利用黄土高原1982年-2006年GIMMS—NDVI数据,以投影寻踪分类评价法、标准差、线性趋势分析和Hurst指数为基础,分析了黄土高原NDVI的变化趋势,并且预测了该研究区植被覆盖变化总体趋势及其空间分异。分析结果表明:①近25年来黄土高原植被覆盖变化波动程度较高,但植被覆盖是向着改善的方向发展的。2002年以后NDVI波动程度总体上在减小,意味着近年来植被覆盖整体情况趋于稳定改善的状态;②近10年黄土高原的Hurst指数的均值为0.7195,数值较高,但存在明显差异,东部的Hurst指数高,西部Hurst指数相对较低;黄土高原地区植被覆盖时间序列主要表现为可持续性,即持续改善或者持续退化。在现有基础上,未来的地表植被覆盖改善区有向北部扩展的趋势;退化区有向南部扩展的趋势;③各种植被类型的Hurst指数除城市外平均值均为0.5〈H〈1,即为可持续性序列。其中,林地可持续性最强,除城市外的其他土地利用次之,草地最弱。
Using 15-day NDVI products of Global Inventory Modeling and Mapping Studies (GIMMS) VGT during the period 1982-2006 and projection pursuit clustering (PPC), standard deviation, linear trend analysis, and the Hurst index, NDVI trends across the Loess Plateau were examined. On the basis of this, general trends in vegetation cover change and its spatial variation in the future were predicted. Results show that 1) over the past 25 years, changes in vegetation cover varied significantly, so it seems to be difficult to draw a definite conclusion about if the coverage of vegetation was improved or degraded. The PPC method can be used to perform evaluation of time series of variables. It was found that vegetation cover over the study region was gradually improved. After 2002, overall, variation in NDVI was found to progressively decrease, which means that vegetation cover improved generally in recent years. It was also found that the PPC method highlights the trend in dependent variables, so it seems to be more suitable for deriving variations of variables. In addition, standard deviation and linear trend analysis were jointly conducted so as to eliminate the inadequacies of the two methods in trend analysis, which can not only reflect the magnitude of changes, but also a change in direction; 2) The Hurst index of NDVI time series in the Loess Plateau over the past 10 years was generally higher, and was significantly different in space. Specifically, it was relatively higher in the east Loess Plateau, but was relatively lower in the west Loess Plateau. It is expected that improvement in vegetation cover will expand to the north Loess Plateau, whereas the degraded areas will likely expand to the south Loess Plateau. 3) The Hurst index of various vegetation types except for the city varied between 0.5 and 1 on average, i.e., a persistent series. Forest land had the strongest persistency; grassland had the weakest persistency. The regional differences resulting from a particular environment and varying managem