Date of Award

2025

Degree Name

Environmental Science

College

College of Engineering and Computer Sciences

Type of Degree

M.S.

Document Type

Thesis

First Advisor

Dr. Mindy Armstead

Second Advisor

Dr. Scott Simonton

Third Advisor

Dr. Greg K. Michaelson

Abstract

The proliferation and frequency of harmful algal blooms (HABs) attributed to eutrophication, storm events and a changing climate have been an increasing concern in both lotic and lentic freshwater ecosystems. Methods of detecting HABs have been explored through fluorescent measurement and sample analysis, but are often expensive and time-consuming. A novel smart device application is in development to detect HABs based on images captured and analyzed by machine learning algorithms trained to distinguish potential cyanobacterial blooms. The HABs App model accurately detected cyanobacteria in strong relationship with biovolume concentrations (R2 = 0.996) within the Greenup Pool of the Ohio River, a system with many factors that may contribute to algal blooms. This accuracy was compared to weaker relationships observed using the bbe AlgaeTorch (R2 = 0.562) and an YSI EXO2 Multiparameter Sonde (R2 = 0.162). The mean ratio of cyanobacteria to green algae biovolume (0.336) was also compared with the EXO2 sonde (0.044), bbe AlgaeTorch (0.500) and HAB App model (0.200). With the EXO2 sonde, immobile and underestimating, and bbe AlgaeTorch, mobile but overestimating, further analysis of a machine learning-based method with low cost and user flexibility is warranted.

Subject(s)

Environmental sciences.

Ecology.

Biotic communities.

Fresh water.

Machine learning.

Cyanobacteria.

Diatoms.

Algal blooms.

Microcystis.

Ohio River.

Ohio.

West Virginia.

Download

Share

COinS