【作者】 骆社周；

【导师】 张贵宾； 王成；

【作者基本信息】 中国地质大学(北京) ， 地球探测与信息技术， 2012， 博士

【摘要】 森林与大气之间碳、水及能量的生态交换过程受到森林冠层与林分结构的影响，因此，冠层与林分结构的定量提取技术在森林动态监测中显得尤为重要。同时，植被冠层是大气与陆地生物圈的重要接口，而叶面积指数(LAI)是植被冠层对全球环境变化响应过程模型的一个非常关键的参数。因此快速、精确、可靠和客观地评价LAI是研究大气与植被相互作用必不可少的过程。植被指数(Vegetation Index, VI)是光学遥感影像评价地表植被覆盖状况的关键参数。植被指数可用于反演叶面积指数(LAI)，但主要缺陷是植被指数存在饱和问题。被动光学遥感多用于探测植被冠层水平结构信息，在植被生长茂盛的地区，植被指数达到饱和状态后，植被指数和LAI不再线性增加，导致LAI被低估。所以，光学遥感数据提取LAI具有一定的局限性，其反演精度受到一定的影响。如何提高LAI反演精度，并且同时能够使反演方法简单易用，是该领域的一个重要研究方向。激光雷达(Light DetectionAnd Ranging, LiDAR)是近年来国际上发展十分迅速的主动遥感技术，可快速获取目标的三维空间信息，能精确地估算森林垂直结构参数(如树高、冠层郁闭度和地上生物量等)。本文主要有四个目标：一是探索机载、星载LiDAR数据提取LAI的理论方法；二是通过LiDAR数据实现LAI反演并进行精度验证，三是对研究区进行基于LiDAR反演模型的LAI制图；四是用光学遥感数据反演LAI，并与LiDAR数据提取LAI的结果进行对比分析。具体主要开展了以下几个方面的研究工作：(1)对机载LiDAR数据进行了分类，计算出了不同空间尺度的激光穿透指数(LPI)。(2)研究了机载LiDAR数据提取LAI的理论和方法，并通过LPI实现LAI反演。(3)比较了不同空间尺度下机载LiDAR提取LAI的精度，获取机载LiDAR数据反演LAI的最优模型是采样半径为10 m空间尺度(R2=0.77)，并对研究区进行LAI制图。(4)深入研究了星载LiDAR数据波形处理方法，并对GLAS波形数据进行了高斯分解，确定了波形特征参数，并计算了地面回波能量与总能量的比值。(5)探索了星载激光雷达提取LAI的方法，建立了星载LiDAR数据反演LAI的模型(R~2=0.80)，并对研究区进行LAI制图。(6)确定光学遥感反演LAI的最优模型，并实现了研究区LAI制图。(7)对机载LiDAR数据和光学遥感数据提取大野口研究区的LAI结果进行了对比分析，并证明了机载LiDAR数据可提高LAI反演精度。(8)对星载LiDAR数据提取LAI和光学遥感提取西藏林芝地区的LAI进行了对比分析，结果表明星载LiDAR数据能提高LAI的反演精度。本文的创新研究成果包括以下几个方面：(1)提出了GLAS波形数据反演LAI的方法。(2)融合GLAS数据与TM影像，实现了研究区LAI反演及制图。(3)提出了基于机载LiDAR数据的简化LAI反演方法。本研究工作的结果表明LiDAR数据可实现LAI的高精度反演，特别是星载LiDAR数据为全球性高精度LAI估算提供了可能性。更多还原

【Abstract】 Carbon, water, and energy ecological exchanged processes between forest and atmosphereare influenced by forest canopy and stand structure. Therefore, exploring the technique andmethod of quantizing canopy structure and stand structure to monitor the change of foreststructural parameters is very significant. Vegetation canopy is the important interface of theatmosphere and terrestrial biosphere. Since leaf area index (LAI) is very often a criticalparameter in process-based models of vegetation canopy response to global environmentalchange, for numerous studies of interaction of atmosphere and vegetation, rapid, reliable andobjective estimations of LAI are essential.Explaining vegetation cover of earth surface from the view of optical remote sensing,vegetation indices are the most valuable. However, when retrieving LAI by vegetation indices(VIs) derived from remotely sensed data, the main problem encountered is the saturation athigh levels of LAI. That is to say, optical remote sensing data only detect horizontal structureinformation of vegetation canopy. At high levels of LAI, the VIs related to LAI reachsaturation and the LAI doesn’t increase linearly with the VI, which will cause theunderestimation of LAI. Therefore, extraction LAI by optical remotely sensed data has acertain limitation and the accuracy of LAI inversion will be affected to some extent. How toimprove the accuracy of forest LAI inversion and make the inversion method simple and easyto use is a very important study area at home and abroad.Light Detection And Ranging (LiDAR) is a new emerging active remote sensingtechnology in recent years, which has developed very rapidly in the world. LiDAR canmeasure both the vertical and horizontal structure of forested areas effectively with highprecision and it can accurately estimate tree height, canopy closure and above-groundbiomass.The main purpose of this dissertation is to investigate the potential and feasibility ofderiving forest LAI from LiDAR data. The specific objectives of this study were:(1) Exploringtheories and methods of using airborne and spaceborne LiDAR data to extract forest LAI;(2)Establishing the forest LAI estimation model from LiDAR data and validating its accuracy;(3)Mapping forest LAI of the study area based on LiDAR data LAI inversion model;(4) Usingoptical remotely sensed data to extract forest LAI, and the comparison and analysis being carried out between the result of LAI LiDAR-derived and based on optical remotely senseddata. And specifically, this dissertation mainly conducted some research as follows:(1) Classifed the airborne LiDAR data and calculated the laser penetration index (LPI) ofdifferent spatial scales.(2) Studied the theory and method of using airborne LiDAR data to extract forest LAI,and inversed the forest LAI by the LPI.(3) Compared the accuracy of the airborne LiDAR-derived LAI inversion model withdifferent spatial scales. And then, the optimum model of LiDAR-derived LAI wasobtained and the determination coefficient (R2) was 0.77, where the samplingradius was 10 m. Finally, Mapped LAI of the study area based on the model.(4) Studied deeply processing methods of spaceborne LiDAR waveform data and the rawGLAS waveform data were decomposed into Gaussian peaks. And then, the ratiosof ground to the entire waveform return energy of satellite borne LiDAR.(5) Explored the method of LAI retrieval based on satellite borne LiDAR, andestablished the model (R2=0.80). Finally, Mapped LAI of the study area based onthe model.(6) Introduced the theory of optical remote sensing retrieving LAI and established theoptimum model of LAI inversion in the study area. And then, mapped LAI of thestudy area using the model.(7) Compared and analyzed the results between airborne LiDAR-derived LAI and basedon optical remotely sensed data in the Dayekou study area. The R2of based onoptical remote sensing was 0.63, while the R2of airborne attained to 0.77. Thisshowed that airborne LiDAR data could improve accuracy of LAI retrieval.(8) Compared and analyzed the results between satellite borne LiDAR-derived LAI andbased on optical remotely sensed data in the study area of Linzhi, Tibet. Themaximum R2of based on optical remote sensing was 0.65, while the R2ofGLAS-derived LAI attained to 0.80.The result showed that satellite borne LiDARdata could improve accuracy of LAI inversion.The main innovations in this dissertation included as followings:(1) The method of GLAS waveform data inversing forest LAI was put forward. (2) Reversed and mapped forest LAI of the study area by integrating GLAS and TM data.(3) The simple method of inversing LAI was developed based on airborne LiDAR data.By the research above, the dissertation found that LiDAR data offer a new way toaccurately estimate forest LAI. Especially, satellite borne LiDAR data open the possibility ofglobal forest LAI estimation with high accuracy.更多还原