Abstract: For better understanding of bidirectional reflectance effects of heterogeneous surface observed via optical remote sensing, and to improve the ability of quantitative retrieval of canopy structure and physiological parameters, this study introduced an improved tower-based automatic canopy multi-angle hyperspectral observation system at Qianyanzhou flux station in Jiangxi Province. This system included a spectroradiometer(UniSpec-DC), a Pan-Tilt Unit(PTU), a power supplying system, a software control system, serial converters, and so on. A 45° angular holder is mounted on the tilt axis of PTU, which allows a full range of view zenith angle(θ_V)for canopy reflectance measurements, while the inherent motion range of tilt axis is only between -37° and 42° around the vertical axis. Several pre-processing methods, for example, sensor calibration, dark currency correction, abnormal data removal, and interpolations of missing angles and outrange data, were applied on the data obtained from the system. The pre-processed data then could be used to compute canopy reflectance, vegetation indices, and do quantitative analyses. Strong relationships(R² = 0.961 4, P<0.001)were observed between “photosynthetically active radiation”(PAR)from multi-angle hyperspectral observation system and tower measurements. Photochemical reflectance index(PRI)derived from reflectance data obtained by the system showed significant logarithmic relationships(R² = 0.514 2, P<0.001)with light use efficiency(LUE)calculated by flux tower data, which was well related to previous studies. Thus, multi-angle hyperspectral remote sensing is not only an important tool in quantitative researches of forest ecosystems, but also an effective way to further understanding the scaling issues of aero space remote sensing and filed measurements.