Abstract:Taking the dominant shrubs,Potentillafruticosa L,Rhododendron ,and Salix cupularis,in the Eastern Qilian Mountains of the Qinghai-Tibet Plateau as the research objects and based on field hyperspectral data and biomass data collected in the field,we selected the characteristic bands to form the hyperspectral vegetation index,four estimation models of biomass (linear,exponential,logarithmic,and polynomial) and five vegetation index (NDVI,RVI,DVI,GNDVI,and PRI) were constructed by single factor regression method,and the accuracy of the model's fitting effect was verified.The results showed that shrub biomass was highly correlated with NDVI and GNDVI (R2>0.8),and the fitting effect of polynomial model was better than that of the linear model,exponential model,and logarithmic model.The polynomial fitting estimation model of alpine shrub biomass constructed by GNDVI was the optimal model,which has the highest simulation accuracy and the best fitting effect (R2=0.884 9).The establishment of the biomass estimation model of alpine shrub had important practical significance for accurate estimation of the alpine shrub biomass on the Qinghai-Tibet Plateau,and can provide a theoretical reference for the estimation of other plant biomass and protection of the Qinghai-Tibet Plateau.