Use and appropriation of space in urban public parks: GIS methods in social geography

Abstract. The research objective is the analysis of the appropriation of space in urban public parks. For this purpose, extensive field observations were conducted in several parks in Zurich, Switzerland, over the span of three years, with records made of the location, assumed age, gender and activity of park visitors. Based on research in environmental psychology and anthropology, a model was developed building on the two concepts of «personal space» and «activity footprints» to represent space appropriation. In line with the view that quantitative spatial analysis methods remain a valid tool for critical, non-positivist research, the model was implemented using kernel density estimations for the spatio-temporal analysis of the observed park use. It is argued that the probability surfaces generated by kernel density estimations are an adequate representation of the specific vagueness of human space appropriation as they remain sensitive to the presence of individual park visitors.


Introduction
This paper presents an approach that uses quantitative methods for the digital representation of human space use and appropriation in public parks. It will show that it is possible to calculate and visualize the use and the appropriation of space quantitatively, while at the same time remaining sensitive to issues of equality, accessibility and gender. In a case study in Zurich, Switzerland, three small urban public parks were systematically observed during the summers of 2005 to 2007. Using this data, several approaches are feasible to represent the actual use and appropriation of space, including potential processes of domination and exclusion, which may not be immediately obvious. The paper conceptualizes and implements the appropriation of space at the micro level of individuals, using concepts from anthropology and environmental psychology, such as personal spaces and crowding. The vague boundaries of space usage and appropriation will not be eliminated but explicitly addressed in the analysis and visualization. In parallel to a visual exploration of the data, detected patterns will be tested using spatial statistics methods.
After presenting the motivation and the objectives, the authors will summarize the current state of the debate on quantitative vs. qualitative methods. The key statement is that GIS methods do not automatically imply positivist research, but instead provide opportunities for critical geography research. Then, the current state of the research is presented and some preliminary results shown.

Motivation, Terminology and Research Objectives
The research project focuses on three specific urban parks in the city of Zurich and is part of the project "Sustainable Design, Management and Appropriation of Urban Public Parks" 3 supported by the National Research Program 54 "Sustainable Development of the Built Environment" of the Swiss National Science Foundation. The aims of this project are to identify design and planning elements as well as management strategies that could foster a socially sustainable appropriation of public parks. Two teams make up the research cooperation. One team focuses on the social aspects of space appropriation (called SOSPA, see contribution of KASPAR in this issue), while the project reported on here deals mainly with the visualization and analysis of space appropriation (called VISPA). Closely collaborating, the two teams aim for an integration of qualitative and quantitative methods, expecting this synthesis of methods to be an important asset of the overall research project.
For this research, space appropriation is defined as the process in which each human constantly, sometimes consciously and sometimes unconsciously, lays claim to surrounding space. On the one hand, this happens for space that one considers his or her personal space, in which intrusion by others can be considered inappropriate. On the other hand, carrying out activities also appropriates space, be it reading or playing some ball game. In the public sphere of urban parks, these spaces and their appropriation engender a constant negotiation process with other, often unfamiliar, people. The authors acknowledge that researchers from the social sciences use the term "space appropriation" somewhat differently, involving more contextual information as well as recognizing the symbolic relationship between individual and place (compare KASPAR, this issue).
In the VISPA team the key research objectives are the development of a model framework for the quantitative analysis of human space use and appropriation, and a toolkit of methods to support decision makers in improving the quality of life of citizens. These objectives require the integration of a theoretical and methodological background ranging from social geography, environmental psychology, and information visualization to geographical information science. The authors intend to conduct the research with a pragmatist, mixed methods approach, using both qualitative and quantitative methods sequentially and iteratively as appropriate (CRESWELL 2003;MORGAN 2007).

Spatial Analysis with GIS and Positivism
Quantitative geography consists of the analysis of numerical spatial data, the development of spatial theories and the construction and testing of mathematical models of spatial processes (FOTHERINGHAM, BRUNSDON ET AL. 2000).
While today it is generally accepted that the physical world is symbolically structured by the social world and society (WERLEN 1993;LÖW 2001), and inseparable from social processes and relations (PAVLOVSKAYA 2006), quantitative geographic research indeed overemphasized space in a determinist, functionalist manner, searching for globally applicable laws. Reproduction of the natural and social world was reduced to a technical problem, where errors are the result of lack of technical skill or unintentional distortion (PICKLES 1994). Consequently, quantitative geography is still strongly associated with positivist epistemology (SHEPPARD 2001;POON 2005). Critics argued that quantitative methods reproduce geographies of primarily white, male, bourgeois power structures. For researchers advocating non-positivist knowledge production, qualitative methods have become an accepted strategy (SHEPPARD 2001;PAVLOVSKAYA 2006). However, many simple spatial analysis functions are actually rather qualitative in nature. Visualization, for example, is a qualitative research approach well suited for use throughout the whole research

Modelling Space Use and Appropriation
In order to detect informal processes of exclusion and domination, it is imperative to develop a method to make the otherwise invisible conflicts in space appropriation visible. While 6 research in sociology (BOURDIEU 1991) and social geography (WERLEN 2000) on space appropriation has focused on patterns at the spatial and/or meso scale, most of the research dealing with individual human space use at the micro scale has been conducted by anthropologists and psychologists (HALL 1966;SOMMER 1969;GOFFMAN 1974;ALTMAN 1975;FREEDMAN 1975;BALDASSARE 1978;JOHNSON 1987). To the knowledge of the authors, there has been only one quantitative spatial study on this issue ( (1966) termed these passive spaces "personal spaces", conceptualizing them as concentric distance zones around a person and observing differences between cultural groups. BAXTER (1970) agrees and concludes from extended observations in natural settings, that also age and gender can modify these interpersonal distances. The basic concept of personal spaces has been extended into the theory of proxemics, which includes additional factors such as type of space and behaviour categories (Littlejohn and Foss 2005). The authors restrict the first modelling approach to informal personal spaces, with the distance zone determined by the activity type. Modifications such as fixed-feature space or the individual sociopetalsociofugal axis (facing) are to be implemented later.
A newly introduced component is the concept of activity footprints, representing active spaces. Each activity requires a specific space termed the activity's footprint. Their size and 7 shape is estimated from literature and observations. It is important to note at this point that these estimations are grounded in empirical evidence, but are in need of refinement and more research before they can be considered accurate enough to contribute to a sufficiently realistic modeling of human space appropriation.

Figure 1: Potential Conflicts in Space Appropriation -Potentielle Konflikte in der
Raumaneignung -Conflits potentiels de l'appropriation de l'espace The assumption is that a potential for crowding and goal interference exists when personal spaces and (incompatible) activity footprints of other park visitors overlap. Consider the example in Figure 1 of two soccer players and a reader and their respective activity footprints and personal spaces. The assumption is that the reader is looking for undisturbed reading experience and thus feels uncomfortable with a soccer player's unpredictable movements when the latter enters her social distance zone. The same would be true in reverse. However, the activity footprint of the reader is very small and the personal space claimed by the soccer players is small due to their dynamic activity -they might feel uncomfortable only by a foul of a fellow player. Therefore, there is no overlap between the reader's activity footprint and the players' personal spaces.
During the extended observations in the public parks of Zurich that took place during more than 140 hours, the activities were classified into seven main categories: Static solitary (sleeping, reading), static interactive (observing, talking, card games), eating (BBQ, picnicking), dynamic regular (football, badminton), dynamic irregular (running around), park infrastructure (park-specific playgrounds), and activities involving water. In addition to the activity type, the observers also recorded each visitor's approximate age, gender, and group 8 affinity. The location and time of the activities were recorded by placing points at the approximate centre of activity, with the ID of the park visitor, activity type and start time of the activity.

Analysis of Space Appropriation and Potential Conflicts
First, it is necessary to acknowledge the uncertainty associated with both data and analysis results. The multitude of terms used for describing uncertainty makes it necessary to briefly clarify and define the usage of the different aspects of "uncertainty" in the work: • Inaccuracy: Errors made during the observations, concerning both spatio-temporal location as well as attributes.
• Incompleteness: Some of the details may not have been recorded.
• Vagueness of the boundaries of the personal spaces and activity footprints.
Pre-tests for intercoder reliability have shown a spatial inaccuracy of less than one meter, and a temporal inaccuracy of about one minute. A careful choice of attribute categories enabled a high accuracy. The data is almost complete, only minor details were omitted during busy periods. The spatial inaccuracy and incompleteness increase with activities that involve a lot of movement, since it was impossible to capture the exact space-time location of every park visitor at all times. The authors consider the inaccuracy and imprecision acceptable for the development and testing of the model and the analysis of the data. The vague boundaries were addressed by the main analysis method, which were Kernel Density Estimations (KDE).
They are a well-researched spatial analysis method that fulfils the project's requirements and has been widely used for point data representing humans, although mostly at an aggregate or meso/macro scale (KWAN AND LEE 2003;LEVINE 2006). Detailed information on KDE can be found in several standard works on spatial analysis (FOTHERINGHAM, BRUNSDON ET AL.

2000).
In the Wahlenpark in the summer of 2006, 842 visitors were recorded in total: 418 male, 402 female and 22 infants of unknown gender. Most visitors were adults (76%), with children second in numbers (17%) and very few teenagers and seniors (3.5% each). To detect potentials for conflict, further analytical steps are necessary, part of which are already implemented and part of which are currently in progress: In a next step, the temporal dimension has been included and overlapping activity footprints and personal spaces were calculated. For each group of visitors, two kernel density estimates were calculated: One for the personal spaces and one for the activity footprints. Assuming that there are no conflicts of space appropriation within one group of visitors, for each visitor group the prevalent activity type was chosen as group activity. The authors adjusted parameters controlling the spread (bandwidth) and height (volume or population) of a group's kernel density estimate to account for the activity-specific sizes of activity footprints and personal spaces. For each moment in time, each group's activity footprint was checked against the personal spaces of all other groups via map multiplications, so that only overlapping grid cells would retain any value at all (for a detailed treatment of the analysis, see upcoming publications). It is important to note here that this method does not "smooth over" individual park visitors: Since the values are multiplied, even a single park user can be represented by a high space appropriation and potential conflict if in proximity to a larger group of other park visitors.
Another benefit to the analysis is the fact that the probability surface generated by kernel density estimations also ameliorates the problem of inaccurate and incomplete data: The closer to the centre of activity (represented by the original point), the more probable and intense this space is used by the park user. It is also important to remember that the parameter values at the current stage of research are assumptions derived from own experience and observations, and have yet to be verified in the evaluation process and refined accordingly.
In a final step, the findings will be synthesized with results gained through the analysis of interviews with park visitors (done by SOSPA), to see where there are similarities and discrepancies.

Conclusion
In this paper spatial analysis methods are applied at the micro-scale of individuals. The goal is to examine the appropriation of space in urban public parks. Extensive field observations in several parks have been conducted over the span of three years, recording the location, age, gender, and activity of park visitors in a database. Based on research in environmental psychology, a model was developed that represents human space use and appropriation 11 building on the two concepts of personal space and activity footprints. Arguing that quantitative spatial analysis methods remain a valid tool for non-positivist research, the model was implemented using kernel density estimates for the spatio-temporal analysis of the observed park use. In conclusion, the probability surface generated by kernel density estimations is an adequate representation of the specific vagueness of human space appropriation and remains sensitive to individual park visitors. The paper also shows that it is possible to use quantitative methodology of geographic information science and the tools of geographic information systems for a critical geography research project.

Summary
Research objective is the analysis of the appropriation of space in urban public parks. We have conducted extensive field observations in several parks over the span of three years, recording the location, age, gender, and activity of park visitors in a database. Based on research in environmental psychology and anthropology, we developed a model that builds on the two concepts of Personal Space and Activity Footprints to represent space appropriation. Arguing that quantitative spatial analysis methods remain a valid tool for critical, non-positivist research, we implemented our model using kernel density estimations for the spatio-temporal analysis of the observed park use. The probability surfaces generated by kernel density estimations are an adequate representation of the specific vagueness of human space appropriation that remains sensitive to the presence of individual park visitors.