Coal mining in northern China's grasslands faces significant environmental challenges due to the fragile semi-arid ecosystem and the abundance of mineral resources. However, current research lacks clarity in both theoretical understanding and effective quantification of its impacts. Insufficient distinction between direct and indirect environmental effects, the oversight of interactions from neighboring activities, and inaccurate measurements of impact intensity all hinder a comprehensive understanding. These limitations also prevent the precise delineation of ecological protection compensation areas. In this study, we developed a comprehensive framework utilizing Landsat time-series images (1989–2017) and field sampling data (228 sample locations in 2017) to assess the boundaries and intensities of mining impacts. First, the k-means clustering algorithm was applied to identify the Normalized Difference Vegetation Index (NDVI) time-series trajectories and map the spatial information of direct impacts. Then, a buffer zone was established and an amplitude model was proposed for determined the cross-impacts and their intensities based on the volatility of NDVI trajectories. Accordingly, a Trend-thresholds model was proposed to automatically extract the indirect impacts boundary of mining-dominated on vegetation under varying intensities. The Soil Quality Index (SQI) ∼ distance curves were fitted and a derivative model was employed for determine the boundary of indirect impacted on soil. We applied this assess framework in Baorixile coalfield (i.e., Baorixile coal mine (BM), DongMing coal mine (DM), Shunxing coal mine (SX)) a typical semi-arid grassland mining area, and quantified three types of impacts: direct, indirect, and cross-impacts. Findings reveal negative coal mining impacts primarily confined to permitted mining areas and decreasing over time. Direct impacts boundary limited in the mining sites and its impact intensity shows a dual nature, with a negative weakening and a positive strengthening trend. Widespread negative indirect impacts exist outward the mining area displays spatial heterogeneity, and its intensity attenuation with distances. In there, the very severe effect (D1) is predominantly focusing on the grassland where near by the mining work face (BM maximum range 1.2 km, DM maximum range 0.55 km, SX maximum range 0.4 km). Moreover, our study indicated that vegetation and soil exhibited consistent responses to coal mining impacts within 1 km, demonstrating the feasibility of using remote sensing data to qualify the impact boundary on the grassland environment. Besides, we further identify significant cross-impacts surrounding the mining areas, characterized by strong negative effects, particularly in the southeastern region of BM. This research offers valuable insights for effective ecological compensation governance and scientific environmental management in similar global contexts.