The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
1031 
quantitative concepts is the best way to learn and evaluate it 
presently. The index weighting method, Analytic Hierarchy 
Process (AHP) and eigenvalue of group decision-making 
method are usually used at present. Principal Component 
Analysis was adopted here. This is a useful technique for 
reducing multidimensional indicators to lower dimensions with 
most information and weighting indicators properly. Meanwhile, 
it could find main contradiction and useful information through 
correlation analysis of indicators. 
4. RESULTS AND ANALYSIS 
With the assessment indicator system, the ecosystem health 
indexes of Ordos area were extracted from the following 
process: 
4.1 Pressure Indicators Analysis 
Based on MODIS/Terra NDVI data of Ordos area, threshold 
extraction was used to get the desertification area and 
distribution, the accuracy assessment of remote sensing 
investigation is 85% with statistic data as reference. 
Meanwhile, according to the 1:250,000 general road map and 
research experience, a buffer with 300m as radius was generated, 
and traffic impact index was got through calculating the ratio of 
the road buffer zone area in assessment units. 
The human footprint index was got from the ratio of artificial 
landscape (farmland, building lot) in one assessment unit, on the 
basis of 1:250,000 land use map. 
Finally, Principal Component Analysis (PCA) of these 
indicators was done in SPSS. From the weight statistic result 
(Table 2), we can see, the main pressures of study area come 
from artificial one, specifically speaking, are human footprint 
index, population density, ratio of theoretical and practical 
stocking rates, that means the study area is facing population 
growth and over-grazing presently. The desertification pressure 
is the main part of natural one. In summary, the regional 
environmental management should focus on coordination of 
population growth, environment protection and desertification 
controlling. Presently, the situation in middle and south-west 
part is serious (Figure 3), because these areas are mainly desert 
plateau, sand and desert. As the desert expanding, the ecosystem 
is in danger. 
Index 
Weight 
Human footprint index 
0.35 
Population density 
0.32 
Stocking rates 
0.34 
Desertification index 
0.25 
Traffic impact index 
0.17 
Ratio of damaged land 
0.23 
Table 2. Pressure indicators weight statistic result 
4.2 State Indicators Analysis 
Combined distribution of grassland pattern with theoretical 
stocking rate, the largest grassland pattern (desert steppe) was 
N 
0 15 » so 90 
Figure 3. Result of Comprehensive Pressure analysis in Ordos city 
chosen and the value of the different grassland pattern was got 
by Eq. (1): 
Ri = — 
S 
Where R ; = the vigor value of grassland i 
Si= the theoretical stocking rate of grassland 
S= stocking rate of desert steppe 
Landscape diversity and patch density index was derived from 
following process. First, ArcGIS was used to overlay and 
integrate land use, DEM and grassland pattern data, after slope 
and roughness extraction, landscape software FRAGSTATS was 
used to get Shannon’s Diversity Index (SHDI), Shannon’s 
Evenness Index (SHEI) and Patch Richness Density (PRD), 
then the two indexes were got from PCA. 
Then, PCA of these indicators was done in SPSS and pressure 
index distribution of study area was got (Figure 4). As we can 
see from Table 3, the contributing value of the pressure 
indicator is mainly from annual precipitation-evaporation ratio 
and the next are grassland pattern and vegetation coverage. The 
whole study area belongs to drought and rainless region, 
afforestation models adaptable to local conditions should be 
adopted, such as salix matsudama, salix mongolica, hippophae 
rhamnoides etc. From the final result, we can see, the state 
indicators in north-east part of study area are the best, and these 
in the middle and south part are the worst. 
Index 
Weight 
Patch density index 
0.10 
precipitation-evaporation ratio 
0.39 
Landscape diversity index 
0.11 
Vegetation coverage 
0.28 
Net Primary Productivity 
0.22 
Grassland pattern 
0.31 
Table 3. State indicators weight statistic result