Plant Spectral Discrimination Based on Phenological Features
ZHANG Lei1, ZHAO Jian-long2, JIA Kun3*, LI Xiao-song1
1. Key Laboratory of Digital Earth Science, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China 2. School of Earth Sciences and Resources, China University of Geosciences, Beijing 100083, China 3. School of Geography, Beijing Normal University, Beijing 100875, China
Abstract：Spectral analysis plays asignificant role onplant characteristic identification and mechanism recognition, there were many papers published on the aspects of absorption features in the spectra of chlorophyll andmoisture, spectral analysis onvegetation red edge effect, spectra profile feature extraction, spectra profile conversion, vegetation leaf structure and chemical composition impacts on the spectra in past years. However, fewer researches issued onspectral changes caused by plant seasonal changes oflife form, chlorophyll, leaf area index. This paper studied on spectral observationof 11 plantsof various life form, plant leaf structure and its size, phenological characteristics, they includedeciduous forest with broadvertical leaf, needleleafevergreen forest, needleleafdeciduous forest, deciduous forest with broadflat leaf, high shrub with big leaf, high shrub with little leaf, deciduous forest with broad little leaf, short shrub, meadow, steppe andgrass. Field spectral data were observed with SVC-HR768 (Spectra Vista company, USA)，the band width covers 350～2 500 nm，spectral resolution reaches 1~4 nm. The features of NDVI, spectral maximum absorption depth in green band, and spectral maximumabsorption depth in red bandwere measured after continuum removal processing, the mean, amplitude andgradientof thesefeatures onseasonalchange profilewere analyzed, meanwhile, separability research on plant spectral feature of growth period and maturation period were compared. The paper presents a calculation method ofseparabilityof vegetation spectra which consider feature spatial distances. This index is carried on analysis of the vegetation discrimination. The results show that: the spectral featuresduring plant growth periodare easier to distinguish than them during maturation period. With the same features comparison, plant separabilityof growth period is 3 points higherthan it during maturation period. The overall separabilityof vegetation spectrumfeatures shows seasonal amplitude>seasonal gradient>seasonal mean during the growth period, but the separabilityof seasonal gradient shows highest value for the plant seasonal NDVI change. Therefore, the features of seasonal NDVI gradient, seasonal amplitude of maximumspectral absorption depth in green band, seasonal amplitude of maximumabsorption depthin red band are effective composition for plant discrimination.
[1] Piron A, Leemans V, Kleynen O, Computers and Electronics in Agriculture，2008，62: 141. [2] Koger C H, Bruce L M, Shaw D R. Remote Sensing of Environment，2003，86(1): 108. [3] LI Xian-feng，ZHU Wei-xing，JI Bin (李先锋，朱伟兴，纪 滨). Transactions of the Chinese Society for Agricultural Machinery (农业机械学报)，2010，41(11): 168. [4] ZHOU Lan-ping，WEI Huai-dong，DING Feng，et al(周兰萍，魏怀东，丁 峰，等). Journal of Arid Land Resources and Environment(干旱区资源与环境)，2013，27(3)：121. [5] LIU Ke，ZHAO Wen-ji，GUO Xiao-yu，et al(刘 克，赵文吉，郭逍宇，等). Acta Ecologica Sinica(生态学报)，2010，30(21)：5853. [6] ZU Qin，ZHAO Chun-jiang，DENG Wei，et al(祖 琴，赵春江，邓 巍，等). Spectroscopy and Spectral Analysis(光谱学与光谱分析)，2013，33(5)：1202. [7] LIN Chuan，GONG Zhao-ning，ZHAO Wen-ji，et al(林 川，宫兆宁，赵文吉，等). Acta Ecologica Sinica(生态学报)，2013，33(4)：1172. [8] Muraoka H，Saigusa N，Nasahara K N, et al. Journal of Plant Research, 2010, 123(4): 563. [9] YANG Ke-ming，GUO Da-zhi，CHEN Yun-hao(杨可明，郭达志，陈云浩). Computer Engineering and Applications(计算机工程与应用)，2006，31：213.
