作为陆地生态系统的重要组成部分，植被是促进地球不同圈层间物质循环和能量流动的关键纽带，对气候变化及人类活动高度敏感。植被变化是自然因素与人为因素综合作用的结果，因此，深入理解植被动态变化及时空分异规律，厘清植被变化的驱动因素，对区域生态保护和可持续发展至关重要。本文以黄河流域陕西段为研究区，基于归一化植被指数（Normalized Difference Vegetation Index, NDVI）、气象数据、社会经济数据等，综合利用趋势分析、偏相关分析、地理探测器和随机森林等方法和模型，分析了2000−2019年研究区NDVI时空变化特征，辨析了NDVI变化的驱动因素，预测了未来情景下NDVI分布及变化。本文主要成果如下：

（1）2000−2019年黄河流域陕西段NDVI总体以0.0086/a的速率呈极显著增加趋势，不同地域NDVI增速为：陕北（0.0109/a）>关中（0.0053/a），不同土地利用类型NDVI增速为：草地（0.011/a）>耕地（0.0081/a）>林地（0.0078/a）。NDVI呈增加趋势的面积占比为86.19%，其中显著增加的面积比例为81.90%，陕北（91.87%）>关中（66.91%），林地（97.48%）>草地（90.38%）>耕地（69.78%）。偏相关分析发现，研究区NDVI与年降水量呈正相关关系的面积占比为90.24%，NDVI与年均温呈正相关关系的面积占比为66.3%。

（2）将自然因素和人为因素进行细化并选择代表性指标，利用地理探测器进行了研究区NDVI驱动因素分析，单因子探测发现，黄河流域陕西段NDVI的主要驱动因子是降水量、耕地面积、林地面积、草地面积，四个因子对NDVI变化的解释力分别为0.73、0.66、0.63和0.38，说明降水和土地利用类型变化是研究区植被变化的主要驱动力，其中土地利用类型变化与退耕还林还草等人类活动是密不可分的；双因子交互探测显示年降水量和人口密度交互可解释82.5%的NDVI变化。生态探测结果显示q值较大的降水量、草地面积、林地面积及耕地面积对NDVI的作用机理与q值较小的其他因子均呈显著差异，进一步表明降水量和土地利用类型变化是植被变化的主要驱动因素。

（3）基于地理探测器重要性排序结果，利用随机森林模型预测研究区2030、2040和2050年不同排放浓度下的NDVI，结果表明：从空间上来看，未来30年植被生长状况整体向好。在不同分区上，低排放情景下，陕北NDVI逐渐增加，关中NDVI基本保持稳定;中排放情景下，陕北和关中NDVI在2040年都出现明显下降;高排放情景下，陕北NDVI出现缓慢下降，关中NDVI则波动上升。在不同土地利用类型上，低排放情景下，耕地、林地及草地NDVI均呈逐渐上升趋势；中排放情景下，耕地、林地及草地NDVI波动上升；高排放情景下，耕地、林地及草地NDVI均呈小幅度下降趋势。建议在加快城镇开发进程的同时，应确保耕地占补平衡，科学编制林地与草地保护、修复及利用规划方案，维护区域植被生态系统稳定。

As an important part of terrestrial ecosystems, vegetation is a key link to promote material circulation and energy flow between different spheres of the earth, and is highly sensitive to climate change and human activities. Vegetation change is the result of the combination of natural factors and human factors. Therefore, it is very important for regional ecological protection and sustainable development to deeply understand the dynamic changes of vegetation and the law of spatial and temporal differentiation, and to clarify the driving factors of vegetation change. In this paper, the Shaanxi section of the Yellow River Basin is taken as the research area. Based on NDVI, meteorological data and socio-economic data, the spatial and temporal variation characteristics of NDVI in the study area from 2000 to 2019 are analyzed by using trend analysis, partial correlation analysis, geographic detector and random forest. The driving factors of NDVI change are analyzed, and the distribution and change of NDVI in the future scenario are predicted. The main results of this paper are as follows:

(1) From 2000 to 2019, the NDVI in the Shaanxi section of the Yellow River Basin showed a significant increase trend at a rate of 0.0086/a. The growth rate of NDVI in different regions was: Northern Shaanxi (0.0109/a) > Guanzhong (0.0053/a). The growth rate of NDVI in different land use types was: grassland (0.011/a) > cultivated land (0.0081/a) > forest land (0.0078/a). The area of NDVI showed an increasing trend accounted for 86.19%, of which the proportion of significantly increased area was 81.90%, Northern Shaanxi (91.87%) > Guanzhong (66.91%), woodland (97.48%) > grassland (90.38%) > cultivated land (69.78%). Partial correlation analysis showed that 90.24% of NDVI was positively correlated with annual precipitation, and 66.3% of NDVI was positively correlated with annual average temperature in the study area.

(2) The natural factors and human factors were refined and the representative indexes were selected. The driving factors of NDVI in the study area were analyzed by using the geographical detector. The single factor detection showed that the main driving factors of NDVI in the Shaanxi section of the Yellow River Basin were precipitation, cultivated land area, forest land area and grassland area. The explanatory power of the four factors to NDVI change was 0.73,0.66,0.63 and 0.38, respectively, indicating that precipitation and land use type change were the main driving forces of vegetation change in the study area. The two-factor interactive detection showed that the interaction of annual precipitation and population density could explain 82.5% of the NDVI change. The results of ecological detection showed that the mechanism of precipitation, grassland area, woodland area and cultivated land area with large q value on NDVI was significantly different from other factors with small q value, which further indicated that precipitation and land use type change were the main driving factors of vegetation change.

(3) Based on the importance ranking results of geographical detectors, the random forest model was used to predict the NDVI of the study area under different emission concentrations in 2030,2040 and 2050. The results showed that: From the spatial point of view, the vegetation growth status in the next 30 years is generally good. In different partitions, under the low emission scenario, the NDVI in northern Shaanxi gradually increased, and the NDVI in Guanzhong remained basically stable. Under the medium emission scenario, the NDVI in northern Shaanxi and Guanzhong decreased significantly in 2040.Under the high emission scenario, the NDVI in northern Shaanxi decreased slowly, while the vegetation in Guanzhong showed a fluctuated upward trend. In different land use types, under the low emission scenario, the NDVI of cultivated land, forest land and grassland vegetation showed a gradual upward trend. Under the medium emission scenario, the NDVI of cultivated land, forest land and grassland vegetation showed a fluctuated upward trend. Under the high emission scenario, the NDVI of cultivated land, forest land and grassland showed a slight downward trend. It is suggested that while accelerating the process of urban development, the balance of cultivated land should be ensured, and the protection, restoration and utilization planning schemes of forest land and grassland should be scientifically compiled to maintain the stability of regional vegetation ecosystem.

[1]丁海勇, 丁昕玮. 基于SPOT_NDVI的甘肃省植被覆盖变化及其与气候、地形因子的关系[J]. 长江流域资源与环境, 2020, 29(12):2665-2678.

[2]秦格霞, 芦倩, 孟治元, 等. 1982-2015年中国北方草地NDVI时空动态及其对气候变化的响应[J]. 水土保持研究, 2021, 28(01):101-108+117.

[3]马新萍, 白红英, 贺映娜, 等. 基于NDVI的秦岭山地植被遥感物候及其与气温的响应关系-以陕西境内为例[J]. 地理科学, 2015, 35(12):125-130.

[4]Pouliot D, Latifovic R, Olthof I. Trends in vegetation NDVI from 1km AVHRR data over Canada for the period 1985-2006[J]. International Journal of Remote Sensing, 2009, 30(1):149-168.

[5]方精云. 我国森林植被带的生态气候学分析[J]. 生态学报, 1991, 11(4):377-387.

[6]金凯, 王飞, 韩剑桥, 等. 1982-2015年中国气候变化和人类活动对植被NDVI变化的影响[J]. 地理学报, 2020, 75(05):75-88.

[7]陈晨, 王义民, 黎云云, 等. 黄河流域1982-2015年不同气候区植被时空变化特征及其影响因素[J]. 长江科学院院报, 2022, 39(02):56-62+81.

[8]IPCC. Climate Change 2013: The Physical Science Basis. Contribution of Working Group Ⅰ to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press, 2013.

[9]张亮, 丁明军, 张华敏, 等. 1982-2015年长江流域植被覆盖度时空变化分析[J]. 自然资源学报, 2018, 33(012):2084-2097.

[10]李晶, 刘乾龙, 刘鹏宇. 1998-2018年呼伦贝尔市植被覆盖度时空变化及驱动力分析[J]. 生态学报, 2022(01):1-16.

[11]刘海, 刘凤, 郑粮. 气候变化及人类活动对黄河流域植被覆盖变化的影响[J]. 水土保持学报, 2021, 35(04):143-151.

[12]税伟, 白剑平, 简小枚, 等. 若尔盖沙化草地恢复过程中土壤特性及水源涵养功能[J].生态学报, 2017, 37(1):277-285.

[13]戚登臣, 李广宇. 黄河上游玛曲湿地退化现状、成因及保护对策[J]. 湿地科学, 2007, 5(4):341-347.

[14]褚琳, 黄翀, 刘高焕, 等. 2000-2010年黄河源玛曲高寒湿地生态格局变化[J]. 地理科学进展, 2014, 33(3):326-335.

[15]杨达, 易桂花, 张廷斌, 等. 青藏高原植被生长季NDVI时空变化与影响因素[J]. 应用生态学报, 2021, 32(04):1361-1372.

[16]伍宜丹, 马悦, 吴浩然, 等. 基于MODIS-EVI指数的四川省植被指数时空演变特征及驱动力[J]. 水土保持研究, 2020, 27(05):230-236+243+2.

[17]徐雪. 基于LAI/NPP的全球保护区植被保护效度分析[D]. 贵州师范大学, 2022.

[18]田庆久, 闵祥军. 植被指数研究进展[J]. 地球科学进展, 1998, 13(4):328-333.

[19]Tucker C J. Red and photographic infrared linear combinations for monitoring vegetation[J]. Remote Sensing of Environment, 1979, 8(2): 127-150.

[20]赵安周, 刘宪锋, 朱秀芳, 等. 2000-2014年黄土高原植被覆盖时空变化特征及其归因[J]. 中国环境科学, 2016, 36(5):290-300.

[21]徐浩杰, 杨太保. 柴达木盆地植被生长时空变化特征及其对气候要素的响应[J]. 自然资源学报, 2014, 29(3):398-409.

[22]贺振, 贺俊平. 基于SPOT-VGT的黄河流域植被覆盖时空演变[J]. 生态环境学报, 2012, 000(10):1655-1659.

[23]袁丽华, 蒋卫国, 申文明, 等. 2000-2010年黄河流域植被覆盖的时空变化[J]. 生态学报, 2013, 33(24):7798-7806.

[24]Wang XH, Piao SL, Ciais P, et al. Spring temperature change and its implication in the change of vegetation growth in North America from 1982 to 2006[J]. Proceedings of the National Acay of Sciences of the United States of America, 2011, 108: 1240-1245.

[25]Jamali S, Seaquist J, Eklundh L, et al. Automated mapping of vegetation trends with polynomials using NDVI imagery over the Sahel[J]. Remote Sensing of Environment, 2014, 141: 79-89.

[26]Pan NQ, Feng XM, Fu BJ, et al. Increasing global vegetation browning hidden in overall vegetation greening: Insights from time-varying trends[J]. Remote Sensing of Environment, 2018, 214: 59-72.

[27]Wei FL, Wang S, Fu BJ, et al. Vegetation dynamic trends and the main drivers detected using the ensemble empirical mode decomposition method in East Africa[J]. Land Degradation & Development, 2018, 1-12.

[28]Lamchin M, Wang S W, Lim C H, et al. Understanding global spatio-temporal trends and the relationship between vegetation greenness and climate factors by land cover during 1982-2014[J]. Global Ecology and Conservation, 2020, 24: e01299.

[29]Bashir B, Cao C, Naeem S, et al. Spatio-Temporal Vegetation Dynamic and Persistence under Climatic and Anthropogenic Factors[J]. Remote Sensing, 2020, 12(16): 2612.

[30]Adepoju K, Adelabu S, Fashae O. Vegetation Response to Recent Trends in Climate and Landuse Dynamics in a Typical Humid and Dry Tropical Region under Global Change[J]. Advances in Meteorology, 2019, 2019.

[31]Liu Z, Liu Y, Li Y. Anthropogenic contributions dominate trends of vegetation cover change over the farming-pastoral ecotone of northern China[J]. Ecological Indicators, 2018, 95: 370-378.

[32]张乐艺, 李霞, 冯京辉, 等. 2000-2018年黄河流域NDVI时空变化及其对气候和人类活动的双重响应[J]. 水土保持通报, 2021, 41(5):276-286.

[33]石智宇, 王雅婷, 赵清, 等. 2001-2020年中国植被净初级生产力时空变化及其驱动机制分析[J]. 生态环境学报, 2022, 31(11):2111-2123.

[34]张思源, 聂莹, 张海燕, 等. 基于地理探测器的内蒙古植被NDVI时空变化与驱动力分析[J]. 草地学报, 2020, 28(05):1460-1472.

[35]徐勇, 黄海艳, 戴强玉, 等. 西南地区陆地植被生态系统NPP时空演变及驱动力分析[J]. 环境科学, 2023, 44(05):2704-2714.

[36]孟琪, 武志涛, 杜自强, 等. 基于地理探测器的区域植被覆盖度的定量影响-以京津风沙源区为例[J]. 中国环境科学, 2021,41(02):826-836.

[37]田智慧, 任祖光, 魏海涛. 2000-2020年黄河流域植被时空演化驱动机制[J]. 环境科学, 2022, 43(02):743-751.

[38]陈安, 李景吉, 黎文婷, 等. 2001-2018年雅砻江流域植被NDVI时空动态及其对气候变化的响应[J]. 水土保持研究, 2022, 29(01):169-175+183.

[39]董镱, 尹冬勤, 李渊, 等. 黄土高原植被的时空变化及其驱动力分析研究[J]. 中国农业大学学报, 2020, 25(08):120-131.

[40]Shi Y, Jin N, Ma X, et al. Attribution of climate and human activities to vegetation change in China using machine learning techniques[J]. Agricultural and Forest Meteorology, 2020, 294: 108146.

[41]Zhou Z, Ding Y, Shi H, et al. Analysis and prediction of vegetation dynamic changes in China: Past, present and future[J]. Ecological Indicators, 2020, 117: 106642.

[42]Gao P, Du W, Lei Q, et al. NDVI Forecasting Model Based on the Combination of Time Series Decomposition and CNN-LSTM[J]. Water Resources Management, 2023: 1-17.

[43]Wu T, Feng F, Lin Q, et al. A spatio-temporal prediction of NDVI based on precipitation: an application for grazing management in the arid and semi-arid grasslands[J]. International Journal of Remote Sensing, 2020, 41(6): 2359-2373.

[44]Zhao F, Yang G, Yang H, et al. Short and medium-term prediction of winter wheat NDVI based on the DTW-LSTM combination method and MODIS time series data[J]. Remote Sensing, 2021, 13(22): 4660.

[45]Rafiei Sardooi E, Pourghasemi H R, Azareh A, et al. Comparison of statistical and machine learning approaches in land subsidence modelling[J]. Geocarto International, 2022, 37(21): 6165-6185.

[46]孙高鹏, 刘宪锋, 王小红, 等. 2001-2020年黄河流域植被覆盖变化及其影响因素[J].中国沙漠, 2021, 41(04):205-212.

[47]陆大道, 孙东琪. 黄河流域的综合治理与可持续发展[J]. 地理学报, 2019, 74(12):2431-2436.

[48]陈晨, 王义民, 黎云云, 等. 黄河流域1982-2015年不同气候区植被时空变化特征及其影响因素[J]. 长江科学院院报, 2022, 39(02):56-62+81.

[49]张君, 延军平. 1982-2013年陕西不同植被类型NDVI变化特征分析[J]. 干旱区资源与环境, 2017, 31(04):86-92.

[50]郑勇, 杨武年, 刘冲, 等. 川西高原近20a植被覆盖变化遥感动态监测及驱动力分析[J]. 遥感技术与应用, 2020, 35(06):1447-1456.

[51]卫宇婷. 山西省1986-2015年植被覆盖变化分析及其对气候因子的响应[D]. 山西农业大学, 2018.

[52]杜加强, 贾尔恒·阿哈提, 赵晨曦, 等. 1982-2012年新疆植被NDVI的动态变化及其对气候变化和人类活动的响应[J]. 应用生态学报, 2015, 26(12):3567-3578.

[53]Qi XZ, Jia JH, Liu HY, et al. Relative importance of climate change and human activities for vegetation changes on China's silk road economic belt over multiple timescales[J]. Catena, 2019, 2019(180): 224-237.

[54]Jiang LL, Guli·Jiapaer, Bao A, et al. Vegetation dynamics and responses to climate change and human activities in Central Asia[J]. Science of The Total Environment, 2017, 599-600: 967-980.

[55]黄森旺, 李晓松, 吴炳方, 等. 近25年三北防护林工程区土地退化及驱动力分析[J].地理学报, 2012, 67(5):589-598.

[56]杨钰杰, 白晓永, 谭秋, 等. 1982-2015年“一带一路”地区NDVI时空演变规律及其影响因素[J]. 山地学报, 2020, 38(02):252-264.

[57]史尚渝. 黄土高原气象干旱时空变化及其对植被的影响[D]. 中国科学院大学(中国科学院教育部水土保持与生态环境研究中心), 2019.

[58]王劲峰, 徐成东. 地理探测器:原理与展望[J]. 地理学报, 2017, 72(01):116-134.

[59]林滢, 邵怀勇. 基于随机森林算法的河南省冬小麦产量预测最佳时间窗和影响因子研究[J]. 麦类作物学报, 2020, 40(07):874-880.

[60]李玲, 张福平, 冯起, 等. 环青海湖地区草地对气候变化和人类活动的响应[J]. 生态学杂志, 2019, 38(04):1157-1165.

[61]李元春, 侯蒙京, 葛静, 等. 甘南和川西北地区草地植被NDVI变化及其驱动因素研究[J]. 草地学报, 2020, 28(06):1690-1701.

[62]宋伟宏, 王莉娜, 张金龙. 甘肃祁连山自然保护区草地时空变化及其对气候的响应[J]. 草业科学, 2019, 36(09):2233-2249.

[63]杜富林, 宋良媛, 赵婷. 草原生态补奖政策实施满意度差异的比较研究以锡林郭勒盟和阿拉善盟为例[J]. 干旱区资源与环境, 2020, 34(08):80-87.

[64]韩炳宏, 周秉荣, 颜玉倩, 等. 2000-2018年间青藏高原植被覆盖变化及其与气候因素的关系分析[J]. 草地学报, 2019, 27(06):1651-1658.

[65]Zhu Z, Bi J, Pan Y, et al. Global data sets of vegetation leaf area index (LAI) 3g and fraction of photosynthetically active radiation(FPAR) 3g derived from global inventory modeling and mapping studies (GIMMS) normalized difference vegetation index (NDVI3g) for the period 1981 to 2011[J]. Remote sensing, 2013, 5(2): 927-948.

[66]张静, 杜加强, 盛芝露, 等. 1982-2015年黄河流域植被NDVI时空变化及影响因素分析[J]. 生态环境学报, 2021, 30(05):929-937.

[67]Piao S, Yin G, Tan J, et al. Detection and attribution of vegetation greening trend in China over the last 30 years[J]. Global change biology, 2015, 21(4): 1601-1609.

[68]Zhao A Z, Zhang A B, Liu X F, et al. Spatiotemporal changes of normalized difference vegetation index (NDVI) and response to climate extremes and ecological restoration in the Loess Plateau, China. Theoretical and Applied Climatology[J], 2018, 132: 555-567.

[69]Geng G, Yang R, Liu L. Downscaled solar-induced chlorophyll fluorescence has great potential for monitoring the response of vegetation to drought in the Yellow River Basin, China: Insights from an extreme event[J]. Ecological Indicators, 2022, 138: 108801.