The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
241 
3. RESULTS AND DISCUSSIONS 
3.1 Landscape dynamic evolution in study areas 
Based on the landscape classification, the landscape dynamic 
change information can be concluded by the ArcGIS 9.1 
software. The individual class area and change statistics for the 
30 years are summarized in Table 1. It showed that grassland 
was the dominant landscape type. Area of grassland in different 
coverage was 8228.86 knr in 2006, a ratio of 59.33%. The main 
green plantations in this region were the grassland. However, 
the water area and construction land only took a small part in 
this region in the last 30 years. 
The construction of dam increases the water area in the region 
around reservoir. And the grassland has turned out to be water 
area for reservoir inundation. The grassland area has dropped 
from 8851.61 km 2 of total area in 1977 to 8228.86 km 2 in 2006 
with a decrease of 7.04%. In addition, there had been some 
opposite changes in it. The thirty years from 1997 to 2006 
witnessed the obvious drop of the middling coverage grassland 
from 4960 km 2 to 3676 km 2 with the ratio of it decreased from 
35.76% to 26.50%; the area of low coverage grass experienced 
a steady increase in amount from 3149 km 2 in 1977 to 3843 km 2 
in 2006 with the ration of it increased from 22.71% to 27.71%; 
the high coverage grass area decreased slightly during the past 
thirty years. The area of woodland ascended from 793 km* to 
901 km 2 , with a 0.8% rise, for the woodland is far away from 
the water area, it can be concluded that the impact of the 
construction of the hydropower station to the woodland is slight 
and acceptable. For the newly constructed reservoir, the whole 
water area including rivers, reservoirs and bottomlands has 
expanded from 470.48 km 2 in 1977 to 703.55 km 2 in 2006 with 
the ratio of water area increased from 3.39% to 5.07%. The 
construction of cascade hydropower station also accelerated 
local economic development, which converted most unused 
land to construction land. The change in farmland area was 
minimal, with a small increase from 2706 km 2 in 1977 to 2775 
km 2 in 2006. The regional landscape structure was changing 
notably with a massive increase of water area. These data 
proved that the main influence area of cascade hydropower 
reservoir to ecological landscape was the surrounding region in 
the last 30 years. 
Year 
Landscapetype—_ 
1977 
1996 
2000 
2006 
Area(km 2 ) 
Ratio(%) 
Area(km 2 ) 
Ratio(%) 
Area(km 2 ) 
Ratio(%) 
Area(km 2 ) 
Ratio(%) 
Woodland 
793.65 
5.72 
829.28 
5.98 
822.71 
5.93 
901.20 
6.50 
Low coverage grass 
3149.20 
22.71 
3572.22 
25.76 
3601.61 
25.97 
3843.52 
27.71 
Middling coverage grass 
4960.11 
35.76 
4407.32 
31.78 
4294.05 
30.96 
3675.92 
26.50 
High coverage grass 
742.30 
5.35 
678.12 
4.89 
636.98 
4.59 
709.42 
5.12 
Water area 
470.48 
3.39 
522.82 
3.77 
650.05 
4.69 
703.55 
5.07 
Construction land 
12.26 
0.09 
150.45 
1.08 
152.56 
1.10 
156.04 
1.13 
Unused land 
1035.16 
7.46 
1099.59 
7.93 
1037.46 
7.48 
1104.26 
7.96 
Farmland 
2706.14 
19.51 
2609.52 
18.82 
2673.87 
19.28 
2775.40 
20.01 
Total 
13869.3 
100 
13869.3 
100 
13869.3 
100 
13869.3 
100 
Table 1 Statistics of landscape change in the study area from 1977 to 2006 
3.2 Regional landscape pattern trajectory analysis 
Landscape structure could be described by its composition and 
configuration. Landscape composition refers to characteristics 
associated with presence and amount of each patch type within 
the research boundary. Landscape element is the basic, 
relatively homogenous and ecological units. So, it is feasible 
and reasonable to assess landscape variation by the patch related 
index. The most important landscape element type is the patch, 
which is a nonlinear surface area differing in appearance from 
its surroundings (W. Ouyang, 2007). 
Besides analyzing rules of the landscape structure change from 
1977 to 2006, the paper also studies the spatial-temporal 
characteristics in landscape evolution during the past 30 years 
grounded on 10 landscapes indexes by using geographic 
information system (GIS) and the landscape analysis tool 
FRAGSTATS. 
• Patch density (PD) 
Where A = total landscape area (m 2 ) 
nj = number of patches in the landscape of class type i. 
• Mean patch size (MPS) 
Where 
Where 
where 
Where 
MPS = 
ay = patch area 
nj = the patch number of different land use. 
Largest patch index (LPI) 
max(a.) 
LPI - — 
A 
ay = patch area 
A = the total area of all patches. 
Landscape shape index (LSI) 
0.25 E 
LSI = 
77 
E = length of total patches borderlines. 
A = the total area of all patches. 
Edge density (ED) 
(2) 
(3) 
(4) 
1 m 
ed = -XL Pi (5) 
A i=1 j=1 
m = total landscape factor types at certain special 
resolution 
Py = total length of borderline that shared by two 
consecutive landscape patches (patch i and patch j). 
CONTAG