Desert shrubs are increasingly recognized as a key component in desertification control across the drylands of China. However, their long-term dynamics and specific contributions to combating desertification remain poorly understood, primarily due to the lack of high-resolution and long-term monitoring data. To address this gap, we developed an optimized classification framework that integrates time-series change detection with time-independent classification methods. Using this approach, we produced an annual desert shrub dataset for the Ordos region, spanning from 1986 to the present. The dataset achieved an average overall accuracy of 90.03% (+/- 2.45%), demonstrating its reliability in capturing the spatiotemporal dynamics of desert shrubs. Analysis revealed a progressive land cover transition, from bare sand to sparse grass and ultimately to desert shrub, highlighting a broader trend of sandy land greening. Over the past 38 years, the desert area has declined by approximately 19,569 km(2), while the extent of desert shrubs has more than doubled, increasing from 14.06% to 31.42% of the landscape. Furthermore, 44.41% of the pixels showed a statistically significant (p < 0.05) expansion of desert shrubs within a 1 km radius. In areas exhibiting significant greening trends, more than half of the sand greening was attributed to the expansion of desert shrubs. These findings underscore the critical role of desert shrub proliferation, supported by favorable climatic conditions, in mitigating desertification. This progress is especially relevant for promoting the sustainable development of global drylands under a changing climate.