Point clouds with ground point classification and individual tree segmentation of 25 northern boreal forest and tundra sites from UAV-based lidar surveys in northwestern Canada in 2022

This dataset represents LiDAR 3D point clouds with centimetre accuracy. It is a collection of processed point clouds that were collected by a UAV-based laser scanner (YellowScan Mapper). The region covered is part of the Northern Boreal Forest and the Tundra-Taiga Ecotone (TTE). The aim is to map the forest structure and in particular to segment individual trees. This allows structural and biophysical information to be derived on forest condition, structure, topography, morphology, disturbance (wildfire, thaw slumps, wind events), thermokarst lakeshore interfaces, etc. The fieldwork was planned and carried out by researchers from the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI). It took place in the peak summer of 2022, from 10 July to 19 August. The study area covers a long bio-climatic gradient from the northern tundra and open forested areas around Inuvik, Northwest Territories, Canada, through the southern Yukon Territory and mountainous terrain across the Mackenzie Mountains and southern Cassiar Mountains, to areas near Fort Nelson, British Columbia, Canada. At 26 sites, we pre-selected locations based on satellite data. At each site, transects at least 50 m wide and 500 m long were flown in a grid of parallel flight lines 20 m apart. At the ground level, one or, in rare cases, multiple 30-metre-diameter plots were inventoried as ground reference data. The recordings were made at an altitude of 70 m above ground liftoff at a speed of 5 m/s, as recommended by the manufacturer. Before each flight, a GPS base station was set up to collect data for later geographic correction. The raw data processing steps involved correcting the flight trajectories in POSPac PP-RTX , flight strip alignment in software YellowScan CloudStation, and colorization with RGB module picture data. The subsequent processing and metadata collection in PC2RCHIVE includes (i) ground classification, (ii) creating Digital Terrain Model (DTM), Digital Surface Model (DSM) and Canopy Height Model (CHM), (iii) tree segmentation, (iv) generating maps and (v) user provided metadata and relevant information from the data processing are summarized in tables. The processing of the resulting point clouds followed methods developed in PC2RCHIVE, a tool to easily standardise the point cloud data products and their metadata.

本数据集为厘米级精度的激光雷达（LiDAR）三维点云数据集，其采集自基于无人机平台的激光扫描仪（YellowScan Mapper），均为经过预处理的点云集合。覆盖区域位于北部寒温带针叶林与苔原-泰加林交错带（Tundra-Taiga Ecotone，简称TTE）范围内，旨在开展森林结构制图，尤其聚焦于单木分割任务。借此可提取森林健康状况、林分结构、地形地貌、植被形态、干扰事件（野火、融滑坍、风灾等）、热喀斯特湖岸界面等相关的结构与生物物理信息。野外作业由亥姆霍兹极地与海洋研究中心阿尔弗雷德·魏格纳研究所（AWI）的科研人员规划并实施，作业时间为2022年盛夏，即7月10日至8月19日。研究区域涵盖一条跨度极大的生物气候梯度带，从加拿大西北地区因维克周边的北部苔原与疏林区域出发，途经育空地区南部、横跨麦肯齐山脉与卡斯卡山脉南部的山地地形，最终延伸至加拿大不列颠哥伦比亚省纳尔逊堡周边区域。研究团队基于卫星数据预先选定了26个采样点，每个采样点均布设宽度不小于50米、长度500米的样带，采用间隔20米的平行飞行航线网格开展数据采集。在地面层面，每个采样点均设置1个（极少数情况下为多个）直径30米的样地，用于开展地面调查以获取地面参考数据。本次数据采集按照设备制造商的建议，设置飞行高度为离地70米，飞行速度为5米/秒；每次飞行前均架设GPS基准站采集数据，用于后续的地理校正。原始数据处理流程包括：在POSPac PP-RTX中校正飞行轨迹，使用YellowScan CloudStation软件完成航带拼接，以及通过RGB模块的影像数据为点云赋予色彩。后续在PC2RCHIVE平台中开展的处理与元数据收集工作包括：(i) 地面点分类；(ii) 生成数字地形模型（Digital Terrain Model, DTM）、数字表面模型（Digital Surface Model, DSM）与冠层高度模型（Canopy Height Model, CHM）；(iii) 单木分割；(iv) 制图生成；(v) 整理用户提供的元数据，同时将数据处理过程中的相关信息汇总为表格。最终点云的处理流程采用PC2RCHIVE开发的方法，该工具可便捷实现点云数据产品及其元数据的标准化。