Abstract: This study aimed to explore the effects of the combination of various vegetation indices and partial least squares regression (PLSR) on improving the evaluation accuracy of aboveground dry biomass of winter wheat. The experiment was based on nitrogen operation test and wasconducted to analyze the correlation between 18 vegetation indices and the aboveground drybiomass of winter wheat. The better vegetation indices were selected to establish the PLSR model as a combined vegetation index, and the model performance was then evaluated. The results showed that, except for the chlorophyll normalized vegetation index (NPCI), a good correlation was observed between the vegetation indices and aboveground dry biomass of winter wheat.Especially, the correlation coefficients of the four indices, i.e., MERIS terrestrial chlorophyll index (MTCI), green normalized difference vegetation index (GNDVI), modified red edge ratio vegetation index (MSR705), and pigment specific simple ratio carotenoids (PSSRc), were greater than 0.800. Among the PLSR models established with vegetation index combination, the model calibration set (R²=0.719, RMSE=0.316) and validation set (R²=0.696,RMSE=0.346) based on the combination of PSSRc, MSR705, and MTCI performed best. Therefore, we concluded that the combination of multiple vegetation indices could improve the estimation accuracy of aboveground dry biomass of winter wheat. This study provides an effective approach for hyperspectral remote sensing estimation of the aboveground biomass of winter wheat.

Key words: soil moisture, cosmic-ray fast neutron method, completely dry soil condition, underlying surface, number of neutrons