矿区生态扰动与恢复监测能够指导人类科学管理煤炭开采活动。当前基于矿区生态问题开展的单一生态环境评价及局部矿区生态扰动与恢复监测对满足区域与生态保护的协调发展具有局限性，因此本研究以黄河流域煤炭区为例，全面考量生态系统格局（Comprehensive Landscape Index，CLI）、生态系统质量评估指数（Remote Sensing Ecology Index，RSEI）和生态系统服务功能评估指数（Comprehensive Ecosystem Service，CES），对三种指标赋权计算1990~2022年长时序生态系统综合评估指数（Comprehensive Ecosystem Assessment Index，CEA），以CEA指数表征黄河流域煤炭区的生态综合状况，进而探究黄河流域煤炭区生态的扰动与恢复，最后定量分析人类活动对生态扰动与恢复的影响。研究结论如下：

      （1）计算黄河流域煤炭区生态系统格局（CLI）、质量（RSEI）、服务（CES），并综合这三项指标计算生态系统综合评估指数（CEA）。结果表明在1990~2022年期间：①中上游蒙陕区CLI均值上升，中游黄陇区CLI均值下降，其余煤炭区CLI均值变化相对稳定，整体呈现中游CLI水平较低，上游和中下游CLI水平较高的分布格局；②各煤炭区RSEI均值均呈现上升趋势，空间上表现为中上游RSEI水平较低，而上游和中下游RSEI水平较高的分布格局；③上游青海区、中游黄陇区CES均值波动上升，中游山西区和中上游蒙陕区CES均值变化较为稳定，而上游宁夏区和中下游豫鲁区CES均值下降，空间上表现为中上游和中下游CES水平较低，上游及中游CES水平较高的分布格局；④中下游豫鲁区CEA指数均值下降，其余各煤炭区CEA指数均值均呈现上升趋势，各煤炭区CEA指数分布呈现较强的空间分异性，即上游CEA水平最高，中下游次之，中上游最低的分布格局，时空演变以显著改善为主。

     （2）利用BFAST（Breaks For Additive Seasonal and Trend，BFAST）算法与LandTrendr算法对黄河流域煤炭区生态进行扰动与恢复监测，结果表明：①BFAST算法监测中上游蒙陕矿区首次突变时间与神东矿区采矿时间较为一致且范围较广，可知BFAST算法对生态的监测较为敏感；②BFAST算法监测的中上游蒙陕区、上游宁夏区、上游青海区、中下游豫鲁区、中游山西区及中游黄陇区发生首次突变的时间分别为2000年、1993年、2006年、2005年、2000年及1997年，最大强度突变时间分别为2008年、2002年、2006年、2005年、2003年及2005年，发生的突变次数分别为2次、3次、2次、2次、2次及2次。

     （3）采用改进的BFAST01模型进一步监测分析黄河流域煤炭区扰动与恢复情况，结果表明：①BFAST01模型共监测出8种扰动与恢复变化趋势类型：中上游蒙陕区以“反转（由增到减）”扰动（2006~2011年）和“单调型增加（正中断）”恢复（1998~2002年）为主；上游青海区以“中断（随着正中断减少）”扰动（1993~2000年）和“中断（随着负中断增加）”恢复（2006~2012年）为主；中游山西区以“中断（随着正中断减少）”扰动（1999~2001年）和“中断（随着负中断增加）”（2006~2012年）为主；上游宁夏区以“中断（随着正中断减少）”扰动（1997~2000年）和“中断（随着负中断增加）”恢复（2006~2012年）为主；中游黄陇区以“中断（随着正中断减少）”扰动（1995~2003年）和“反转（由减到增）”恢复（2008~2011年）为主；中下游豫鲁区以“单调型减少（负中断）”扰动（1994~1998年）和“反转（由减到增）”恢复（2005~2014）；②除上游青海区和上游宁夏区的生态变化以扰动为主，其余煤炭区的生态均以恢复为主，变化前后主要表现为显著，针对扰动区域可进行生态修复措施进行改善，恢复区域可采用开采与修复并行的方式协调发展；③典型平朔露天矿扰动与恢复的空间分异性较为明显，矿区逐步实施土地复垦促使西南部生态以“反转（由减到增）”恢复为主；排土场向东部扩张导致东北部生态以“反转（由增到减）”扰动为主；持续的采矿活动也引起布尔台井工矿生态以“反转（由增到减）”扰动为主；④选择代表性像元的扰动与恢复变化类型与谷歌影像实际变化进行对比验证可知，真实影像所表现的扰动与恢复状况，与BFAST01监测出的扰动与恢复特征相吻合，即用BFAST01模型进行长时间序列的生态扰动与恢复监测是适用的，可靠性较强。

     （4）采用随机森林建模预测1999-2022年自然因素条件下的CEA' ，结合残差定量区分自然因素和人类活动对生态恢复与扰动的影响。结果表明：①随机森林建模的可决系数R²为0.899，均方根误差RMSE为0.053，精度较高；②1990-2022年自然因素影响下CEA' 呈现上游水平最高，中下游次之，中上游最低的分布格局，时空演变以稳定不变为主；③1990-2022年人类活动影响下，各煤炭区残差均值呈上升趋势，在退耕还林（草）与矿区生态修复措施的共同作用下，生态逐渐向好发展，黄河流域煤炭区生态以显著改善为主，这也与BFAST01监测出的煤炭区以恢复为主的结果较为一致。

        Ecological disturbance and restoration monitoring in mining areas can guide human scientific management of coal mining activities. The current single ecological environment assessment and local ecological disturbance and restoration monitoring based on ecological issues in mining areas have limitations in meeting the coordinated development of regional and ecological protection. Therefore, this study takes the coal mining area in the Yellow River Basin as an example, comprehensively considers the Comprehensive Landscape Index (CLI), Remote Sensing Ecology Index (RSEI), and Comprehensive Ecosystem Service (CES), and calculates the Comprehensive Ecosystem Assessment Index (CEA) from 1990 to 2022 by weighting the three indicators. Using CEA index to characterize the ecological comprehensive status of the coal mining area in the Yellow River Basin, and based on this, exploring the ecological disturbance and restoration of the coal mining area in the Yellow River Basin, Finally, quantitatively analyze the impact of human activities on ecological disturbance and restoration. The research conclusion is as follows:

       (1) Calculate the ecosystem pattern (CLI), quality (RSEI), and services (CES) of the coal mining area in the Yellow River Basin, and calculate the comprehensive ecosystem assessment index (CEA) by integrating these three indicators. The results indicate that between 1990 and 2022:①The average CLI level in the middle and upper reaches of the Inner Mongolia Shaanxi region has increased, while the average CLI level in the middle and upper reaches of the Huanglong region has decreased. The average CLI level in other coal mining areas has remained relatively stable, showing an overall distribution pattern of lower CLI level in the middle reaches and higher CLI level in the upstream and middle and lower reaches;②The average RSEI of each coal mining area shows an upward trend, with a spatial distribution pattern of lower RSEI levels in the middle and upper reaches, while higher RSEI levels in the upstream and middle and lower reaches;③The average CES levels in the upper reaches of Qinghai and Huanglong regions fluctuate and increase, while the average CES levels in the middle reaches of Shanxi and Inner Mongolia Shaanxi regions remain relatively stable. In contrast, the average CES levels in the upper reaches of Ningxia and the middle and lower reaches of Yulu regions decrease, showing a spatial distribution pattern of lower CES levels in the middle and upper reaches, middle and lower reaches, and higher CES levels in the upper and middle reaches;④The average CEA index in the middle and lower reaches of Yulu District has decreased, while the average CEA index in other coal areas has shown an upward trend. The distribution of CEA indices in each coal area shows strong spatial differentiation, with the highest CEA level in the upstream, followed by the middle and lower reaches, and the lowest in the middle and upper reaches. The spatiotemporal evolution is mainly characterized by significant improvement.

       (2) Using BFAST algorithm and LandTrender algorithm to monitor ecological disturbance and restoration in the coal mining area of the Yellow River Basin, the results show that:①The first mutation time of Mengshan mining area in the middle and upper reaches of the Yangtze River is consistent with the mining time of Shendong mining area;②The BFAST algorithm monitors the first mutation occurring in the middle and upper reaches of Inner Mongolia and Shaanxi, Ningxia, Qinghai, Yulu, Shanxi, and Huanglong regions in 2000, 1993, 2006, 2005, 2000, and 1997, respectively. The maximum intensity mutation occurred in 2008, 2002, 2006, 2005, 2003, and 2005, with 2, 3, 2, 2, 2, 2, and 2 mutations, respectively.

       (3) The improved BFAST01 model is used to monitor and analyze the disturbance and restoration of coal area in the Yellow River basin. The results indicate that: ①The BFAST01 model monitored a total of 8 types of disturbance and recovery trends (Table 4.6): the Inner Mongolia and Shaanxi regions in the middle and upper reaches were mainly characterized by "reversal (from increase to decrease)" disturbances (2006-2011) and "monotonic increase (positive interruption)" recovery (1998-2002); The upstream Qinghai region is mainly affected by "interruptions (decreasing with positive interruptions)" disturbance (1993-2000) and "interruptions (increasing with negative interruptions)" recovery (2006-2012); The middle reaches of Shanxi region are mainly characterized by interruptions (decreasing with positive interruptions) from 1999 to 2001 and interruptions (increasing with negative interruptions) from 2006 to 2012; The upstream Ningxia region is mainly affected by "interruptions (decreasing with positive interruptions)" disturbance (1997-2000) and "interruptions (increasing with negative interruptions)" recovery (2006-2012); The middle reaches of Huanglong District are mainly characterized by "interruption (decreasing with positive interruption)" disturbance (1995-2003) and "reversal (decreasing to increasing)" recovery (2008-2011); The middle and lower reaches of the Yulu region experienced a "monotonic decrease (negative interruption)" disturbance (1994-1998) and a "reversal (from decrease to increase)" recovery (2005-2014);②With the exception of Qinghai and Ningxia, the ecological changes in the upper reaches are mainly disturbed, while those in the rest of the coal-mining regions are mainly restored, the restoration area can be developed in a coordinated manner by mining and repairing in parallel;③The spatial differentiation of disturbance and restoration in typical Pingshuo open-pit mines is more obvious. The gradual implementation of land reclamation in the mining area has led to a "reversal (from decrease to increase)" restoration of the ecology in the southwest; The expansion of the dumping ground to the east has led to a "reversal (from increase to decrease)" disturbance in the ecology of the northeast; The continuous mining activities have also caused a "reversal (from increase to decrease)" disturbance in the ecology of the Boltai mine;④By comparing the types of disturbance and restoration changes of representative pixels with the actual changes in Google Images, it can be concluded that the disturbance and restoration status of the real images are consistent with the disturbance and restoration characteristics monitored by BFAST01. Therefore, using the BFAST01 model for monitoring the disturbance and restoration of long time series is applicable and has strong reliability.

      (4) Using random forest modeling to predict CEA'  under natural factor conditions from 1999 to 2022, and combining residual quantification to distinguish the impact of natural factors and human activities on ecological restoration and disturbance. The results indicate that:①The Determinable Coefficient R² of Random Forest Modeling is 0.899, with a root mean square error RMSE of 0.053, indicating high accuracy;②Under the influence of natural factors from 1990 to 2022, CEA'  showed a distribution pattern with the highest level in the upstream, followed by the middle and downstream, and the lowest in the middle and upstream. The spatiotemporal evolution was mainly stable and unchanged;③Under the influence of human activities from 1990 to 2022, the average residual values in various coal mining areas showed an upward trend. With the joint action of returning farmland to forests (grasslands) and ecological restoration measures in mining areas, the ecology gradually improved. The ecological improvement in the coal mining areas of the Yellow River Basin was mainly significant, which is consistent with the results of BFAST01 monitoring that the coal mining areas mainly focus on restoration.