Partitioning biomass in Canadian boreal forests with fine-scale structure characterization from terrestrial laser scanners

精确估计树木一级的生物量分配对于编制森林资源清单、碳核算或木材产品至关重要。如果生物量库存没有确定性，就很难对生物经济项目进行资本投资，并限制了森林部门的潜力。到目前为止，一些省份和地区依靠传统的森林清查来确定生物量估计数的空间分布。该项目旨在从两种类型的陆地激光雷达（TLS）数据中产生高价值的信息：1）木材和树叶生物量分配，以及2）森林火灾可燃物估计。因此，它是由两个研究问题：1）TLS可以用来取代目前的破坏性测量，以提供生物量分配（茎，分支，叶）的物种特异性模型，同时保持15%的精度要求？2）更广泛和更好的精细尺度结构表征能否支持下一代火灾行为产品，以更好地表征和映射燃料类型，从而更好地预测全国范围内的火灾行为和风险？主要目标是将一套经过验证的算法应用到一个可操作的、开放访问的软件管道中，该软件管道具有系统的输出控制（人工辅助追溯），以促进TLS数据的处理，从而表征森林结构和估计树木生物量分区。该管道将实现四种主要算法来处理森林地块的TLS点云：（1）单个树冠的分割，（2）木材与树叶材料的分类，（3）干形和分支结构的表征，以及（4）树木叶面积指数（LAI）的估计。这里介绍的合作，跨学科的研究项目将解决知识差距，并为加拿大林业部门产生重大成果，包括：a）改善生物质供应的获取，安全性和竞争力，B）改进预测和应对森林野火风险，以及c）培训未来的劳动力，重点是多样性。

Precise tree-level estimates of biomass partitioning are fundamental to compiling forest resource inventories, carbon accounting or wood products. Without certainty in biomass inventories, it becomes difficult to make capital investments in bioeconomy projects and limits forest sector potential. As of today, several provinces and territories rely on traditional forest inventory for spatializing biomass estimates. This project aims at producing high value information from terrestrial lidar (TLS) data of two types: 1) wood and leaf biomass portioning, and 2) forest fire combustible estimation. It is thus guided by two research questions: 1) Can TLS be used to replace current destructive measurements to provide species-specific models of biomass partitioning (stem, branch, and leaf) while retaining an accuracy requirement of 15%? And 2) Can broader and better fine-scale structure characterization support the next generation fire behavior products to better characterize and map fuel types, which in turn would allow better predictions of fire behavior and risk at national scale? The principal objective is to implement a suite of validated algorithms into an operational, open-access, software pipeline with systematic output controls (human-assisted retroaction) to facilitate the processing of TLS data for characterizing forest structure and estimate tree biomass partitioning. The pipeline will implement four main algorithms to process TLS point clouds of forest plots: (1) segmentation of individual tree crown, (2) classification of wood vs. leaf material, (3) characterization of the stem form and branching structure, and (4) estimation of the leaf area index (LAI) of trees. The collaborative, inter-disciplinary research project presented here will address knowledge gaps and generate significant outcomes for the Canadian forest sector, including: a) improved biomass supply access, security and competitiveness, b) an improved anticipation and response to forest wildfire risks, and c) training the future workforce, with an emphasis on diversity.