地基与机载激光雷达协同的阔叶林聚集度指数反演

The spatial distributions of leaves in canopies are often non-random because they are organized in various structures. The non-random leaf distribution of vegetation can be characterized by a dispersion parameter called the clumping index (CI). Accurate retrieval of CI over forest canopies would obtain much accurate LAI and improve our understanding of forest carbon and nitrogen cycle, aboveground biomass estimation and forest growth monitoring. Based on the vegetation canopy transfer modeling, this project is designed to explore the vertical leaves clumping characteristics description based on terrestrial and airborne light detection and ranging (LiDAR) data.. With long term plot level observations, algorithms for estimating effective and true leaf area index model of terrestrial LiDAR data are improved, and the effects of zenith angle and azimuth angle to CI retrieval were eliminated. The methods and models for accurately retrieving CI and the vertical distribution of broadleaf forest canopy are designed. The airborne LiDAR point clouds are sparse, and the characteristics of the leaf point cloud cannot be extracted accurately. With the support of woody-to-total area ratio factors based on terrestrial LiDAR, synergistic CI retrieval model and algorithm based on terrestrial and airborne LiDAR will developed by the Beer-Lambert, and gap fraction from airborne LiDAR. The accuracy and uncertainty of the CI estimation model are evaluated by comparing the results of separated LiDAR data sets with those of the combined method.

聚集度指数（CI）是用来描述植被冠层叶片的聚集状态的一个重要指标，其准确反演对于叶面积指数准确估算、森林生长监测等具有重要意义。项目将在辐射传输理论的支持下协同地基和机载激光雷达（LiDAR）数据探索阔叶林冠层叶片聚集状态垂直分布特征表达方法。.通过大量观测实验，改进地基LiDAR数据的有效与真实叶面积指数反演模型，消除天顶角与方位角影响，构建准确表达阔叶林冠层CI及其垂直廓线的地面观测方法与模型；针对机载LiDAR点云稀疏，无法准确提取叶片点云的特点，耦合地基LiDAR的木质比反演模型，结合Beer-Lambert定律与机载LiDAR间隙率反演模型，构建地基与机载LiDAR协同的CI反演方法与模型，实现大面积阔叶林冠层CI及其垂直分布的准确表达；对比地基与机载LiDAR的有效LAI与间隙率反演模型，结合CI垂直分布廓线，评价地基与机载LiDAR数据反演CI的影响因素与不确定性特征。

聚集度指数（CI）是用来表征植被冠层叶片空间分布集聚特征的重要参数，对于研究森林碳、氮循环、叶面积指数和生物量估算、森林生长监测等具有重要意义。本项目开展了基于地基与机载激光雷达（LiDAR）数据的阔叶林冠层CI反演模型与方法研究，主要取得以下成果：.1.建立了基于地基LiDAR数据的阔叶树枝叶分离和真实叶面积指数反演方法，为CI反演与验证提供了方法与数据支撑。基于图论的单木枝叶分离方法仅使用点云几何位置，可充分利用最短路径优势特征，鲁棒性强，分类精度高。基于包络面拟合的阔叶树真实叶面积指数反演方法利用了枝叶形态特征，反演精度高、鲁棒性强。.2.构建了基于蒙特卡洛模拟射线方法和地基LiDAR数据的阔叶林冠层间隙率反演方法，充分发挥了林木冠层三维点云数据高密度、高精度的优势；提出了基于冠层上下边界掩膜方法和基于栅格映射法的绝对路径长度反演方法，实现了基于地基LiDAR数据的阔叶林冠层CI反演。.3.构建了基于形态学理论的机载LiDAR树木点云提取方法，提出了一种无需先验知识，基于点云三维分布特征的机载LiDAR单木分割算法，提升了森林样地数据处理效率。.4.为充分利用阔叶林冠层三维激光点云数据，结合体元法和投影法优势建立了基于激光点面积模型的阔叶林冠层间隙率估算方法，克服了传统方法对点云垂直信息利用不足所产生的信息损失。基于理想状态下叶片在空间中的分布模式可以视作泊松分布原理，建立了在无需冠层间隙率的情况下直接反演阔叶林冠层CI的模型与方法。协同地基与无人机载LiDAR数据，分别利用激光点面积模型和泊松分布模型实现了阔叶林冠层CI及其垂直分布的准确表达。.在以上成果支撑下，项目在遥感期刊和重要学术会议上发表了12篇学术论文，其中SCI检索期刊论文6篇。申请国家发明专利8项，已授权国家发明专利2项。