Abstract:During July and August 2009, an ASD Fieldspec3 spectroradiometer and a Japanese Fuji digital camera were used to measure hyperspectral data and vegetation coverage on the Hulunbeier Grassland, Inner Mongolia, China. Regression analysis methods were used to establish a ground spectral relationship between the local measured vegetation coverage and normalized difference vegetation (ASD NDVI). Then, the relationship between NDVI got from MODIS/TERRA (MODIS NDVI) and ASD NDVI was analyzed to establish a MODIS Spectral Model to estimate vegetation coverage with MODIS NDVI. Finally, the accuracy of MODIS spectral model was validated. The MODIS Spectral Model was just a linear equation but had higher prediction accuracy than sub-pixel models with 11.58% standard error and 88.75% average prediction accuracy.