Understanding Suburb’s Morphology: a Cognitive Approach in GISMishra A. School of Mathematical and Geospatial Sciences RMIT University Melbourne Victoria Arrowsmith C. School of Mathematical and Geospatial Sciences RMIT University Melbourne Victoria Chhetri P CR-SURF, School of Geography Planning and Architecture University of Queensland Brisbane, Qld. Phone: +61422036653 s3061589@student.rmit.edu.au Abstract This paper discusses a research project that canvassed how people perceive relative distance as a medium to understand how they construct cognitive spaces. The study area chosen for this project was Melbourne Metropolitan area. The sample consisted of individuals aged between 18 and 60. Each of them carried through a short questionnaire, which required them to draw a cognitive map as well as array a list of suburbs in terms of the preference of living and perceived cost of property. Various other questions like age, education, household income, ethnic background etc were also there in the questionnaires. These questionnaires were then used to build a spatial database in Geographical Information Systems. The perceived locations of suburbs as marked by the respondents were digitized and there ‘mean centers’ and ‘standard distances’ were calculated for each suburb. Following this, a cognitive map of Melbourne was created using these ‘mean centers’ and ‘standard distances’. In this research, GIS was used to compare cognitive space with the objective space. 1. Introduction When people use or try to remember information about the environment, such as objects, events, or phenomena situated within a geographic space, they tend to produce systematic errors (Kitchin et al, 1977). This suggests that their cognitive representations of those geographic spaces and the acquired knowledge are systematically distorted (Downs et al, 1977). To act effectively in the complex environment, people need more accurate understanding of the way things and processes are structured, organized and connected over a space. This research paper presents an approach based on cognitive mapping that contribute to the investigation of mental processing of spatial knowledge about geographic spaces. In this research, an investigation has been made to find out how people perceive distance and how they construct cognitive spaces. The study area chosen for case study in this research is the Melbourne metropolitan area. An attempt has been made to study the distortion between a perceived and the actual arrangement of suburbs. Understanding the way people perceive distance can help identifying the reasons why certain suburbs are more preferred. It can enable us to develop a better understanding of the impression that various suburbs create on people and how such impression leads to residential location decision choices. Furthermore, it is also vital to know if age, gender or education affect people’s perception of distance and influence their preferences for certain suburbs. In this paper, we have used GIS to compare cognitive space with the objective space. The cognitive space has been generated from data collected through a survey. This research involves the study of the mapping ability of people as well as investigating the factors, which may influence the perception of people about the way suburbs are situated in relation to one another in Melbourne. 2. Cognitive Maps The term Cognitive map was introduced by Tolman (1948), which latter became the basis for cognitive psychology research. Cognitive mapping is “a process composed of a series of psychological transformations by which an individual acquires, codes, stores, recalls, and decodes information about the relative locations and attributes of phenomena in their everyday spatial environment." (Downs and Stea, 1973, p. 6) Cognitive maps in cognitive psychology are considered as ‘dynamical schemes’ inside human mind. They are parts of our reflection of the physical world and participate on motivation to decision making and creating and changing our attitudes. Whether in real or virtual space we form cognitive maps to deal with, and process the information contained in the surrounding environment. 3. Objectives The principle aim set out for this research is to generate a cognitive map of Melbourne. In order to achieve this aim, a number of objectives are developed which are as following:
4. Study Area The study area selected for this research is Melbourne, which is the capital city of the State of Victoria. The city is located in south-eastern Australia, on the Port Phillip Bay at the mouth of the Yarra River. Melbourne is Australia’s second most populous city, after Sydney, and is a major economic, cultural, and administrative center of Australia. Sixteen suburbs from Metropolitan Melbourne were selected for this research that include Frankston, Werribee, Doncaster, Broadmeadows, Box Hill, Sydenham, Caulfield, Footscray, Greensborough, Coburg, Sandringham, Epping, Cranbourne, Prahran, Dandenong and Belgrave.  Figure 4.1: Figure 4.1: Map showing locations of 16 suburbs selected for the research 5. Data Collection Third year RMIT University students were given the task to interview and collect information on cognitive maps from the residents of Melbourne. In all a total of 143 respondents, aged 18 or above were chosen and interviewed. In this research we obtained cognitive maps by asking the respondents to locate 16 systematically distributed suburbs of Melbourne on a hard-copy map of Melbourne statistical boundary map. Most of social and behavioral data was collected by questionnaire. Each of the respondents carried through a short questionnaire, which required them to draw a cognitive map as well as array a list of suburbs in terms of the preference of living and perceived cost of property. Moreover, three lifestyle questions were asked with reference to the lifestyle in the western and eastern suburbs of Victoria. Various other questions like age, education, household income, ethnic background etc were also there in the questionnaires. These questionnaires were then used to build a spatial database in Geographical Information Systems. 6. Methodology The first step was to create two separate databases for spatial and attribute data. Spatial database was created in ArcGIS software. A relational database for the attribute data collected in questionnaire was created in Microsoft Access database management software. After that both these databases were integrated in ArcGIS software to create a GIS database.  Figure 5.1: Data and Information flow path The perceived locations of suburbs as marked by the respondents were digitized and there ‘mean centers’ and ‘standard distances’ were calculated for each suburb. Following this, a cognitive map of Melbourne was created using the mean centers and standard distances. This cognitive map of Melbourne was then compared with the actual Melbourne statistical boundary map in ArcGIS (GIS software) to analyze and understand the reasons of distortion. 6.1 Spatial Database Creation The hard-copy maps collected in the survey were digitized along with the locations marked on them in GIS software. For this analysis ESRI’s ArcGIS software package was used. The location, to which the participants placed a particular suburb on the map, was estimated in ArcMap in meters, with relation to the actual location of that same suburb in the layer with actual location of each suburb. This was undertaken by digitizing all the data using the digitizing tools in ArcGIS software. This data was then categorised and separated in to 16 different layers. After digitization of the sixteen layers created in ArcGIS, the next step is to process them creating a GIS database. It was made sure that all the 16 layers are georeferenced. Georeferencing describes the process of locating an entity in 'real world' coordinates (ESRI’s ArcGIS Manual). The data in every suburb layer has 142 points, and they were numbered accordingly in their attribute tables, to develop a relational database. The development of relational database is an important step, as it allows all the layers to be linked by means of a common field. 6.2 Database Creation in Microsoft Access Microsoft Access was used to store the attribute data in a format suitable for analysis. Microsoft Access, a database management software program, is widely used in GIS applications. The first step was to design the structure of the database. Three tables were formed with a common “unique” column “Survey_id”. Table 1: This table contains data on the living preferences of the respondents among the given 16 suburbs in the questionnaire. The table was given the name “Living Preferences” Table 2: This table contains data on ranking of suburbs with respect to perceived cost of property by the respondents. The table was named “Suburb Ranking”. Table 3: This table contains all the socio-economic and behavioral data of the questionnaire. For better comprehension and analysis of the data, numerical point values were used to convert the data into more meaningful attribute table. The Microsoft Access database was then imported in ArcGIS and integrated with spatial database of the sixteen suburb layers using the data integration tools. 7.0 Calculation of Mean Centers and Standard Distance For all the 16 suburb layers, calculations were done to find the mean centers and standard distance of each suburb location and compare the differences between the actual suburb locations and perceived suburb locations. 7.1 Calculation of Mean Centers The first distance-based measurement that can be applied to a dot point symbol measures the central tendency of the pattern and is called the mean center. It is found simply by calculating the arithmetic means x’ and y’ of the spatial coordinates of the n points (Unwin 1981). Unwin (1981) gives a methodology to calculate the mean center which is explained below:- In symbols, the mean center is calculated as:  Mean centers are useful summaries of point pattern, particularly because we need to summarize differences in the distribution of point data themes. To calculate weighted mean the following equations was used to calculate weighted mean center of points (as in our 16 suburb themes):-  Where fi = frequency or weighting factor. The data was processed in ArcView 3.2a for finding out the mean centers of the 16 suburb locations as marked by the respondents. 7.2 Calculation of Standard distance After calculating the mean centers, we need to find the standard distance around all the mean points to know the concentration/ density of the data and its spread. Just as in ordinary statistics we measure the scatter of observation on a single variable about their mean by the standard deviation, so we can measure the spatial dispersion of a point pattern by its standard distance ds (Unwin 1981). We first need to calculate the variance along the x and y axes independently and then combine the two results to give the overall standard distance.  7.3 Calculation of distances between actual point positions of suburbs and mean-center points. After calculations of mean centers and locating them on the maps, these mean centers were converted into a separate single spatial layer named “Mean Centers”. These calculations were performed in MapInfo 5.0 GIS software. Suburbs which are on fringes of Melbourne were bit difficult for respondents to locate and this is illustrated in table 7.1 which shows the distances between mean centers and the actual suburb locations. As the suburbs move further away from the city or a known feature on the map, such as bay, the distance between mean centers and actual suburb locations is increasing. This is especially evident with Belgrave, Dandenong, Cranbourne and Frankston. It is very clear that the closer the suburb to city the more accurately it is represented by the respondents, for example Footscray, Sandringham, Epping and Caulfield. The average distance that perceived suburbs were from actual suburbs is 4709.93 m. Many people who live in certain areas may not interact with facilities or people in other areas. So people, who live in the outer west, may not generally have much interaction with people who live in the outer eastern suburbs and the opposite is also true. This tends to decrease as people live closer to the central business district because more people share the same cultural and social space, and allows for more interaction between people from different areas.  Table 7.1: Distance between mean centers and actual suburb locations in meters 8. Generation of Cognitive map This step includes a comparison of the actual suburb locations and mean centers on the map and the distortion of Melbourne statistical division map according to the shifted position of the 16 suburb locations (i.e. mean centers). Procedure:
 Figure 8.1: Cognitive map of Melbourne overlaying actual Melbourne map As can be seen in the figure 8.1, cognitive map of Melbourne has shifted northwards in comparison to actual Melbourne map. 9. Analysis and Querying in GIS database The GIS database was used to analyze and understand the arrangement of suburb locations by the respondents, in order of their preference. This database was also analyzed to find out an opinion relative to the lifestyle preferences of the people living in the eastern and western suburbs of Melbourne. Finding out the cognitive mapping ability of respondents in terms of the region where they live, there age, highest education they received was also an important part of this research study. Some findings of this analysis are listed below:
10. Conclusions The results reveal a positive relationship between desirability of living and cost of property. It is apparent that experience of a particular area may have played a vital role in terms of influencing mapping ability. Thus middle-aged adults seem to have a better and more accurate in mapping tasks when compared with younger adults. It is also evident that the majority of people who undertook survey preferred Eastern suburbs in comparison to western suburbs. Almost 90 percent of participants felt that quality of life is better in eastern suburbs. This is also clear, perhaps in a more subtle way, in the way people have placed these suburbs on the cognitive map. It can be seen in the cognitive map that suburbs in the eastern part of Melbourne are perceived to be closer to the CBD. It gives the impression that people preferences are in part are affected by the perceived distance. Distance, therefore, may seem to be a factor in the desirability for a particular suburb. The cognitive map generated for Melbourne has been slightly compressed in comparison to actual map of Melbourne that has resulted a shift in the northeastern part of Melbourne. The cognitive map suggests that people tend to remember landmarks such as bay, coastline, and city center and perhaps align suburbs based on the configuration of these landmarks. This research provides supplementary information into the choice structure and preference patterns of people’s decision-making and may help to develop better urban design, spatial connectivity and layout configurations. References |