The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
Up to now, there is no systematic analysis research on the 
impact of vegetation against the cascade hydropower station 
construction. Hence the goal of this paper is to assess the trend 
and temporal-spatial distribution of vegetation base on the 
remote sensing imagines, and then analysis the impact of the 
cascade hydropower station construction on the vegetation 
fraction. 
2. STUDY AREA 
In this paper, the study area reaches to 425 km long longyang 
gorgre to Liujia gorge reservoir (including Longyangxia 
Reservoir 105 km) , the total area is about 16730.9km , mainly 
covered by grassland (59.95%) , crop land (17.73%), unused 
land (8.80%)and water (4.30%). This area locates in the 
northwest inland plateau of China, east edge of northeastern 
Qinghai-Tibet Plateau, acrossing Qinghai and Gansu provinces 
(Figure l).This region is typical continental climate with plateau 
natural condition. The coldest and warmest average monthly 
temperature is 0-0.5°C in January and 15-15.5°C in July. Mean 
annual precipitation ranges from 194 to 357 mm, most occurs 
between June and August. However, the mean annual 
pan-evaporation is between 1500 and 2131 mm, exceeding 
precipitation nearly up to 10 times. The study area is the upper 
region of Yellow River watershed, which average altitude is 
2800-4800 m above sea level with a sharp eastward gradient. 
The Yellow River watershed is characterized by both rich 
natural resources and fragile eco-environments. 
+ 
l4Mxya 
I ! 
“ __ 
^ &***•i j 
Jt} 
I'-'»*»» t.k'v «tto» 
is»*#* *?**•>.' 
*»»<■“" ■ :w.r«K i 
* •: 
s .« x « « « c; - 
m. m wmmamm . ammmmm 
wme* ****** 
Figure 1 The location of study area 
3. DATA AND METHODOLOGY 
The model for vegetation index converts to vegetation coverage: 
= —-P-Yl-..- N .9. VI ™. (2 ) 
NID VI+ ND VI ■ 
max min 
Where / ndvi = vegetation coverage; 
NDVI max = the maximum NDVI value; 
NDVI mm = the minimum NDVI value. 
DEM(1:250 000)data, roads data(l: 250 000), residents data(l: 
100 000) and the collecting field data were used to analyze the 
vegetation distribution character. The ARCGIS 9.0 software 
were used to convert data types and did the overlay, 
reclassification, zonal statistic analysis. 
3.2 Method 
3.3.1 The classification of vegetation coverage change 
According to the Criterion of Classification of Soil Erosion, 
vegetation coverage was divided into five kinds: high-coverage, 
moderate-high coverage, moderate coverage, moderate-low 
coverage, low coverage. The increase of the vegetation 
coverage could be described as the conversion of high-grade 
coverage to low-grade coverage, while the decrease of the 
vegetation coverage could be described as the conversion of 
low-grade coverage to high-grade coverage. In a certain 
regional area, the rate of the increase vegetation coverage area 
and the total regional area was called as the vegetation increase 
rate, while the opposite was the vegetation decrease rate. 
3.3.2 Vegetation coverage change analysis 
The cause analysis of vegetation change is based on the GRID 
module in ARCGIS WORKSTATION, ©generating aspect by 
DEM data, aspect was divided into nine categories: north, 
northeast, east, southeast, south, southwest, west and northwest; 
take elevation 100m as the distance, elevation data will be 
divided into 11 bands ; taking into account the relation of the 
altitude and aspect, superimposed altitude data and slope data; 
©extracting the data of settlements and roads from the digital 
topographic maps (scale 1:250 000), taking residents and roads 
as the centre and 1 km as the unit respectively , producing 19 
and 14 buffer zones separately. On those bases, we obtained the 
number of the residents and the vegetation cover change area in 
different aspects, elevation and distance buffer zones, and then 
calculated the increase rate and the decrease rate of vegetation 
coverage. The correlation analysis between vegetation coverage 
change rate and other factors was completed in SPSS. 
3.1 Data source 
According to the construction time of each power plant, the 
remote sensing images in the same month of 1977, 1996 and 
2006 for research area were selected to monitor the change of 
regional vegetation distribution in the area of cascade 
hydropower development for 30 years. 
NDVI was extracted from the MSS, ETM+ and TM imagines, 
and then the vegetation fraction was estimated from NDVI 
through the dimidiate pixel model. Before estimating the 
vegetation fraction, we eliminated the errors that the NDVI 
from the different instruments, allowing vegetation indices from 
one instrument to be inter calibrated against another (Michael D. 
Steven, et si, 2003). 
The calculation formula of NDVI could be expressed as 
follows: 
NDVI = NIR ~ RED (1) 
NIR + RED 
4. RESULTS AND DISCUSSIONS 
4.1 Vegetation coverage change analysis in study area 
In both two periods, from 1977 to 1996 and from 1996 to 2006, 
the vegetation coverage change mainly rose in study area. The 
proportion of increased vegetation area accounts for 27.04% 
and 37.77% of the total area separately, while those of 
decreased vegetation area are 12.74% and 7.51% respectively. 
The distribution of the improved vegetation area was sporadic 
and patchy, mainly to a slight increase. And the distribution of 
the decreased vegetation area was patchy. From 1977 to 1996, 
the regions in which the vegetation coverage increased were 
surrounding the Longyangxia reservoir and Guide wetland. The 
regions in which the vegetation coverage decreased were mainly 
inside the Longyangxia reservoir. 
234