Date of Award

5-2025

Document Type

Thesis

Degree Name

Master of Forest Resources (MFR)

Department

Forestry and Environmental Conservation

Committee Chair/Advisor

Nilesh Timilsina

Committee Member

Jessica A. Hartshorn

Committee Member

Federico Iuricich

Committee Member

Brook T. Russell

Abstract

Callery pear (Pyrus calleryana Decne. Rosaceae) is a rapidly spreading invasive species in the southeast United States (US) that hampers the function of native tree species and the forest ecosystem. Native to Southeast Asia, the species has encroached on the natural and secondary vegetation habitats of the Southern US in recent years. This study attempted to 1) map the Callery pear across the southeast US and 2) model the aboveground biomass (AGB) of individual Callery pear trees.

We used Callery pear locations and winter imageries of Sentinel-2, Sentinel-1, L-band radar, and terrain imagery for mapping purposes. In addition, we also derived features such as vegetation indices, texture, slope, and aspect in Google Earth Engine (GEE). We used random forest (RF) and support vector machine (SVM) classifiers with four scenarios to determine the usefulness of remote sensing data in classifying Callery pear. Radar images and derived indices were more important than multispectral images for discriminating Callery pear. The scenario with all the imageries and their derived indices was most accurate for RF (Accuracy = 92.6%) and SVM (89.6%). The final model detected Callery pear spread in the northeastern part of the Southeast US.

Similarly, we collected discrete laser points from an unmanned aerial vehicle (UAV) in South Carolina. We calculated the point cloud metrics, such as height, intensity, and crown diameter, to model the AGB of Callery trees. The individual Callery tree AGB ranged from 17.41 kg to 1098.48 kg ( = 184.99 kg), where point clouds overestimated tree height (%bias = -0.19, RMSE = 2.25 meters) and crown diameter (%bias = -1.56, RMSE = 3.81 meters) from field measurements. On average, point clouds overestimated the AGB by 411.34 kg (%bias = -5.71, RMSE = 897 kg). Our RF model identified tree height as the best predictor for estimating AGB from point clouds, while crown diameter was also one of the best predictors. Overall, this work demonstrates the positive role of satellite imagery in monitoring Callery pear. At the same time, the AGB findings contribute to the existing literature on point cloud-based AGB estimation for invasive species.

Available for download on Sunday, May 31, 2026

Included in

Life Sciences Commons

Share

COinS