Classification Results 
The best result of the classification was calculated from 
the Local-Mean-Matching procedure of the SPOT XS'95 
and the IRS-1C '97 fusion product. The classification 
accuracy of the original data sets, the layerstack with 
additional information and the illumination classes was up 
to 12 % more worse than the LMM-result. 
Calculation of Landscape Indices : 
For a discrimination of landscape structure calculation of 
landscape indices carried out. 
The landscape indices were calculated for a subset of the 
investigation area. The indices, especially patch-, class- 
and landscape indices were calculated with the Raster- 
Version of Fragstats, Vers. 2.0. 
  
Figure 2: Local-Mean-Matching Classification - 
City area Dahn , Palatine forest 
Legend Figure 2: 
    
Patch-Indices are geostatistical descriptions of every 
homogenous object in the classified image, regardless 
which class it refers to. Patch-indices are for example : 
Area(ha), Perimeter(m), Shape Index, Core Area(ha), 
Core Area Index(%), Landscape Similarity(%), Edge 
Contrast(%), Fractal Dimension, Number of Core Areas. 
Class-Indices are calculating geostatistical values for 
every class in the classified image ( Table 1.), while 
landscape indices describing the total analysed area 
regardless of each class. Typical landscape indices are 
the Shannon Diversity Index, Simpson Diversity Index, 
Patch Richness, Patch Richness Density (#/100 ha), 
Relative Patch Richness, the Shannon - and Simpson 
Evenness Index. 
  
Patch Type: conifers,young 
Class Area (ha): 936.835 
Total Area (ha): 3029.522 
Percent of Landscape (%): 30.924 
Largest Patch Index (%): 6.890 
Number Patches: 1386 
Patch Density (#/100 ha): 45.750 
Mean Patch Size (ha): 0.676 
Patch Size SD (ha): 7.123 
Patch Size CV (96): 1053.874 
Total Edge (m): 540090.000 
Edge Den (m/ha): 178.276 
Landscape Shape Index: 25.535 
Mean Shape Index: 1.381 
Area-Weighted Mean Shape: 8.158 
Double Log Fractal: 1.330 
Mean Patch Fractal: 1.073 
Area-Weighted Mean Fractal: 1.297 
Core 9^ of Landscape (96): 24.828 
Total Core Area (ha): 752.183 
Number Core Areas: 1008 
Core Area Density (#/100 ha): 33.273 
Mean Core Area 1 (ha): 0.543 
Core Area SD 1 (ha): 6.073 
Core Area CV 1 (%): 1119.037 
Mean Core Area 2 (ha): 0.746 
Core Area SD 2 (ha): 7.111 
Core Area CV 2 (96): 952.888 
Total Core Area Index (96): 80.290 
Mean Core Area Index (96): 14.242 
Intersper/Juxtapos (96): 42.784 
  
  
  
Table 1: Class indices of the class 
* conifers, young" calculated 
with FRAGSTAT , Vers.2.0: 
For the interpretation of landscape indices it is important, 
that the size, form and the topological characteristics of 
different landscape objects is captured in standard 
values. 
O'Neill et al. (1988) described the demanded 
characteristics of landscape indices : 
1. the indices should describe similar structures in the 
landscape by identical values. 
2. indices should be independent and uncorrelated from 
other indices 
3. indices should be scale invariant, that means they 
should be independent from the absolute area of the 
investigated region. 
374 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
In th 
are \ 
For 
TRA 
Inde 
uncc 
calci 
area 
calc 
The 
bord 
Inde 
indic 
calci 
line. 
area 
For 
Frac 
a mi 
scal 
very 
2 f 
perir 
The 
the 
inde 
Juxti 
neig! 
If the 
class 
inde: 
The 
First 
the « 
high 
Con 
The 
satel 
data 
land: 
desc 
land: 
evali 
geoc 
Refe 
Cha 
Cripf