基于2000-2015年的NDVI、NPP和GPP多源数据,并结合人类干扰强度等级的构建,采用线性回归分析和地理加权回归分析方法,对比分析了在不同干扰等级下三江源区生态工程实施前5年和后10年草地的动态变化及其驱动因素。结果表明:(1)人类干扰强度和NDVI、NPP、GPP在空间上具有显著的区域差异,均表现为从西北向东南增加的趋势,且2006-2010年和2011-2015年草地NDVI、NPP和GPP值明显高于2000-2004年;(2)与2000-2004年相比,2006-2010年草地NDVI、NPP和GPP值的草地退化面积均减少,而2011-2015年退化面积又均增加,表明生态工程建设后的5年较前5年草地退化状况得到改善,然而在下一个5年草地又出现退化趋势;(3)在各人类干扰等级下生态工程实施后的5年内70%以上的草地处于改善状态,均出现退化面积比例显著减少,改善面积比例显著增加的趋势,而在2011-2015后5年草地退化面积比例增加,表明生态工程的实施促进了草地生态系统的改善,尤其在重度人类干扰区,草地由2000-2004年的52.71%退化区域到2006-2010年84.95%的改善区域。(4)影响草地动态变化的主要因素为距离道路的距离、距离居民点的距离、人口密度、干旱、温度和坡度等,各影响因素分布具有明显的空间异质性。研究结果将有助于为三江源区草地生态系统恢复、管理和战略制定提供科学指导和决策依据。;Using Normalized Difference Vegetation Index (NDVI), Net Primary Productivity (NPP) and Gross Primary Productivity (GPP) data from 2000 to 2015 and combining with of human interference level, linear regression analysis and geographically weighted regression (GWR) analysis were used to compare and analyze the grassland dynamics and their driving factors in the Three-River headwaters region before and after the implementation of ecological construction projects. Results showed that:(1) spatially, human interference and NDVI, NPP and GPP had significantly regional differences, which all showed increasing trends from northwest to southeast, and the grassland NDVI, NPP and GPP values during 2006-2010 and 2011-2015 were significantly larger than that during 2000-2004. (2) Compared with 2000-2004, the NDVI, NPP and GPP values of grassland have been improved during 2006-2010 after ecological construction project, while NDVI, NPP and GPP values showed degrading tendencies again during 2011-2015. (3) In all human interference levels, 70% of the grassland was improved within 5 years after the implementation of ecological projects, and the area ratio of degraded grassland significantly decreased and that of the improved grassland showed significant increasing trend, indicating that the implementation of ecological projects has contributed to the improvement of grassland ecosystems. Especially in areas of severe human disturbance, grassland changed from 52.71% degraded areas in 2000-2004 to 84.95% improved areas in 2006-2010. (4) The main driving factors affecting the dynamic changes of grassland were distance to roads, distance to settlements, population density, drought, temperature and topographic slope, and the distribution of each influencing factor had obvious spatial heterogeneity. The study will help to provide scientific guidance and decision-making basis for grassland ecosystem restoration, management and strategy formulation in the Three River Source Region.