Frontiers of Architectural Research (2015) 4, 35–45

H O S T E D B Y Available online at www.sciencedirect.com

http://dx.doi.2095-2635/& 2

nCorrespondE-mail addPeer review

www.elsevier.com/locate/foar

RESEARCH ARTICLE

Planting design for urban parks: Space syntaxas a landscape design assessment tool

Ayman Hassaan Mahmouda,n, Reham H. Omarb

aDepartment of Architecture, Faculty of Engineering, Cairo University, Giza 12211, EgyptbThe Canadian International College, Egypt

Received 27 February 2014; received in revised form 31 August 2014; accepted 1 September 2014

KEYWORDSLandscape design;Space syntax;Planting;Urban parks;Cairo

org/10.1016/j.foar.2014014. Higher Education P

ing author. Tel.: +20 10ress: arch.ayman@yahoounder responsibility of

AbstractTrees are a major factor in defining the spatial qualities of outdoor spaces. This studyinvestigates the influence of spatial configuration produced by the proposed tree plantingdesign on the visual fields of an urban park using space syntax theory. Space syntax theoryassumes that an urban area can be represented as a matrix of connected spaces. Thequantitative properties of this matrix in the form of syntactic measures can be measuredusing computer simulations. This research investigates how space syntax techniques can helpassess the effect of tree configurations on the social structure of a small-scale garden in anurban park. Such techniques are assumed useful in predicting the social structure of theproposed space and in assessing design alternatives. An experimental study using threedifferent planting design proposals for an urban park is conducted. Data are analyzed usingspace syntax techniques. Results reveal a significant effect of planting configurations onsyntactic measures across the three proposals. This study suggests that space syntax techniquesmay have a significant value in the assessment of schematic planting design, especially at thepedestrian movement level.& 2014. Higher Education Press Limited Company. Production and hosting by Elsevier B.V. Allrights reserved.

1. Introduction

The current literature has investigated the influenceof landscape vegetation on the frequency and range of

.09.001ress Limited Company. Productio

06670260..co.uk (A.H. Mahmoud).Southeast University.

movement (Shriver, 1997; Foltête and Piombini, 2007;Chamberlain and Meitner, 2013) and on the choices ofroutes in specific places such as public gardens (Loiteronand Bishop, 2005). However, these studies are focused onpedestrian behavior in small and closed spaces (Zacharias,1997; Moskell and Allred, 2013) and in unknown areas(Zacharias, 2006). Generalizing these observations in large,familiar environments is necessary to distinguish the pro-cesses of space perception that affect movement. Research

n and hosting by Elsevier B.V. All rights reserved.

A.H. Mahmoud, R.H. Omar36

on movements in an intraurban environment must integratea large amount of qualitative data to generate useful inputsfor urban planning (Røe, 2000; Stahle and Marcus, 2010),such as landscapes. Such approach will help in understand-ing how individuals choose their routine trips in an environ-ment and how social structure is affected bypedestrian trips.

Visual representation and simulation techniques aretraditionally used to help designers in the design process.These techniques help in visualizing design concepts andin exploring how proposed layouts might work in reality.This study investigates the use of identifying the visualproperties of spatial configurations initiated by plantingdesigns in a landscape setting.

1.1. Aim of the study

This study investigates the effect of the planting design ofurban parks based on the assumptions of a space syntaxmodel, especially at the pedestrian movement level. Thismethod may have an extensive value in the evaluationprocess at the schematic planting design stage. The mainquestion motivating this study is whether a planting designmethod can be developed by integrating space syntaxtechniques. In this context, the space syntax technique ofthe visibility graph analysis (VGA) is adopted. This techniqueoffers the possibility of estimating theoretical accessibilityor “natural movement” by measuring integration. VGAmeasures can be used to study the effect of the spatialconfiguration of the proposed tree planting design forlandscape spaces on visual fields and expected pedestrianmovement. This technique is supposed to help in predictingthe visibility and accessibility of planting design schemes ofan urban park.

2. Background

2.1. Planting design

Robinson (2004) showed that plant size relative to thedimensions of the human figure is critical when designingspaces for people. An important design stage is distinguish-ing areas on a plan using canopy height because plant heightestablishes much of the spatial framework and controlsvision, movement, and physical experience (Robinson, 2004,p. 28). Preben Jakobsen (1977), a Danish landscape archi-tect, identified the most useful size categories for designersas ground level, up to knee height, knee to waist height,and below or above eye level. Dividing trees according tosize (small: mature height, 5–10 m; medium: 10–20 m; tall:20 m) is helpful for design purposes. The present researchfocuses on small mature and medium trees, which create avisual and physical barrier. Medium trees can create spacesthat contain small structures and therefore have a signifi-cant influence on the spatial structure of urban landscape(Austin, 1982; Conway and Bourne, 2013).

Trees are capable of defining space and increasing thedegree of enclosure (Booth, 1983; Bell, 2004). The spatialarrangement of trees affects the visual field of a space.Salingaros (1999) reported that the use of urban space is

linked to the information field generated by surroundingsurfaces and to how easily the information can be receivedby pedestrians. Successful urban spaces also offer tactileinformation from local structures meant for standing andsitting. The total information field in turn determines theoptimal positioning of pedestrian paths and nodes.

Tree placement is a key element in urban landscapearchitectural design. Trees are one of the tools for definingoutdoor open space. They provide shape and configurationto spatial environments. Many aspects must be consideredwhen proposing sites for tree planting in urban areas(Wua et al., 2008). The important site conditions relatedto tree selection include climate factors, soil characteristics,environmental conditions, planting space, site location,existing vegetation, esthetics, land ownership and regula-tions, social influences, and maintenance requirements(Bassuk and Trowbridge, 2004).

Gilman (1997) recommended various minimum widths ofplanting sites for trees with different full-grown sizes: 3–4 ft(0.90–1.20 m) for small trees, 4–6 ft (1.20–1.80 m) for med-ium trees, and over 6 ft (1.80 m) for large trees, with 1 ft(30 cm). These recommendations were later supported byKirkpatrick et al. (2011). Arnold (1980) encouraged thecollective use of trees in groves, rows, and symmetricalunits in urban design and explained the esthetic principlesused in grouping trees in a variety of settings. Wu et al.(2008) developed a computer program to iteratively search,test, and locate potential sites for tree planting by virtuallyplanting large, medium, and small trees on plantable areas,with large trees given priority because of the great benefitsaccrued from them.

2.2. Simulation of visual fields in environmentaldesign

In environmental design studies, visibility techniques havebeen incorporated with advances in agent-based simulationand virtual reality (VR) technology. For example, Penn et al.(1997) used “agents” that navigate through VR environ-ments to retrieve isovists, which represent the measures ofvisible space throughout configurations and the associatedvisual fields through the space that they produce (Benedikt,1979a, 1979b). On the contrary, Batty and Jiang (1999)developed a parallel “agent system” to construct isovists.Extending the work on isovists, Conroy (2001) investigatedthe formal visual properties of paths that people takewithin controlled experiments in VR environments. Fisher-Gewirtzman and Wagner (2002) suggested that visibilityanalysis should cover and quantify the third dimension asBenedikt (1979a, 1979b) managed to achieve for twodimensions. They proposed a three-dimensional analysisof the “spatial openness” (or isovist volume) of viewsfrom flats. However, they restricted themselves to a two-dimensional isovist area because of inadequate computingprocessing capacity. Teller (2003) suggested an innovativesolution to overcome the problem of computational com-plexity in a three-dimensional analysis. He investigated theangular area of the sky observable throughout the urbanenvironment to analyze “sky opening” from locations. Thismethod offers a scale-invariant measure of the environ-ment, which Teller recommended as a practical way to

37Planting design for urban parks

avoid the scale dependency of many measures of isovists,such as area or visual integration. He suggested that apartfrom offering examples of typical values, the method is alsouseful for classifying town squares and investigating theeffect of new buildings on the perception of the environ-ment. A related approach was established by Ratti (2002),who mapped the distribution of perceived sky area fromvarious positions that give a weighted directional measureof visual openness.

Turner (2003a, 2003b) argued that although these inno-vative techniques add a three-dimensional quality, theycannot be broken down into the recent uses of visibilityanalysis, which can be divided into two categories: (1) clas-sification and (2) assessment of the perceptual qualities ofthe environment. The present study extends the argumentthat the visibility analysis of the proposed planting designscould detect the specific effects of spatial propertiescaused by various planting arrangements in landscapedesign.

2.3. Space syntax theory

Space syntax is a research program that studies thecorrelation between human societies and space from theperspective of the general theory of the structure ofpopulated space in all its different forms: buildings, settle-ments, cities, and landscapes (Hillier, 1996; Hillier andHanson, 1998; Bafna, 2003; Koohsari et al., 2013). Spacesyntax is also defined as a graph-based theory used byarchitects and urban designers to examine how the spatiallayout of buildings and cities influences the economic,social, and environmental outcomes of human movementand social interaction (Dawson, 2002). Its techniques offerprecise quantitative descriptions of the way in which thebuilt spaces of a setting are organized (Hillier et al., 1983).According to Hillier and Hanson (1984), the social meaningof the environment arises from spatial composition, and thetopological structure of an environment is a primary ele-ment by which a society creates and establishes roles todevelop some types of social relationships. Therefore,constructed environmental spatial patterns incorporateand give shape to social patterns.

Space syntax is built on two recognized ideas thatattempt to reflect both the objectivity of space and ourintuitive engagement (Hillier and Vaughan, 2007). The firstidea is that we should think of space not as the backgroundto human activity, as we think of it as the background toobjects, but as an essential aspect of everything humanbeings do in the sense of moving through space. Movementis essentially linear, and interface requires a “convex” spacein which all points can see all others and a variably shaped,often sharp, visual field referred to as an “isovist” can beseen from any point in space (Benedikt, 1979a, 1979b). Thesecond idea, called “configuration of space,” indicates thathuman space is not only about the characteristics ofindividual spaces but also about the interrelations betweenthe many spaces that make up the spatial layout of abuilding or a city.

The current study investigates the effect of the spatialconfiguration of trees on the development of design propo-sals for urban parks using space syntax theory. The results

of this work could be used to generate a strategic designframework based on a detailed analysis of spatial con-figuration.

2.4. Space syntax methods

A fundamental principle within space syntax theory is thatsocial structure is fundamentally spatial and, inversely, theconfiguration of populated space has a primarily social logic.This principle is very important in landscape design, whichseeks to create an outdoor environment for people toaccommodate their social interaction.

Converting the landscape space to a discrete configura-tion is helpful because different labels can be applied to itsindividual parts. These parts can then be allocated todifferent groups, people, or activities. Various rules ofbehavior and conventions can be related to various partsof the landscape space. Individual parts of a landscapespace can carry a specific symbolic or cultural value. Thispoint of view would usually imply that the planned land-scape space improves the readability of existing socialstructures. Space syntax theory rejects the straightforwardspace-as-form and society-as-content distinction (Hillierand Hanson, 1984). The fundamental concepts of spacesyntax include “integration” and “connectivity.”

2.4.1. IntegrationThe “integration” of space is a function of the mean numberof lines and changes in direction required to go from thatspace to all other spaces in the spatial system. Integration isaccordingly about “syntactic” and not “metric” accessibil-ity, and the expression “depth” rather than “distance” isused to illustrate how far a given space is from anotherspace. All lines in a spatial layout have certain depth valuesfrom every other line. The integration value of a line is amathematical way of expressing the depth of a line from allother lines in the system. These values significantly differfrom one line to the next but are one of the most significantproperties of architectural, urban, and landscape spatialconfigurations (Hillier and Hanson, 1984, 1998; Hillier,1996). According to Ortega-Andeane et al. (2005), a spaceis “integrated” when the other spaces have a relativeshallowness in relation to it. A space is “segregated” whenthe other spaces have a relative depth in relation to it.

2.4.2. ConnectivityThe connectivity of a node can be defined as the number ofother nodes directly connected to it. Connectivity, pathlength, and clustering coefficient are three key measuresfor a topological analysis of a given space. They constituteessential measures for exploring small-world and scale-freeproperties. Some studies have pointed out that integrationis closely related to human spatial behavior. A consistentrelationship has been found to exist between the spatialintegration measure of an urban space and the observedhuman movement that flows in it (Hillier et al., 1993a,1993b). Kim and Penn (2005) reported a consistent relation-ship between spatial cognition and syntactic integration inurban areas. A pedestrian movement rate in an outdoorspace can be predicted with other syntactic results fromspace syntax analysis. Several studies have been conducted

250 5000

Figure 1 Aerial photograph of the study area. Hatched areashows the experimental area.

A.H. Mahmoud, R.H. Omar38

to compare measured pedestrian movement rates andspatial integration (Hillier, 1996; Dawson, 2003; Raford,2003; Read, 2001). The results endorse such relationships.Evidence suggests that the space syntax methods of repre-sentation and measurement of the spatial configuration ofoutdoor spaces could be useful in studies of planting designbecause they influence pedestrian movement patterns inurban parks.

3. Modeling outdoor spaces using spacesyntax

Many methods of spatial analysis have been developed toimprove our understanding and modeling of real-worldexperience. Space syntax theory assumes that any urbanarea can be represented as a matrix of connected spacesand that the quantitative properties of this matrix can bemeasured using computer simulations. Space syntax modelsthe spatial configurations of outdoor urban spaces by using a“connectivity graph” representation. Such a configurationof space identifies patterns that can be used to study urbanlayouts and human behaviors (Jaing et al., 2002).

Visibility and visual perception significantly influence howpeople behave, appreciate, and experience the environ-ment. They could serve as determining factors in designinglandscape spaces. Space syntax studies of pedestrian beha-vior in building and open urban environments have found aconsistent correspondence between the configuration ofspace and the patterns of usage found within such space.Researchers in the space syntax community (Hillier et al.,1993a, 1993b; Penn et al., 1998; Desyllas and Duxbury,2001) claimed through several empirical studies that bothpedestrian and vehicle movement can be predicted by localintegration measure.

3.1. Visibility graph

Benedikt, 1979a, 1979b suggested isovists or viewshedpolygons as objectively determinable basic elements foranalyzing the spatial characteristics of spaces. Isovistsdepict spatial properties by illustrating the visible areafrom a single observation point. Several quantitativedescriptors that describe the physical properties of thecorresponding space can be derived from these viewshedpolygons; examples of these descriptors include area,perimeter length, number of vertices, and length of openor closed boundaries (Franz and Wiener, 2008). The short-coming of the method by Benedikt is that it only considersthe local properties of space and disregards the visualrelationship between the current location and the entirespatial environment (Turner et al., 2001). Turner et al.(2001) suggested that the experience of space is related toan overlay of isovists and that more than a single isovist isneeded to quantify the actual perceived space because theway an individual experiences and uses space involvesmotion. To effectively describe the spatial characteristicsof environments, Turner et al. (2001) developed VGA as atechnique that permits the integrative analysis of severalpositions inside an environment by calculating the intervi-sibility of positions regularly distributed over the entireenvironment. This technique suggests additional second-

order measurands, such as “clustering coefficient” and“integration” (global topology-oriented characteristicvalues). Another advantage of VGA is that its processanalysis can be completely automated (Benedikt, 1979a,1979b). Empirical evidence shows that isovists capture theenvironmental properties of space that are relevant forspatial behavior and experience. Desyllas and Duxbury(2001) reported that in open and less-defined outdoorspaces, a more significant correlation exists between VGAvisibility and pedestrian movement than between otherspace syntax analysis techniques, including an axial maprepresentation of space.

3.2. Pedestrian movement and agent-basedmodel

The agent-based model is among the most important spacesyntax techniques used to simulate pedestrian movement inan environment. Turner and Penn (2002) compared thenumber of visitors moving through an indoor architecturalspace with the number of “agents” progressing through anagent-based model of the same environment. They reportedthat the agents use natural movement rules by randomlyselecting a position from the current field of view andmoving toward it for several steps before redefining thedestination. Turner (2003a, 2003b) applied a similar modelto an urban environment. He recommended that in anoutdoor environment, arrival zones and duration periodsin the system that are closely correlated with pedestrianmovement patterns must be carefully selected. The sig-nificance of the analytic method of Turner (2003a, 2003b) isthat the agent-based model suggests to focus on the humanaccessible topology as invoked through the process ofinhabitation, accordingly simplifying the tools an architector environmental designer requires to assess the socialadequacy of a specific project. The current study attemptsto extend this assumption to the context of landscapedesign by assessing the effect of a proposed planting layouton space inhabitation and its feedback on design proposals.

39Planting design for urban parks

4. Methods

4.1. Study area

El-Qanater Gardens is a large rural environment located20 km northeast of Greater Cairo, Egypt. It is popular amongmiddle-class families as a destination for one-day trips andoutdoor recreation. Figure 1 shows a satellite image of thestudy area and the immediate surroundings of El-QanaterGardens. The area highlighted by a hatched pattern is thespecific area of study. This area is a primary destination andhas a high density of users and activities. It can beconsidered in the practical application of space syntaxtheory in landscape design. Three different plantingschemes are proposed as subjects of this study. A detailed

5 0 5 10 15m North 5

5 0 5 10 15m

Figure 2 (a) Planting design proposal no. 1. (b) Planting d

analysis of space use pattern and a quantitative analysis ofeach planting design proposal for El-Qanater Gardens areconducted using the VGA technique (Figure 2).

4.2. Measures and techniques

The spatial analysis technique applied in this study is basedon the analysis of the visibility polygons drawn from eachgrid of tiles covering all accessible areas (Turner et al.,2001). One global measure, “visual integration,” and onelocal measure, “visual clustering coefficient,” are used toinvestigate a large landscape area, that is, El-QanaterGardens, Cairo. Similar to global integration, visual integra-tion explains the relationship of each space to the spacenetwork as a whole. It describes the relative visibility of

0 5 10 15m North

North

esign proposal no. 2. (c) Planting design proposal no. 3.

A.H. Mahmoud, R.H. Omar40

a point location to all the other points within the space.This “visual clustering coefficient” measure is importantbecause it has been correlated with “pedestrian movementgate counts” in previous studies (Batty et al., 1998; Turneret al., 2001; Desyllas and Duxbury, 2001).

The clustering coefficient of a given location quantifiesthe degree of convexity or converse multidirectionality ofthe isovists generated from that location. It is a result of theratio of the actual direct connections of nodes within theisovist to the maximum possible connections of the samenumber of nodes if they were in a convex arrangement. Itsvalue ranges from 0 (extremely “rough” multidirectionalisovists) to 1 (perfectly convex isovists). The clusteringcoefficient is an important spatial indicator of the potentialperceptuality of copresence and, therefore, the potential toform groups or to interact (Turner et al., 2001; Franz andWiener, 2008).

This study conducted VGA using Depthmap 9.0, a compu-ter program for visibility analysis, to investigate the land-scape (Turner, 2001, 2004; Mahmoud and Brown, 2009).A comparative method was developed to investigate theeffect of planting pattern, including the visibility propertiesof the three planting design proposals in a single statisticalmodel. Multiple regression and correlation analysis wasperformed using SPSS to test the hypothesis.

Three planting design proposals were developed. Allthree proposals used the same number of trees. Therecommendations of Gilman (1997) for tree spacing werefollowed. The first proposal applied rectangular array(Arnold, 1980). The second proposal used curved, free-form configuration. The third proposal applied rectangulararray with the addition of a central space along the mainwalkway to test the effect of Bell's proposition that “node”can increase the balance and visual inertia of landscapespace (Bell, 2004).

4.3. VGA of the study area

A visibility graph of the pedestrian movement spaces of thestudy area was developed using Depthmap 9.0 to identify thespace syntax measures. The visibility graph was constructedfor all locations (nodes) specified on an evenly spaced grid of1.50 m, covering publicly visible and accessible areas. Aplanar isovist—visibility or accessibility field—was generatedfrom each grid point to produce a network of all directconnections between nodes. A visibility graph was thenderived and analyzed to acquire the patterns of configura-tional and visual properties (Turner et al., 2001). The land-scape features one level of publicly accessible space. In thegray scale, the most integrated and shallowest nodes onaverage are shown in dark gray, and the least integrated anddeepest nodes on average are depicted white. The visualintegration shows the pedestrian spaces within the proposedpark design. Circulation spaces connecting the park spacesserved as the focus of integration in the area.

Intelligibility analysis using the correlation between con-nectivity and integration provided a clue of how the entirespatial configuration is clear to its users (Syntax, 2004).Previous studies have shown that visual integration issignificantly correlated with movement (Parvin et al.,2007; Desyllas and Duxbury, 2001).

4.4. Gate count analysis

In this context, the calculations were applied to theproposed environment. A theoretical gate count must beused to estimate the proposed movement pattern. TheDepthmap “agent” tool was used to calculate the expected“gate count” by predicting pedestrian movement within thethree proposals based on the spatial configuration of trees.This tool was developed by Turner (2003a, 2003b) and hasbeen found useful in predicting theoretical movement(Hillier and Iida, 2005; Bada and Guney, 2009).

Numerical data were exported from Depthmap in atabulated format and inserted in the SPSS software todetect the difference between the proposals using spacesyntax measures. The data of each design proposal weregrouped into a single SPSS file. Statistical analysis ofvariance (ANOVA) was performed to reveal the effect ofspatial configuration on space syntax measures.

5. Results

5.1. Intelligibility analysis

The relationships between planting design and the charac-teristics of spatial space and pedestrian movement wereanalyzed based on the results from the VGA and the data onthe case study area. Figure 3 shows a graphical comparison ofthe proposed planting designs based on the results of thespace syntax analysis. According to the results of the VGA,the second proposal showed the highest value of globalintegration shown as curved pathways (Table 1). Both thefirst and the third proposals showed nearly equal values ofglobal integration. In the first proposal, the measure of visualintegration varied from 13.23 at the entrance and reached itspeak at 19.57 at the space in front of the kiosk. In the secondproposal, an emergent warm color in the visual integrationmap resulted in the left portion of the park to have a highervalue of visual integration (22.55) compared with the lowmeasure at the entrance space (14.77).

The measure of clustering coefficient was mostly consis-tent along the pathway in the first proposal. In the secondproposal, the measure of clustering coefficient decreasedalong the pathway starting from 0.545 at the entrancespace to 0.494 at the space in front of the kiosk. Bycontrast, the third proposal showed a consistent increasein clustering coefficient from 0.465 at the entrance to 0.527at the kiosk.

The measure of mean depth in the VGA was stronglycorrelated with visual integration. By comparing the coeffi-cients of determination (R2), we found that the firstproposal was the highest (R2=0.9710), followed by thethird proposal (R2=0.9637) and the second proposal(R2=0.9511). The measurement of mean depth was alsostrongly correlated with connectivity in the three proposals(R2=0.9998, 0.9989, and 0.9997). However, the strengthof the correlation between mean depth and clusteringcoefficient varied. The correlation was stronger in thethird proposal (R2=0.2592) than in the first proposal(R2=0.2283). By contrast, the correlation was weak in thesecond proposal (R2=0.0019).

Proposal 1 Proposal 2 Proposal 3C

onne

ctiv

ityC

lust

erin

g co

effic

ient

Vis

ual I

nteg

ratio

n (H

H)

Gat

e C

ount

s

Figure 3 Graphical representations of the VGA of three planting design proposals.

41Planting design for urban parks

Figure 4a–c shows the results of the intelligibility analysis(i.e., correlation between connectivity and visual integra-tion). The correlation coefficient (R2) between the visualconnectivity and the visual integration of point was sig-nificant in the first proposal (R2=0.9737) and in the thirdproposal (R2=0.9719). It was not significant in the secondproposal (R2=0.9559). The correlation between visual sta-bility (clustering coefficient) and connectivity was strongerin the third proposal (R2=0.2535), which integrates gridplanting with a central curved space, than in the firstproposal (R2=0.1146), which includes a rectangular array

of plantation. The correlation was inadequate in the secondproposal (R2=0.0009).

5.2. The effect of spatial configuration on visualfields

The effect of spatial configuration on visual fields (repre-sented by space syntax measures) was analyzed with ANOVAusing SPSS software (Table 1). Spatial configuration wasfound to significantly affect all visual fields (po0.001).

Table 1 Analysis of variance, influence of spatial configuration on syntactic measures, and comparisons of mean scores.

Factor Levels Mean F-value

Connectivity VisualintegrationH–H

Visualclusteringcoefficient

Connectivity VisualintegrationH–H

Visualclusteringcoefficient

Gatecount

Spatialconfiguration

Rectangular array 1525.08 15.8304 0.5161 6.934** 43.500** 60.110** 3.278**

Curved composition 1515.57 16.2719 0.5262Central node 1498.34 15.9128 0.5168

nnpo0.001.

0 20000

19.8792R² = 0.973755

y = 0.0063267 x + 6.1817

Connectivity

Vis

ual I

nteg

ratio

n [H

H]

0 2208

0

23.2982 y = 0.00713647 x + 5.45608

Connectivity

Vis

ual I

nteg

ratio

n [H

H]

0 20180

20.6451 R² = 0.967196y = 0.00648342 x + 6.1981

Connectivity

Vis

ual I

nteg

ratio

n [H

H]

- -R² = 0.955907

Figure 4 (a) Correlation between visual integration and connectivity measures of the first proposal. (b) Correlation between visualintegration and connectivity measures of the second proposal. (c) Correlation between visual integration and connectivity measuresof the third proposal.

A.H. Mahmoud, R.H. Omar42

The different alignments of trees in the proposed designsresulted in different spatial configurations. This condition isexpected to influence the visual fields experienced by userswhen they encounter the spaces in reality.

6. Discussion

This study describes a method for determining how plantingdesign can influence the visual fields through three plantingdesign proposals of an urban park using space syntaxtechniques. The proposals were developed using three

different approaches. The first proposal followed thesymmetric, rectangular array of trees and straight path-ways. The second proposal followed a curved pathway and arandom grid pattern. The third proposal applied a combina-tion of symmetric rectangular array and a circular space atthe center of a straight pathway.

Is it possible to predict the social structure of space usingspace syntax techniques? The answer appears to be positive.The first proposal, with the straight linear pathway, did notshow significant changes in the visual integration analysisthrough Depthmap. The straight line and symmetric layoutdid not generate a variety of visual fields. In the second

Figure 5 (a) Location of the most connected area of the design proposal. (b) Connectivity analysis of second proposal.

43Planting design for urban parks

proposal, the visual integration analysis showed an emer-gent warm color, which appeared as a spot resembling amain node connected to the western side of the pathwaythrough the designed park. Salingaros (1999) stated thatspatial network appears to have structure and a strongintegration with the global space at this portion of land-scape space. Conway and Bourne (2013) suggested that thisresult could inform the designer to insert an attractivevisual component, e.g., colorful flower composition or apiece of sculpture. Bell (2004) suggested that the height ofthis element should not exceed eye level to avoid theobstruction of the visual field.

In the third proposal, visual integration analysis showedan emergent warm color, which appeared like a main nodealong the pathway through the designed park. The spatialnetwork appeared to have structure and strong integrationat this space. This finding agrees with that of Bell (2004)given that the circular arrangement of trees created adegree of enclosure defining the space. This condition couldmotivate the designer to add a visual element, e.g., a waterfountain below eye level to maximize and emphasize thisvisual integration.

The results of the intelligibility analysis showed that thethird proposal integrated a grid-planting pattern witha central curved space and thus provided a strongercorrelation between visual stability (clustering coefficient)and connectivity compared with the first proposal, whichincluded a rectangular array of plantation. The secondproposal showed inadequate correlation. This resultrevealed an area within the second proposal that showeda high degree of connectivity within the space (Figure 5aand b). This outcome may guide designers in making use ofthis advantage and adding some design elements within thisarea to enhance design esthetics. Furthermore, the thirdproposal created a space system with a potential forcopresence and interaction at the central space. The twoother proposals did not show sufficient clustering potential.

6.1. The social structure of space

The literature indicates that visual integration is signifi-cantly correlated with movement (Parvin et al., 2007;Desyllas and Duxbury, 2001). The “agent” facility in theDepthmap software was used to show the traces of 500sighted agents with 170 degrees of vision who randomlyselect a point within their field of vision, move three pixels(2.4 m) toward it, and repeat the process (Turner (2003a,2003b); Bada and Guney, 2009). In the first proposal, thedensity of traces followed the central pathway structure toa remarkable degree. In the second proposal, the densityshifted to the western side of the park and away from themain pathway, thus allowing further exploration of the parklandscape. This process reflected the local scaling of spacesrather than the overall configuration. In the third proposal,the density of traces was the highest in the central spacealong the pathway, thereby emphasizing the focal point atthe central space and reflecting the local scaling of theoverall configuration.

The curved and central node alternatives will satisfyseveral visual aspects, including connectivity and increasedintegration in the global context. They are also expected toencourage social activities in large portions of the park. Bycontrast, the grid alternative will develop a rather staticspace where the central walkway is isolated and does notpermit further exploration of the rest of the park.

Although trees are one of the most important factorsinfluencing landscape preference, their spatial configurationsinfluence visual fields. This finding confirms that of Wua et al.(2008). This contribution will affect pedestrian movementand the social structure of the landscape. Landscape archi-tects should not limit park planting designs to symmetriclayouts because curved and free-form arrangements mayenable enhanced visual and social experiences within a space.Such considerations correspond to the findings of Bell (2004)that curved free-form shapes contribute to a highly dynamic

Tree Planting Proposal

CAD version of the proposal

UCL Depthmap Visibility Graph Analysis

Test the Tree Planting

Develop Tree configuration

Final Tree Planting Proposal

Predict the Social Space

Not Compatible

Compatible

Figure 6 A developed process of planting design of an urbanpark.

A.H. Mahmoud, R.H. Omar44

and interesting landscape experience. Figure 6 shows thesuggested development of the design phase. This methoddevelops the designers' understanding and prediction of theoutcome of their form-making decisions. The proposedpattern for tree planting was tested using UCL Depthmap todiscover visibility and social structure. This assessment willguide designers in predicting the visual and social structure ofthe design. In this way, the process of design enhancementreaches an optimum solution based on the prediction of thesocial structure of space.

This research is limited to the two-dimensional configura-tion of the landscape pattern. However, it is still useful in theearly schematic design stages, in which designers explore theconsequences of the various geometric forms of their designs.Further research is required to investigate the effect of thethree-dimensional configuration of tree canopies.

7. Conclusions

Tree planting design was found to have a significant effecton visual fields in urban parks. Although the experimentalstudy sought to understand different issues, the results ledto a general conclusion that the curved free-form arrange-ment of trees and, to a certain extent, the central spatialnode of trees generated relatively integrated and con-nected spaces within the park design. The space syntaxtechniques used in this study could be used to enhancedesign proposals. A careful study of visual fields embeddedin design proposals could help predict pedestrian movementand social structure of proposed designs. The results of suchstudy could aid landscape designers in understanding howtheir designs will work. In this context, the space syntaxapproach can be used in future research to investigate otherspatial parameters, including tree heights, colors, texture,and form.

Acknowledgement

This work was carried out using Depthmap 9.0. The aca-demic license was obtained with kind permission from theSpace Syntax Organization.

References

Arnold, H.F., 1980. Trees in Urban Design. Van Nostrand, Reinhold,NY.

Austin, R.L., 1982. Designing with Plants. Van Nostrand, Reinhold,NY.

Bada, Y., Guney, Y.I., 2009. Visibility and spatial use in urban plazas:A case study from Biskam Algeria. In: Proceedings of the 7thInternational Space Syntax Symposium. Koch, D., Marcus, L.,Steen, J., (Eds.). Stockholm, Sweden.

Bafna, S., 2003. Space syntax: a brief introduction to its logic andanalytical techniques. Environ. Behav. 35 (1), 17–29.

Bassuk, N.L., Trowbridge, P.J., 2004. Trees in the Urban Landscape:Site Assessment, Design and Installation. Wiley, Hoboken, NJ.

Batty, M., Conroy, R., Hillier, B., Jiang, B., Desyllas, J., Mottrom,C., Smith, A., Turner, A., 1998. The Virtual Tate, CASA WorkingPapers (5), Centre of Advanced Spatial Analysis, UCL. London,UK.

Batty, M., Jiang, B., 1999. Multi-agent simulation: new approachesto exploring space – time dynamics within GIS. CASA WorkingPaper (10), Centre for Advanced Spatial Analysis, UniversityCollege London. London.

Bell, S., 2004. Elements of Visual Design in the Landscape. SponPress, London.

Benedikt, M.L., 1979a. To take hold of space: isovists and isovistsfields. Environ. Plan. B: Plan. Des. 6, 47–65.

Benedikt, M.L., 1979b. To take hold of space: isovist & isovist fields.Environ. Plan. B: Plan. Des. 6, 47–65.

Booth, N., 1983. Basic Elements of Landscape Architectural Design.Wavefront, Illinois.

Chamberlain, B.C., Meitner, M.J., 2013. A route-based visibilityanalysis for landscape management. Landsc. Urban Plan. 111,13–24.

Conroy, R.A., 2001. Spatial Navigation in Immersive Virtual Envir-onments (Ph.D. thesis). Bartlett School of Graduate Studies,University College London, London.

Conway, T.M., Bourne, K.S., 2013. A comparison of neighborhoodcharacteristics related to canopy cover, stem density andspecies richness in an urban forest. Landsc. Urban Plan. 113,10–18.

Dawson, P.C., 2002. Analysis of Central Inuit snow houses. J.Anthropol. Archaeol. 21, 464–480.

Dawson, P.C., 2003. Analyzing the effect of spatial configuration onhuman movement and social interaction in Canadian Articcommunities. In: Proceedings of the 4th International SpaceSyntax Symposium. London.

Desyllas, J., Duxbury, E., 2001. Axial maps and visibility analysis: acomparison of their methodology and use in models of urbanpedestrian movement. In Proceedings of the 3rd InternationalSpace Syntax Symposium, Georgia Institute of Technology,Atlanta, GA (p. 27-1).

Fisher-Gewirtzman, D., Wagner, I.A., 2002. Spatial openness as apractical metric for evaluating built-up environments. Environ.Plan. B: Plan. Des. 30, 37–49.

Foltête, J., Piombini, A., 2007. Urban layout, landscape featuresand pedestrian usage. Landsc. Urban Plan. 81 (3), 225–234.

Franz, G., Wiener, J.M., 2008. From space syntax to spacesemantics: a behaviorally and perceptually oriented methodol-ogy for the efficient description of the geometry and typology ofenvironments. Environ. Plan. B: Plan. Des. 35, 574–592.

Gilman, E.F., 1997. Trees for Urban and Suburban Landscapes.Delmar Publishers, Albany, NY.

Hillier, B., 1996. Space is the Machine. Cambridge University Press,Cambridge.

Hillier, B., Hanson, J., 1984. The Social Logic of Space. Cambridge:Cambridge University Press, Cambridge.

45Planting design for urban parks

Hillier, B., Hanson, J., 1998. Space Syntax as a research pro-gramme. Urban Morphol. 2, 108–110.

Hillier, B., Hanson, J., Peponis, J., Hudson, J., Burdet, R., 1983.Space syntax. A different urban perspective. Architect's J.November, 51–63.

Hillier, B., Iida, S., 2005. Network effects and psychological effects:a theory of urban movement. In: Proceedings of the 5thInternational Space Syntax Symposium, University of Delft.Delft, the Netherlands, pp. 553–564.

Hillier, B., Penn, A., Hanson, J., Grajewski, T., Xu, J., 1993a.Natural movement: configuration and attraction in urban pedes-trian movement. Environ. Plan. B: Plan. Des. 20, 29–66.

Hillier, B., Penn, A., Hanson, J., Grajewski, T., Xu, J., 1993b.Natural movement: or, configuration and attraction in urbanpedestrian movement. Environ. Plan. B: Plan. Des. 20, 29–66.

Hillier, B., Vaughan, L., 2007. The city as one thing. Prog. Plan. 67(3), 205–230.

Jaing, B., Claramunt, C., Klarqvist, B., 2002. Integration of spacesyntax into GIS for modelling urban spaces. Int. J. Appl. EarthObs. Geoinf. 2, 161–171.

Jakobsen, P., 1977. Shrubs and Ground Cover. In: Clouston, B. (Ed.),Landscape Design with Plants. Heinemann, London.

Kim, Y.O., Penn, A., 2005. Linking the spatial syntax of cognitivemaps to the spatial syntax of the environment. Environ. Behav.36 (4), 483–504.

Kirkpatrick, J.B., Daniels, G.D., Davison, A., 2011. Temporal andspatial variation in garden and street trees in six easternAustralian cities. Landsc. Urban Plan. 101 (3), 244–252.

Koohsari, M.J., Kaczynski, A.T., Giles-Corti, B., Karakiewicz, J.A.,2013. Effects of access to public open spaces on walking: isproximity enough? Landsc. Urban Plan. 117, 92–99.

Loiteron, D., Bishop, I., 2005. Simulating visitor movement: auton-omous agents in urban public garden. In: Proceedings ofCUPUM'05: Computers in Urban Planning and Urban Management(University College of London, London. ⟨http://128.40.111.250/cupum/searchpapers/index.asp⟩ (accessed 25.05.09).

Mahmoud, A.M., Brown, F.E., 2009. The urban grid and its con-sequences Assiut: an Egyptian case-study. In: Koch, D., Marcus,L., Steen, J. (Eds.), Proceedings of the 7th International SpaceSyntax Symposium.

Moskell, C., Allred, S.B., 2013. Residents' beliefs about responsi-bility for the stewardship of park trees and street trees in NewYork City. Landsc. Urban Plan. 120, 85–95.

Ortega-Andeane, P., Jime´nez-Rosas, E., Mercado-Dome´nech, S.,Estrada-Rodrı´guez, C., 2005. Space syntax as a determinant ofspatial orientation perception. Int. J. Psychol. 40 (1), 11–18.

Parvin, A., Ye, M.A., Jia, B., 2007. Effect of visibility on multilevelmovement: a study of the high-density compact built environ-ment in Hong Kong. Urban Des. Int. 13 (3), 169–181.

Penn, A., Conroy, R., Dalton, N., Dekker, L.,Mottram, C.,Turner, A.,1997. Intelligent architecture: new tools for the three dimen-sional analysis of space and built form. In: Proceedings of the 1stInternational Symposium on Space Syntax University CollegeLondon. London, pp. 30.1–30.19.

Penn, A., Hillier, B., Banister, D., Xu, J., 1998. Configurationalmodeling of urban movement networks. Environ. Plan. B: Plan.Des. 25, 59–84.

Raford, N., 2003, Looking Both ways: space syntax for pedestrianexposure forecasting and collision risk analysis. In: Proceedingsof 4th International Space Syntax Symposium (London, URL:⟨http://www.spacesyntax.org/symposia/SSS4/fullpapers/51Rafordpaper.pdf⟩ (accessed 25.06.09).

Ratti, C., 2002. Urban Analysis for Environmental Prediction (Ph.D.thesis). University of Cambridge, Cambridge.

Read, S., 2001. Thick Urban Space: shape, scale and the articula-tion of ‘the urban’ in and inner-city neighborhood of Amster-dam. In: Proceedings of the 3rd International Space SyntaxSymposium, Georgia Institute of Technology. Atlanta.

Robinson, N., 2004. The Planting Design Handbook, 2nd ed.Ashgate, Hants, UK.

Røe, P.G., 2000. Qualitative research on intra-urban travel: analternative approach. J. Transp. Geogr. 8 (2), 99–106.

Salingaros, N.A., 1999. Urban space and its information field. J.Urban Des. 4, 29–49.

Shriver, K., 1997. Influence of environmental design on pedestriantravel behaviour in four Austin neighborhoods. Transp. Res. Rec.1578, 64–75.

Stahle, A., Marcus, L., 2010. Compact sprawl experiments fourstrategic densification scenarios for two modernist suburbs inStockholm. J. Space Syntax 1 (1), 59–76.

Syntax S., 2004. Fifth Framework Programme in AGORA cities forPeople. The London Institute, London.

Teller, J., 2003a. A spherical metric for the field-oriented analysisof complex urban open spaces. Environ. Plan. B: Plan. Des. 30,339–356.

Turner, 2001. Depthmap: a program to perform visibility graphanalysis. In: Proceedings of the 3rd International Symposium onSpace Syntax, Georgia Institute of Technology, GA.

Turner, A., 2003b. Analysing the visual dynamics of spatial mor-phology. Environ. Plan. B: Plan. Des. 30, 657–676.

Turner, A., 2004. UCL Depthmap. Version 5.09. University CollegeLondon, London.

Turner, A., Doxa, M., O'Sullivan, D., 2001. From isovists to visibilitygraphs: a methodology for the analysis of architectural space.Environ. Plan. B: Plan. Des. 28, 103–121.

Turner, A., Penn, A., 2002. Encoding natural movement as an agent-based system: an investigation into human pedestrian behaviourin the built environment. Environ. Plan. B: Plan. Des. 29,473–490.

Wu, C., Xiao, Q., Gregory, E., McPherson, E.G., 2008. A method forlocating potential tree-planting sites in urban areas: a casestudy of Los Angeles. USA Urban For. Urban Green. 7 (2), 65–76.

Wua, C., Xiaoa, Q., McPherson, E.G., 2008. A method for locatingpotential tree-planting sites in urban areas: a case study of LosAngeles, USA. Urban For. Urban Green. 7, 65–76.

Zacharias, J., 1997. The impact of layout and visual stimuli on theitineraries and perception of pedestrians in a public market.Environ. Plan. B: Plan. Des. 24, 23–35.

Zacharias, J., 2006. Exploratory spatial behaviour in real and virtualenvironments, Landsc. Urban Plan. 78: 1–13.