In situ measurement of water spectrum is one of the indispensable basic works in the research of water optical properties and watercolor remote sensing inversion modeling. The conventional oblique observation method is restricted by its strict observation geometric conditions, it is necessary to constantly adjust the observation angle according to the position of the ship and the azimuth of the sun, especially for the spectral observation of river water, need to consider the river direction, shoreline shelter and other conditions. Therefore, only a few stations can be set for observation of discrete sample points, it is not easy to carry out rapid observation for continuous navigation in the river water with complex environments around the shoreline. The rapid observation for continuous navigation of the field water spectrum can obtain the reflectance spectrum of large samples of water from different local times, enrich the understanding of the bidirectional reflection characteristics of water, and establish more accurate inversion models, which plays an extremely important role in the study of watercolor remote sensing. Because of this, a rapid observation method for continuous navigation of reflectance spectrum of inland river surface water based on vertical observation geometry is designed in this study, and then obtained the full-wavelength remote sensing reflectance data of the whole river through Spatio-temporal matching technology. The experiments in some sections of XiXiao River in Hangzhou show that the correlation coefficient between remote sensing reflectance obtained by this method and the watercolor components such as chlorophyll concentration and turbidity measured synchronically is strong, the determination coefficients R² are all greater than 0.855 at the characteristic bands selected in this study. The observation zenith angle of Sentinel-2 is close to 0, close to vertical observation. In this study, use the spectral response function of Sentinel-2B to perform equivalent spectral simulation of the measured spectrum and converts it into the equivalent remote sensing reflectance of the corresponding bands. The inversion result is modeled with the sensing reflectance data after atmospheric correction based on Sen2Cor. The analysis results show that the remote sensing reflectance data after atmospheric correction based on Sen2Cor is obviously overestimated. At the same time, use of Sentinel-2B’s atmospheric apparent reflectance to deduce the radiance, then used FLAASH atmospheric correction to obtain the remote sensing reflectance of Sentinel-2B. By modeling and analyzing the equivalent remote sensing reflectance, the remote sensing reflectance data after atmospheric correction based on FLAASH is also overestimated when it is less than 0.02 sr^-1, but obviously underestimated when it is greater than 0.02 sr^-1. Research shows that the large sample measured remote sensing reflectance data obtained by this method has the application potential to verify the authenticity of satellite reflectance products.