Geography Reference
In-Depth Information
8
<
0,
x ¼
0
0
:
1
x
0
μ
HPH
ðÞ¼
ð
15
Þ
, 0
<
x
<
0
:
1
:
:
1
x
0
:
:
1,
1
All the remaining coefficients are calculated similarly with the exception of coef-
ficient
C
4
, which has not been calculated yet because the system is not connected to
the on-line meteorological data.
The final deceleration coefficient
is calculated from the individual files
cxfuzzy
with the help of tool Fuzzy Overlay. The first five coefficients enter
the calculation with the help of relation max{
C
1
,
C
2
,
C
3
,
C
5
,
C
6
}. The resulting
value is multiplied by a raster of coefficient
C
7
by reason of assigning meaning of
the individual communications according to traffic importance (highways, first-
class roads, forest roads, etc.). The result is the cost map that can be an input for
searching of an optimal route in a decision-making process in CCM.
Results and Discussion
As it was mentioned at the beginning, the aim of the task was to describe thoroughly
the creation of process models for calculation of deceleration coefficients within
verification of the theory of terrain passability with the use of crisp set and fuzzy
theory and also to show the problems we met.
For verification of functionality of the created models, database DM
´
25 was
chosen, in the quality of 2010. Brno and its closest surroundings were used as the
area of testing. For the chosen area, first, the individual coefficients were calculated,
then the resulting cost map. It was followed by the verification of results directly in
the terrain.
One of the main problems of creation of the cost map proved to be the length of
calculations of the individual coefficients, which was very long, much longer than
for crisp set calculation. The complete calculation with fuzzification took place
with a basic pixel size of 25 m. Then we came to a pixel size of 5 m. In both given
cases, the time consumption of the calculation was acceptable. From the size of
mean error in position of the individual object, however, the necessity to use a pixel
size of 1 m became apparent. When we used this resolution level for the simplest
calculation, the time of calculation prolonged to an unbearable length. Coefficient
C
1
was the most time-consuming for calculation. It took 22 min on a desktop PC
Dell Precision 490, processor Intel Xeon 5130, 2,6 GHz, RAM 4GB, operation
system 32 bit Windows 7, Enterprise. For example calculation of coefficient
C
5
took 17 min,
C
6
took 15 min. The quickest was the calculation of coefficient
C
7
,
which lasted 3 min. The complete calculation of the cost map together with the
already calculated coefficients then took another 45 min. A sample of a part of the
calculated cost map is shown in Fig.
2
. If it was necessary to calculate in a higher
