基于连续小波分析的植物理化参数反演中光谱分辨率影响分析
投稿时间:2018-04-10  修订日期:2018-05-09  点此下载全文
引用本文:
摘要点击次数: 103
全文下载次数: 
作者单位E-mail
张竞成 杭州电子科技大学 生命信息与仪器工程学院 zhangjc_rs@163.com 
刘 鹏 杭州电子科技大学 生命信息与仪器工程学院 zzwliupeng@163.com 
王斌 杭州电子科技大学 生命信息与仪器工程学院 wangb_rs@163.com 
张雪雪 杭州电子科技大学 生命信息与仪器工程学院 zhangxx_rs@163.com 
黄文江 中国科学院遥感与数字地球研究所 数字地球重点实验室 huangwj@radi.ac.cn 
吴开华 杭州电子科技大学 生命信息与仪器工程学院 wukaihua@hdu.edu.cn 
基金项目:浙江省科技计划项目(2016C32087)和国家自然科学基金(41671415)
中文摘要:光谱连续小波分析(continuous wavelet analysis,CWA)自提出后被认为在植物生理生化参数反演方面具有很大潜力。光谱分辨率作为遥感器观测能力的重要指标,同时也是CWA分析数据的关键参数,但该指标对基于CWA的植物生理生化参数反演影响尚不明确。为此,本文基于一套由PROSPECT模型模拟的包含叶绿素含量(Cab)、类胡萝卜素含量(Car)和叶片水含量(LWC)等重要植物生理生化参数及其光谱的数据,通过对光谱进行系列梯度的重采样和CWA分析,详细研究了光谱分辨率对植物生理生化参数反演的影响。结果表明:(1)采用CWA能够成功提取对Cab, Car和LWC等参数敏感的特征并建立具有较高精度的反演模型;(2)随着光谱分辨率的降低,敏感小波特征的数量、相关性以及反演精度总体均呈下降趋势,但下降的幅度、拐点均不相同,体现出分辨率对不同参数影响的差异性;(3)采用CWA反演建模时,不同植物生理生化参数对光谱分辨率敏感性差异较大,LWC敏感性较低,Cab次之,Car敏感性较高。根据这一结果,采用CWA反演Car,Cab和LWC时光谱数据在分辨率不低于8 nm,32 nm和64nm时能够得到较理想的结果。上述研究能够为实际中进行植被生理生化参数监测时的传感器选择提供依据。
中文关键词:连续小波分析,高光谱遥感,植被生理生化参数,光谱分辨率
 
Impact analysis of spectral resolution on retrieving plant biophysical and biochemical parameters based on continuous wavelet analysis
Abstract:Continuous wavelet analysis (CWA) has been proposed to have great potential as a spectral analysis tool for retrieving plant biophysical and biochemical parameters. Spectral resolution, as an important indicator of hyperspectral sensors, is also a key parameter of CWA analysis. But the impact of this index on retrieving plant biophysical and biochemical parameters is unknown yet. To solve this issue, a simulated spectral dataset based on the PROSPECT model was obtained with corresponding biophysical and biochemical parameters including chlorophyll content (Cab), carotenoid content (Car), and leaf water content (LWC). According to the resampled spectra at varying spectral resolution and CWA, the impact of spectral resolution on the retrieving of plant biophysical and biochemical parameters was then studied. The results show that: (1) CWA can be used to successfully extract sensitive features and to establish retrieving models of parameters including Cab, Car and LWC with high accuracy. (2) With decline of spectral resolution, the number of sensitive features, their correlation, and retrieving accuracy tend to decrease. However, the decline amplitude and the inflection point of the decline curves are all different, which reflected the different impact of the spectral resolution different for different parameters. (3) There is a significant difference in the sensitivity to spectral resolution among different plant biophysical and biochemical parameters, with the LWC appeared the most insensitive, followed by Cab, and with the most sensitive of Car. Based on this result, when to retrieve Car, Cab and LWC with CWA, a reasonable result could be expected in case the spectral resolution is no lower than 8 nm, 32 nm and 64 nm, respectively. The present study can provide a basic understanding in selection of hyperspectral sensors for retrieving and monitoring of plant biophysical and biochemical parameters using the CWA method.
keywords:Continuous wavelet analysis, Hyperspectral remote sensing, Plant biophysical and biochemical parameters, Spectral resolution
  查看/发表评论  下载PDF阅读器

版权所有:《红外与毫米波学报》编辑部