星载光子计数激光雷达森林参数反演模型研究

We will focus on spaceborne photon-counting lidar system for forest observation purpose. A photon counting Lidar signal simulator will be established based on the vegetation radiative transfer model and airborne full-waveform Lidar data. In order to reveal the mechanism of the relationship between the scattering elements and the photon signal, also the noise generation and distribution patterns, we intended to develop an adaptive noise filtering algorithm and vegetation photon separation algorithm based on ellipse local outlier factors. Then a series of suitable photon counting lidar indices for forest structures characteristic parameter will be constructed with the combinations of photon point cloud quantiles, varied-scale synthetic waveform parameters and other variables. We will develop and establish the extendable inversion models for the estimation of forest parameter such as tree height and biomass, combined with factors such as statistical scale and model accuracy of photon data, develop uncertainty evaluation methods for inversion of forest structural parameters, and establish photon-counting Lidar data processing and forest parameter inversion systems. Based on the motion characteristics of the satellite platform and sensor observation capabilities, the effects of factors such as footprint distribution, incident energy, footprint size, overlap, and data signal-to-noise ratio on the inversion accuracy of forest parameters will be studied and proposes a satellite-borne photon counting Lidar solution for forest observations. The parameter optimization solution could provide the necessary theoretical and data support for the application of photon counting Lidar in China and serve the verification of terrestrial ecosystem observation satellite missions.

针对采用单光子探测体制的星载光子计数激光雷系统，基于植被辐射传输模型和机载全波形Lidar数据建立光子计数Lidar信号模拟器，从机理上揭示林木散射单元与光子信号的关系、噪声的产生和分布规律；发展基于椭圆离群因子的自适应噪声滤除算法和植被光子分离算法；结合光子点云分位数、变尺度合成波形参数等变量构建适于森林结构信息表征的光子计数lidar数据特征参数集，发展并建立可推广的树高、生物量等森林参数反演模型，结合光子数据的统计尺度、模型精度等因子，发展森林结构参数反演的不确定性评价方法，建立光子计数Lidar数据处理和森林参数反演系统；根据星载平台的运动特点和传感器观测能力，研究光斑分布、入射能量、光斑大小、重叠度、数据信噪比等因素对森林参数反演精度的影响，提出面向森林观测的星载光子计数Lidar参数优化方案，为我国光子计数Lidar的应用和陆地生态系统观测卫星任务的载荷验证提供理论与算法支持。

项目针对采用单光子探测体制的星载光子计数激光雷系统在林区探测的高噪声、算法可推广性差等问题，基于植被辐射传输模型和高精度机载激光雷达数据、ICESAT-2 ATLAS观测数据，开展了林区光子自适应分类算法和森林参数估测模型研究。主要结果为：（1）在深入研究单光子探测体制森林观测机理的基础上，建立了森林地物的光子计数LiDAR信号模拟器，通过系统剖析传感器可能的探测方式以及森林参数场景，从机理上揭示林木散射单元与光子信号的关系、噪声的产生和分布规律，提出了面向森林观测的星载光子计数LiDAR参数优化方案，最优光斑尺度应该设计在10-20 m的范围，采样间隔不超过0.8米。（2）发展了自适应旋转离群因子的噪声滤除算法和植被光子分离算法LOFR，该算法通过地形自适应旋转椭圆搜索区域，具有优良的地形自适应性和参数自优化能力，提高了复杂地形条件下林木及林下地形信息的分离和提取能力，在复杂地形条件下优于美国宇航局的官方产品。（3）构建了适于森林结构信息表征的光子计数LiDAR数据特征参数集，发展并建立了可推广的树高、生物量等森林参数反演模型，在我国典型林区均表现出很好的适应性，树高和生物量模型的验证精度解释能力分别为85%、80%左右，进而建立了光子计数LiDAR数据处理和森林参数反演系统，生产了全国ATLAS光子点云分类及森林参数产品。本项目的星载光子计数LiDAR参数方案和数据处理算法，为我国陆地生态系统观测卫星任务的光子计数LiDAR载荷验证提供理论与算法支持。项目在国内外主流学术期刊发表学术论文12篇，其中SCI 检索6篇，EI检索3 篇；申请算法发明专利4项，已授权2项；培养研究生4名，博士后2名。