1982-2012年中亚地区植被时空变化特征及其与气候变化的相关分析

Spatial-temporal variation of vegetation and its correlation with climate change in Central Asia during the period of 1982-2012

干旱区植被生态系统对气候变化极为敏感,并且干旱区的植被变化研究对全球碳循环具有重要意义。然而近几十年来,中亚干旱区植被对气候变化的响应机制尚不甚明朗。利用归一化植被指数NDVI数据集和MERRA(Modern-Era Retrospective Analysis for Research and Applications)气象数据,采用经验正交函数(EOF,Empirical Orthogonal Function)和最小二乘法等方法系统分析了31a(1982—2012年)来中亚地区NDVI在不同时间尺度的时空变化特征。进一步分析和研究NDVI与气温和降水的相关性,结果表明:1982—2012年,中亚地区年NDVI总体呈现缓慢增长趋势,而1994年以后年NDVI呈现明显下降趋势,尤其在哈萨克斯坦北部草原地区下降趋势尤为突出。这可能是由于过去30年间,中亚地区降水累计量的持续减少造成的。NDVI的季节变化表明春季NDVI增长最为明显,冬季则显著下降。与平原区相比,中亚山区的NDVI值增长幅度最大,并且山区年NDVI与季节NDVI呈现显著增加趋势(P<0.05)。中亚地区年NDVI与年降水量正相关,而年NDVI与气温变化存在弱负相关。年NDVI和气温的正相关中心在中亚南部地区,负相关中心则出现在哈萨克斯坦的西部和北部地区;NDVI和降水的相关性中心刚好与气温相反。此外,在近30年间的每年6月至9月,中亚地区NDVI与气温存在近一个月的时间延迟现象。本研究为中亚干旱区生态系统变化和中亚地区碳循环的估算提供科学依据。

Dry-land plant ecosystems are particularly sensitive to climate change. Despite their importance in the global carbon cycle, the responses of vegetation dynamics to climate change in recent decades in Central Asian dry-lands remain unclear. In this study, we used normalized difference vegetation index （NDVI） data to analyze spatial-temporal changes in vegetation and to determine the correlation between vegetation and climatic variables over the period of 1982-2012 in Central Asia using the empirical orthogonal function （EOF） and least square methods. The results showed that the annual NDVI in Central Asia experienced an overall weak increasing trend during the study period, and a sharp decrease since 1994, especially in the grasslands of northern Kazakhstan, which might have been caused by decreased precipitation during the last three decades. In terms of seasonal variation, the NDVI in spring had a greater increasing trend than did the other three seasons. A significantly declining trend （P 〈 0.05） was found in winter. Compared to a weak increase in the NDVI in the plains areas, the mountainous areas experienced the greater increasing trend and the trend was significant at the annual and seasonal scales （P 〈 0.05）. The EOF results illustrated that the annual NDVI showed a uniform spatial pattern during 1982-2012. Further, the NDVI showed long-range dependence characteristics, which indicated the increasing trend may be continuous in the future. In terms of the relationships between the NDVI and climate variables over Central Asia, the annual NDVI was positively correlated with annual precipitation, and a weak negative correlation between annual NDVI and temperature was detected. The positive correlation coefficient center appeared between the annual NDVI and annual temperature over the southern Central Asia, and the negative center appeared over the areas between Kazakhstan, Turkmenistan, and Uzbekistan. However, the opposite correlation coefficients were found over these areas between the annual NDVI and annual precipitation. In addition, one-month time lags were found between NDVI and temperature from June to September in Central Asia during 1982-2012. Our study provides important information on vegetation variation and would facilitate better understanding of the natural ecosystem changes over Central Asia for the past three decades.