Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Geography & Environmental Studies

Faculty/School

Faculty of Arts

First Advisor

Not Applicable

Advisor Role

Not Applicable

Abstract

The urban landscape is an interspersed mixing of residences, shops, theaters, parks, natural areas, and a multitude of other uses. From the early days of the central markets, to the planned downtown, to the heavily planned super-regional shopping complexes, commercial landscapes evolve. There has been considerable research conducted on analyzing the commercial structure of urban environments in an attempt to better understanding the nature of retailing and its resultant impacts on the geography of the city.

This research has three broad goals: a) to develop a technique that makes operational, in a systematized and objective manner, an approach to analyzing the structure of the commercial environment; b) to apply the approach within a GIS environment, and; c) to develop a generalized typology of urban commercial structure. The systematized analysis is a series of guidelines and statistics which can be applied in an objective manner. The development of the nearest commercial neighbor as a statistical measure of proximity to other commercial operations was the foundation of the approach to clustering commercial operations in to retail areas.

To achieve the overall goals, three census metropolitan environments (Sudbury, Kitchener and Ottawa) were used as study areas. These cities represent small, medium and large census metropolitan environments, respectively, within Canada. Commercial locations for each city were extracted from a national database of locations and mapped in a GIS environment. For each study area, the nearest commercial neighbor values were generated and the appropriate statistics extracted.

Commercial clusters were generated by using the average nearest commercial neighbor value and multiples of the median commercial neighbor value. These nearest neighbor and median values were inputted into a buffering routine as the buffer size. The resulting clusters were then compared to ortho-imagery and in the case of Kitchener, land use planning documents. Two approaches for cluster generation were employed; 1) Point-only where all individual addresses were used on the clustering, and; 2) Point plus Polygon where those commercial operations that existed within polygons (malls and central business districts) were removed from the dataset, the remaining points were then clusters and the polygons added back to the results. Finally the results from both clustering approaches were compared to land use parcels to assess accuracies of the technique.

The results indicated that the overall method proposed was effective in determining commercial zones, and that the 2x iteration of the median nearest commercial neighbor technique yielded the most accurate results. Moreover, three main conclusions were drawn. The first was that there was a difference, and in some cases significant differences, between the land use planned commercial areas and areas that have grown larger through agglomeration. Secondly, there are density variations between core and suburban areas that, at times, resulted in a larger definition of a commercial area within the core because the lesser dense suburban areas having an impact on the nearest commercial neighbor values. Thirdly, there was considerable over-capturing of commercial areas when the buffer multiples were greater than 3x. In addition, the point plus polygon clustering technique indicated that while the defined areas were more accurate when the polygons were used, it was only in areas where those polygons were the main commercial cluster. In mixed areas, there was no discernable advantage to using the polygons. Furthermore, the removal of points had a strong impact of the nearest commercial neighbor values generated. Lastly, when dealing with polygons, the geographic arrangement of the commercial type became important.

Based on the findings of the commercial zone analysis, a typology of commercial development was detailed. This typology contained three main geographic components, namely the core, suburb and gateway areas of the urban environment. Within each geographic location, a series of commercial forms were identified. This new typology allowed for the inclusion of historical remnants of landscapes and consequently allows for a comparison against older typologies. The typology employed a three part urban classification system which is applicable to any type of urban environment and, finally, the focus on geographic form removes the impact of store changes and the changes in the nature of commercial zones over time.

This research has operationalized a systematic and replicable method of examining urban commercial location data for the purpose of determining commercial structure. This technique can be applied to future datasets easily and objectively allowing for a readily updatable typology; thus rendering it less static than previous typologies. It is the use of the technology, namely GIS, that adds this dynamism to the analyses. Furthermore, it has been demonstrated that the potential exists for using GIS to analyze commercial location data.

This research has contributed to this evolution by analyzing the geography of commercial development during a snapshot in time. However, by developing a series of operational and repeatable techniques that focus on the geographical organization of commercial locations, it is hoped that the results will function as the conceptual and practical framework for commercial structural analysis of urban environments for future studies.

Convocation Year

2008

Download

Share

COinS