The single greatest operational error in my data is
derived from the conversion of the data from vector to raster. This
is because raster cells by nature cannot reproduce the exact shape of the
vector polygon. I attempted to reduced this raster stepped effect by
choosing a high resolution. However, my 1 pixel for every 17 meters
of ground is significantly higher than original minimum mapping scale of 2
ha. This creates the illusion of greater detail that does not match
the scale of the original data.
Methodological errors are generally derived from my
choice of variables for my analysis.
Initially
and throughout much of this project, I felt that
Also, there was also various types of error in
my data before I started my analysis. While the gathering of this
particular data followed the industry standard procedures, error was
created in all of the stages of the data gathering data. For
example, the forest layer attributes was derived largely from areal photo
interpretation, which is subject interpretation error. While the
industry has statistical methods of determining reliability (in this case,
attribute accuracy was determined to be > 85%), I can
only guess at what kind of errors were propagated and magnified through my spatial
analysis.
As already mentioned, the
forest industry as well as the Ministry of Forests utilize far more
sophisticated methods of analysis and a greater decision support system
than is used here. Further, I
had to use my own intuitive sense to create a reasonable result, due to
the lack of information. For
example, crown closure was one of the major factors I used to determine
potential logging sites instead of more direct measurements of density and
volume. I initially believed
such a fined grained analysis was possible, but the further I got into my
analysis and the more I learned about forestry, I realized that this was
not possible.
My analysis would have been
more complete had I had a more extensive road network of the island.
As it was, it included only the major roads on the island, and the
vast majority were located a small north west section of the island where
most people live. Forest proximity
to a road network is an important variable in determining the cost of
timber extraction, proximity of timber areas to roads would have influenced
the outcome of my analysis.
