The vast majority of theses in this collection are open access and freely available. There are a small number of theses that have access restricted to the WWU campus. For off-campus access to a thesis labeled "Campus Only Access," please log in here with your WWU universal ID, or talk to your librarian about requesting the restricted thesis through interlibrary loan.

Date Permissions Signed


Date of Award

Spring 2022

Document Type

Masters Thesis

Department or Program Affiliation

Environmental Science

Degree Name

Master of Science (MS)


Environmental Sciences

First Advisor

Wallin, David O.

Second Advisor

Yang, Sylvia

Third Advisor

Bunn, Andrew Godard


Eelgrass is an important part of coastal environments in the Pacific Northwest as it provides crucial habitat and moderates storm surge. Padilla Bay, Washington is home to one of the largest eelgrass meadows in North America and contains the native Zostera marina and the non-native Z. japonica. The relationship between these two species is of interest due to the influx of Z. japonica in previously unvegetated areas. I used an uncrewed aerial system (UAS) and a multispectral camera to study species dynamics. I compared the ability of values derived from the band data, two vegetation indices, and a principal components analysis (PCA) to predict eelgrass cover using pixel- and object-based methods. In the pixel-based analysis, using a red and red edge band in a multiple linear regression was the best way to estimate overall percent cover for the full season (R2 = 0.79). Regressions used to predict Z. marina (R2=0.55) and Z. japonica (R2=0.32) cover individually performed poorly. In the object-based analysis, using band means and standard deviations in a support vector machine (SVM) classification resulted in an overall accuracy of 78.3%. This method performed the best at classifying segments based on dominant species, with user’s accuracies for Z. marina and Z. japonica of 80% and 100%, respectively. PCA-informed segmentation and classification also performed well, with an overall accuracy of 70.6%. Conducting object-based image segmentation with the Micasense Dual Camera System and SVM classification may be a promising method for identifying spectral differences between Z. marina and Z. japonica.




remote sensing, uncrewed aerial systems, multispectral imagery, support vector machine, image segmentation, eelgrass, Padilla Bay, drone, marine


Western Washington University

OCLC Number


Subject – LCSH

Eelgrass--Washington (State)--Padilla Bay--Identification; Remote sensing--Equipment and supplies

Geographic Coverage

Padilla Bay (Wash.)






Copying of this document in whole or in part is allowable only for scholarly purposes. It is understood, however, that any copying or publication of this document for commercial purposes, or for financial gain, shall not be allowed without the author’s written permission.