Public spaces revisted: a study of the relationship between patterns of stationary activity and visual fields Abstract Maria Beatriz de Arruda Campos and Theresa Golka Space Syntax Limited, UK b.campos@spacesyntax.com, theresag@gmx.de This paper investigates the relationship between visual fields and patterns of stationary activity in several London public spaces. The aims of this paper are two-fold: first to examine whether there is a direct relationship between the two, and second to determine the most useful method of visual field analysis for the study of informal stationary activities. To meet this objective, visibility graph analysis, overlapping point isovists, and convex isovists methodologies are used. The present study involves the analysis of static activity data collected during the summers of 2003 and 2004 for six public spaces in central London, ranging from tourist dominated spaces to public spaces used almost exclusively by office workers. The spaces observed displayed a wide range of sizes, geometry, morphology, and surrounding land uses. Correlation between the static activity patterns and visual fields shows that stationary activity is more likely to happen in relatively segregated spaces off-set from the main access points. These findings support those of a previous study (Arruda Campos, 2000, 1999), which suggested that people s preferred location follows a distinct pattern, with people selecting more secluded areas compared to the exposed ones for informal static activities. 1. Introduction The suggestion of people preferring more secluded areas for static use seems to be plausible. Already when deciding to visit a public space, a person is aware of the fact of meeting people and activity. A public space embodies characteristics of urban settings like copresence with strangers and randomness of encounter (Doxa, 2001). Its use as a place to meet, rest, learn, observe and potentially interact with others is part of the public culture. Not missing the main activity or attraction and watching people plays an important role in the use of a public place. People like to observe people walking by and pay attention to any activity in the public space but not necessarily being active themselves. Therefore it seems to be clear that people might prefer areas with the largest possible field of vision over the place, but at the same time do people prefer more exposed or enclosed places for their stationary occupation? 2. The case studies The present study involves the analysis of static activity for six public spaces in London which vary in size, geometry, morphology, and surrounding land uses. They are: Euston Square and Regents Place in the West End, Paternoster Square and Finsbury Square in the City of London; and Parliament Square and Trafalgar Square in central London.
546 M. B. de A. Campos and T. Golka Regents Place, Paternoster Square and Finsbury Square are all semi-enclosed pedestrianised spaces surrounded by office buildings, and therefore used almost exclusively by office workers. The explicit location of Euston Square, also a semi-enclosed pedestrianised space, in front of Euston Station leads to a high number of people coming from or getting off the train using the public space for waiting, besides a high number of office people occupying the large offer of seats and tables at lunchtime. Much larger in size and richer in design Parliament and Trafalgar Square are located further south towards the Thames and are characterised by their great views and location close to relevant tourist sights, being Trafalgar Square in itself, a very popular tourist attraction in London. In contrast to the other four public spaces, both Trafalgar and Parliament squares are exposed spaces, surrounded by busy vehicular roads. Out of the six cases, only Parliament Square does not have coffee shops or sandwich bars within the premises. All squares provide seating capabilities that differ in number, shape and arrangement. 3. Methodology To investigate whether there is a relationship between visual fields and patterns of stationary activity in public spaces, three methods have been used and compared: overlapping point isovist (OPI), visibility graph analysis (VGA) and convex isovists. A) The overlapping point isovists methodology has been used in the past in a study of 12 squares in the City of London (Arruda Campos, 2000, 1999). The overlapping point isovists is based on the axial break up of the urban grid in which the public spaces are embedded. The intersection point of two or more axial lines from which any segment of the public space can be seen defines the isovist vantage point 1. Each isovist is the entire visual field from this point respecting buildings and walls and unusable space, such as columns, flowerbeds and sculptures, but not sitting places as barriers of vision. For each square, the point isovists were overlapped resulting in convex spaces with different degrees of exposure, denominated the coverage density. Higher coverage density signifies a higher degree of visual exposure. The coverage density for each public space was divided into a scale with six bands - dark grey to light grey - and dark grey areas have the highest coverage density levels. To see if there is a dependent rate of static occupancy, the levels of overlapping isovists have been set in context to number and position of static people found in each of the overlapping levels. This method analyses the exposure related to the surrounding of the public space, putting it in context to the urban grid where the public space is embedded in, i.e., the observer is considered to be outside the public space. B) A visibility graph analysis for each square has been used to calculate the visual integration of the space itself (Turner, 2001). Likewise for overlapping point isovists, only unusable space has been omitted from the graph. For the comparison of the results of the overlapping point isovist method to the results of the VGA, similarity in way of analysis was given and therefore similar factors considered. The first version of the VGA including fixed furniture but excluding unusable space from the calculation matches the method of the overlapping isovists. However, as an experiment, a second version of the VGA model has been created where all fixed furniture has been excluded of the analysis, and is referred to as V GA R. 1 Limited to 250 meters radius from the square.
Public spaces revisted 547 Figure 226: Overlapping point Isovist of 6 squares
548 M. B. de A. Campos and T. Golka While the overlapping point isovists is mainly used to focus on the visibility of the public space in context to the urban grid where the public space is embedded, the VGA was limited to the boundary of the public space. The observer is considered to be inside the public space and the urban grid is not part of the analysis. VGA models including the surrounding grid were tested for all public spaces but then street junctions, which had no visual relationship with the public space, often became the focus point of the system diverting the aim of the analysis. The same limitation was noted when vehicular roads were excluded, i.e., street junctions became once again the focus point of the system. The VGA image of each square has been set to an illustration of equal count of six colour levels, from dark grey to light grey, where dark grey areas have the highest visibility levels, and then been analysed in context to position and number of people in stationary use of the square for each of the six bands. C) As the third method, convex isovists have been used to analyse the size of the visual field from hot spots for stationary activity inside the square. The visual fields areas, which were not restricted to the public space, have been quantified and compared individually for each of the public spaces examining whether stationary people s location is related to the size of the visual field. As before, unusable space has been omitted from the analysis. The observer is considered to be inside the public space but the urban grid is part of the analysis. 4. Spatial analysis Using direct observation, information on stationary people was recorded over one weekday during the summers of 2003 and 2004 for two time periods: mid morning (off peak for static occupancy) and lunch time (peak time for static occupancy). The snap shot information on stationary activity was overlaid against the VGA and overlapping point isovists maps. People using the facilities of the cafes and sandwich shops were excluded from the analysis. For the convex isovists, isovists were drawn from locations with a concentration of stationary people and the geometric area of the isovist was subsequently quantified. Table in fig. 228, summarises the number of static people observed according to time period. The next step was to correlate the number of static people for both time periods according to the VGA and point overlapping isovists. For each separate public space, the coverage density and VGA integration levels were further normalised into a three level scale (low, medium and high) to make the statistical analysis more meaningful due to a high number of areas with no static people recorded 2, as summarised in Table 229. Table in fig. 230 summarises the visual fields areas according to the convex isovist analysis. 4.1. Overlapping point isovists Analysis shows that for the six surveyed public spaces the pattern of static distribution is inversely related to coverage density levels. Although in two cases (Euston Square and Paternoster Square) the number of static people in the low and medium areas of coverage density are similar (especially for Euston Square), in all public spaces, the areas with high coverage density are the ones with the lowest levels of stationary activity. Averaging out 2 During the off-peak time there were areas where no one was observed.
Public spaces revisted 549 Figure 227: VGA of 6 squares
550 M. B. de A. Campos and T. Golka Figure 228: Total number of static people observed according to time period. Figure 229: Total number of static people observed according to OPI or VGA analysis and bands. Figure 230: Total number of static people observed according to OPI or VGA analysis and bands.
Public spaces revisted 551 all public spaces, only 10% of people sat in areas with highest levels of exposure compared to 50% in the areas with the least coverage density. Finally, detailed visual inspection of the data suggests an important second underlining trend. When considering all the areas of low coverage density, the ones which seem to receive the majority of static people are the ones close to, or in the view of, areas of high coverage density. Likewise, the same behaviour seems to be found for the areas of medium coverage density levels. In summary, when the data was analysed together, the findings suggest that people s preferred location actually follows a distinct pattern, with people selecting more secluded areas compared to the exposed ones for unprogrammed static activities. The number of static people for all cases showed a preference for locations with low levels of exposure (50%), followed by medium levels (40%) followed by high levels of coverage density (10%). 4.2. Visibility graph analysis Firstly assessing the information based on all usable sitting areas not been considered for the calculation, the results of the visibility graph analysis shows a trend where the pattern of static distribution is inversely related to the visual integration. Of the six surveyed public spaces, with exception of Euston Square, all the remaining squares did presented higher number of people in areas of lower visibility. Although in further two cases the number of people in areas of low, medium and high visibility is not statistically significant, the number of stationary people in areas of highly visibility is the lowest for the three remaining studied cases. For all six cases, the percentage distribution of static people for areas with low visibility levels is 45%, followed by medium levels with 31% followed by high levels of coverage density with 24%. When the analysis is extended to the model where sitting areas have been considered for the calculation, once again the same pattern emerges. For all six cases, the percentage distribution of static people for areas with low visibility levels is 50%, followed by medium levels with 35% followed by high levels of coverage density with 15%. However, by making the sitting areas a visual obstruction, the results are biers towards areas of low visibility. 4.3. Convex isovists Finally, the analysis of point isovists showed a tendency for people to congregate in areas of large visual fields. However, there was not a direct correspondence between the size of the isovist; and the number of stationary people and the location of hot spots for stationary activities and the size of the visual field. 5. Discussion When comparing the different methodologies, this analysis suggests that the convex isovist can be a difficult approach as the results vary substantially despite minor changes in the size and location of the isovist. The comparison of overlapping point isovists and VGA, both highly objective approaches, however, highlights a fundamental difference between the two methodologies; while overlapping point isovists assesses the visibility of the public space from the outside - how visible the public space is from the surrounding area; VGA, in this particular research excludes the surrounding grid. The results of the overlapping point isovists analysis confirm previous findings (Ibid, 2000) that people seem to avoid very exposed spaces and prefer areas that provide good
552 M. B. de A. Campos and T. Golka views but still have some level of privacy for unprogrammed static activity. In addition, it is suggested that a second underlying principle exits where users prioritise locations close to high coverage density. After having arrived at public spaces through the linear properties of space, users may at this point choose locations that provide them with a reasonable degree of privacy. Total exposure, the concern of the public gaze (Valentine, 1998) is something undesirable. The first area of the public space seen when approaching by the user while moving around in the urban environment is the one to be avoided and a more secluded location is selected, preferable nearby. Hence, the user is in control of how far one wants to be visually exposed but without losing the ability to see. It is further suggested that the selection of the particular location in the proximity of the ones close to high coverage density areas is further shaped by the internal configuration of the public space. This process is well documented by the VGA analysis which results suggest a trend for people to select slightly secluded areas. In summary, both overlapping point isovist and visibility graph analysis are functional methodologies for studying patterns of space use in public spaces. Further studies with a larger sample will help to clarify any differences in the results. Literature Arruda Campos M. B. (2000) Urban Public Spaces: A Study of the Relation Between Spatial Configuration and Use Patterns, PhD thesis, University of London. Arruda Campos M. B. (1999) All That Meets the Eye: Overlapping Isovists as a Tool for Understanding Preferable Location of Static People in Public Squares, in: Proceedings of the Second International Symposium on Space Syntax, Brasilia: University of Brasilia. Arruda Campos M. B. (1997) Strategic Spaces: Patterns of Use in Public Squares of the City of London, in: Proceedings of the First International Symposium on Space Syntax, London: University College London. Benedikt, M. L. (1979) To Take Hold of Space: Isovists and Isovists Fields. Environment and Planning B, 6: 47-65. Conroy Dalton, R and Bafna, S. (2003) The Syntactical Image of the City, in: Proceedings of the Forth International Symposium on Space Syntax, London: University College London. Cutini, V. (2003) Lines and Squares: Towards a Configurational Approach to the Morphology of Open Spaces, in: Proceedings of the Forth International Symposium on Space Syntax, London: University College London. Doxa, M. (2001) Morphologies of Co-presence in Interior Public Space in Places of Performance, in: Proceedings of the Third International Symposium on Space Syntax, Atlanta. Georgia Institute of Technology. Hillier, B. (1996) Space is the Machine. Cambridge: Cambridge University Press. Hillier, B. (1984) Mansion House Square Inquiry - Proof of Evidence. Unit for Architectural Studies, Bartlett School of Architecture and Planning, University College London. Hillier, B. et. al (1990) Broadgate Spaces - Life in Public Spaces, Unit for Architectural Studies, Bartlett School of Architecture and Planning, University College London.
Public spaces revisted 553 Hillier, B. et. al (1990) The Public Space of Paternoster Square, Unit for Architectural Studies, Bartlett School of Architecture and Planning, University College London. Turner, A. (2001) DepthMap: A Program to Perform Visibility Graph Analysis. In: Proceedings of the Third International Symposium on Space Syntax, Atlanta. Georgia Institute of Technology. Valentine, G. (1998) Food and the Production of the Civilised Street, in: N. Fyfe (ed.), Images of the Street, Planning, Identity and Control in the Public Space, London: Routledge. Whyte, W. (1988) City - Rediscovering the Center, New York: Doubleday.