【摘要】 NDVI是监测地表植被变化的重要工具，但不同来源的NDVI数据集在不同尺度下的应用存在较大差异，需要对其进行一致性评估。本研究选取了SPOT、MODIS、GIMMS以及CDR AVHRR等四套NDVI数据集，采用相关分析和趋势分析的方法，评估了2000—2015年间四套数据集在三江源地区年际变化的一致性和差异性，结果表明：(1)从NDVI平均值的年际变化趋势来看，SOPT NDVI和MODIS NDVI均呈现出增加的趋势，且两者之间具有最强的相关性（R=0.72,p<0.001），而GIMMS NDVI和CDR AVHRR NDVI则呈现出减少的趋势。(2)从栅格尺度的NDVI年际变化趋势来看，SPOT NDVI、MODIS NDVI以及GIMMS NDVI均呈现出整体变绿的趋势，变绿像元占比分别为86.94%、69.52%以及66.95%，而CDR AVHRR NDVI数据则呈现出整体变褐的趋势，变绿像元占比仅为33.38%。(3)从不同海拔梯度下四套NDVI数据集的差异来看，四套数据集中平均NDVI随海拔增加的变化趋势相同，但GIMMS的平均NDVI值要显著大于其他三套数据集。从NDVI年际变化速率来看，SPOT数据显著高于其他三套数据集，虽然MODIS和GIMMS捕捉到的变化速率相似，但依然呈现出MODIS NDVI>GIMMS NDVI的特征。本研究结果显示，2000—2015年间，三江源区植被表现出整体变绿的趋势，但变绿的范围和幅度在不同数据集之间却存在显著差异，在实际使用中应谨慎选择，而CDR AVHRR NDVI数据至少在三江源地区可用性存疑。更多还原

【Abstract】 NDVI is an essential tool for monitoring changes in surface vegetation, but the applications of NDVI datasets from different sources at different scales are quite different, and consistency assessment is required. In this study, four NDVI datasets including SPOT, MODIS,GIMMS and AVHRR were selected, and the methods of correlation analysis and trend analysis were used to evaluate the consistency and difference of the interannual changes of the four datasets in the Three River Source Region(TRSR) from 2000 to 2015, the results show that:(1) From the change trend of NDVI average value, both SOPT data and MODIS data show an increasing trend, and there is the strongest correlation between them(R=0.72, p<0.001),however, GIMMS and AVHRR show a decreasing trend.(2) From the change trend of pixel scale, SPOT, MODIS and GIMMS data all show an overall trend of greening, and the proportion of green pixels is 86.94%, 69.52% and 66.95%, while the AVHRR data shows an overall trend of browning, and the proportion of green pixels is only 33.38%.(3) From the difference between the four NDVI datasets under different altitude gradients, the variation trend of the centrally averaged NDVI with increasing altitude is the same, but the average NDVI value of GIMMS is significantly larger than that of the other three datasets. However,from the perspective of the change range of NDVI, the change range captured by SPOT is significantly higher than that of the other three datasets. Although the change range captured by MODIS and GIMMS is similar, it still shows the characteristics of MODIS>GIMMS. The results of this study suggest that from 2000 to 2015, the vegetation in the TRSR presents an overall greening trend, but the range and magnitude of greening are significantly different in different datasets. Data availability of AVHRR at least in the TRSR is questionable.更多还原