A Remote Sensing Method Of Estimating Grazing Intensity Of A Typical Steppe

China has nearly 400 million hm2 of natural grassland, or 13% of the world's grassland area, which occupies about 41% of the total land area. Excessive grazing in the grassland during the past 30 years has lead to its degradation with currently over 90% of the grassland experiencing varying levels of degradation. Effects of grazing in natural grasslands of northern China have been studied since 1980. These studies produced many interesting findings, however approaches to quick and efficient estimation of grazing effects over large geographic areas have been lacking. Our study attempted to resolve this shortcoming by investigating a typical steppe of Inner Mongolia and focusing on the scale of family ranch. We analyzed changes in biomass and developed a method to assess effects of different grazing intensities over a spatially extensive region. Our approach involves the use of field biomass data and remotely sensed derived vegetation indices (Ⅵ) which were analyzed with Detrended Correspondence Analysis (DCA) and regression modeling. Our conclusions are as follows:(1) The first and the second DCA axes represented effects of soil and atmosphere. Soil was found to be the most important factor that affectsⅥ. DCA divided 14 VIs into 4 categories and helped us select the best category (NDVI, SAVI and OSAVI) which minimizes the influence of soil and the atmosphere.(2) Regression models for estimating productivity based on biomass depend on biomass levels. For biomass lower than 370 g/m2, a simple linear model performs the best. Both linear and exponential models predict productivity equally well for biomass levels between 370 and 720 g/m2 while exponential model is more robust for biomass levels greater than 720g/m2.(3) Different types of grassland have different productivity levels. Plant species that constitute grassland communities also have different tolerance to grazing. Grassland communities characterized by high productivity and dominated by species with high grazing tolerance can withstand greater grazing intensity.(4) We usedΔB7-8 and G to construct a model that estimates effects of different grazing intensities on grassland communities of a typical steppe:G=0.135ΔB7-8-0.171 (R2=0.461, p<0.01). The model is robust statistically and can be used for moderately accurate predictions across our study area.