Operational Problems
The main operational problem I encountered was the inability to use 'roads' as a factor. I was unable to use the roads data, which was originally one of the factors for locational analysis (see Methodology ). The raster layer created was basically useless, since virtually every pixel within Vancouver was assigned a roads value. I don't however, think that this factor as its own layer is crucial: since transit routes are generally along major arteries, the transit layer compensates for this.

Methodological Problems
1) Library size: each library was given equal value, although in reality the libraries in Vancouver vary tremendously in size and circulation capacity, from the 6-story Central Branch to the incredibly tiny Carnegie and Strathcona Branches. Initially, I was going to at least create a separate layer for the Central Branch, but since there were not very high suitability values in that area anyway, it would not have altered the locational analysis.

2) Land use: only property that 1) belongs to the city, or 2) is not in use and can be purchased by the city could actually be considered. Unfortunately, as of September of 2002, this data costs money to obtain. However, based on a PDF map available on the City of Vancouver website (see Coloured Zoning Map ), a decent amount of land in that area is zoned as CD1 (comprehensive development district), meaning that at the area could be potentially used as a site for a new library.

Another issue involving landuse is that the Grandview Cut runs right through the area I have determined is the most suitable for a new library. This does reduce the contiguous area in the area that is suitable. Again, however, since the area is so large, the impact of this is slight.

3) Subjectivity: There is inherent subjectivity involved in both the reclassification of data and the weighting process (these steps were outlined earlier in Methodology ). The relative importance between the different factors is qualitative, not quantitative, but because of the nature of GIS, the values had to be quantified.

4) Time lapse between data sources: The census data is from 1996, while the survey I used is from 2000. The demographics will most likely have changed slightly over time.