Semester

Fall

Date of Graduation

2022

Document Type

Dissertation

Degree Type

PhD

College

Eberly College of Arts and Sciences

Department

Geology and Geography

Committee Chair

Brenden McNeil

Committee Co-Chair

Nichola Skowronski

Committee Member

Aaron Maxwell

Committee Member

Timothy Warner

Committee Member

Jamison Conley

Abstract

Light detection and ranging (LiDAR) and terrestrial laser scanning (TLS) sensors are powerful tools for characterizing vegetation structure and for constructing three-dimensional (3D) models of trees, also known as quantitative structural models (QSM). 3D models and structural traits derived from them provide valuable information for biodiversity conservation, forest management, and fire behavior modeling. However, vegetation studies and 3D modeling methodologies often only focus on the forest canopy, with little attention given to understory vegetation. In particular, 3D structural information of shrubs is limited or not included in fire behavior models. Yet, understory vegetation is an important component of forested ecosystems, and has an essential role in determining fire behavior.

In this dissertation, I explored the use of TLS data and quantitative structure models to model shrub architecture in three related studies. In the first study, I present a semi-automated methodology for reconstructing architecturally different shrubs from TLS LiDAR. By investigating shrubs with different architectures and point cloud densities, I showed that occlusion, shrub complexity, and shape greatly affect the accuracy of shrub models.

In my second study, I assessed the 3D architectural drivers of understory flammability by evaluating the use of architectural metrics derived from the TLS point cloud and 3D reconstructions of the shrubs. I focused on eight species common in the understory of the fire-prone longleaf pine forest ecosystem of the state of Florida, USA. I found a general tendency for each species to be associated with a unique combination of flammability and architectural traits. Novel shrub architectural traits were found to be complementary to the direct use of TLS data and improved flammability predictions.

The inherent complexity of shrub architecture and uncertainty in the TLS point cloud make scaling up from an individual shrub to a plot level a challenging task. Therefore, in my third study, I explored the effects of lidar uncertainty on vegetation parameter prediction accuracy. I developed a practical workflow to create synthetic forest stands with varying densities, which were subsequently scanned with simulated terrestrial lidar. This provided data sets quantitatively similar to those created by real-world LiDAR measurements, but with the advantage of exact knowledge of the forest plot parameters, The results showed that the lidar scan location had a large effect on prediction accuracy. Furthermore, occlusion is strongly related to the sampling density and plot complexity.

The results of this study illustrate the potential of non-destructive lidar approaches for quantifying shrub architectural traits. TLS, empirical quantitative structural models, and synthetic models provide valuable insights into shrub structure and fire behavior.

Download

Share

COinS