GEOG 355 Final Project - Urban GVRD Analysis & Livability Index
Project Design by: Peter Stein (301032002)
Part #1: Introduction & Conceptual Outline
"Analyzing livability also means transposing political ecology debates about sustainability and social justice from fields and forests to the streets, factories and sewers of the built environment" (Evans, 3). There is no stopping the tide of urbanization and the influx of a greater number of people to urban areas in the 21st century. For the first time in human history, the urban has eclipsed the rural in terms of the total proportion of inhabitants (>50% of the world's population is now living in cities. Therefore, as more and more individuals look to find prosperity in the urban environment, it is essential that the right tools are present to accommodate them. This issue is of overwhelming ecological, political, economic and social importance and its implications, whether good or bad, will depend on the preparedness of cities to integrate these human needs into the pre-existing urban structure. Knowing that the environment is already being pushed to the brink, a wholistic approach is necessary to offer the right types of informative capabilities to persue rationally sound policies. Providing the technical expertises to seize and act upon this window of opportunity is of utmost concern. Thus, proactive decision-making is made possible by providing an accurate assessment of the spatial problem on hand. This project strives to demonstrate how a hypothetical livability scenario based on real data can go along way towards understanding the challenges that must be overcome in order to avoid a mismanagement of human and financial resources in a specific urban environment.
The Greater Vancouver Regional District (GVRD) is a collection of twenty-one municipalities and one of Canada's finest city conglomerates. But what specifically allows one to make such assertions? What infrastructural variables define the relative liviability scale of a specific urban area? Where are infrastructural improvements most needed and required for a better overall quality of urban life?
This goal of this project is to conduct a spatial analysis investigation of the GVRD using GIS (Geographic Information Systems) software to perform a multi-criteria analysis (MCE) of the region. This decision support system "is a common method for assessing and aggregating many criteria" (Clark Labs IDRISI tutorial, p102). The GIS criteria can be divided into 2 categories: constraints and factors. Each are of equal value when analyzing the relative distribution of opportunity across an urban space. In addition, the combined output is vital for determining an accurate representation the "fuzzy" boundaries that will constitute the Hypothetical Assessment of GVRD Livibility Map that is the final visualization output. This is derived from the collection/preperation of data from perticular sources and the subsequent spatial analyses of each and their respective inputs into IDRISI Decision Wizard. Thus, this project attempts to chart the apportionment of ideal, good, average and poor living spaces and extrapolate from the data any prevailing trends from the creation thereof to discuss relevant conclusions on the matter.
Part #2: Data Collection
Data acquisition and management is the first step of any relevant project design. First, an appropriate base map needed to be used that fulfilled certain requirements. The gvrd_lu_2001.rst file (S: Drive > GEOG 355 > Lab 8) was deemed permissible because it had a suitable projection of the entire GVRD (UTM_10N) and was in the proper file format (raster). Data also needed to be collected for the feature layers (points/lines/polygons) that were to compose the various factors for analysis. Located in the GVRD folder on the S: Drive\Data Warehouse were golf_courses.shp; greenspaces2.shp; parks.shp. In addition, two other shapefiles were collected from the S: Drive > GEOG 355 > Lab 8 folder BCedu_unprj.shp and BCtrs_unprj.shp.
The gvrd_lu_2001.rst file is a landuse file that is subdivided into sixteen categories. Certain categories were deemed acceptible and others unacceptable for potential liviability areas. The suitable land use categories deemed suitable were to undergo the MCE are as follows: Cat2 - Open & Undeveloped; Cat5 - Residential (Rural); Cat6 - Residential (Single Family); Cat10 - Residential (Townhouse & Low-rise Apartments); Cat13 - Commercial (Residential/Mixed); Cat14 - Agricultural and; Cat16 - Residential (High-rise Apartments). The unsuitable areas, RECLASS 0, included: Cat0 - Transportation, Communication and Utilities; Cat1 - Lakes and Water Bodies; Cat3 - Recreation and Protected Natural Areas; Cat4 - Protected Watershed; Cat7 - Industrial Extractive; Cat8 - Institutional; Cat9 - Commercial; Cat11 - NA; Cat12 - Industrial; and Cat15 - Harvesting and Research. The rationale behind what qualified as a suitable land use area was primarily determined by the logic of preconditionality and whether residents would actually reside in such a place and where optimal future development might occur.
A reclass function was preformed on the dataset to assign suitable areas = 1 and unsuitable = 0. These various landuse areas are the constraining values for the MCE.
Part #3: Methodology
IDRISI's Macro Modeller was useful for organizing the linear methodology of the project design; below is a screen-capture of the necessary data preparation and manipulation before the utilization of the Decision Wizard:
All the relevant map features or layers were required to have the same projection. (UTM_10N) was chosen because it is the most accurate projection system of urban areas) and each layer had to be standardized using a compatible format to enable a MCE (raster and not vector). Several shapefiles used in the project needed to be processed accordingly to enable a proper "meshing" of the layers. Therefore, the first step in the data manipulation required that each shapefile be brought in ArcMap. Because the gvrd_lu_2001.rst landuse file had a predefined coordinate system (UTM_10N), the ArcToolbox tool "Project" was used for standardization purposes (Data Management Tools > Projections and Transformations > Feature > Project) on all factor layer files. Then, because of the vector format of the shapefiles, RASTERVECTOR needed to be used to transform vector data into the appropriate raster data set for data congruency.
Five specific factors were chosen to frame the study based on their relevance as a livibility factor and on their availability and accessibility. Several different categories of livability factors needed to be addressed for this project. Access to public transportation, proximity to educational facilities, adjacency to allocated natural and recreational areas were all estimated to enhance the quality of urban livability standards. "Do you see New Westminster as a Liveable City? It is a city where the voices, priorities and rights of children are an integral part of public policies, programmes and decisions. It is, as a result, a city that is fit for all." (http://www.lcncf.ca/checklist.html) In terms of a broader spatial analysis (i.e. the GVRD) the same criterion should apply and be incorporated on a regional scale to facilitate the positive growth of livable cities.
Each factor had a corresponding map layer: 1) Train stations; 2) Schools; 3) Golf clubs; 4 & 5) Greenspaces & Parks. Below is the screen captures of the Macro Modeller process to prepare the 5 factors:
Raster conversion obscures the visualization (bctrstops.rst) but the point data is still there...
The first factor alludes to the necessity of having a reliable transportation service to meet logistical demands. Moving about the city in a timely manner is a prerequisite for efficient living. The world is moving faster than ever and, therefore, transportation is a key factor in determining urban livability. Transit helps maintain the vitality of major cities' central business districts, connects workers to jobs in suburban and rural areas, relieves traffic congestion, stimulates economic development around stations and reduces energy consumption and achieves clean air standards (National Business Coalition for Rapid Transit, 2003). A community's strong commitment to schools is "taking on more and more importance as the performance of an area's public schools in key to successful economic development efforts" and is central to raising a new generation of well-educated and properly informed citizenry (BNet, 2007). Finally, recreational and outdoor areas (golf courses, greenspaces and parks) are integral to promoting a livable city. "Devoting some time for recreation on a daily basis helps in the long run in maintaining health and achieving a peace of mind. Recreational activities serves as a means of relaxation. Research has shown that recreation on a daily basis reduces risks of diabetes and hypertension, enhances physical and mental health and improves quality of life" (Buzzle.com, 2008). Therefore, the evidence speaks for itself, but how can one amalgamate these considerations into one comprehensive format using the spatial analysis capabilities of a GIS?
Part #4: Spatial Analysis
The effective representation and visualization using spatial analysis tools is the primary strength and proficiency of any GIS. Customization of the data set enables a flexible approach to synthesizing the various components into one working unit. By using a variety of analytical methods, one can perform a series of spatial manipulations and accomplish pertinent spatial inquries of real world phenomena. To enable a "fuzzy" MCE, various preferences needed to be accorded to each factor. The following summary table provides the basic information of Decision Wizard specificities in IDRISI with a more detailed discussion to follow:
Factor | Min | Max | Membership Function Shape | Membership function Type | Control Points (a/b, c/d)
#1 BCdst | 0 | 37755.5 | Symmetric | Sigmodial | 0 500 , 2000 5000
#2 BCdst1 | 0 | 34568.7 | Monotonically | Decreasing | J-Shaped | 0 4000
#3 gc_dist | 0 | 45112.8 | Monotonically Decreasing | Linear | 0 7000
#4 gs_dist 0 | 20896.9 | Monotonically | Decreasing | Sigmodial | 0 1500
#5 parks_dist | 0 | 20896.9 | Monotonically Decreasing | Sigmodial | 0 3000
BCdst is the distance array map calculated from the point information obtained from the SkyTrain stations across the GVRD. The relationship is best represented by a symmetrical distribution because the ideal location is not directly beside the train station, due to noise issues, but within a range of 500m to 2km. Distances further than 2km up to 5km are less ideal because of increased travel times.
(Fig. bcdst.rst)
Suitability Map bcdst:
bcdst1 is the distance array map calculated from various point distance information collected concerning the proximity of schools. It is logical that the closer one is located to a school, the better (monotonically decreasing). Therefore, the control points were determined as those areas closest to 0 (most suitable) while suitability gradually decreases up to 4 km.
(Fig. bcdst1.rst)
Suitability Map bcdst1:
gc_dist, gs_dist and parks_dist are all reflective of the importance of having recreational/outdoor areas in the vicinity, therefore, all 3 are monotonically decreasing. Respectively, the control points were assigned at 7 km for golf courses, 1.5 km for greenspaces and 3 km for parks.
(Figs. gc_dist1.rst; gs_dist.rst; parks_dist.rst)
Suitability Maps (gc_dist_suit, gsdist_suit, parkdist_suit)
Factor weighting was also incorporated in the model to emphasize essential vs. non-essential services. The relative distribution of the factor weighting classification is as follows: BCdst = 0.3; BCdst1 = 0.4; gc_dist = 0.1; gs_dist = 0.1; and parks_dist = 0.1 where total = 1.0.
Using Decision Wizard in IDRISI, a final "fuzzy" map was produced combining the constraints and the relative weight values of the 5 factor raster layer maps into one. (Fig. Hypothetical Assessment of GVRD Livability Index.rst)
Then the values denoting suitability where transfered into four equal-interval categories (ideal, good, average & poor) to yield the final livability map using a custom palette in Symbol Workshop, each representing a quarter or 63/255 of potential variations (blue=ideal, green=good, yellow=average, red=poor).
Classified Livability Index & Final Map:
Part #5: Discussion & Results
"For most developing cities the problem is connecting growth to livability" (Evans, 4). Juxtaposing the Greater Vancouver Regional District Map enables a comparative discussion for the final map output results. The primary corridor of ideal/good livability runs through Vancouver, Burnaby/New Westminster and Coquitlam. Several smaller clusters of livability are located in North/Central Surrey, North Richmond, Pitt Meadows and Maple Ridge. Areas of good/average livability are situated in North Vancouver City, south part of the District of North Vancouver, Port Coquitlam and West Vancouver (University Endowment Lands). Relatively speaking, the Township of Langley, most of Surrey, White Rock, Belcarra, Anmore, Port Moody and Delta are ranked poor to average. Non-applicable areas are scattered throughout the map as areas of land use that did not meet the specified constraints and are interestingly located sometimes in regions surrounded by high livability arrays. This is an important feature because it provides a warning to potential inhabitants that potentially unwelcoming land use activities might be occurring nearby. In an effort to harmonize suitable and non-suitable landuse areas, it is a key facet of the project to make urban developers accountable for the distribution of less wanted land-use purposes in
areas of development and growth linked to livability preconditions.
Part #6: Methodological & Operation Problems
The scope or "compactness" of the study area (GVRD) is one operational problem that may lend ambiguity to the project. While GIS allows for an examination of large (small scale) geographic regions, the emphasis on attention to detail is of fundamental consequence to any project. The "compactness" would be improved if smaller unitary measures were used (i.e. block faces or dissemination areas). Specificity is also an issue (i.e. what type of schools are being represented? Are they public or private, elementary or high school) and therefore some of the preconditions may seem over-simplified or lumped into one amorphous group). Many additional factors could have been included in the livability computation (i.e. Health care facilities, libraries etc). The final legend (poor, average, good, ideal) is continuous and the appropriate labels were not attached directly to the legend. This detracts from the presentation of the final product. In Decision Wizard, variations with regards to factor specifications (weighting etc.) could have a dramatic impact on the final suitablility map, but numeric values were chosen as to be as closely approximate to a realistic simulation as possible.
References & Sources
American Public Transportation Association, National Business Coalition for Rapid Transit (2003) "The Economic Importance of Public Transit" Retrieved November 15, 2008. <http://www.apta.com/research/info/online/economic_importance.cfm>
BNet Business Network. "Public schools' importance in economic development growing" (June 11, 2007) Retrieved November 18, 2007. <http://findarticles.com/p/articles/mi_qa5277/is_200706/ai_n21244995>
Buzzle.com "Importance of Recreation" Retrieved November 20, 2008. <http://www.buzzle.com/articles/importance-of-recreation.html>
Evans, P., "Livable Cities? Urban Struggles for Livelihood and Sustainability" Berkeley: University of California Press, 2001.
Liveable Cities need Children and Families (2008) Retrieved November 20, 2008. <http://www.lcncf.ca/checklist.html>
Contact Info: pms1@sfu.ca