我国西北地区东部时间序列NDVI数据集重建方法比较研究  

Comparative studies of reconstruction methods for the long term NDVI dataset in the east of Northwest China

作　　者：王玮[1] 郭铌[1] 沙莎[1] 胡蝶[1] 王小平[1] 李耀辉[1] 

机构地区：[1]中国气象局兰州干旱气象研究所,甘肃省干旱气候变化与减灾重点实验室,中国气象局干旱气候变化与减灾重点实验室,甘肃兰州730020

出　　处：《草业学报》2016年第8期1-13,共13页Acta Prataculturae Sinica

基　　金：甘肃省气象局气象科研项目(GSMAMs2016-10);中国博士后科学基金项目(2015M582734);公益性行业(气象)科研专项(重大专项)(GYHY201506001-5)资助

摘　　要：高质量、长时序归一化植被指数(NDVI)数据集不仅是连续监测陆地表面特征的基础,也是研究气候与陆地生态系统变化的重要参数。本研究以生态环境较为脆弱的西北地区东部为例,借助多种时间序列重建方法对LTDR NDVI数据集中的噪声进行拟合重建,并结合农业气象资料和高质量NDVI数据,对不同重建方法的拟合结果开展适用性评价分析,结果表明,1)下垫面类型是影响重建方法拟合效果的重要因素。根据不同植被类型或作物生长特点,每种重建方法对其噪声消除能力有所不同;2)在年均NDVI较高(NDVI≥0.3),且NDVI曲线具有明显季节变化的草地、林地以及牧草等作物种植区域内,经过D-L拟合重建的NDVI具有较高的保真能力和适应性;3)在年均NDVI较低(NDVI<0.3),且植被季节生长变化不明显或NDVI曲线不呈季节对称性变化的稀疏植被区,以及以冬小麦为典型作物种植的区域内,经过S-G滤波重建的NDVI数据表现出相对较好的保真能力和适应性。A high-level time-series NDVI dataset is not only the basis for continuous monitoring of the land sur-face,but also an important tool for studying change related to climate and land use factors in terrestrial eco-systems.We reconstructed the noise component of the LTDR NDVI data for the east of Northwestern China where the ecosystem is fragile,using various time-series reconstruction methods.This paper use agrometeoro-logical data and high-level NDVI data to evaluate the accuracy of different reconstruction methods.The results show that：1）The vegetation or crop land cover is an important factor affecting fitted results of the various re-construction methods.Each reconstruction method has a different noise reduction ability depending on differ-ences in vegetation or crop growth characters;2）The D-L reconstruction method has a better noise reduction ability and applicability in those areas of grassland,and woodland for which the annual average NDVI data is higher （NDVI≥0.3）and the NDVI curve has obvious seasonal changes;3）The S-G reconstruction method has better fidelity ability and applicability in some areas of crop land in winter wheat and in areas of sparse vegeta-tion for which annual average NDVI data are lower （NDVI〈0.3）and where the NDVI curve have no obvious seasonal changes.

关 键 词：NDVI 时间序列重建 植被遥感 AVHRR 

分 类 号：Q948[生物学—植物学]