山西省2003-2012年植被时空变化格局及对气候因子的响应

Spatiotemporal Dynamics of Vegetation and Its Response to Climatic Factors in Shanxi Province from 2003 to 2012

利用2003—2012年卫星遥感数据归一化植被指数(NDVI)、山西省17个气象站逐日气温、降雨量资料和土地覆盖数据,分析了山西省近10a来NDVI的时空变化特征,及其对气候因子的响应。结果表明:近10a月NDVI变化呈现单峰型;7—8月份的植被整体呈中度改善;各个行政区中临汾、吕梁地区为明显改善;太原、忻州、晋中为中度改善;朔州、运城、长治、阳泉为轻度改善;大同与晋城地区基本不变;自然植被覆盖类型方面,林地、灌丛、草地3种自然植被覆盖类型都呈现改善趋势,林地为轻度改善,灌丛与草地为中度改善;山西省东南部地区的NDVI普遍高于西北部地区,且西北部的植被增长状况比东南部较好。基于月尺度的NDVI与气温呈显著正相关,而降雨量对NDVI变化存在一个阈值;基于年际尺度的NDVI与气温、降雨量的单变量相关不显著;基于月尺度的NDVI与同期的气温和降雨量的复合相关模型优于单变量相关模型。

Based on the normalized difference vegetation index（NDVI） data from moderate resolution imaging spectroradiometer（MODIS）, the daily temperature and precipitation data from 17 meteorological stations, and the land cover data in the period from 2003 to 2012, this paper analyzed spatiotemporal variation patterns of the vegetation and their correlations with climatic factors in recent ten years in Shanxi Province. The re- suits showed that the vegetation index of months indicated a single peak; in recent decade, the vegetation in- dex of Shanxi Province indicated moderate improvement in July and August; the vegetation index improved in all regions, the obvious improvement was observed in Linfen and Lvliang, the moderate improvement was found in Taiyuan, Xinzhou and Jinzhong, and the slight improvement was found in Shuozhou, Yuncheng, Changzhi and Yangquan; with respect to different land cover types, the vegetation index showed moderate improvement in shrub land and grassland, but slight improvement was found in woodland;the values of ND- VI in the southern and eastern regions were high, but low values were observed in the northern and western regions,and the improvement of NDVI in the northwest was better than the southeast. There is a significant positive linear correlation between the monthly average NDVI and the monthly mean temperature in the prov- ince. A threshold value of precipitation affecting NDVI existed, and NDVI did not increase significantly. There is a weak correlation between the NDVI and the interannual change of temperature and precipitation; the effect of univariate correlation models of monthly average NDVI and the monthly mean temperature/pre- cipitation is better than the composite variables correlation models.