Materials for the course of virtual forest twins based on marteloscope point cloud data in VirtualForest project. The materials prepared by Viet Nguyen (Eberswalde University for Sustainable Development) with R scripts was compiled with reference to materials from the course LiDAR data collection and analysis taught by Dr. Nikolai Knapp (Thuenen Institute of Forest Ecosystems).
The tutorial provides guidance on estimating individual tree attributes based on marteloscope point cloud data. The acquisition of basic tree attributes, including tree height, DBH, and crown area, contributes to the development of virtual forest twins. The tutorial includes 05 lectures along with R scripts:
This section aims to give a short introduction to the R programming language and Rstudio application as an integrated development environment for R. A step-by-step guide to setup R and Rstudio was also given, the environment that will be further used for LiDAR point cloud processing.
Users are recommended to use R version 4.2.2, and reproduce R environment using the renv.lock file. For more information, please find Introduction to renv
LiDAR or Light Detection and Ranging is an active remote sensing technology that can be used to measure 3-dimensional distribution of vegetation within forest canopies. The section introduce fundamental of LiDAR (or lidar) including:
- What is LiDAR?
- Advantages of LiDAR
- Disadvantages of LiDAR
- LiDAR components
- LiDAR Principle
- LiDAR Platforms
- Data structure
- Important definitions
- LiDAR metrics
- Software for LiDAR processing
The section provide guidance to preprocess the raw point cloud data, including:
- Point cloud clipping to area of interest
- Point cloud thinning for homogeneous density
- Point cloud filtering to remove noises or outliers
- Terain normalization
For extracting tree attributes at a single tree scale, a segmentation of tree crowns is required. The Individual Tree Segmentation (ITS) classifies LiDAR point clouds into single trees with corresponding IDs. After segmentation, each point has an ID corresponding to the tree it belongs to. In this section, raster-based and point cloud-based segmentation algorithms were introduced.
After individual tree segmentation, tree attributes at single tree level can be calculated. In this section, methods to estimate some basic tree inventory attributes were introduced:
- Crown area as area 2D convex hull polygons
- Tree height derived from the highest point in each segmented tree
- Tree location as barycenter of all points in each segmented tree
- DBH estimate using allometry
