GS in terms of accessibility indicators:
of time or distance. Moreover, choices about where to locate new
GSs should aim to minimise social disparities of access, to avoid - an origin place (typically a point or area) representing the location
low-income, minority or other social groups lacking access to the of the users/population that can potentially have access to the
benefits provided by GSs (Pham et al., 2012). service/GS;
The following paper presents a set of accessibility GIS based indi- - a destination place (typically a point or area), representing the
cators aimed at quantifying different measures of accessibility to location of the service that the users/population can potentially
open spaces (OSs) in the city of Catania, south Italy, an urban con- access;
text characterised by a general lack of GSs and high density of urban - a distance measure between the aforementioned places.
settlements. Particularly, accessibility indicators are calculated to
rank existing OSs – currently not zoned or used as GSs – according The location of the users/population can be modelled using typ-
to their scores of accessibility, in order to understand which are ical census geo-referenced data such as census tracts, which are
more suitable to be addressed for new GSs by the municipality’s geographical units where users (residents or specific social groups)
land-use plan. are located. Census tracts also allow specific social groups (children,
The paper is structured in the following way. Approaches for elderly people, ethnic minorities) to be taken into account when
modelling the accessibility to GSs are discussed in Section 2. The evaluating the accessibility to urban services (Barbosa et al., 2007;
study area and used geo dataset are described in Section 3. The Jansson and Persson, 2010; Wright Wendel et al., 2012). Thanks to
proposed indicators are presented in Section 4. Results of indica- their periodic update, census data can also provide a chronological
tors’ calculation are given in Section 5 and than discussed in Section evolution of the socio-economic system. They are usually proxied
6. Finally, conclusions are summarised in Section 7. by the centroid of each polygonal tract which represents the entire
population living there. To obtain a more precise distribution of the
population, additional ancillary data, such as land-use maps con-
2. Modelling accessibility to urban greenspaces taining information about residential areas, can be used (Apparicio
et al., 2008; La Rosa and Privitera, 2013). In this way, it is possible to
Spatial approaches and GISs are widely used to examine the evaluate the number of people living inside each residential area as
pattern of accessible urban GSs. Some of the main research fields a percentage of the number of people living inside the entire census
cover the identification of areas suffering from lack of accessibility tract.
(Barbosa et al., 2007), social inequities deriving from ethnic minori- The location of the destination place (i.e. in this paper, a GS)
ties with limited access to GSs (Boone et al., 2009; Lindsey et al., depends on the geographical primitives (point or polygons) that
2001; Pham et al., 2012; Talen and Anselin, 1998), and the rela- are used. Geometric centroids of the polygons representing the GSs
tionship between availability of and accessibility to GSs and health are often used, as well as their entrances or access points. The dis-
outcomes (Apparicio et al., 2008; Paquet et al., 2013). tance between origin and destination will be different depending
In recent years, the modelling of accessibility to GSs has evolved on the chosen destination place; many studies have investigated
substantially, thanks to enhanced GIS features (particularly the the influence on obtained results of this choice with different out-
development of GIS network analysis modules, such as ArcGis Net- comes (Nicholls and Shafer, 2001; Comber et al., 2008; Higgs et al.,
work Analyst) and personal computers’ computational capabilities. 2012), even though it might often be very difficult to have or to
From simple radius-buffering methods aimed at evaluating the create detailed geographic data such as about access points to GSs
number of facilities and the proportion of the population within a (e.g. by manual digitalisation).
fixed area (Nicholls and Shafer, 2001; Moseley et al., 2013), recent Three types of distance are typically used for calculating
research has developed to integrate urban landscape character- accessibility measures (Apparicio et al., 2008): Euclidean distance
istics or socio-economic data with more advanced methods for (straight line), Manhattan distance (distance along the two sides of
distance evaluation (Pham et al., 2012; Sander et al., 2010). a right-angled triangle opposed to the hypotenuse) and network
There are two common approaches in current research streams distance (shortest time or distance). Euclidean and Manhattan
that are used for measuring accessibility to GSs or other urban distances can easily be calculated with every GIS. Network distance
facilities with GIS (Apparicio et al., 2008; Higgs et al., 2012). is usually more precise but requires an additional geographical
The first is to use indicators/measures which count the num- dataset (a roads network layer) that might be not be available or
ber of services (e.g. green areas, hospitals, other urban services) that might need additional work in order to be created. In previous
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx 3
Fig. 1. The study area of Catania (Italy) and the distribution of open spaces within the municipality.
work, different types of distance have shown some correlation important and unused source of urban biodiversity and ecosystem
when applied to urban contexts: for instance, Apparicio et al. services (La Rosa and Privitera, 2013). These OSs will be the geo-
(2008) found a strong correlation between Euclidean or Manhat- graphical units for calculating the indicators in the next sections.
tan distances and network distances across the metropolitan area In the current study, the following geo dataset was used. The
of Montreal (even if with some local differences in suburban areas). city’s street network, available as a shapefile, was converted into
One important issue that is usually less remarked in accessibility a network in order to calculate the network distances (see Section
modelling is that measures and indicators can be attributed both to 4).
the origin place (i.e. in this paper, census tracts) or destination place The census tracts from the 2001 national census were available
(i.e. GSs). In the first case, indicators will describe a characteristic as a polygonal shapefile.
of the urban environment (the presence of GSs) and then attribute The layer of OSs obtained by Urban Atlas land-use layer (EEA,
this characteristic to the place where people or users are located 2010) for the city of Catania had an average scale of detail of
(census tracts). Indicators such as the number/area of GSs within 1:12,000. The source of the land-use classification was a SPOT 5
a fixed distance from census tracts or the minimum distance from image (resolution 2.5 m) acquired in the summer of 2005. OSs
GSs are example of this case. In the second case, indicators will included the following land-use categories of the Urban Atlas:
quantify a characteristic of the potential users in an urban context agricultural, semi-natural and wetland areas; forest; land without
and then attribute it to a GS. The number of people living within current use. To update and check the Urban Atlas land-use layer, a
a fixed distance from a GS is a typical example of this case. The visual inspection of high-resolution regional orthophotos (Regione
choice of one of the two approaches depends on the needs of the Sicilia, 2009) and recent Google Maps images was performed. The
analysis: if the aim is to understand the geographical distribution land-use layer was spatially adjusted in order to match the Italian
or supply of urban GSs, the first approach would be preferable; if national geo-referencing system of the street network and census
the analysis aims at understanding the potential demand of GSs for tracts. In all, 150 control points were used.
planning, the second approach might be preferred. In the rest of this The street network, census tracts and OS layers are shown in
paper, all indicators refer to GSs, because the aim of the research Fig. 2.
is to characterise GSs in terms of accessibility and highlight the
implications for urban planning of using one indicator over another.
4. Methods and proposed indicators
3. Study area and available geo datasets Different indicators are proposed to investigate the role of dis-
tance and the way accessibility to OSs can be modelled. Indicators
In this work, the study area is Catania, one of the main cities are divided in two classes, with reference to the two different
in southern Italy (Fig. 1), with an administrative area of 180 km2 approaches introduced in Section 2: simple distance indicators (SIs)
and a population of 293,458. The city is in the centre of a and proximity indicators (PIs). SIs measure the number of people
large conurbation that represents the largest metropolitan area at fixed distances from each OS that can have access to each OS. PIs
in Sicily, a settlement system characterised by extensive urban quantify the number of people at fixed distances from each GS and
sprawl (La Rosa and Privitera, 2013). Catania is characterised by weight them with their distance from GSs (gravitational model).
a relevant shortage of public spaces and services, especially GSs. Geometric centroids of GSs and census tracts will be used as origin
Currently the amount of public GSs is about 3 m2 /inhabitant, much and destination places, respectively.
less than the minimum amount specified by national legislation Indicators are calculated based on two types of distance:
(9 m2 /inhabitant) (Privitera et al., 2013).
Apart from existing GSs, a significant number of other OSs,
currently unmanaged, are present in the city. These areas are - Euclidean distance (E).
quite heterogeneous in shape, size, land-use, function, bio-physical - Road network distance (N). This is based on Dijkstra’s algorithm
features, and ecological and landscape value, but represent an (Zhan and Noon, 1998), which solves single-source shortest path
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
4 D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx
Fig. 2. Available data for indicators’ calculation: layers of streets network, census tracts and open space.
Table 1 of the network analysis returned a .dbf table containing more than
Proposed indicators of accessibility to greenspaces, type of distance and distance’s
250,000 records, approximately the same number returned by the
thresholds.
calculation of Euclidean distance.
Euclidean distance Network distance
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx 5
Fig. 3. Example of proximity modelling: in the configuration A the OS in green has a higher proximity to the 3 census tracts in red than in configuration B. (For interpretation
of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Source: La Rosa and Privitera, 2013
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
6 D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx
Fig. 4. Maps of simple indicators’ (SI) scores. The indicator expresses the number of people within the different distance thresholds. Classes are made with a quintile
classification method.
Table 3 Table 4
Spearman’s rank correlations for simple indicators (SI). P values < 0.001. Spearman’s rank correlations for proximity indicators (PI). P values < 0.001.
Correlations Correlations
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx 7
Fig. 5. Maps of proximity indicators’ (PI) scores. The indicator expresses the number of people within the different distance thresholds weighted for the squared distance to
open spaces. Classes are made with a quintile classification method.
from 300 m to 600 m, but then slightly decrease when considering 600 m), lower values were found when correlating SI N and PI N
the entire municipality without fixing a threshold. ( = 0.585 for SI N 300 and PI N 300; = 0.648 for SI N 600 and
Slightly lower correlations were found when comparing SIs PI N 600). The lower correlation between SI and PI might show how
with PIs (Table 5). At the same distance threshold (300 m or the inclusion of sociological variables (i.e. the census tract’s popu-
lation) has an influence on the accessibility rank – and consequent
spatial location – of OSs.
Table 5
Spearman’s rank correlations for simple and proximity indicators (SI and PI). P val-
ues < 0.001. 6. Discussion
Correlations
6.1. Indicators’ choice and methodological issues
SI E 300 0.765
PI E 300 The primary issue to be discussed is how to choose the most
SI E 600 0.783
PI E 600
suitable indicators to describe the accessibility of people to GSs.
SI N 300 0.585 This might depend on several factors, such as the aim of the research
PI N 300 or application, the number and types of available geo datasets, and
SI N 600 0.648 the methodological assumptions/choices (type of distance, use of
PI N 600
centroids or census tracts) that are made.
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
8 D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx
Fig. 6. Maps of more accessible open spaces (5th quintile) calculated by simple indicators.
Fig. 7. Maps of more accessible open spaces (5th quintile) calculated by proximity indicators.
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx 9
The service or typology of GS involved in the accessibility eval- walked or driven by users between an origin (i.e. in this study,
uation should be considered. If a study aims to understand which a census tract) and a destination (i.e. an OS). Network distances
OSs can be used for particular GSs such as playgrounds, a simple are more precise and closer to real distances. Since they are
distance indicator may be most appropriate, because playground calculated between centroids, they do not represent the real dis-
areas tend to be highly localised in the urban fabric (Talen and tances that an user has to walk/drive to get inside the OS, but
Anselin, 1998). On the other hand, it is also true that SIs (e.g. the rather they are proxies. However, network distances have the
number of people at a fixed distance from the GS) are not able to ability to minimise this centroid effect because, when generat-
take into account people who live outside the fixed distance but ing the network in the GIS, OS centroids are moved to the closest
who may be willing to travel or drive to reach a particular service element of the street network before the distance matrix is calcu-
(La Rosa and Privitera, 2013). For this reason, proximity indicators lated.
calculated with the gravity model might be more able to capture all Considering the same distance thresholds, the results obtained
the information about potential users inside a larger urban context for the indicators show how network distance is always more
(conurbations of towns and suburbs belonging to several munici- restrictive in terms of numbers of reachable OSs at that dis-
palities or a metropolitan area). Thus, if the aim of the evaluation tance. Euclidian values were almost double when passing from
is to understand which of the existing OSs might be more suitable 300 m to 600 m threshold and this is obviously related to the
to be a GS (as in this research), PIs might allow a more complete longer distances that it is necessary to walk/drive to reach the
picture of accessibility to OSs in the entire urban context, as they GSs along the road network. This result is relevant and highlights
are able to take into account the spatial configuration of all existing as urban analyses conducted for different aims (implementation
OSs, considering the entire network of OS (Koohsari et al., 2013; of policies/investments or market evaluation of real estate) might
Moseley et al., 2013) for their calculation. require the indicators to be calculated using different distance
The choice of the exponent for the distance in the gravita- thresholds.
tional model formula (1) should be based on the study of users’
travel behaviour. The square distance might be suitable for con-
6.4. Limitation on the use of census tracts
sidering a walking travel mode, the transport mode more used
by people having access to GS. If users look for high quality ser-
Indicators were calculated inside the municipal area of Cata-
vices (i.e. urban parks with particular features or playgrounds
nia, because no spatial data about OSs was available outside the
with very new or well maintained equipments), they might accept
area; for this reason, there may be a ‘border effect’ in the indica-
travelling longer distance to access the service of the sought-
tors’ scores. Lower scores for OSs near borders (especially in the
for quality. In this case, the exponent of the gravitational model
north-west part of the municipality) might also be due to the fact
formula can be fixed at 1. On the other hand, less-than-ideal ser-
that there are no census tracts outside the municipalities. This con-
vices might be accepted if services of the desired quality are not
firms the findings by Apparicio et al. (2008) that also showed some
available within an acceptable threshold of travelling distance. In
local variations in the correlation between Euclidean and network
this other case, the exponent of the formula can be increased to
distances, especially in suburban areas where Euclidean distances
2.
are less precise. These results highlight the importance of also hav-
ing data about the population outside the area of interest in order
6.2. Limitation on the use of SI
to get a more realistic picture of accessibility to GSs also by people
living outside the city administrative borders but close to OSs. This
The results obtained from SIs confirm the findings and limita-
need appears as even more relevant in contemporary metropolitan
tions on the use of buffer distances to estimate the accessibility to
areas, where conurbations are continuous and may include several
GSs. The use of SIs may lead to under-provision of functional GSs
municipalities.
and their benefits, because these indicators are not able to evaluate
The use of centroids as proxies for the census tracts can be also
the different users of GSs and thus highlight spatial inequalities in
lead to an underestimation of the number of accessible OSs. This
accessibility (Moseley et al., 2013).
occurs for larger OSs when using a fixed distance Euclidean thresh-
The available geographical dataset strongly influences the
old (especially for a low threshold like 300 m), because the buffer
choice of indicators to use. As already discussed, indicators based
from centroids has to exceed the real size of OSs. This problem does
on network distance require a road network. Even if today there
not occur for network distance, because the centroids of the OSs are
are several open source road datasets at different scales and reso-
always snapped to the closest element of the network (see Section
lutions, such information is not always available and it can be costly
4).
to build for the purpose. The same constraints can be extended to
the use of other additional information that might increase the pre-
cision of results, such as about access points to OSs. Moreover, if it 6.5. Implications for urban planning
is necessary to plan for OSs to become future GSs, the location of
entrances cannot be known at the time of the research (as in the Beside the considerations discussed above, the results of this
case in this research). study show a considerable correlation among indicators. Par-
ticularly, a high correlation among indicators calculated with
6.3. The way distance is calculated Euclidean and network distances has been found, as also reported
by Apparicio et al. (2008). The correlation was higher for pairs
Euclidean distances are easy to calculate with all available of indicators calculated with the same distance type (higher than
GIS software and have historically been used in many applica- 0.7, < 0.001) than for indicators with different distance type.
tions (Sander et al., 2010). They do not need a network layer of Moreover, results from correlation analysis showed that corre-
streets and also require shorter computing times when calculat- lation varied when increasing distance thresholds but exhibited
ing the distance matrix between centroids of census tracts and a maximum for distance threshold of 600 m. In any case, the
centroids of OSs. Also, some additional work is usually needed overall high correlation between indicators might prompt a care-
when setting the network parameters, like the manual snapping ful consideration of the increasing operational times and costs
of points to the available network. On the other hand, Euclidean related to use a network layer, especially whether this is not avail-
distances usually fail to represent adequately the actual distances able.
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
10 D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx
However, land-use choices about which OS can be planned as concluded that different combinations of GS characteristics are
new GSs might be more interested in understanding spatial con- likely to influence each behaviour, because the likelihood that a per-
figurations of most accessible OS than which distance threshold is son uses different types of GS depends on the individual behaviours.
more suitable to be used. In other words, it might be more relevant As a consequence of this, there is a need to integrate quantita-
to localise where are the most accessible OSs (in relative terms) in tive and qualitative aspects in the complex decision making that
a city than to quantify how many people can have access to a single is usually invested in urban planning processes, in order to better
OS (as the number of people will change with the used the distance understand which factors influence the daily use of GSs (Jansson
threshold). and Persson, 2010; McCormack et al., 2010).
This type of understanding is important because, as stated previ-
ously, some land uses may require a higher proximity than others; 7. Conclusions
areas with a higher proximity are likely to be reached more easily
than areas with a lower proximity. Specific types of GS might be Urban GSs are considered by recent researchers to have positive
chosen for high-accessibility areas, especially those characterised implications for health, ecosystem service provision and cultural
by a high proximity to residential settlements. Examples include enrichment. Understanding which existing OSs can be addressed
allotment gardens, playgrounds and other informal green areas by urban planning towards new GSs needs support tools able to
(DTLR, 2002). Allotment gardens or playgrounds are typical urban quantify their degree of accessibility within urban areas.
land uses which constitute small patches but with good proxim- The present work proposes and discusses a set of indicators
ity to residential areas. For the activities in these areas, people that can be used in urban planning, highlighting the pros, cons
usually need be within short times and distances. For example, and limitations of their use. Proposed indicators are divided into
children or older people need to walk short distances to get to two main categories: simple distance indicators (SIs) and prox-
playgrounds or allotment gardens. One explanation underlying the imity indicators (PIs). The first account for the number of people
attractiveness of a land-use is the utility associated with the activ- or users that can have access to a particular OS, while the second
ity to be conducted at the destination (Gahegan, 1995). This is weight these people or users with the distance from their location
particularly relevant in urban settlements characterised by high to the OSs. Indicators are calculated using Euclidean and network
population density, the presence of multi-storey apartments and a distance. The results show that more precise and accurate results
lack of public and private gardens. are obtained when using network distance, although this type of
For particular OSs that present lower accessibility indicator distance requires more data and analysis (e.g. a network layer);
scores, other land uses can be proposed, including new forms of results obtained by using Euclidean distance tend to overestimate
urban agriculture (Wells and Gradwell, 2001; Zasada, 2011) or the amounts of persons that can reach the single OS. A high corre-
urban greenery (Young, 2010). The final choice should depend also lation in indicator scores might suggest in some circumstances the
on other specific characteristics such as size and the presence of use of indicators that require less data (e.g. indicators calculated
trees covering the land (La Rosa and Privitera, 2013). with Euclidean distance). The presence of a border effect in indica-
tors’ scores also suggests that more appropriate results from PIs are
6.6. Further research on the inclusion of different social group obtained in highly urbanised areas or metropolitan areas. For this
reason, the spatial extent of the dataset should not be limited to the
Further research might include the refinement of indicators so area under examination but should include also census tracts that
as to include particular social groups (e.g. children and elderly are outside the city’s administrative boundaries and that represent
people) or different ethnic groups (Comber et al., 2008) and the the places of origin of people who can potentially have access to
interaction between urban environment and socio-ecological sys- OSs.
tems (Cook et al., 2012; Giles-Corti, 2006). This implies the need Different OSs can be identified as being the more accessible
of socio-economic data to further characterise – e.g. using them as depending on the indicators used and, overall, the variation in
weights for indicators – the particular type of new GS: those OSs indicators’ scores should be carefully considered by researchers
with higher scores of indicators calculated using particular census and planners in terms of the methodologies used, assumptions
variable (i.e. families with young children) can be prior for new accepted, and limitations. In any case, indicators for evaluating the
playgrounds. accessibility of GSs represent very useful tools for urban planning
Accessibility indicators might also be further developed in and might be able to provide municipalities or other metropolitan
order to specify which particular types of GS can be proposed by public bodies (provinces, metropolitan areas) a sound base from
urban planning. Schipperijn et al. (2010) focused on the relation which to develop differentiated policies for a more liveable-in and
between different types of GSs and users’ behaviours. These authors healthy urban environment.
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx 11
ID OS SI E 300 SI N 300 SI E 600 SI N 600 PI E 300 PI N 300 PI E 600 PI N 600 PI E TOT PI N TOT
0 1955 301 5280 1282 54,519 32,846 77,516 37,227 130,824 63,189
1 1316 1316 5599 1818 74,847 363,291 96,862 365,467 148,708 394,298
2 1868 705 6789 1409 80,811 63,153 109,455 67,971 163,791 90,778
3 1057 2 5780 2287 25,019 42 51,975 10,749 99,542 41,005
4 397 388 1945 784 12,257 13,116 22,465 14,725 67,370 37,499
5 475 378 2011 475 40,970 21,243 48,401 21,718 75,671 33,756
6 137 67 3658 222 4487 6845 20,488 7685 50,874 27,762
7 475 182 1470 475 22,392 174,179 27,253 176,393 56,353 188,386
8 2736 1248 7455 2536 109,956 83,267 130,078 91,247 188,769 125,311
9 1721 365 6811 968 45,080 11,832 68,226 14,881 153,067 64,443
10 3123 1592 10,669 5454 177,677 33,938 219,393 51,598 270,533 86,015
11 1159 750 7833 2350 130,356 3,376,889 164,815 3,388,045 239,539 3,433,005
12 3135 285 4744 661 73,974 10,925 83,639 12,947 125,209 37,081
13 183 183 1728 899 319,864 2910 327,681 6241 349,891 23,408
14 1045 530 5093 2149 60,795 31,429 81,225 37,220 157,457 87,872
15 2 – 292 – 28 – 2673 – 45,369 22,212
16 59 – 512 – 767 – 2698 – 37,321 16,944
17 1143 – 5319 705 15,518 – 35,697 2644 81,867 20,703
18 773 237 3490 644 21,689 21,962 34,894 23,647 98,938 61,193
19 2094 80 8345 2454 12,1996 1301 150,656 12,206 218,220 58,067
20 1234 299 6671 1022 52,132 4687 79,906 8296 168,424 48,634
21 1981 – 5471 705 49,862 – 66,762 2461 119,134 30,783
22 1730 999 4296 1341 91,441 176,199 102,880 178,229 133,406 199,823
23 122 4 2579 95 3224 292 12,347 1066 44,294 17,661
24 1542 105 4118 932 70,406 1199 81,659 5683 111,732 24,423
25 613 – 3770 1204 79,503 – 96,135 5804 169,973 46,820
26 1230 202 3768 1624 95,805 114,477 107,590 121,229 139,499 143,833
27 1981 965 5701 1575 86,344 193,314 104,303 197,116 158,159 228,048
28 1664 310 7492 2416 121,228 5228 151,095 16,191 198,032 49,420
29 464 237 3472 864 18,771 65,882 35,543 68,129 104,149 108,320
30 2321 915 8657 2459 268,547 10,282 297,528 19,047 386,927 74,501
31 2396 781 16,540 4530 87,797 64,249 161,645 82,370 260,482 148,588
32 341 226 2125 692 7463 11,927 16,143 13,727 52,463 38,651
33 1389 214 6348 2324 152,851 1,678,873 177,000 1,689,724 257,614 1,735,560
34 522 237 2819 244 37,503 10,613 46,733 10,639 116,380 44,883
35 390 26 4001 468 7168 416 23,941 2683 69,253 34,322
36 471 165 9713 3595 219,649 215,544 265,503 230,437 360,380 288,269
37 1766 119 5472 1536 79,776 42,782 99,397 49,491 154,718 77,391
38 1062 166 7013 1069 31,153 10,250 61,719 15,375 136,610 64,294
39 279 212 2370 1458 5352 3783 16,746 9555 51,533 31,719
40 3094 1023 17,816 8213 93,920 16,527 167,054 48,475 264,365 117,821
41 450 – 4290 1134 19,873 – 41,295 6913 71,066 27,572
42 2870 755 15,556 6872 69,491 14,509 131,452 43,366 221,163 100,129
43 3645 2008 17,396 8416 129,297 49,202 202,613 81,560 301,875 155,474
44 1540 620 6185 2194 37,263 15,048 62,737 20,789 89,931 39,362
45 574 71 2265 508 20,450 36,327 28,725 38,261 51,746 56,373
46 1474 722 3661 1084 77,467 28,061 88,412 30,093 112,381 45,684
47 419 – 8597 1994 7904 – 49,169 9368 127,484 45,522
48 172 – 1903 946 1925 – 11,413 6638 46,617 22,485
49 7 – 2727 388 327 – 17,019 2258 61,322 21,265
50 719 282 4475 1134 15,424 4170 32,375 7437 113,464 55,530
51 695 2818 273 26,713 – 37,358 1077 98,342 30,332
52 569 293 4069 1585 7421 8232 27,323 13,667 67,710 41,865
53 63 – 378 321 1250 – 2733 1729 28,658 16,011
54 1428 1254 6355 2464 53,337 31,247,034 79,382 31,251,845 166,759 31,305,418
55 1056 695 2964 1211 52,663 22,159 61,295 25,127 95,157 43,850
56 1068 466 3227 1118 44,516 18,426 55,125 21,438 112,854 45,194
57 761 484 4350 1745 37,496 14,058 57,547 18,932 131,063 63,499
58 809 62 2715 532 56,004 58,668 65,767 60,379 103,968 80,670
59 1119 310 5006 1462 6,364,585 780,477,850 6,385,271 780,485,383 6,461,686 780,519,374
60 2657 102 5845 1218 62,019 1396 76,209 8888 127,589 29,537
61 2079 1126 8050 4065 53,414 250,469 87,445 266,437 141,430 302,405
62 1145 421 4314 1118 99,511 364,822 113,904 370,387 166,377 394,670
63 18 18 75 22 6206 0 6477 38 39,841 12,162
64 442 277 2672 880 10,186 3635 21,676 5632 75,016 36,278
65 276 59 4596 1217 10,214 846 31,327 6497 80,922 35,181
66 987 440 8214 4108 23,392 126,896 60,827 144,437 152,879 201,683
67 388 – 2861 388 4496 – 13,968 4287 55,047 25,411
68 18 – 22 – 454 – 471 – 31,075 –
69 120 202 2362 1397 30,033 25,360 40,216 29,633 74,244 49,720
70 2846 1166 12,170 4094 93,766 133,586 139,035 147,687 238,987 211,924
71 604 384 4978 1503 40,775 121,689 61,263 127,310 132,068 176,250
72 1697 494 7203 2221 75,224 15,776 104,968 25,230 192,171 70,885
73 3548 651 11,169 3336 113,639 17,618 154,722 30,379 240,585 80,683
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
12 D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx
Annex I (Continued )
ID OS SI E 300 SI N 300 SI E 600 SI N 600 PI E 300 PI N 300 PI E 600 PI N 600 PI E TOT PI N TOT
74 2277 229 8722 1924 125,998 55,013 158,239 63,641 179,232 82,922
75 981 – 3247 202 41,000 – 53,917 1457 92,955 26,367
76 727 216 2705 852 13,322 13,220 22,370 18,054 59,474 40,845
77 – 477 2036 946 – 7531 12,229 10,561 49,060 27,470
78 2242 1065 5251 1065 45,424 99,713 59,657 99,713 107,300 125,168
79 1324 881 4788 1483 142,782 176,636 157,444 180,088 222,983 214,245
80 1333 397 4654 712 22,991 8975 37,882 10,549 84,140 30,003
81 2897 427 14,830 3778 78,793 34,522 138,783 49,843 238,031 119,349
82 1399 163 5972 1218 47,653 921 71,458 6588 129,096 37,557
83 188 4548 1283 5489 25,655 5679 71,815 40,035
84 1610 168 4649 2331 74,304 1925 94,083 15,200 157,153 41,173
85 1704 625 7543 1121 127,880 3930 156,682 5818 228,478 47,494
86 1372 968 8205 3357 26,474 23,033 55,139 32,943 149,235 83,551
87 1289 – 6228 705 30,728 – 60,276 3085 109,861 21,646
88 231 57 1989 378 14,307 – 23,683 1743 58,737 12,306
89 840 – 6026 886 12,529 – 36,463 3529 109,722 37,451
90 1077 246 7319 2283 36,963 – 67,540 19,747 154,462 70,331
91 2078 683 7848 2067 963,689 13,171,798 994,765 13,179,819 1,072,602 13,213,049
92 321 321 754 571 15,891 16,223 17,861 17,340 41,992 30,373
93 615 46 4296 46 95,908 5642 110,589 5642 151,014 5642
94 843 – 2634 695 18,215 – 29,670 3434 62,447 18,532
95 2047 1335 8299 4062 81,083 1,981,369 114,097 1,992,744 159,187 2,026,384
96 801 11 5498 1550 11,271 152 37,282 7916 105,878 35,030
97 1188 735 4462 2222 27,244 37,235 44,367 43,840 69,892 59,809
98 1004 633 2759 1418 26,709 947,790 36,926 952,522 64,646 966,458
99 2229 881 5629 1187 32,456 802,835 49,486 803,832 102,996 832,242
100 264 – 2356 453 6455 – 19,193 3156 68,326 26,417
101 2935 817 12,471 5720 66,881 15,752 111,641 40,649 196,724 92,263
102 14 228 1350 1207 259 5048 7904 9141 46,040 32,202
103 1119 370 6488 1118 53,761 7660 88,519 11,108 135,225 32,849
104 1282 1075 4770 1430 32,445 122,704 47,239 124,193 130,668 159,704
105 1603 451 3167 1127 31,572 14,553 37,662 19,466 83,349 43,434
106 1537 – 3314 705 40,105 – 46,696 2461 92,354 30,734
107 28 – 7565 886 489 – 35,923 3952 113,052 39,003
108 2 – 950 453 1411 – 5993 3098 48,284 26,318
109 1663 1187 9295 1500 2,778,068 33,242 2,819,027 34,669 2,885,571 64,565
110 646 355 5127 355 32,246 148,925 52,503 148,925 118,371 173,701
111 1453 423 10,310 3125 26,645 4,660,811 73,480 4,673,613 150,035 4,715,862
112 4 – 201 77 212 – 1038 363 32,644 15,013
113 2384 588 14,811 4331 97,258 20,065 1,58,978 41,230 245,099 107,342
114 359 – 2180 620 5241 – 12,226 2776 77,158 34,132
115 – – 2203 194 – – 8467 618 33,314 14,637
116 – – 501 – – – 2408 – 57,151 21,413
References
Gahegan, M., 1995. Proximity operators for qualitative spatial reasoning. In: Frank,
Apparicio, P., Abdelmajid, M., Riva, M., Shearmur, R., 2008. Comparing alternative A.U., Kuhn, W. (Eds.), COSIT’95 Proceedings: Spatial Information Theory – A The-
approaches to measuring the geographical accessibility of urban health services: oretical Basis for GIS. Lecture Notes in Computer Science, 988. Springer, Berlin,
distance types and aggregation-error issues. International Journal of Health Geo- pp. 31–44.
graphics 7, 7. Giles-Corti, B., 2006. People or places: what should be the target? Journal of Science
Barbosa, O., Tratalos, J., Armsworth, P., Davies, R., Fullers, R., Johnson, P., Gaston, K., and Medicine in Sport 9, 357–366.
2007. Who benefits from access to green space? A case study from Sheffield, UK. Halden, D., Jones, P., Wixey, S., 2003. Accessibility Analysis Literature Review.
Landscape and Urban Planning 83, 187–195. Measuring Accessibility as Experienced by Different Socially Disadvan-
Boone, C.G., Buckley, G.L., Grove, J.M., Sister, C., 2009. Parks and people: an envi- taged Groups. Working Paper, Available at http://home.wmin.ac.uk/
ronmental justice inquiry in Baltimore, Maryland. Annals of the Association of transport/projects/samp.htm (last accessed 29.06.2013).
American Geographers 99, 767–787. Higgs, G., Fry, R., Langford, M., 2012. Investigating the implications of using alterna-
Comber, A.J., Brunsdon, C., Green, E., 2008. ‘Using a GIS-based network analysis tive GIS-based techniques to measure accessibility to green space. Environment
to determine urban greenspace accessibility for different ethnic and religious and Planning B 39, 326–343.
groups. Landscape and Urban Planning 86, 103–114. Jansson, M., Persson, B., 2010. Playground planning and management: an evaluation
Cook, E.M., Hall, S.J., Larson, K.L., 2012. Residential landscapes as social-ecological of standard-influenced provision through user needs. Urban Forestry & Urban
systems: a synthesis of multi-scalar interactions between people and their home Greening 9, 33–42.
environment. Urban Ecosystems 15, 19–52. Jo, H.K., 2002. Impacts of urban green space on offsetting carbon emissions for middle
Dai, D., 2011. Racial/ethnic and socioeconomic disparities in urban green space Korean. Journal of Environmental Management 64, 115–126.
accessibility: where to intervene? Landscape and Urban Planning 102, 234–244. Koohsari, M.J., Kaczynski, A.T., Giles-Corti, B., Karakiewicz, J.A., 2013. Effects of access
Department for Transport Local Government and the Regions (DTLR), 2002. Improv- to public open spaces on walking: is proximity enough? Landscape and Urban
ing Urban Parks, Play Areas and Green Spaces. Improving Urban Parks, Play Areas, Planning 117, 92–99.
London, 216 pp. Kowarik, I., 2011. Novel urban ecosystems, bio diversity, and conservation. Environ-
EEA, 2010. Urban Atlas, Available at http://www.eea.europa.eu/ mental Pollution 159, 1974–1983.
data-and-maps/data/urban-atlas (last accessed 29.06.2013). La Rosa, D., Privitera, R., 2013. Characterization of non-urbanized areas for land-use
El-Geneidy, A.M., Levinson, D.M., 2006. Access to Destinations: Development of planning of agricultural and green infrastructure in urban context. Landscape
Accessibility Measures. Report # 1 in the Series Access to Destinations Study, and Urban Planning 109, 94–106.
Available at http://www.lrrb.org/PDF/200616.pdf (last accessed 29.06.2013). Lindsey, G., Maraj, M., Kuan, S., 2001. Access, equity, and urban greenways: an
Fang, C.F., Ling, D.L., 2003. Investigation of the noise reduction provided by tree belts. exploratory investigation. The Professional Geographer 53, 332–346.
Landscape and Urban Planning 63, 187–195. McCormack, G.R., Rock, M., Toohey, A.M., Hignell, D., 2010. Characteristics of urban
Frank, A.U., 1992. Qualitative spatial reasoning about distances and directions in parks associated with park use and physical activity: a review of qualitative
geographic space. Journal of Visual Languages and Computing. 3, 343–371. research. Health and Place 16, 712–726.
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011
G Model
ECOIND-1742; No. of Pages 13 ARTICLE IN PRESS
D. La Rosa / Ecological Indicators xxx (2013) xxx–xxx 13
Moseley, D., Marzano, M., Chetcuti, J., Watts, K., 2013. Green networks for people: Sugiyama, T., Leslie, E., Giles-Corti, B., Owen, N., 2008. Association of neighbourhood
application of a functional approach to support the planning and management greenness with physical and mental health: do walking, social coherence and
of greenspace. Landscape and Urban Planning 116, 1–12. local social interaction explain the relationship? Journal of Epidemiology and
Nicholls, S., Shafer, C.S., 2001. Measuring accessibility and equity in a local park Community Health 62 (5), e9, http://dx.doi.org/10.1136/jech.2007.064287.
system: the utility of geospatial technologies to park and recreation. Journal of Talen, E., Anselin, L., 1998. Assessing spatial equity: an evaluation of measures of
Park and Recreation Administration 19 (4), 102–124. accessibility to public playgrounds. Environment and Planning 30, 595–613.
Nowak, D.J., Crane, D.E., Stevens, J.C., 2006. Air pollution removal by urban trees and Thompson, C.W., 2002. Urban open space in the 21st century. Landscape and Urban
shrubs in the United States. Urban Forestry & Urban Greening 4, 115–123. Planning 60, 59–72.
Paquet, C., et al., 2013. Are accessibility and characteristics of public open spaces Ward Thompson, C., 2011. Linking landscape and health: the recurring theme. Land-
associated with a better cardiometabolic health? Landscape Urban Planning, scape and Urban Planning 99, 187–195.
http://dx.doi.org/10.1016/j.landurbplan.2012.11.011. Wells, B.L., Gradwell, S., 2001. Gender and resource management: Community
Pham, T.-T.-H., Apparicio, P., Séguin, A.-M., Landry, S., Gagnon, M., 2012. Spatial dis- supported agriculture as caring-practice. Agriculture and Human Values. ,
tribution of vegetation in Montreal: an uneven distribution or environmental 107–119.
inequity? Landscape Urban Planning 107, 214–224. Wright Wendel, H.E., Zarger, R.K., Mihelcic, J.R., 2012. Accessibility and usability,
Privitera, R., Martinico, F., La Rosa, D., Pappalardo, V., 2013. A proposal for an Urban Green space preferences, perceptions, and barriers in a rapidly urbanizing city
Green Infrastructure in a medium sized city in southern Italy. Nordic Journal of in Latin America. Landscape Urban Planning 107, 272–282.
Architectural Research (in press). Yang, J., McBride, J., Zhou, J., Sun, Z., 2005. The urban forest in Beijing and its role in
Regione Sicilia, 2009. Volo fotogrammetrico, lidar, ortofoto ecartografia, air pollution reduction. Urban Forestry & Urban Greening 3, 65–78.
http://www.sitr.regione.sicilia.it/images/docs/volo sicilia digit.pdf (last Yao, X., Thill, J.C., 2005. How far is too far? – A statistical approach to context-
accessed: 29.06.2013). contingent proximity modeling. Transactions in GIS 9 (2), 157–178.
Sander, H.A., Ghosh, D-, van Riper, D., Manson, S.M., 2010. How do you measure dis- Young, R.F., 2010. Managing municipal green space for ecosystem services. Urban
tance in spatial models? An example using open-space valuation. Environment Forestry & Urban Greening 9, 313–321.
and Planning B 37, 874–894. Zasada, I., 2011. Multifunctional peri-urban agriculture – a review of societal
Schipperijn, J., Ekholm, O., Stigsdotter, U.K., Toftager, M., Bentsen, P., Kamper- demands and the provision of goods and services by farming. Land-Use Policy
Jørgensen, F., Randrup, T.B., 2010. Factors influencing the use of green space: 28, 639–648.
results from a Danish national representative survey. Landscape and Urban Zhan, F.B., Noon, C.E., 1998. Shortest path algorithms: an evaluation
Planning 95, 130–137. using real road networks. Transportation Science 32 (1), 65–73,
Shin, D-h., Lee, K-s., 2005. Use of remote sensing and geographical information sys- http://dx.doi.org/10.1287/trsc.32.1.65.
tem to estimate green space temperature change as a result of urban expansion.
Landscape and Ecological Engineering. 1, 169–176.
Please cite this article in press as: La Rosa, D., Accessibility to greenspaces: GIS based indicators for sustainable planning in a dense
urban context. Ecol. Indicat. (2013), http://dx.doi.org/10.1016/j.ecolind.2013.11.011