Project Design:
This project was divided into three separate, yet equally important, phases:
- Literature Review
- Data Preparation
- Spatial Analysis
Literature Review:
For this project I focused primarily on literature pertaining to homelessness enumeration methodology. As stated in the introduction, there have been many different attempts to try and enumerate homeless populations each of which manages to do some things better than others. One underlying theme, however, is that most methods rely on some form of local knowledge to accurately identify areas where the homeless may congregate (Bogard, 2001; Berry, 2007). Methods which rely solely on estimates based on use of services, such as shelters and food banks, have an inherent bias as a large portion of homeless people are known not to utilize these services on a regular basis. In fact, according to the Social Planning and Research Council of BC’s (2008) homeless count, as many as 50% of the homeless population living on Metro Vancouver streets do not use the shelter services that are readily available. Although this is likely a combination of choice as well as a lack of available shelter space this is simply a number that cannot be overlooked. This factor naturally leads to the question of “where do these homeless people sleep at night?”.
Upon reviewing the literature it became clear that relatively little is known about the specific characteristics of non-service locations which homeless people may use as refuge. Most sources identified broad spaces such as abandoned buildings or public parks as places of sanctuary (Bogard, 2001). These findings suggest that the focus should be placed on the social and environmental characteristics which make up the areas in which these homeless people are found. By placing the emphasis on broadly categorized environments we greatly undervalue the large volumes of data that are available to the modern researcher. If one were to define areas not simply based on remoteness, but as a factor of intertwined variables such as crime, health care use, service availability, population density, land use and other data sets representative of the issue we could then have the ability to produce truly meaningful spatial output. Based on the current state of knowledge in this area, one would simply have to grab a generic landuse map and circle all the areas relating to a few types of land use to create a map of potential homeless refuge sites. This project, albeit a very simplified version, lays the framework for a better understanding of the social and environmental factors governing the spatial distribution of the homeless population.
Data Preparation:
The data preparation involved in this project was relatively straightforward, however, as with all GIS work, it was imperative that it be performed methodically and in a well documented manner. The following describes the steps involved in preparing the data sets for use in the Multi criteria analysis:
Prior to any geoprocessing in Idrisi, ArcGIS was used to project and clip a few of the data sets. Once this was done they were imported into Idrisi as Idrisi Vector files and the geoprocessing model (see above) was developed.
A multi-criteria analysis is simply a mathematical overlay of a series of factors and constraints. In this case, we had one constraint and four factors. Below is a description of how we developed each of these data sets beginning with the constraint and then finishing with the four factors.
Green Space and Open and Undeveloped Land Use Layer (constraint):
This layer was generated by simply reclassifying the GVRD land use layer.
GVRD Landuse
Reclassified:
Commercial Distance Layer (factor):
This layer was developed in a two stage process. First, we used the reclass tool to isolate all of the commercial and mixed commercial/residential land use from the GVRD land use layer. The last step involved in getting this layer ready for analysis was passing it through the distance tool to produce the Commercial Distance layer.
Commercial Distance
Road Distance Layer, Railway Distance Layer (factor):
Both of these layers were developed using the same steps. First, we needed to convert the vector files (roads, railways) to raster using the lineras function and then create boolean layers. Note that the street and railway files are complete and that the incomplete appearance in the images below are a result of scale and the lineras function.
Roads (click to view larger image):
Railway Line (click to view larger image):
Road_Dist:
Rail_Dist:
Slope Layer (factor):
This layer was created by applying the slope tool to the GVRD_DEM file
GVRD_DEM:
Slope: