选用19822013年GIMMS NDVI数据,运用变异系数法、Theil-Sen median趋势分析耦合Mann-Kendall检验以及Hurst指数法,研究了中亚干旱区植被覆盖的空间格局、不同维度的空间变异性、时间变化特征和未来趋势预测。同时,结合CRU降水、气温资料和MODIS土地覆盖数据,对植被覆盖时空格局及其变化的驱动因素进行分析。结果表明: ① 19822013年中亚干旱区植被覆盖有较强的空间异质性,有植被覆盖的区域占总面积的85.32%,无植被覆盖的区域占14.68%。受降水量控制,植被覆盖呈山区高平原低、西部高东部低、北部高南部低的特点;受河流和人工灌溉的影响,绿洲区的植被覆盖高于荒漠区。②近32 a全区植被覆盖的波动变化较明显,各变异程度的面积比例:中等波动变化>相对较高的波动变化>高波动变化>相对较低的波动变化>低波动变化。受降水变率和人工种植的影响,植被覆盖的高波动变化主要在荒漠区和诸流域绿洲区;低波动变化主要在植被生长良好的区域。③ 32 a间全区NDVI呈增长趋势,NDVI距平的变化率为0.01·(10a) ~(-1)。基于像元尺度的分析也表明,全区植被覆盖变化趋势以增加为主,各类变化趋势的面积比例:轻微增加>显著增加>轻微减小>显著减小>无法确定。植被改善是区域气候增湿增暖和绿洲土地覆盖变迁所致。④全区NDVI的Hurst指数均值为0.63,Hurst指数大于0.5的范围所占比例为75.17%,即未来全区植被覆盖的变化趋势以持续性增加为主,其中25.23%的区域未来变化趋势无法确定。
英文摘要:
Using the GIMMS NDVI data that were concentrated between 1982 to 2013, in this study the Coefficient of Variation method,coupling of Theil-Sen median trend analysis,Mann-Kendall test and Hurst index method were used to investigate the spatial pattern of vegetation coverage, spatial heterogeneity of different dimensionalities, characteristics of temporal variation,and future trends in the arid regions in Central Asia. Moreover, the precipitation and temperature data of CRU and MODIS land cover type data were used to analyze the factors that drove the spatial pattern and change of vegetation coverage. The results are as following: ① There was a strong spatial heterogeneity of vegetation coverage over the arid regions in Central Asia during the period from 1982 to 2013, the areas covered by vegetation and of bare ground accounted for 85.32% and 14.68% of the total area respectively. Under the influence of precipitation, the vegetation coverage in the mountainous regions was higher than that in the plains,and it was higher in the west region than that in the east and higher in the north area than that in the south. Affected by rivers and artificial irrigation, the hydrothermal conditions in oases were better, so the vegetation coverage in these areas was higher than that in the deserts; ② Variation of vegetation coverage in the study area over the past 32 years was obvious. The proportion of areas with different variation degree was in an order of medium variation > relatively high variation > high variation > relatively low variations > low variation. Under the influence of precipitation variability and artificial cultivation, the high variation of vegetation coverage occurred mainly in the deserts and oases, but the low one mainly in the regions where the vegetation grew well; ③ NDVI in the study area was in an increase trend in the past 32 years,and the change rate of NDVI anomaly was 0.01·(10a) ~(-1). The pixel-based analysis also indicated that the vegetation coverage was dominated by an increase trend,and the area proportion of the variations was in an order of slight increase > significant increase > slight decrease > significant decrease > unidentification. The improvement of vegetation coverage was driven by the increase of both precipitation and temperature, especially the change of land cover types in the oases; ④ The Hurst index average of the study area was 0.63,and the proportion of the areas with Hurst index higher than 0.5 accounted for 75.17% of the total area, that means the vegetation coverage in the study area will be continuously increased in the future, in which the change trend in 25.23% regions could not be determined.