A Sampling Strategy for Remotely Sensed LAI Product Validation Over Heterogeneous Land Surfaces
作者:Zeng, YL (Zeng, Yelu)[ 1,3 ] ; Li, J (Li, Jing)[ 1 ] ; Liu, QH (Liu, Qinhuo)[ 1 ] ; Li, LH (Li, Longhui)[ 2 ] ; Xu, BD (Xu, Baodong)[ 1,3 ] ; Yin, GF (Yin, Gaofei)[ 1,3 ] ; Peng, JJ (Peng, Jingjing)[ 1,3 ]
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
卷: 7  期: 7  页: 3128-3142  特刊: SI
DOI: 10.1109/JSTARS.2014.2312231
出版年: JUL 2014

摘要
The development of efficient and systematic ground-based spatial sampling strategies is critical for the validation of medium-resolution satellite-derived leaf area index (LAI) products, particularly over heterogeneous land surfaces. In this paper, a new sampling strategy based on high-resolution vegetation index prior knowledge (SSVIP) is proposed to generate accurate LAI reference maps over heterogeneous areas. To capture the variability across a site, the SSVIP is designed to 1) stratify the nonhomogeneous area into zones with minimum within-class variance; 2) assign the number of samples to each zone using Neyman optimal allocation; and 3) determine the spatial distribution of samples with a nearest neighbor index. The efficiency of the proposed method was examined using different vegetation types and pixel heterogeneities. The results indicate that the SSVIP approach can properly divide a heterogeneous area into different vegetation cover zones. Whereas the LAI reference maps generated by SSVIP attain the target accuracy of 0.1 LAI units in cropland and broadleaf forest sites, the current sampling strategy based on vegetation type has a root mean square error (RMSE) of 0.14 for the same number of samples. SSVIP was compared with the current sampling strategy at 24 VALERI sites, and the results suggested that samples selected by SSVIP were more representative in the feature space and geographical space, which further indicated the reasonable validation over heterogeneous land surfaces.
通讯作者地址: Li, J (通讯作者)
 Beijing Normal Univ, Chinese Acad Sci, Inst Remote Sensing & Digital Earth, State Key Lab Remote Sensing Sci, Beijing 100101, Peoples R China.
地址:
 [ 1 ] Beijing Normal Univ, Chinese Acad Sci, Inst Remote Sensing & Digital Earth, State Key Lab Remote Sensing Sci, Beijing 100101, Peoples R China
 [ 2 ] Univ Technol Sydney, Plant Funct Biol & Climate Change Cluster C3, Sydney, NSW 2007, Australia
 [ 3 ] Univ Chinese Acad Sci, Coll Resources & Environm, Beijing 100049, Peoples R China