Problems and Discussion


Introduction	Problems and Discussions Design: Unable to obtain the exact location of current farms – It is obvious that an area is not suitable for a new farm if there is an existing one already. But since these farms probably do not have a postal address, there is no way to reference them on the map. So an existing farm may already occupied in where is “suitable” in the current result already. A problem with the scale – Analyzing over such a large area is not the most appropriate way to pinpoint where a salmon farm should locate. All the analysis can do is to show where in general is good and where is not. Yet if the project was designed to concern a small area, i.e. the east coast of Queen Charlotte Island, then there are not enough data points to provide a good interpolation at that scale. So this is an inherit problem. Location of cities and towns are disregarded – For example, it is obvious that there will not, and should not have any salmon farms near the Greater Vancouver region, but it may be very desirable to have some farms near Port Alberni. Either way, the effect of cities and towns are not taken into consideration. Limited temporal coverage – Limited by the time allowed, it is most reasonable to pick a season or month to represent. An advantage of the current way is that data are very consistent in terms of ways they are prepared. Almost all the data fits within one time frame. Or on the other hand, the amount of work can be easily doubled or tripled (by guess!) if a time series is taken into account. Methodology - Operational: Extremities affected the depth layer – The originally intended depth from 0m - (-500m) will exclude most area around Queen Charlotte. This is a problem due to a few extremities just out of bound (-3500m). Alternatively, the value from those three points can be ignored, but for the consistency of the data, and because -3500m does not constitute as an unknown value, I will consider them legitimate. Interpolation –The largest problem is probably the accuracy (to model the reality) interpolation (with the currently available sample points) can provide. IDW and spline both has its strengths and weaknesses, but neither seems to be the best to make a very reasonable surface. Although there are many weight and neighbor combinations possible for both method, and there are also many sophisticated methods availabe (i.e. Krigging), the most simplest ones are used but the truthfulness remains unknown. Legend – For some reason, the scale bar does not come out properly. Thus all the maps do not have a scale bar. Improvement: More data points will provide better surfaces. Average data where appropriate to take into account of natural variations, instead of taking spot data (temporally).
Data Collection
Methodology
Spatial Analysis
Interpolation
Overlay
Fuzzy
MCE
Result
Discussion
Conclusion


Email me