The population within the AMC limits was 4.5 million in 2001 (Census of India 2001). The Sabarmati River divides the city in two parts. The city s tra

EXPLORING THE RELATIONSHIP BETWEEN URBAN ROADS, URBAN FORM AND ECONOMIC ACTIVITY IN AHMEDABAD USING SPACE SYNTAX Arindam Majumdar 1, Richard Sliuzas 2, Talat Munshi 1, Mark Brussel 2 1 Center for Environmental Planning and Technology (CEPT), Ahmedabad, India (majumdar20477, munshi)@itc.nl 2 International Institute for Geoinformation Science and Earth Observation, (ITC), The Netherlands KEY WORDS: Spatial configuration, changing urban form, space syntax accessibility ABSTRACT: Urban growth and form are outcomes of various underlying factors such as population growth, migration of people, policies of local government, local and global market forces, individuals reactions to these interventions, etc. Recent studies have indicated that spatial configuration and the relationship between different streets are also factors that influence human activity and agency, reflecting the differences in accessibility and connectivity over the city as a whole and also within different localities within the city. Space syntax analysis provides various measures in using space as a means of investigating the structure of a city, directed at how different components of this structure are related with the configuration of the built environment and activities which take place within it. This configuration influences people s perception about their built environment and therefore their behaviour, such as their movement and activity patterns within the city. This research performs a Space syntax analysis of the city of Ahmedabad, India that is then used to examine relations between accessibility and land use, job distribution and building height for five different parts in the city. We use the AXWOMAN extension tool for ArcGIS to generate the axial map for syntax analysis and the global integration and local integration values for the street network. Spatial interpolation techniques are used to generate a surface of syntax values of the streets to estimate those values on the building level for easy comparison with other indicators. One conclusion of this research is that land use patterns do vary with spatial configuration and accessibility. Commercial land uses concentrate in highly accessible or movement-rich localities whereas residential land uses tend to be found in movement-poor localities within Ahmedabad. Another conclusion of this research is that the city s syntax centre, i.e., the best integrated locality in terms of road network, has shifted towards the western side of the city, while the concentration of jobs is still near the old centre which is on the eastern side. This shift points to a potential mismatch between the road layout, road typology, traffic distribution and employment, possibly contributing to an increased congestion on the city s roads. The third conclusion is that high rise buildings are associated with highly accessible localities. This type of analysis can help planners and policy makers to better appreciate the relationships between road network design, urban form and human activities, providing an additional tool for the ex-ante assessment of road layout for town planning schemes or other improvement projects. INTRODUCTION Cities are in a continuous process of change. The physical city can be viewed as the combined product of numerous decisions taken by its inhabitants and organizations, adapting and adjusting themselves over time to the global and local changes, problems and opportunities. The factors influencing this change process differ for different cities; however, one of the major factors is the city s existing urban form. We define the urban form of a city as its three dimensional form characterised as represented by variations in population density, land use, and floor space. Previous research has shown the relation between urban sprawl, higher mobility, travel distance and travel demands which are causing air pollution, land and energy consumption (Batty, Chin et al. 2002). Such studies show that the configuration of the built environment, its urban form, has a direct link with the efficiency of the transportation services observed in any region. This research aims at revealing how the street layout of Ahmedabad influences the pattern of job availability, the location of land uses and building height. The street layout is understood as the morphology of the street network in the region (Kim and Sohn 2002). Space syntax is used to analyse the structure of the street layout. This layout is then represented in an axial map for the region as a series of intersecting lines called the axial lines. The case study region is Ahmedabad city located in Gujarat state of western India. It has two urban authorities, the Ahmedabad Municipal Corporation (AMC, 460 km 2) and the Ahmedabad Urban Development Authority (AUDA, adjoining areas. 5 3 2 1 1 = Relief Road 2= Ashram Road 3 = CG road 4 = Kankaria 5 = S.G. Highway /Drive in road Figure 1: Google Earth image showing Ahmedabad with major roads (Source: Google Earth) 4 51

The population within the AMC limits was 4.5 million in 2001 (Census of India 2001). The Sabarmati River divides the city in two parts. The city s traditional core dates from 1411 AD and is located to the east of the river. The urban form of this small core is very different from the rest of the city. The buildings are generally low rise and either residential or mixed land use with very big markets at certain places. It is a high density, low rise type development which has one of the highest movements of people and goods within the city. By contrast the western parts are characterised by well planned developments with wider roads. Almost all major educational and financial institutions are located within this part. Thus the city is expected to provide an interesting case for study as it has variable urban form in different parts with both organic and planned type developments. The major commercial areas are found at Relief Rd. on the eastern side and Ashram and CG Roads on the western side of the river. There is also a newly developing area on the western side near SG Rd. which is characterised by big shopping malls and offices (see Figure 1). Primary data was collected on the road network of the city. In addition, 12640 buildings were sampled in five different areas within the city. Figure 1 shows the approximate location of these sample areas. Literature Review Urban form of a city can be represented as a pattern of urban units of growth which are linked together through transport routes (Batty 2004). One of the most important factors considered in the functioning of an area as a CBD is accessibility that can be defined as relative proximity of a space from all other spaces in the system where the impedance to move between two points can be a function of costs, benefits, time, etc. (Batty 2009). It is believed that activities which require an accessible location such as business, retail, etc., tend to cluster near the CBD to increase their economic and functional viabilities. This growth is discussed through many models like Alonso s bid rent model of urban land use which is based upon the relationship between accessibility and land rent (Pacione 2005). However, this simple concentric ring model has its limitations: it does not consider the element of and variations in urban morphology; it assumes a homogeneous terrain (Burgess 1925; Pacione 2005; Sarma 2006). A limitation of the orthodox urban models is the assumption that individuals navigate within a network based on metric distance whereas the cognitive distance, calculated on the geometric, topological and network factors, also plays a role (Hillier and Iida 2005). Rather than using only metric distance, network users also tend to perceive distance as a function of changes in the direction of movement. A limitation of such an approach is that sets of individual changes in direction are very difficult to aggregate and to model as network flows. Space syntax models have the ability to model geometric and topological factors. Moreover, they go further to show that when individual decisions are aggregated into an urban flow they show dependency on the urban grid structure of the city thus suggesting that the urban grid influences the movement of traffic, pedestrian and vehicles. Space syntax theory originated as a non discursive theory of architecture that later was applied in many studies relating space configuration and movement patterns observed in such configurations. Initially it was used at a very small scale, e.g. to understand movement patterns inside a museum or hall. Recently, it has been developed further with the integration of GIS and is now being used to study many urban phenomena at larger scales. The theory of space syntax is based on the central theme of correlation of spatial configuration or layout of open spaces and the corresponding movement observed in that space. Space syntax study considers spatial configuration as an independent variable in the analysis of social systems (Kim and Sohn 2002). The approach is based upon principles of natural movement and the theory of movement economy (Hillier, Penn et al. 1993; Hillier 2007), which are the basis for space syntax s theory of spatial configuration. Hillier (1998; 2007), discussed the correlation between difference in values of spatial configuration in parts of London and the corresponding crime rates observed and uses measures of spatial structure to explain why higher crime rates are observed in certain parts of the city. He also discussed how spatial diffusion of pollution from vehicles is affected by the configuration of the street grid and how it potentially affects pedestrians. The space synatx model also showed good predictions of pollution levels in the urban grid environment. Kim and Sohn (2002) studied the effect of urban street layout on the pattern of land use density of office buildings in Seoul. These studies show how space syntax analysis helps to explain the behaviour of people through the influence of the configuration of the urban grid. The key parameters produced using this method are identified as connectivity, depth, mean depth, relative asymmetry, real relative asymmetry, global integration, local integration and intelligibility ( explained in (Manum, Rusten et al. ; Hillier 1996; Hillier 2007). Each is briefly describe below. Connectivity and Depth explain the relative accessibility from other lines in the system. Connectivity is defined as the number of lines to which a particular line in the axial map is connected and Depth is the number of steps required to pass from one line to another. The depth of an adjacent line is one, higher depth values imply that the line is more deeply embedded within the network than lines with lower depth values. Depth is calculated at three steps known as the local depth and also for the entire graph (the graph of axial network) called as total depth / global depth. Mean depth is computed as the average number of steps to reach from each of the lines in the graph to that particular line explaining relatively difficulty in terms of movement turns to access a particular place.. Relative asymmetry (RA) and Real relative asymmetry (RRA) describe the integration of a node by a value between 0 and 1, where a low value describes high integration. Global integration is given by the reciprocal of RRA. It shows the integration of that particular line with the rest of the lines in the system. Values more than 1 indicate high integration while lower values indicate segregation from the rest of the lines in the system. Local integration is the integration values of a line with the lines which are only 3 steps away from it. It is also called radius-3 integration. Intelligibility is the coefficiency of correlation between local parameters and global parameters. So local parameters like connectivity, control, local integration can be correlated with global integration to see the intelligibility of the system. A local area is said to be more intelligible if the local integration is higher than the global integration values (Hillier 2007). 52

The notion of relative nearness can be analysed in space syntax through variables of global and local integration. Thus if a street shows a high global integration it means that it has better connectivity to the whole system than others. If a street shows high local integration it means that the street is better connected to its immediate neighbouring streets. The axial representation, and the measures derived from it, have been researched and found successful in the study of the social and cultural roles of space, particularly for the evaluation of the impact of spatial configuration on pedestrian and vehicular movement patterns and also for evaluation of design proposals (Figueiredo and Amorim 2005). Thus, to understand if and how urban form is affected by the configuration of road network in the city of Ahmedabad we relate the space syntax indicators mentioned above with urban form details as given by income levels, building heights and land use configuration. presence of a building at a place and hence the diffusion from adjacent points will be slightly different from that shown through the interpolation surface (Batty 2004). Figure 3 explains the method used to generate the interpolation surface. The space syntax values for each parameter mentioned earlier were generated using AXWOMAN 4 (Axwoman). The values thus generated were converted to point features for interpolation. The Inverse Distance Weighted Interpolation Method (IDW) was used to generate the interpolated surface of urban form variables mentioned above from point locations. A circular zone of influence of 40m was used and a total of 6 neighbouring points were used to calculate the value of each point on the interpolated surface. Developing the space syntax model for Ahmedabad The conceptual diagram of the space syntax model developed for Ahmedabad is shown in figure 2. The first step was to prepare an axial map for the region wherein the convex space is mapped through isovist lines (Hillier and Hanson, 1984). These lines are also called the axial lines and in our case represent the unobstructed line of sight for public roads of movement. An axial map of the city is defined as the least set of straight lines that pass through all the open space (Hillier and Hanson 1984). Analysis of the axial map enables the computation of relative nearness or accessibility, in common terms (Batty and Carvalho 2003) Figure 3: Interpolation of Space Syntax Parameters Analysis of the relationship between spatial configuration of the urban road layout and urban form. Figure 2: Space Syntax Model for Ahmedabad The values of the space syntax parameters were generated at the street level. However, the data for the building height and land use were at building level. Thus to bring them to the same spatial level of analysis interpolation techniques were used. The advantage of using an interpolation technique is that it gives an immediate sense of spatial variations in the space syntax accessibility values. It also makes it very easy to compare these values with any other types of indicators such as land use and building height, at any level of analysis unit. The major limitation of this technique is that it does not take into account the effect of buildings and obstructions in evaluating the visibility of the space. In reality visibility may change due to the Space syntax analysis reveals differences in accessibility levels of different parts in the city. There is a clear indication that the accessibility on both sides of the city is very different. The global integration values on the eastern side of the city are lower compared to that in the western parts. Moreover, there are certain pockets on the eastern fringe which are showing little integration with the rest of the city. The prominent role of the bridges connecting both parts of the city is clearly visible. The central and the western parts of the city are showing very high global integration values, namely, the areas near Ashram Rd., CG Rd. and SG Rd.. One major finding is that the global integration value near Relief Rd. is lower than that of both CG Rd. and Ashram Rd.. The local integration values on these three streets are also high, which indicates their importance within the city. Relief Rd. had developed as the commercial hub of the city in the past. However, other roads are now showing higher global integration values than Relief Rd., probably indicating the 53

Spatial Configuration of Jobs Co-relation Jobs -Global Integration Values Figure 4 : Relating Urban Form with Space Syntax Values 54

change in the position of the functional core of the city with the increase in its size. To analyze how these syntax values relate to urban form indicators discussed earlier the data on the three urban form indicators used in this research was collected through primary survey at the building level while the job availability data was collected from secondary sources and was available on grid level of cell size 100m x 100m. Global integration and building height were found to be significantly positively correlated (r =.198; p <.01) indicating that if the global integration value increases the building tends to be high rise. By contrast, local integration is not correlated to the variation of building height in Ahmedabad. To understand the spatial variation in the correlation between global integration and building height, five sample areas in different parts of the city were used. The correlation results obtained for the sub regions also show that there is a positive correlation between building height and global integration. Among the five locations, Ashram Rd. and CG Rd. areas have the highest correlation between building height and global integration. These are also the areas where most of the new commercial development and increase in jobs can be observed (Munshi 2003). In terms of land use variations in land use for the entire city and for the five areas of Ahmedabad in relation to syntax parameters theoretically it can be said that commercial, industrial, institutional and mixed land uses are expected to locate at places that have higher global integration values and residential land use will tend to locate at places with low global integration values. Figure 5: Chart showing the mean global integration associated with each land use Figure 5 shows the mean global integration value associated with each land use in the 5 study areas. As expected the mean value of global integration associated with residential land use is 0.39, which is significantly lower than that of any other land use. Commercial land use is having the highest mean global integration associated with it (m = 0.407), followed by industrial (m = 0.405) and institutional, mixed and open spaces with almost same values (m = 0.401). However spatial variations can be seen across the five different locations. Areas that are predominantly residential like Satellite-Drive-in Rd. in the west of the city and Kankaria to the east, show less variation in global integration values amongst the land uses, and also lower overall values as compared to other more commercialised areas like CG Rd. and Ashram Rd.. We would expect to find more pressure to convert non-commercial land uses into commercial in the 2 latter areas (Munshi and Zuidgeest paper under review). Similarly a positive relation between local integration and the different land uses was also observed. The residential land use showed least association with the mean local integration value where as commercial showed the highest association. In line with the bid rent theory the number of jobs near places with high accessibility will be higher than at places with low accessibility. A positive correlation was observed between availability of jobs and global integration values (r = 0.289; p <.01 level). Though not strong, it does point to a degree of connection between the location of jobs and places with higher global integration. In terms of the five specific areas Satellite- Drive-in Rd.showed the highest positive correlation with global integration (r = 0.60; p<.01) followed by Ashram Rd. (r = 0.25; p<.01) and Kankaria (r = 0.22; p<.05). The integration values although, do not follow the primacy in number and density of jobs with the five case areas. The job density is highest in the Relief Rd. area followed by SG Rd. and CG Rd., however the mean global integration value is highest at SG Rd. followed by CG Rd. and Relief Rd. This is indicative that through the effect of the Relief Rd area, the city s traditional city centre is degenerating into sub-centres and areas where higher integration values are observed are the potential attractors of the new jobs and commercial establishments. This is also in line with the observations of Munshi and Zuidgeest (paper under review). Conclusion The purpose of the research described here was twofold: first, to generate a space syntax analysis for the city of Ahmedabad; second, to use this analysis as means to analyse the effect of accessibility on change in urban form for different parts of the city. It was found that the western part of the city, which is more affluent in terms of its quality of life (Munshi, Brussel et al. 2008), has higher integration values, and also that the river and the connectivity across it played a very important part in determining these integration values. The correlation analysis of the space syntax parameters with urban form indicators for Ahmedabad showed that the spatial configuration of streets is correlated to building height, land use and availability of jobs. Positive and significant correlation was observed between global integration and building height. In case of land uses, residential land use was observed to be associated with the least values of global integration in all the five sample areas, whereas commercial was observed to be associated with highest global integration values. Similarly, commercial land use was associated with high values and residential with low values of local integration. A significant, positive correlation was found between job availability and global integration for all areas except Relief Rd.. Incidentally, the Relief Rd. area has the highest number and density of jobs among the 5 areas but the global integration value is lower than in the Satellite-Drive-in Rd. and CG Rd. areas. This shows the increasing importance of newly developing areas for job creation. It was also observed that generally relatively weak correlation results were found. It is possible that the strength of the correlation will increase if urban form data is available at a finer resolution. The effects of the interpolation method on the correlation analysis should also be investigated. Despite these 55

reservations the results show the use of the space syntax concept in understanding the spatial configuration of a city and future wok is planned to extend the analysis into other areas and deepen the understanding and interpretation of such analyses in order to better support decision making on land use, transport and urban form. Acknowledgement: The research presented in this thesis is part of the research project Land, urban form and the ecological footprint of transport: application of geo-information to measure transportrelated urban sustainability in developing countries with a case study of Ahmedabad, India, which received a project grant (SP-2006-09) from Volvo Research and Educational Foundations (VREF). References Axwoman. Retrieved 12th October, 2009, from http://www2.hig.se/~bjg/axwoman.htm. Batty, M. (2004). "A New Theory of space Syntax." UCL Working Paper Series (Paper 75): 35. Batty, M. (2009). "Accessiblity: In search of a unified theory." Environment and Planning B: Planning and Design 36: 4. Batty, M. and R. Carvalho (2003). "A rigorous definition of axial lines: ridges on isovist fields." UCL Working Paper Series 69: 19. Kim, H. K. and D. W. Sohn (2002). "An analysis of the relationship between land use density of office buildings and urban street configuration: Case studies of two areas in Seoul by space syntax analysis." Cities 19(6): 10. Manum, B., E. Rusten, et al. AGRAPH, Software for Drawing and Calculating Space Syntax Graphs. Munshi, T. (2003). Planning Public Transport of Ahmedabad City in India. Unpublished M.Sc.thesis ITC, Enschede. Munshi, T., M. Brussel, et al. (2008). Developing a Geo-Spatial Urban Form - Travel Behaviour Model for the City of Ahmedabad, India CODATU XIII. Hô Chi Minh city: 11. Munshi, T. and M. Zuidgeest (paper under review). "Polycentricity and travel behavior: A case of Ahmedabad city, India." Urban Studies. Pacione, M. (2005). Urban Geography: A global perspective. New York, p.140-149, Routledge. Penn, A. and A. Turner (2002). Space syntax based agent simulation: 16. Sarma, A. K. (2006). The social logic of shopping: a syntactic approach to the analysis of spatial and positional trends of community centre markets in New Delhi. Built environment, University College London. Master's thesis: 58. Batty, M., N. Chin, et al. (2002). Sprawling cities and transport: from evaluation to recommendations. State of the art review of urban sprawl impacts and measurement techniques: 52. Burgess, E. (1925). "Concentric zone model." Retrieved 16/3/2009, from http://en.wikipedia.org/wiki/concentric_zone_model. Figueiredo, L. and L. Amorim (2005). Continuity lines in the axial system. Fifth Space Syntax International Symposium,. Delft University of Technology, Delft: 13. Hillier, B., Ed. (1996). Space is the Machine, Cambridge University Press. Hillier, B. (1998). "The common language of space: A way of looking at the social, economic and environmental functioning of cities on a common basis." 23. Hillier, B. (2007). Space is the machine, University of Cambridge: 343. Hillier, B. and J. Hanson (1984). The social logic of space. Cambridge. pp. 108-110, Cambridge University Press. Hillier, B. and S. Iida (2005). Network and Psychological Effects in Urban Movement: 16. Hillier, B., A. Penn, et al. (1993). "Natural movement: or, configuration and attraction in urban pedestrian movement." Environment and Planning B: Planning and Design 20(1): 38. 56