Discussion and Results:
(Click on the image to view larger version)
The above image was produced using the raster layer output of our MCE. Areas in orange and red have been identified as most suitable and areas in dark blue are least suitable. One area that jumps out right away is the region surrounding the Coquitlam River on the north east portion of the map. Having worked in that area for 7 years, I have seen the homeless population grow dramatically. The wooded buffer regions surrounding the Coquitlam River are dotted with homeless camps and offer great seclusion just a stone’s throw away from some key features. The area surrounding this is largely commercial with several restaurants and a supermarket. Moreover, this region is also home to a major rail works yard with trains passing through on a very regular basis. Another factor which compounds the homeless problems in this area is the fact that it borders the site of Riverview Hospital, which, until fairly recently was home to many people suffering from mental health issues. These factors combine to make this area a very suitable location for homeless refuge sites. Other sites that were identified by this model include, Burnaby Lake Park, a region of undeveloped forest in north Surrey (by the Patullo bridge) and, to a lesser extent, Stanley Park.
An area that the model picked up quite well was the downtown eastside. This was a bit of a surprise because that area is primarily developed and I did not think the model would identify it as well as it did. I do, however, believe that with a more comprehensive landuse layer, such as a BCAA data set, the model can better handle areas such as the downtown eastside.
One area that was under represented by this model was Pacific Spirit Park in UBC. I think this is due to two major factors. First, with the exception of perhaps a few minor stores or restaurants on the UBC campus, there is little in the way of commercial landuse surrounding the park. A more comprehensive land use dataset would allow us to look closer at the UBC area to determine land use types which may provide access to key necessities such as food. The second major factor is simply a lack of any rail tracks. The closest rail tracks are approximately 15 kilometres away running parallel to Arbutus Street. What this under representation shows is that our model should be refined and I believe the addition of datasets including crime, detailed landuse, and trail networks would greatly improve the accuracy of this analysis.