面向机载LiDAR森林地上生物量反演的数据采集方式与模型优化研究

Airborne LiDAR-estimated forest aboveground biomass (AGB), cooperated with satellite imageries, are usually treated as “ground truth data” in the large-scale mapping of forest AGB. However, there is an important question remains to be resolved. That is how to improve the spatial and statistical representativeness of airborne LiDAR data meanwhile to develop the most optimized AGB estimation model that suits multiple forest types on the premise of saving economic cost. In this project, several discrete airborne LiDAR data collecting strategies will be simulated to investigate how to improve the spatial and statistical representativeness as well as use ratio of airborne LiDAR data. Based on these, new LiDAR metrics with certain ecological meaning will be developed through characterizing forest spatial structure using airborne LiDAR data. The response of the new LiDAR metrics to forest AGB will be investigated. Finally, the most optimized AGB estimation model that suits multiple forest types will be developed with various LiDAR metrics as input data, during which the reliability of estimated AGB to be the “ground truth data” will be assessed using field-measured AGB. The implementation of this project will provide new ideas for optimizing the estimation procedure of forest AGB, providing reliable “ground truth data” for large-scale forest AGB mapping.

机载LiDAR数据反演的森林地上生物量（AGB）常被当作“地面真值”与星载遥感数据结合用于大范围AGB制图。如何在节省经济成本的前提下，提高LiDAR数据的空间和统计学代表性并获取适用于多种森林的AGB最优反演模型是机载LiDAR反演的森林AGB作为“地面真值”之前亟待解决的关键问题。本项目通过模拟多种离散的机载LiDAR数据采集方式来探究如何提高LiDAR数据的空间和统计学代表性以及利用率；在此基础上，以冠层结构特征化的方式利用小光斑点云和大光斑伪波形数据构建具有一定生态学意义的LiDAR特征变量，并探索其与AGB的响应机理；以多类LiDAR特征变量集为输入源构建适合不同森林类型的AGB最优反演模型，并通过地面实测数据评价AGB反演结果作为“地面真值”的可靠性。本项目有望为优化机载LiDAR森林AGB反演过程提供新思路，为大范围森林AGB制图提供可靠的森林AGB“地面真值”。

机载LiDAR遥感常被用于定量反演植被结构和地上生物量(AGB)等重要碳循环因子。针对如何在节省经济成本的前提下，提高机载LiDAR数据的空间和统计学代表性以获取适用于多种森林的AGB最优反演模型的问题，本项目通过模拟了多种离散的机载LiDAR数据采集方式来探究如何提高LiDAR数据的空间和统计学代表性以及利用率；在此基础上，研究了如何利用小光斑机载LiDAR数据和大光斑伪波形数据本身的空间结构、高程和强度等属性的分布特征构建了能够表征森林冠层结构空间异质性和复杂度的LiDAR特征变量，以提高森林AGB反演过程中模型信息输入源的丰富度；以多类LiDAR特征变量集为输入源构建了适合不同森林类型的AGB最优反演模型，并通过地面实测数据对AGB反演结果作为“地面真值”的可靠性进行了全面评价。在项目执行期间，以第一作者身份发表相关论文4篇，包括国内高水平英文论文1篇。此外，待投稿论文1篇。完成项目年度进展报告2份；结题报告1份。参加国内会议3次。