The main problem that I encountered in my analysis was that the Vancouver Island dataset that I used was at a relatively small scale.  When searching for a suitable location over a large area, such as Vancouver Island, small scale data such as this can have the benefit of being somewhat general and not overloading the analysis with information.  However, data with such a large resolution can lead to problems when doing spatial manipulations.  Distance and buffer operations can only be as precise as the data's resolution.  Since the data's resolution was 1000 m,  the distance and buffer operations performed in my analysis are only accurate to the nearest kilometer.

        The large resolution also affects the accuracy of the slope analysis, as a pixel's slope is calculated from the elevations of its surrounding cells.  Elevation, and therefore slope, can change quite dramatically in 1000 m, especially in British Columbia.  Looking back, I would have given less weight to the slope factor in the cost/preference analysis.  The resolution of the raster data also introduces error when converting a vector line coverage to raster.  The lines (ex. major roads) can only be as thin as each pixel, meaning that the roads become a kilometer wide.

 
        Besides a larger scale dataset, the project could have also benefited from the inclusion of a few more data coverages, namely a layer showing the location of other eco-tourism and tourism resorts on Vancouver Island.  The resort could have found a niche area away from competition or located closer to others in an attempt to steal away their clientele.  But, this is why hindsight is 20/20.

        All considered, I believe that the project was a relative success.  The location selected was a cluster of suitable sites, increasing the chance that the area is somewhat spatially correlated according to the criteria imposed, and not just an anomaly caused by the large resolution of the data.  Of course this will not truly be known unless the site is ground truthed--a whole other chapter to my project that will mostly likely never be completed.

THE END

---

BACK     INDEX