A systems approach for estimating forest attributes from LiDAR
Loading...
Files
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of New Brunswick
Abstract
Light detection and ranging (LiDAR) is a popular technique in landscape-level forest inventory design, analyses, and implementation. The application of LiDAR in forestry has developed a set of standard analysis methods. However, many current methods ignore the importance of sample design, which is replaced with opportunistic samples, and focus on machine learning imputation to independently estimate different forest attributes leading to inconsistency across estimates that limit the practical applicability of LiDAR in forest management decision-making. In this dissertation, a workflow for LiDAR-assisted forest inventory is developed. Starting with sample designs (Chapter 2), variable probability sampling methods, previously limited by insufficient prior information. are explored. Appropriate LiDAR sampling designs obtain higher accuracy estimates at a lower cost. Rather than independent single forest attribute modeling, methods to obtain simultaneous estimates of multiple forest attributes is developed. A systems approach, composed by three allometric equations with five unknowns, is developed to obtain compatible and consistent forest inventory estimates in central New Brunswick (Chapter 3). To test the portability, this systems approach is applied to a large landscape in eastern Nova Scotia, Canada (Chapter 4). Both regions show that using this systems approach eliminates inconsistency and produce estimates with biological relevance among forest attributes. Finally, copulas are used to demonstrate the potential of obtaining stand Height – Diameter distributions with the use of LiDAR (Chapter 5). Although less precise point cloud co-registration is required, the current results varied among actual field conditions. In the future, applying the systems approach in the areas with different periods of LiDAR data collection might be able to detect changes in stand structure. In addition to area-based estimates, the combinations of the systems approach with the copulas could improve the stand Height – Diameter distribution estimates.