Abstract: The Yellow River source region, as the headwater of the Yellow River, where the ecological environment is sensitive and vulnerable, is located in hinterland of the Qinhai-Tibet Plateau. It is of significant importance to protecting and constructing natural protection of “ThreeRiver Headwater” by using remote sensing to monitor dynamically the grassland degradation in different temporal and spatial scale in this region. Grassland degradation refers to the overall reduction in grassland productivity as a consequence of human activities and natural processes. It represents the decline of grassland quality, density of grass cover, productivity, service function, or an increase in unpalatable grass species, and even as denudation of underlying soil. Compared with conventional grassland degradation researches, which are through field investigation to identify contributing factors and then each of the factors was graded and combined, remote sensing is much more efficient in assessing grassland degradation. Accorded to the national standard of China(GB19377-2003) and actual condition of grassland degradation in the Maduo County in the source region of the Yellow River, this study selected vegetation cover, pasture plant height, aboveground biomass, palatable pasture proportion, soil organic matter as the 5 evaluation indices. Then the calculation models were established based on the remote sense and in situ measured data, to monitor and evaluate grassland degradation in that region. Because of differential sampling sizes on the ground and space, the 5 evaluation indices cannot be directly quantified from TM imageries based on concurrently collected samples over 1 m2 plots. So, significant regression models were used to set relation between in situ measured data and TM bandsderived index values. However, for the evaluation indices of pasture plant height and palatable pasture proportion, there is weak correlation between in situ measured data and remote sensing. In order to solve this problem, the calibration of the TM images was used before the evaluation models establishing. After weighting stack of layers, the grassland degradation processes and distribution character during 1994-2009 were discussed and analyzed systematically. The result shows that the grassland degradation pattern in the Maduo County initially took shape in 1994, since that year this degradation process has been taken place continuously. The grassland degradation is severe especially in the areas which are sensitive to climate change, along traffic artery and close to settlement with frequent human activities. The situation of grassland degradation during 1994-2001 is most serious and the area of severe degradation is 135.59×104 hm2, accounting for 86.53% of the total grassland area. Compared with the four periods with severe grassland degradation, the trend of grass degradation has been alleviated with fewer observations of severe and medium degradation. It means that grassland ecosystem has begun to restore in the Maduo County. Although there are several physical reasons leading to the grassland degradation in Maduo County since 1994, the most important driven force is human activities. The effective way to resolve the grassland degradation problems is to decrease the number of sheep and yak, and to keep the sustainable development, there is an urgent need to enhance the management and dynamic process of grassland ecosystem better.

Key words: grassland degradation, remote sensing monitor, temporal and spatial pattern, the source region of Yellow River, Maduo County