| The
IDRISI macromodel below outlines the procedure used
to get all data layers into fuzzy/constraint representations
that could subsequently be applied to IDRISI's Decision
Wizard:
| 1 |
Constraint
Layers |
|
|
The
first step in arriving at the intended suitability
surface was getting the data into IDRISI and the
creation of two Boolean constraint layers of areas
outside the analysis. Both Park Boundary and Lake
shape data was imported from ARC INFO using the
IDRISI import application ARCSHAPE and rasterized
with POLYRAS. Park Boundary was not altered and
renamed as Parkbndry_con. In the case of Lakes,
assign was used to exclude lake features (i.e. value
0) from their surroundings (i.e. value 1). A composite
overlay of these two removed all areas not to be
considered in the final MCE analysis. |
| 2 |
Slope
and Aspect |
|
|
Both slope and
aspect layers were derived from the 10m Crater
Lake DEM using IDRISI modules of the same name,
and standardized using FUZZY.
The thumbnails
provided show the result of the FUZZY application
on an initial image giving slope in degrees to
a standardized 0-255. Similarly, aspect (third
down) was scaled from 0-360.
Slope_fuzzy was
derived using a sigmoidal decreasing function
with c and d control points of 15 and 65 degrees.
This would seem to be a reasonable limit for popular
recreational trail. Aspect_fuzzy is the result
of a sigmoidal symmetrical function with c and
d control points favouring a range of 30 degrees
around the eastern slopes.
Both slope and
aspect fuzzy use a RADAR scale palette to provide
better visual separation
|
| 3 |
Vegetation
Types |
|
|
In order to give
added weight to the desired vegetation types, the
initial vegetation layer was reclassed using ASSIGN,
giving Lodgepole and Ponderosa Pine, Mountain Hemlock,
and Shasta Red Fir an equal value of 15, Egelmann
Spruce and Douglas Fir (the only remaining vegetation
zones that are trees) a value of 10, and all remaining
classes a value of 5 in order to give proper focus
to forested zones and particularly the Pine/Hemlock/Fir
Group.
Following a reprojection of the layer to match the
number of rows and columns to the rest of the data,
the FUZZY module was applied to the image with a
linear increasing function with control points a=0
and b=15 in order to standardize the image to a
0-255 scale. |
| 4 |
Streams,
Roads and Trails |
| |
|
Stream, Road and
Trail data provided by the National Park Service
were rasterized and reclassed to create separate
binary representations of all feaures on each coverage.
Using DISTANCE a new image was then calculated for
each layer expressing distance from features on
a continous scale. Roads required a reprojection
to meet a matching row/column resolution. Unique
FUZZY parameters were then applied to each of these
images according to the stated criteria and are
summarized in the table below.
Since the intention of minimizing trail routes along
stream lines was to reduce the cost that would accompany
building footbridges, a steep linear increasing
function and control points a and b at 5 and 50
were used. Thumbnails 2 shows a zoomed in image
of the buffer around the streams for better comparison.
Simlarly the buffer around trail routes was narrow,
with points at 20 and 100. Roads however were given
a relatively wide berth using a sigmoidal increasing
function due to their strong negative impact on
the hiking environment.
|
Table:
Summary of All Fuzzy Functions and Control
Points
|
| Layer |
Function |
Control
Points (a,b,c,d,) |
| Slope_fuzzy |
sigmoidal decreasing |
15, 65 (degrees) |
| Aspect_fuzzy |
sigmoidal symmetrical |
0, 181, 202, 255 |
| Vegetation_fuzzy |
linear increasing |
0, 15 |
| Stream_fuzzy |
linear increasing |
5, 50 |
| Road_fuzzy |
sigmoidal increasing |
300, 3000 |
| Trail_fuzzy |
linear increasing |
50, 300 |
|
The final output
of this project ultimately fails in it's intention to
separate a line of best fit for a n
|
)
)
)
)
)
)
)
)
)
)
)
)
)
)