AUGMENTING TIDAL WETLAND VEGETATION AND ELEVATION MONITORING USING UNOCCUPIED AERIAL SYSTEMS (UAS)

Abstract

Tidal wetlands provide numerous ecosystem services that promote coastal resilience. Live vegetation coverage and elevation are key metrics for assessing the health of these crucial systems. However, traditional monitoring methods can be resource- intensive, intrusive, and lack adequate spatial and temporal resolution. This thesis explores the immense potential of Unoccupied Aerial Systems (UAS, or “drones”) for expanding coastal monitoring capabilities. Chapter 1 compared two tree-based classifiers against in situ observations for estimating live vegetation percent cover. While agreement with field observations varied among both model types, random forest models proved to be more robust than simple thresholding decision stump models when applied to validation data. Chapter 2 evaluated the accuracy of Digital Surface Models (DSMs) generated from drone imagery with Structure-from-Motion, and the influence of vegetation presence on vertical error. While vegetation presence significantly increased vertical error rates, it did not explain all differences in elevation model accuracy across sites. Together, this work underscores the role drones can play in connecting researchers and management practitioners with meaningful data to drive decision- making.