ED
e
A
k
m
ik
=
(
)
=
∑
1
10 000 ,
(2)
where
e
ik
is the total lenght of the edges in the landscape that
contains the
i
patch type, and
A
is the total area of landscape (m
2
)
The Largest Patch Index is found by dividing the area of
the largest patch of the relevant patch type by the total land-
scape area, multiplied by 100 to translate it into hundreds.
LPI
is equal to the percentage of landscapes created by the largest
patch and calculated with the following Equation 3,
PI
a
A
j
n
ij
=
( )
( )
=
1
100
max
(3)
where
a
ij
is total area of patch
ij
(m
2
), and
A
is the total area of
landscape (m
2
).
Euclidean Nearest Neighbor Distance Distribution is mean
value of Euclidean Nearest Neighbor Distance which is aver-
age of distance values to the nearest patch of all patches in
the landscape in Equation 4:
ENN MN_
=
=
∑
j
n
ij
h
n
1
(4)
where
h
ij
= euclidean nearest neighbor distance, and
n
= num-
ber of patches of the same class.
The Area-Weighted Mean Fractal Dimension Index express-
es the area weighted average value of the fractal dimension
values of all the patches belonging to the same class. The frac-
tal dimension index is calculated by the following Equation 5:
FRAC
p
a
ij
ij
=
2 0 25 ln( .
)
ln
(5)
where
p
ij
= the perimeter of the patch
ij
(in meters), and
a
ij
=
area of patch
ij
(m
2
)
The value range is 1
%
FRAC
%
2. The simpler the patch
forms are, the closer they are to value 1; the more complex the
patch forms are; the closer to 2.
The Contagion (
CONTAG
) index refers to the full likelihood
that the cell of a particular type of patch will be adjacent to
the same type of cells, and calculated with Equation 6:
CONTAG
P
g
g
ln P
g
g
i
m
k
m
i
ik
k
m
ik
i
ik
k
m
= +
=
=
=
=
∑ ∑ ∑
∑
1
1
1
1
1
*
*
*
ik
m
( )
2
100
ln
( )
(6)
where
p
i
is the amount of area that patch
i
covers,
g
ik
is the the
number of contiguous pixels of patch type
i
and
k
.
The values obtained are 0 <
CONTAG
≤
100.
When the patch types are separated at the maximum level,
CONTAG
approaches 0. When all patch types combine at maxi-
mum level, they take
CONTAG
100 value.
Results and Discussions
LCLU
maps were created for the Center and Safranbolu dis-
tricts of Karabuk adhering to the Urban Atlas class definitions
and methodology with object-based classification of Spot6/7
images. The landscape metrics were used to compare the
spatial structure and forms of these two districts using these
LCLU
data. This research was conducted for two different
scales, the first scale is focusing on urban areas of Safranbolu
and Karabuk Center which was determined by municipal
boundaries and shown in Figure 4 and Figure 5. Whereas, the
second scale covers whole district boundaries considering the
fact that object textures, geometries and characteristics are dif-
ferent in the urban and rural areas.
LCLU
maps of Safranbolu
and Karabuk Center districts shown in Figure 6 and Figure 7.
Two scales were useful to understand landscape metrics and
landscape characteristics relationships.
After the classification process, using the results of clas-
sifications, spatial statistical values of classes were calculated
with
GIS
software. For both urban areas and whole districts,
each of land use classes’ area and percentage values are
shown in Table 3.
For urban areas, it was seen that the percentage of the ur-
ban fabric classes for the Safranbolu is higher than the Center.
Percentage of industrial, commercial, public, military, and
private units in the Center is higher. However, in Safranbolu
this ratio is significantly high.
Table 3.
LCLU
classes, area calculation and percentage results for urban areas.
Class
SUA CUA SUA CUA Safranbolu Center Safranbolu Center
Area (ha)
Area (%)
Area (ha)
Area (%)
Continuous Urban Fabric (S.L. > 80%)
53.98 106.14 1.54% 1.53% 56.13 107.39 0,07% 0,15%
Discontinuous Dense Urban Fabric (S.L. 50% - 80%)
128.60 128.41 3.67% 1.85% 141.95 139.05 0,18% 0,19%
Discontinuous Medium Density Urban Fabric
(S.L. 30% - 50%)
98.85 175.07 2.82% 2.52% 136.48 227.82 0,18% 0,32%
Discontinuous Low Density Urban Fabric (S.L. 10% - 30%) 153.71 250.23 4.39% 3.60% 244.24 329.19 0,31% 0,46%
Discontinuous Very Low Density Urban Fabric (S.L. < 10%) 113.20 128.77 3.23% 1.85% 322.69 356.54 0,41% 0,50%
Industrial, commercial, public, military
and private units
219.33 741.70 6.26% 10.68% 242.67 778.80 0,31% 1,08%
Other roads and associated land
362.19 560.98 10.34% 8.08% 1044.41 1154.31 1,34% 1,61%
Railways and associated land
*
18.13
*
0.26% *
43.03
*
0,06%
Mineral extraction and dump sites
49.97 29.63 1.43% 0.43% 99.66 117.07 0,13% 0,16%
Construction sites
12.60 13.87 0.36% 0.20% 18.77
24.78 0,02% 0,03%
Land without current use
26.91 62.46 0.77% 0.90% 26.91
62.46 0,03% 0,09%
Green urban areas
42.31 126.71 1.21% 1.82% 42.31 126.71 0,05% 0,18%
Sports and leisure facilities
13.51 18.96 0.39% 0.27% 13.51
18.96 0,02% 0,03%
Agricultural areas, semi-natural areas and wetlands
1954.39 3091.06 55.82% 44.52% 15898.00 13148.94 20,42% 18,29%
Forests
252.34 1363.79 7.21% 19.64% 59327.35 55012.43 76,19% 76,54%
Water
19.63 127.67 0.56% 1.84% 255.17 226.94 0,33% 0,32%
Total
3501.52 6943.60 100.00% 100.00% 77870.25 71874.44 100.00% 100.00%
716
November 2018
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
