【作者】 沈艳；

【导师】 缪启龙； 牛铮；

【作者基本信息】 南京信息工程大学 ， 气象学， 2006， 博士

【摘要】 陆地表面的70%为植被所覆盖,植被是陆地生态系统的基本组成成分。植物体内所含的叶绿素、水分、蛋白质、木质素和纤维素等组分以及碳、氮、氢等微量元素统称为植被生化组分。这些组分的含量和构成,能直接或间接影响并制约着生态系统中与人类生存息息相关的许多重要生态、生理过程。高光谱遥感作为目前研究的前沿和热点,除具备常规遥感大面积、适时监测和非破坏性等优点外,还具有常规遥感所不具备的优势:具有10^-2λ数量级范围内的光谱分辨率。因此通过其精细光谱特征在提取植被生化组分的研究中表现出明显优势并越来越受到关注。利用高光谱遥感数据实现对植被生化组分的快速实时监测,是生态学、农学、全球变化研究以及粮食估产、精细农业等应用行业的迫切需求。 本论文围绕高光谱遥感定量提取植被生化组分这一前沿课题,以基于星载高光谱遥感图像提取冠层生化组分为研究重点。在阐明高光谱遥感估算植被生化组分意义的基础上,第一章对该研究的原理、研究层次、方法和试验基础等现状进行了综述,并由此引出了本论文的研究重点。论文的第二章主要介绍了数据资料的获取。详细介绍了在西双版纳地区开展的Hyperion星地同步试验,包括试验设计、样品采集、光谱测量及图像数据获取等。第三章主要介绍了利用高光谱响应特征提取叶片生化组分的研究,其中将含水量的定量提取作为重点内容之一,这章是冠层生化组分反演研究的基础和铺垫。论文第四章重点讨论了高光谱遥感图像处理,包括几何精纠正和大气校正。其中大气校正是遥感定量化研究中必需的。第五章是本论文的重点,主要探讨了利用Hyperion星载高光谱图像数据提取冠层生化组分并依此获得了该地区生化组分分布图。论文的第六章,主要是对全文进行概括总结,归纳了作者的主要研究成果并指出了今后的研究方向。 论文的主要研究成果和结论如下: 1、课题组在西双版纳地区成功开展了Hyperion星地同步试验,获得了生化组分数据、实测光谱数据以及高光谱遥感图像数据。论文中总结了试验过程的注意点并为今后相关试验的顺利开展提出了合理化建议。试验过程中“两个同步”的提出以及叶片保鲜和试验流程的规范化是至关重要的,这体现了论文的试验特色。 2、利用LOPEX数据集中叶片生化数据和光谱数据,研究了叶片层次生化组分的估算。用一阶导数极值和面积归一化的一阶导数极值两个参数估算植被更多还原

【Abstract】 Vegetation is one of the most important components of the ecosystem with 70 percent of the land surface coverage. Components such as chlorophyll, water, protein, lignin, cellulose and elements including carbon, nitrogen, hydrogen in the vegetation are defined as the vegetation biochemistry composition. The content and structure of these compositions can be directly or indirectly influenced on the ecological and physiological processing which is important to human being activities. As hot point and frontier in remote sensing, hyperspectral remote sensing technique not only has the advantages of traditional remote sensing that can be timely and nondestructively used to detect large vegetation area, but also has special advantages with the very high spectral resolution. More delicate spectral difference can help us to accurately retrieve vegetation biochemistry compositions and to monitor. Vegetation biochemistry extraction based on the remote sensing So composition estimation based on hyperspectral remote sensing technique is urgent not only for study in ecology, agronomy and global change but also for application in yield predict and precision farming.This thesis focuses on vegetation components retrieval with remote sensing especially hyperspectral remote sensing, and puts great emphasis on the study of extracting canopy biochemistry from the satellite-based hyperspectral imagery. The first chapter mainly introduced the principle, method and experiment basis of hyperspectral remote sensing in vegetation biochemistry, and then the emphasis of the article. In the second chapter, we primarily introduced the data obtainment, particularly discussed the satellite-land simultaneous experiment of XiShuangBanNa in Yunnan province. The experiment is included the design, sample collection, spectra measurement and imagery order. The third chapter is a foundation research for the canopy biochemistry retrieval. We mainly consider the leaf-level extraction using the hyperspectral response characteristics. The fourth part studied the hyperspectral imagery processing including the geometrical registration and atmospheric correction. The fifth part is the most important in this thesis. The relationship between the Hyperion spectra and the biochemical components is established and further obtain its distribution maps.The last chapter summarized the whole thesis and listed the achievement of this study, as same as, pointed out the research direction in the future. Main development and conclusion as follows:1. Satellite-land simultaneous experiment of XiShuangBanNa in Yunnan province collected the biochemical, measured spectra and hyperspectral remote sensing imagery data. We summarized the emphases and proposed some reasonable suggest for the related experiment. It indicated the experimental feature.2. Using spectral position (wavelength) variable analysis technique, two parameters, namely, first derivative extremum and area normalized first derivative extremum are brought forward and employed to retrieve leaf total nitrogen, cellulose, lignin, starch and water content with good results. Research indicates that the parameters can effectively remove the background influence.3. Leaf water content is retrieved and tested using the LOPEX dataset. The research results indicate that model inversion precision can be improved based on the leaf type and water expression. The spectra indices SR and Ratio975 is the best for FMC and EWT extraction, respectively.4. Simulation results by ACRM model exhibit that the short-wave infrared not the near-infrared wavelength is sensitivity to water content, namely the 1600nm and 820nm. So the two bands combined can strengthen the vegetation water content information. A soil adjusted water index (SAWI) based on the two bands is proposed to compute the canopy water content. SAWI simulation results under different LAI and Cw, different LAI and N combination showed that the index is a possibly applicable in water calculation.5. Hyperspectral imagery processing is studied including geometrical registration and atmospheric correction. Atmospheric correction of Hyperion image is managed both the dark/bright empirical method and 6S physical model. The key of the former is the selection of the calibration objects which need large enough, near Lambert and short of vegetation cover. 6S model atmospheric correction based on MOD IS data can overcome the difficult obtainments of water vapor, ozone concentration and aerosol optical depth. The two methods achieve the satisfactory correction accuracy.Through the partial least squares method, canopy chlorophyll, water, total carbon, total nitrogen, hydrogen and carbon-nitrogen ratio are retrieved and mapped using the processed Hyperion image and measured biochemistry data. It indicated that 1) the carbon-nitrogen ratio in this area is around 30;2) the reflectance in the wavelength of 2052nm and 2173nm is sensitive to canopy total nitrogen, carbon-nitrogen ratio and carbon content. From the practical point of view, using the sensitive bands can produce the instruments to detect the canopy nitrogen and carbon-nitrogen ratio timely and nondestructively. This part established a good foundation for vegetation hyperspectral remote sensing and carbon or nitrogen cycle research in our southwest tropical and sub-tropical humid climate.更多还原