ESTIMATING GRASSLAND YIELDS USING REMOTE-SENSING AND GIS TECHNOLOGIESIN CHINA

From green herbage yield, environment, and remote sensing (RS) data re corded in different grassland types in Fukang County, Xinjiang from 19 91 to 1996, correlation analyses and grassland yield estimates were ob tained using remote sensing and geographic information system (GIS) te chnologies. Methods of processing images, analysing information, and l inking of remote sensing data with ground grassland data were explored . Results showed correlation between fresh herbage yields and ratio ve getation index (RVI) and normalised difference vegetation index (NDVI) (P < 0.01) in four grassland types with correlation coefficient (r) > 0.679. Fresh herbage yields correlated better with RVI than with NDVI for lowland meadow, hill desert steppe, and mountain meadow, but not f or plains desert steppe. Optimum non-linear models for estimating yiel d were selected from six curves, and estimated total yields were verif ied by ground truth large-plot investigations and statistical analyses . The effects of estimating green herbage yields using non-linear mode ls were better than those using linear models in all four grassland re gions. The total accuracy of estimating yields by remote sensing was > 75% over large areas in the four grassland types using a combination o f remote sensing and GIS. Remote sensing, along with GIS, is a new app roach to the use, development, and management of grasslands.