CORRELATION ANALYSIS BETWEEN VEGETATION NEAR-GROUND REFLECTANCE SPECTRAL CHARACTERISTICS AND BIOMASS FOR INNER-MONGOLIA STEPPE
内蒙古草原植被近地面反射波谱特征与地上生物量相关关系的研究

The spectral reflectance characteristics and biological parameters were measured for 15 grassland community types in Inner_Mongolia from 1994 to 2001. In addition, for seven community types, the seasonal variation of these parameters were measured in 1996 and 2001. At large spatial scales, the spectral reflectance was significantly different among meadow_steppe, typical_steppe and desert_steppe and could be distinguished by PCA with lower than 20% mean error. At medium to small scales, the ability to discriminate among community types in typical steppe was higher than at larger spatial scales (the mean error was about 15%) but lower in both the meadow_steppe and desert_steppe. The results of correlation analysis indicated that the spectral reflectance characteristics of biomass and vegetation indices showed strongly significant correlations with spatial scales and community types. The strength of the correlation tended to decrease from small spatial scales to large scales. Linear models were best able to predict biomass from the spectral reflectance characteristics of communities that had high vegetation cover or biomass with an estimated reliability greater than 90%. Non_linear models were the best predictors of communities with low vegetation cover (<40%) and had an estimated reliability of about 85%. From June to September, there were no remarkable differences among the estimated biomass models among different months which implied that we can utilize a common model to monitor the seasonal change of biomass for these grassland types.