Improving Tree Crown Mapping using Airborne LiDAR with Genetic Algorithms

Landscape-scale mapping of individual trees derived from LiDAR (Light Detection And Ranging) data have been found to be valuable for a wide range of environmental analyses including carbon inventories; fuel estimations for wildfire risk assessment and management. These mapping efforts use individual tree crown (ITC) recognition algorithms applied to LiDAR point clouds, which have complex parameter sets. Genetic algorithms (GA) have been demonstrated to be excellent function optimizers for very complex search spaces and perform well for parameter tuning. Here, we use GAs to identify the best of a set of published ITC models and their optimal parameters for airborne LiDAR of forested plots in the Sierra Nevada Mountains of California. We assessed the accuracy of these ITC models in terms of the F-score and percentage bias for tree crown prediction. GA-optimization generally improved on ITC default parameters and showed that these models typically perform better for detecting overstory trees.

Level thesis:degree_level

Master's Degree

Year dc:date.issued

2023

Author dc:creator

• Onyegbula, Johanson

Advisor dc:contributor.advisor

• Greenberg, Jonathan

Committee members dc:contributor.committeemember

• Hanan, Erin
• Tavakkoli, Alireza

• Airborne LiDAR
• Genetic algorithms
• Individual Tree Crown Recognition
• Light Detection and Ranging
• Parameter tuning
• Tree crown mapping

Statement dc:rights

• Creative Commons Attribution-NonCommercial-ShareAlike 4.0 United States

Handle dc:identifier.uri

http://hdl.handle.net/11714/10518

OAI identifier oai:identifier

oai:scholarwolf.unr.edu:11714/10518