Anderson, Connor2024-02-092024-02-092023-06https://hdl.handle.net/11299/260636University of Minnesota Ph.D. dissertation. June 2023. Major: Natural Resources Science and Management. Advisor: Joseph Knight. 1 computer file (PDF); xvii, 176 pages.Surveying methods for invasive plant species need modernization. Cataloging the exact location of invasive plant populations has traditionally focused on in situ monitoring. Such methods are often hindered by resource limitations, access to private property, and physical inaccessibility of remote locations. Surveying for invasive Phragmites australis is particularly difficult due to the limited accessibility in the wetland environments it invades. Remotely sensed data offers the ability to detect invasive Phragmites australis without the need for extensive physical mapping. Uncrewed aircraft systems (UAS) are a popular tool when surveilling for invasive plant species due to their ability to image challenging environments and their high spatial and temporal resolution. Additionally, UAS are unique in that the imagery can be used to create 3D models of the Earth’s surface.This dissertation demonstrates the efficacy of UAS for the identification of invasive Phragmites australis in multiple Minnesota and Michigan wetlands. The primary goal is to provide resource specialists managing for invasive Phragmites australis the information needed to implement UAS for Phragmites australis detection. Three major components are included within this dissertation. First, an object-based image analysis workflow was developed to classify Phragmites australis within Minnesota and Michigan wetlands from three-band (i.e., red, green, blue; RGB) UAS imagery. Second, a study investigating the ability of different machine learning algorithms to identify Phragmites australis within Minnesota wetlands is presented. Methods for improving machine learning classifications with object-based techniques are also described. The final study investigated the ability of five-band (i.e., red, green, blue, red edge, and near-infrared) multispectral UAS imagery to identify Phragmites australis. Voting-based ensemble classifiers were employed to classify Phragmites australis within five Minnesota wetlands. Classifications using the multispectral UAS imagery were compared to classifications using RGB UAS imagery. This research provides critically needed information on the data and methodology required to accurately identify Phragmites australis using UAS.enThesis or Dissertation