[Objective] Differences in above ground biomass (AGB) estimation based on remote sensing images with different spatial resolutions were analyzed, and a basis for the selection of different spatial resolution images for remote sensing estimation of AGB in desert ecosystems was provided.[Methods] Based on a ground-based AGB survey, the AGB-MSAVI statistical model was established by combining Landsat 8 and Sentinel-2 images to estimate AGB in a feldspathic sandstone area by remote sensing, and to analyze the differences between the two estimates in different vegetation coverage areas (high, medium, and low).[Results] Both Landsat 8 and Sentinel-2 images could estimate AGB well, and the spatial distributions of AGB estimation results were consistent. The mean relative errors for the AGB estimation models based on Landsat 8 and Sentinel-2 data were 13.41% and 11.42%, respectively, and the accuracy of AGB estimation based on Sentinel-2 data was relatively higher. There were some differences in AGB estimated by Sentinel-2 and Landsat 8 data in different vegetation coverage areas.[Conclusion] In the low and high vegetation coverage areas, the differences between AGB estimated by Sentinel-2 and Landsat 8 data were relatively small. In contrast, in the medium vegetation coverage area, the spatial heterogeneity was relatively significant, the remote sensing data were constrained by the spatial resolution, and the differences between AGB estimated by the two images were relatively large. The high spatial resolution remote sensing images were effective for improving AGB estimation accuracy.