The spectral reflection curve characteristics of vegetation are different from other substances, such as soil and water, and are closely related to the performance of their physiological traits. The actual angle causes errors in the coverage information obtained by the sensor due to the inconsistency of the incidence effect when looking down and sideways at the canopy vegetation. In this paper, we conducted a real-time multi-angle observation experiment of grassland vegetation in the desert grassland of Siziwang Banner, Inner Mongolia, by using a field online multi-angle spectrometer designed and assembled by ourselves. The four indices, namely the Normalized Vegetation Index (NDVI), Ratio Vegetation Index (RVI), Optimized Soil Adjustment Vegetation Index (OSAVI) and Photochemical Vegetation Index (PRI), were used for multi-dimensional data comparison. We analyzed the correlation characteristics between the spectral diversity of grass physiological traits at different sensor observation angles (SVA) and solar altitude angles (SEA). It was found that the greater the sensor observation angle, the weaker the intra-day variability of canopy reflectance at different wavelengths, showing obvious variability in observation angles and differences in angular sensitivity, and the standard deviation of daily variability of vegetation reflectance was smaller when the sensor observation angle (SVA) was near 75° or observed vertically downward. The reflectance of vegetation was positively correlated with the solar altitude angle for a fixed sensor observation angle. Similarly, the angular effects of different vegetation indices also differed, with the OSAVI index being most sensitive at a sensor observation angle (SVA) of 45°, with the lowest value occurring in different months and the maximum value of the PRI index occurring below a sensor observation angle (SVA) of 60°. It was also found that selecting a certain sensor observation angle for different vegetation growth stages or for different solar altitude angles (SEA) is more beneficial to obtain valid and accurate data. The analysis results of data collected from the field online multi-angle spectrometer in this paper provide scientific data support for the correction of satellite image products, accurate monitoring of vegetation remote sensing, and accurate estimation of grassland biomass.