Data Collection:
    With any GIS project I have worked on in the past where data has to be sought out to be used for a specific software systems, many problems exist. When using Idrisi 32, the data should be in raster format for analysis purposes. Unfortunately, there was not a vast amount of Idrisi data readily available for download off any web site in which I have encountered. To fully answer my spatial question, I had to use FME universal translator and importing/exporting tools on many occasions to convert my data to Idrisi format. For example, my digital elevation model was in ShapeFile format. Exporting the file as a ASCII file was necessary for it to be readable in Idrisi. Learning the process of converting the data was also a problem because required a lot of time and most importantly, a lot of patience and persistence. As for vectors, when using a vector data set in a raster based program, points and lines are generalized and attributes are lost in the conversion (this occurred when i converted my coastline vector file in Idrisi).

Scale and Projection:
    Scale and projection was a problem throughout this project. All Data layers must be converted to a common reference system (utm-10n) and to a common X-Y dimension (same number of rows and columns) in order for any of the MCE analyses employed in this project to be functional.

Quality of Data:
    Problems regarding the quality on data and the depth of background research has been limited by cost ($) and by time (allotted to complete this project. For example, I found out that good quality statistical data from statistics Canada regarding detailed census population requires a fee for it to be obtained (click here for more details - then scroll down to find out). Another example was, that a  waiting period of 3 days was required to access to data from Land Data BC.

Arbitrary use of Data:
    Another area of concern came about when using assumptions like I did in my project were  very error prone. For example, by stating that the minimum value of  urban development near protected watersheds at 100 meters, anything urban development occurring one meter less is not correct, according to the model. The point being is that there is no perfect way to set thresholds especially for environmental settings because continuous variations do occur from one location to another. Related to this fact is that there is fuzzy logic with setting these thresholds. If a given site is 99.9999 meters from the protected watershed, that site will not been counted for (compared to a site located 100.0001 meters from the protected watershed which will be counted for).   In addition, the calculated weighted factors are to be concerned with because it was my decision alone that determined the weights of each factor (more errors and uncertainty!). Decisions were arbitrary, depending on how one perceives the overall biological/physical/chemical conditions that the North Shore Mountains are exposed to on a daily basis.

    Another area of the project to be concerned about is with regard to the weighted factors because it was my decision alone that determined the weights of each factor (more errors and uncertainty!).  Therefore, in summary all decisions made by me in this project may not represent real world phenomena.

Limited Complexity:
    In terms of assessing  urban development suitability ot a given location, sources of uncertainty and error are inherent. All weight factors regarding the elevation factor are derived directly from the values within the 25 meter resolution DEM of the North Shore Mountains region. The resolution of display on  a typical computer monitor lacks sufficient detail to exhibit the subtle changes of the land (i.e. continuous variation) regarding suitability indexes produced by the Idrisi 32 software (however, the limited ability of the human mind to process numerous amounts of information from visual displays is also in part, to blame) . Higher resolution of display is required for the precise display of the distribution of urban development suitability across the North Shore Mountains (most single houses in the Greater Vancouver Metropolitan area are not even 20 meters in width) or for an specific plot of land of the North Shore Mountains. Thus, the derivation of slope, and the development of elevation values data processes generates uncertainty and error. The arbitrary nature of the project and reclassification schema further increases the uncertainty of results as error occurs during each of the computational stages (please refer to the cartographic models for reference). Furthermore, the simple nature of the suitability model  provides only a general spatial distribution of  suitable locations for  regions but the results would still require serious field validation.

         Additional problems regarding the methodology and the spatial analysis processes employed have been discussed earlier, throughout the project.

   Further studies/extension of this project is highly recommended, particularly in terms of a multi-objective analysis, in which there is more than one scenario for development. For instance by comparing a capitalistic urban development scenario versus a sustainable urban development scenario versus a anti urban development, one could possibly generate a more realistic, optimal suitability urban development model for the North Shore Mountains (A Master Thesis Topic?)

    Many thanks for the many people who gave me support and aid to make leading to the completion of this project:

- Barry Carman (Land data BC)

- Dr. Nadine Schuurman (Assistant Professor - Geographic Information Science - Simon Fraser University)

- Jasper Stoodley (Spatial Information Systems Lab Administrator  - Simon Frasar University)

- Aparna Deshpande (Teaching Assistant - Geographic Information Science - Simon Frasar University)
 

- And finally, to my fellow classmates of the Technical Issues to Geographic Information Systems (Geography 355-4) course offered at Simon Fraser University during the 2001-3 semester (you know who you are!)