Date of Award

5-2015

Document Type

Thesis

Department

Environmental Resources Engineering

Thesis Advisor

Lindi J. Quackenbush

Keywords

Purple loosestrife, Japanese knotweed, phragmites

Abstract

As the world becomes increasingly connected, the risk of species spreading across previously impermeable boundaries becomes much higher. Many plant and animal species are establishing themselves in entirely new environments, often with significant consequences for indigenous species. This study focused on three non-native plant species-Lythrum salicaria (purple loosestrife), Fallopiajaponica (Japanese knotweed), and Phragmites australis (phragmites or common reed)-- invasives abundant in Onondaga County. This study aimed to test the feasibility of using readily available oblique imagery to identify the presence of these invasive species. Imagery used in the study came from Google Maps Street View photographs of addresses reported to have dense populations of phragmites, Japanese knotweed, or purple loosestrife. Images were uploaded to ERDAS Imagine 2015 and extraction of plant species was tested using both pixel- and object-based classification approaches. Results largely showed these images were not suitable for positive species identification. The oblique images used for testing varied in spatial resolution and time of capture, which may have impacted success of the classifiers. For example, purple loosestrife is more easily separated from its surroundings when blooming; but the available images did not allow us to capitalize on such phenological distinctions. Future research requires a more targeted data set with confirmed species identification, ground validation, and more consistent image quality. Identifying the presence of invasive plant species within imagery is suggested as an important step toward invasive species mitigation efforts.

Share

COinS