Abstract: To tackle the tricky problem of closely tracking different vegetation types in semi-dry mining areas and study the influencing mechanism of restoration projects on the ecosystem’s structure and function, this study, taking a typical coal mine area in northern Shaanxi as a example, figured out the coverage of three things: greening vegetation, browning vegetation, and bare soil over those ten years by using Sentinel-2 (10m resolution) and Landsat-8 (30m) satellite images from the growing season (July-August) between 2015 and 2024. We used Sen’s slope estimation and Mann-Kendall trend tests to map out how each of these three changed over time and space, even comparing different restoration zones at the pixel level. We found that the area is mostly covered by browning vegetatio. The patches of lots of greening vegetation are clearly clustered in valleys with more water due to topography. In the parts of the restoration projection, the greening vegetation coverage shows increasing trend, especially in the restoration area of the starter the year of 2015. The systematic projection with "digging, refilling, and reclaiming land" set the stage for greening vegetation to grow better by reshaping the land and fixing the soil. On the other hand, projects with "hardened slopes" quickly kept the land stable, but made the restoration area harder for natural plants growing. Therefore, this study showed that the pixel trisection model works well for closely tracking vegetation changes in the messy, dry mining areas. It clearly shows how greening vs. browning vegetation shift during restoration. This gives us a way to measure and check restoration success with real data, which is super helpful for making smart decisions. In a word, these findings matter can help us make better restoration plans, tailor management strategies to different areas, and boost the mine’s ability to store carbon and provide ecosystem service.