摘要
第五代植被指数(C5)对16-Day的Terra/MODIS-NDVI(MOD-NDVI)和Aqua/MODIS-NDVI(MYD-NDVI)采用"错位8 d"合成,理论上可以将时间分辨率提高到8-Day。为了探讨该算法可行性,利用中国东北2008年生长季C5的MOD和MYD数据进行了研究。首先利用线性滤波分别对MOD和MYD时间序列进行平滑,同时对MOD和MYD的合成时间(DOY)进行了统计分析。DOY分析发现:MODIS C5采用MOD与MYD"错位8 d"合成出现的8-Day重叠,可能导致时间错位问题:按照"前8-Day∶后8-Day"统计,比例为55∶45,如果直接合并,会导致像元DOY时间错位比例为7%,这导致MYD-NDVI与MOD-NDVI错位,而且这样的错位不能通过滤波恢复。最后,提出了基于实际DOY的合成算法,可以把C5的两个16-Day产品合并为8-Day时间序列。判定的优先顺序为:DOY>QA>NDVI。
This paper explored the feasibility of developing an 8-day time serial MODIS-NDVI by merging the phased 8-Day Terra-MODIS-NDVI (MOD-NDVI) and Aqua-MODIS-NDVI (MYD-NDVI) 16-Day time serials in Collection 5 (C5). First,MOD-NDVI and MYD-NDVI time serials were filtered separately based on Quality Assessment (QA) and phenology,which ensured cleaned NDVI time serials. Second,the actual day of pixel in each composite period was calculated,and the results showed that the percentage of pixels in first 8-Day of a 16-Day composite period was 55~. The further analysis indicated that if MOD-NDVI and MYD-NDVI were simply merged,7~ of pixels would suffer from time disorder. This time disorder might lead to NDVI disorder, which might not be reasonably restored by filtering. Finally, a conceptual framework was presented to double MODIS-NDVI temporal resolution based on C5 Terra and Aqua. The priority to select a pixel for an 8-Day composite period was actual Day of Year (DOY) ,QA,and NDVI,respectively.
出处
《遥感技术与应用》
CSCD
北大核心
2011年第4期437-443,共7页
Remote Sensing Technology and Application
基金
资源与环境信息系统国家重点实验室自主研究课题"MODIS植被指数质量改进关键问题研究"