旱区植被对人类活动和气候变化影响敏感且脆弱，主导着全球碳循环年际变异。中国是旱区面积最大的国家之一，旱区占国土面积的56.48%，区域内人口约为5.8亿。1970年以来，中国政府在旱区开展了大规模的生态修复以应对荒漠化和生态退化。人类活动与气候变化的共同作用使得中国旱区植被变化过程更为复杂。因此，亟需量化分析其植被动态特征，明确变化过程和机制，为旱区生态修复和管理提供科学依据。

本研究从植被绿度年际尺度动态变化和月尺度植被生产力对干旱的恢复力入手，开展中国旱区植被动态特征和影响因素分析研究。基于GIMMS3.1 NDVI，ERA5 气候数据，MODIS GPP，GLDAS和TerraClimate气候数据，以及ESA-CCI-LC土地利用类型数据，综合使用BFAST断点检测、主成分回归（PCR）、残差趋势分析（RESTREND）方法，基于RStudio、谷歌地球引擎（Google Earth Engine）和ArcGIS等软件平台，从年尺度和月尺度计算和分析了中国旱区植被动态特征及影响因素。得到的主要结果如下：

（1）1982-2015年间，中国旱区植被绿度显著增加（41%的区域显著增加，整体速率为+0.60×10-3/年），其中黄土高原（+1.70×10-3/年）和三北地区（+0.60×10-3/年）植被绿度的增加最为显著。除稀疏植被和建设用地无显著增加外，其他土地利用类型的绿度均显著增加（增幅为2.63-6.6%）。旱区12.85%（约62.18万公顷）的区域植被绿度主要在1997-1999和2005年前后发生趋势转变，趋势转变类型以绿化速率增加为主，转折点前的植被绿度增加趋势约为+0.21×10-3/年，转折点后为+0.42×10-3/年。

（2）人为影响（+0.12×10-3/年）和二氧化碳施肥效应（+0.55×10-3/年）共同主导了中国旱区的植被绿度增加。黄土高原（+1.01×10-3/年）和三北地区（+0.23×10-3/年）的人为影响尤其显著。中国旱区6.73%（约31.64万公顷）区域上的绿度趋势转变可直接归因于2000年前后人为活动的变化。当人为活动造成的正影响加强时，植被动态从无变化转变为显著变绿，当人为影响减小时，植被动态由显著绿化转化为无显著变化，这类阶段性的绿化使整体时段表现为绿度增加，表明2000年以来的生态恢复显著加强了1980年代以来的中国旱区植被绿度增加。

（3）中国旱区月尺度植被生产力主要受土壤水分限制。2001-2019年间，土壤干旱造成了植被生产力的损失约为6.9gC/m2，其中约1/4的GPP距平在干旱事件结束后依然为负，平均恢复时长为1.69个月（区间为1-6.14个月）。空间上，中国旱区中东部和西北部等地区具有较低的恢复力，东北部（内蒙古东北部和黑龙江西部）则具有较高的恢复力。统计分析发现干旱引起的GPP损失越大时，干旱结束后GPP损失往往不能立即恢复，但平均所需恢复时长越短。不同土地利用类型中，建设用地、土地利用转变区和裸地恢复力较高；耕地、草地、稀疏植被和镶嵌植被恢复力较低。林地GPP在干旱期间很少出现损失，但在延迟恢复的情况下，则需要更长的恢复时长；不同区域中，黄土高原土地利用转变区和草地的恢复力较低。

研究结果明确了中国旱区植被变化的时空格局，揭示了中国旱区的植被绿度变化动态和趋势转变的主导因素，量化了不同区域和植被类型在干旱下的抵抗和恢复能力。研究加深了对中国旱区植被的动态变化的理解，为区域生态修复和干旱条件下的环境管理提供依据。

Dryland vegetation is sensitive and vulnerable to human activities and climate change, and its carbon source sink dynamics dominate the interannual variability of global carbon source sinks. China is one of the countries with the largest dryland areas, accounting for 56.48% of the national territory and a population of about 580 million, and since 1970, the Chinese government has led large-scale ecological restoration in drylands to combat desertification and ecological degradation. The combined effects of human activities and climate change have made the vegetation change process in China's drylands more complex. Therefore, there is an urgent need to quantify and analyze the dynamic characteristics of its vegetation, clarify the change processes and mechanisms, and provide scientific basis for ecological restoration and management.

In this study, the interannual-scale dynamics of vegetation greenness and monthly-scale resilience of vegetation productivity to drought were used as an introduction to carry out studies related to the characteristics of vegetation dynamics and analysis of its influencing factors in China’s drylands. Multi-source remote sensing data: GIMMS3.1 NDVI, ERA5 climate data, MODIS GPP, GLDAS and TerraClimate climate data, and Climate Change Initiative (CCI) land use type data were used, and a combination of BFAST breakpoint detection, principal component regression (PCR), and residual trend analysis (RESTREND) methods, based on RStudio, Google Earth Engine (GEE) and ArcGis software platforms to calculate and analyze the characteristics of vegetation dynamics and influencing factors in China’s drylands at annual and monthly scales. The main results obtained are as follows:

(1) The vegetation greenness in China’s drylands increased significantly from 1982 to 2015 (41% of the area increased significantly with an overall rate of +0.60×10-3/year), with the most significant increase in vegetation greenness in the Loess Plateau (+1.70×10-3/year) and the Three Norths (+0.60×10-3/year). Except for sparse vegetation and built-up land, which did not increase significantly, the greenness of all land use types increased significantly by 2.63-6.6%. 12.85% (about 621,800 ha) of China’s drylands vegetation greenness mainly underwent a trend shift around 1997-1999 and 2005, and the type of trend shift was dominated by an increase in greenness rate, with a trend of about +0.21×10-3/year of vegetation greenness increase before the turning point and +0.42×10-3/year after the turning point.

(2) During 1982-2015, anthropogenic impacts (+0.12×10-3/yr) and CO2 fertilization effects (+0.55×10-3/yr) jointly dominated vegetation greening in China’s drylands. Anthropogenic impacts were particularly significant in the Loess Plateau (+1.01×10-3/yr) and in the Three Northern Areas (+0.23×10-3/yr). The shift in greenness trend over 6.73% (about 316,400 ha) of the China dry zone can be directly attributed to the changes in anthropogenic activities around 2000. When positive impacts from anthropogenic activities intensified, vegetation dynamics shifted from no change to significant greening, and when anthropogenic impacts decreased, vegetation dynamics shifted from significant greening to no significant change. These types of phases of greening resulted in an overall increase in greenness over time, suggesting that ecological restoration since 2000 has significantly enhanced the greening of vegetation in China's drylands since the 1980s.

(3) Monthly-scale vegetation productivity in China’s drylands is limited by soil moisture; soil drought caused a loss of vegetation productivity of about 6.9 gC/m2 during 2000-2019, of which about a quarter of the GPP anomaly remained negative after the end of the drought event, with an average recovery time of 1.69 months (interval 1-6.14 months). Spatially, areas such as the central-eastern and north-western parts of China’s drylands have low recovery, while the north-eastern part (northeastern Inner Mongolia and western Heilongjiang) has high recovery. Statistical analysis found that the greater the drought-induced GPP loss, the less the GPP loss tended to recover immediately after the end of the drought, but the shorter the average recovery time required. Among the different land use types, the resilience of built-up land, land use conversion areas and bare land is higher; the resilience of arable land, grassland, sparse vegetation, and mosaic vegetation is lower. Forest land GPP is rarely lost during drought, but requires longer recovery time in case of delayed recovery; among the different regions, the resilience of the Loess Plateau land-use conversion area and grassland is lower.

The study clarifies the spatial and temporal patterns of vegetation change in China's drylands, and reveals the dominant factors in vegetation greening dynamics and trend shifts in China's drylands. By analysing the drought response characteristics of vegetation, the resistance and recovery capacity of different regions and vegetation types under drought were quantified. The study deepens the understanding of the dynamics of vegetation in drylands and provides a basis for regional ecological restoration and environmental management under drought conditions.