Abstract: To evaluate the spectral fidelity of urban water body on the remote sensing images after fusion, GF1 and GF2 images were used as the data sources. Four classical fusion methods such as Brovey transform, Gram-schmidt pan sharpening, NNDiffuse pan sharpening and PCA transform, were applied to fuse the multispectral and panchromatic images of GF1 and GF2. Based on the urban water samples collected from the images before and after image fusion, the mean value, root mean square error (RMSE) and mean relative error (MRE) were used to analyze the wave band deviation and waveform fidelity of urban water body before and after image fusion. Results show that the NNDiffuse Pan Sharpening method is the best for both the GF1 and GF2 images on wave band deviation. The NNDiffuse Pan Sharpening method is best for the GF1 images and the Gram-schmidt Pan Sharpening method is best for the GF2 images on waveform fidelity. Overall, the spectral fidelity of fused urban water is better and quantitative research of water quality parameters can further carried out.