Part I:
Part II: Non-Forest and
Non-productive Forest Area
PART III: Determining
Areas of Highest Value Timber
Overview
The design of my project is divided up into three
separate stages. The first stage o determines
the number of constraints that eliminate areas from the
timber supply regardless of their suitability. These included a coastal
riparian buffer, environmentally sensitive areas and steep sloped
areas. Secondly, I determined those areas that should be
eliminated as potential sites based on measures of the forest itself,
based on minimum criteria regarding density, average height, and age.
These three criteria, height, age and crown cover are the chief factors I
use in my final analysis to determine the best potential logging sites.
A fourth criteria, slope, was also used.
Below is a more detailed description of
methodology, my reasoning behind my methodology, as well as a description
of the spatial analyses I performed on the data to reach my goal.
PART I: Constraints
Purpose: To determine those areas within the
land base not available for timber extraction due to various constraints.
My spatial analysis consisted of creating a Boolean mask for the
three constraining factors. The three constraining factors are discussed
below:
Riparian
Buffer: Removal of forest limited to within a
set distance of the ocean to protect sensitive shoreline habitat. This
distance is set at 50 meters. Protecting
Nearshore Habitats Habitats lying along the shoreline strip of shallow water and
the land immediately adjacent to it can be extremely productive. These nearshore
habitats form an essential link in the food web of all the shared waters.
Shallow waters shelter the sensitive young of many deep-water fish and shellfish
and are home to vital prey for fish, birds and mammals. These areas are
extremely vulnerable to degradation and destruction and so a buffer is a
necessary constraint to provide some protection from soil runoff, and provide
shelter and habitat for birds and mammals, such as the bald eagle, that depend
on nearshore and shoreline resources for food and shelter. This was
a fairly simple procedure. First
I created a Boolean image of my Island so that all land areas = 1 and
all water areas = 0. I
utilized this map a great deal in subsequent analysis.
I re-classed my image ISLANDBOOL so that all water areas = 1 and
land areas = 2. Running
the Buffer image in Idrisi returned an image where all land areas
where again =1 except for my fifty meter buffer.
2. Poor
soil areas: Areas of poorly established, thin soils are very
sensitive to timber extraction. Logging in these areas is not
permitted due to poor forest regeneration when these areas are logged.
Creating a Boolean mask of all areas with severe soil stability
problems was also fairly straightforward. The attribute values file that imported from ArcView
contained this information. MoF
had coded sensitive soils as a sting and the field also contained
information I did not need, specifically environmentally sensitive areas
delineated as "areas with important scenic or recreational
value". After cleaning up my data table in Excel, I imported
the appropriate field as an attribute file and assigned it to my forest
coverage. This image was re-classed into a Boolean image with all
poor soil areas as 'o'. See POORSOILBOOL.
3.
Steep
slope areas: Slopes too steep for logging because
these areas are vulnerable to erosion of soil and slope instability once
trees are removed. The threshold for slope is set to 30 degrees or
greater, a typical industry standard. Creating the slope mask was much more complicated, and took a lot
of time and trial and error. The
contours vector file was imported into Idrisi, converted to a TIN,
converted then again into a DEM
and then finally converted into a
slope
image. See the
cartographic
diagram outlining these steps. Because of the 'bridges and tunnels' effect created through
errors in the Vector-TIN conversion, my DEM had elevation values over
water areas. I eventually gave up on trying remove this error though
specific Idrisi modules and instead ran Overlay Multiply with my
ISLANDBOOL map in order to mask these areas out.
Part II: Non-Forest and Non-productive Forest Area
The
next step toward determining the best logging sites on Read Island was to
remove all non-forest and unproductive forests from consideration. These include areas that
are non-forested, areas where the forest
is too young, and or not dense enough to be considered for extraction.
As with the constraints, the Boolean mask was performed to achieve
this.
-
Forest
too Young – It is considered generally in B.C. forest industry
literature, including the MoF, that coastal forest younger than 80
years is immature. Following
these guidelines, I simply created a Boolean mask that eliminated all
immature forest areas from the land base.
To do this I assigned a newly created values file containing
forest age information to my ISLANDFOREST
image. I re-classed this
new image, AGEFOREST so that all
pixels with a value less than 80 returned a value of 0 on my new map, AGEBOOL
-
Low
Density forest - Crown
closure is used as a measure of stand density and competition. It may
also be used in equations that predict stand volume.– After
examining the attributes files on forest data, I arbitrarily decided
to remove all forest areas on the land base with less than 10 percent
crown coverage. MoF
usually divides Crown Closure into 5 classes, where 0-10 percent is
considered 'very sparse'. I
did not want to go beyond 10% in case I eliminated valuable timber
areas of large old growth trees that had a low crown closure.
I repeated the same procedure as above to create an image
called CROWNBOOL.
-
Trees
too short – Forest areas that may be old enough and dense enough to
be included in the land base may still may be below standard with
regard to timber quality and volume because they are too short.
So I removed forest areas from the land base that were too
short as well, in order to mask out poor timber areas not removed by
the above two masks. I
arbitrarily chose all trees less than 18 meters tall.
I used the same procedure as in the above mentioned creation of
AGEBOOL to create a Boolean image, HEIGHTBOOL
PART III: Determining
Areas of Highest Value Timber
The third and most important leg of this project was
to attempt to accurately estimate the value of the forest using the
available information contained in the forest cover attribute file.
There was no way of determining this directly as useful attributes
such as volume, density and site class were missing from my data table
(although these attributes had their own field, they were empty of
information). My methodology
consisted of determining the most valuable forest areas based on height,
age, and crown closure, the same attributes I used to remove areas for
consideration. My reasoning
for utilizing these three attributes is this:
Firstly, all three of these attributes are important components of
determining forest volume and potential annual yield, although they are
not inclusive in the process. In
other words, while a more fine grained methodology would include other
factors such as volume and site class, I believe that these three
variables are generally adequate to the task.
While there is a very
strong correlation between age class and height class for obvious reasons,
the correlation between crown class and the other two variables is
somewhat less strong. This is because that Read Island has already been extensively
logged, and new forests re-grown in previously cut areas are mostly of the
same age and height. This
often creates a dense forest with a closed canopy.
While this is certain to introduce some error into my analysis,
including crown closure is an essential variable in my assessment as it is
my only means of estimating the volume of timber in conjunction with the
other two variables.
I also included slope as a variable in
my analysis of the best potential logging sites on Read Island.
Slope is quite unrelated to the other criteria, but is an important
component in overall site selection This
is because the steeper the slope, the greater the cost in extracting the
wood. Road building is more
expensive in steep sloped areas, as is both the cutting, removal, and site
clean-up. While I included
this as a factor in determining the overall assessment of potential
logging sites, I stress that it is a much less important criteria than the
above mentioned three.
The Idrisi tools I utilized to determine the best
potential logging sites was the Weighted Linear Combination in Idrisi.
The WLC allowed me to standardize my criteria on a continuous
scale, thereby permitting the combining and comparison of the criteria
to create a final continuous suitability map of potential logging sites.
WLC also permitted of weighing each of the factors
against each other in accordance to their overall importance in
determining site suitability. Therefore
I was able to give slope a much lower weight than the other three
criteria. .Forest height, age
and crown closure were given equal weight.
My reasoning for this was that I could not determine if one factor
was more important than another in determining overall value of a forest
area.
SPATIAL ANALYSIS
Below is a description of the spatial analysis I
employed using Idrisi's WLC in order to create a continuous suitability
map of potential logging areas. All
criteria were scaled to between 0 and 255.
1.
Forest Height – Height is an overall measurement of volume and
therefore forest value; to
state the obvious, as height increases, so does value (all else being
equal). I re-classed HEIGHTFOREST using Fuzzy MCE using the linear
function. My output image was
called HEIGHTFUZZ.
2.
Forest Age - Age is also a
measurement of potential volume. Also,
older trees cost less to process, and they are potentially more valuable
due to tighter grain. The
value of the forest age in determining suitability does not increase in a
linear fashion. Trees, depending on species, achieve their most rapid growth
in the first 200 years. After
this, growth continues, but at a much slower rate.
Therefore, the sigmoidal curve is more appropriate for considering
forest age.
I used Fuzzy again on my AGEFOREST image using the
increasing sigmoidal curve. I
entered 80 (the threshold of immature forest) and 200 as my two
thresholds. Output image was
renamed AGEFUZZ.
3. Crown Cover – Generally speaking, the greater
the crown cover, the greater volume of wood given a forest of similar age,
species, and height. Therefore I used the linear increasing scaling of Fuzzy to
create an output image of continuous Crown Cover suitablility, called
CROWNFUZZ
4.
Slope – Slopes greater than 30 degrees are considered as suitable
areas. However, between 0 and
15 degrees there is little or no increased cost associated with slope.
Slopes beyond 15 degrees, however, ad an increasing overhead cost.
Using the decreasing sigmoidal curve, I set 15
degrees as my lower threshold and 30 degrees as my upper threshold.
I named the output image SLOPEFUZZ.
SEE my Cartographic
Diagram showing the above steps.
The final stage was to create a coverage of
continuous suitability image based on the aggregation of the four above weighted
(by parewise comparison) and scaled factors
along with the six Boolean masks completed earlier.
All of this was performed in MCE, and my resulting image contained
areas of various sizes scaled between 0 and 255.
I called this image FORESTCUT.
My weights were as follows:
Despite the high resolution of my land cover, as is
evident from the image, there area relatively large continuous areas of
homogeneous suitability. I
believe there are two reasons for this.
Firstly, The data originally consisted of over four hundred large
polygons, each of which represented (and assumed) a homogeneous forest
coverage. Secondly, except for slope, my continuous factors were overall well correlated.
This image constitutes a continuous suitability, from
least to most suitable. In
the final stages of my analysis, I had to choose a threshold between
suitable and unsuitable. Because
of the fuzzy nature of my analysis, any threshold on my part will be
mostly arbitrary. Several
suitability thresholds could in fact be produced and utilized.
I chose, quite arbitrarily, 150 (out of 255) as my suitability threshold and
called this image SUIT150, which includes potentially the best sites for
logging on read with a total area of 317 hectares.
I feel that 150 was somewhat of a conservative choice, but I wanted
to be sure to exclude all poor sites. A
threshold at 150 would likely eliminate all areas that did not meet ALL of
the three main factors. A lower threshold would have allowed more
tradeoff between factors. Depending on the threshold, this would
have resulted in the possible inclusion of areas of low density. If in fact these sites all turned out to be extemporary, I would
reduce my suitability further.
I performed one last analysis on my suitable areas.
I rated the five largest suitable areas based on tree species.
My reasoning behind this is that some species are worth more than
others. To do this I created
a Boolean image of all tree species that are considered marketable, namely
Douglas Fir, Western Red Cedar, Western Hemlock, Sitka Spruce and
Lodgepole Pine. Each of these
images was re classed to determine those areas where they were the
dominant species. In other
words, for any given species layer, I re classed the image giving all
areas with greater than 55% coverage as 1 and all else 0.
Each species coverage was re-classed using a different number to
uniquely identify them (Douglas fir 1, Western Red Cedar 2, etc, etc) To do this I created
a 'Boolean' image each of these species.
Then in image calculator I combined each of the species images with
my SUIT150 image using the operator OR.
The result returned an image that differentiated the most suitable
logging sites by species, in areas where a particular species dominates.
Douglas Fir and Western Cedar, the most valuable of the species,
had the largest areas on the map. (see
image SUIT150SPECIES) The
overall area of suitability was reduced by fifty hectares (due to no
leading species with over 55 percent in that area). The largest single area is
50.2 hectares and is dominated by Western Red Cedar. Douglas
fir, however, dominates most of the other sites and is the leading species
in 168 out of a total of 263 hectares of suitable logging
sites.
