[Objective] Extraction methods of spectral information of inland surface water in the Yellow River basin were elucidated and compared to provide guidance for the extraction of spectral information of water bodies with large sediment concentration in the Yellow River basin. [Methods] Two methods that were thought effective at present were chose to extract the spectral information of water body. They were modified normalized difference water index(MNDWI) and linear spectral mixture model. Landsat 8 OLI imagery of reservoir, wetland, lake and river in the Yellow River basin was exemplified to analyze the accuracies of the two methods, and to discuss the regional applicability. In which, the study area was divided into two categories: water and non-water, and high resolution imagery was referred.[Results] The accuracy of linear spectral mixture model in extracting spectral information of reservoir, wetland and lake was higher than that of the MNDWI. The two methods performed better in the large-area water bodies, such as lakes and reservoirs, than in the linear like body as rivers. [Conclusion] In high resolution image, mixed pixels were also existed. Based on that, the linear spectral mixture model had covered the effect of mixed pixels on the spectral information extraction from water bodies, whereby it remarkably improved the extraction precision. The linear spectral mixture model is superior to the modified normalized difference water index.