Monitoring of Open-pit Mining Disturbance and Restoration Using GEE-LandTrendr: A Case Study of Huolinhe Coalfield in Inner Mongolia, China

Increasing open-pit mining is considered to have a remarkable impact on the regional ecological environments due to drastic surface excavation and vegetation damage, therefore there is an urgent need for large-scale disturbance monitoring and restoration effect evaluation. Taking the Huolinhe Mining Coalfield as an example, this study employed the LandTrendr algorithm integrated with the Google Earth Engine (GEE) platform to detect mining disturbances and vegetation recovery using the Normalized Difference Vegetation Index (NDVI) calculated from long-term Landsat images. The results show that: (1) the extent of mining disturbance expanded before 2010 and remained broadly stable until 2017, after which it decreased following the implementation of the restoration project from 2018 onwards. (2) Breakpoints were identified in all years within the study period, and the detection rate for mining-disturbed areas exceeded 96% in the validation years of 2000, 2010, 2017, and 2021. (3) Vegetation recovery started in two periods, 2010-2012 and 2017-2019, the earlier period followed the severe drought of 2010 and is attributed to natural recovery, whereas the larger latter period coincided with the implementation of artificial restoration projects. (4) A lag of approximately one growing season was found between the implementation of reclamation engineering and the first detectable vegetation recovery, so that detection in the implementation year identified only 14.9% to 30.0% of the implemented restoration area, whereas a one year lag adjustment increased the value to 70.0% to 85.1%. (5) NDVI increased in the mining area between 2017 and 2022, and the increase was spatially concentrated in the restoration project area, which is consistent with a positive vegetation-cover response associated with the restoration projects. The results indicate that LandTrendr is an effective tool in change detection and can support rapid identification of large-scale disturbance in mining areas. The results provide a basis for subsequent monitoring of vegetation cover and for the sustainable development of a typical mining area, and they define the conditions under which annual mining-disturbance monitoring can be interpreted at face value.