Monitoring bull kelp (Nereocystis leutkeana) remotely at high resolutions

Bull kelp (Nereocystis leutkeana) is a foundational seaweed in nearshore, marine systems, traditionally supporting economically important fisheries, extracting nutrients in its vicinity, and sequestering carbon and transporting it great distances to subsidize shallow and deep-sea environments. Bull kelp is also a culturally important species for some indigenous peoples along the northeastern Pacific coastline and provides a harvestable product with growing demand. Global trends in kelp distribution show patterns of decline, and the immense loss of northern California bull kelp forests in response to climatic changes have highlighted the need for alternative sources of kelp, both for the product itself, as well as the environmental benefits it provides. Seaweed aquaculture has been a growing industry nationwide and Humboldt Bay has two of the first open-water, commercial kelp farms in California growing bull kelp. In both natural and farmed settings, the need for consistent, accurate monitoring is critical for timely management interventions when populations are observed to be declining, and to reduce long-term costs for California farmers, who already have immense permitting expenses to overcome. Fortunately, low-cost and high-resolution remote monitoring technologies are available through unoccupied aircraft vehicles (or UAVs). The goal of this project was to evaluate the use of UAVs to monitor the growth of bull kelp in farms in Humboldt Bay, and nearby, wild populations along the coast in Trinidad. The second chapter also aims to use the National Agricultural Imaging Program (NAIP) imagery to detect kelp at high resolutions and, combined with recent surveys from UAV, build a time-series of kelp change through the past nine years. I tested various classification methodologies and identified some to be relatively successful in farm settings and others better suited for coastal settings. In terms of accuracy, the classifications were deemed reliable in detecting the presence of bull kelp but varying tidal heights across surveys presented a confounding factor in kelp measured at both the farm and the coast. Generally, detecting kelp from NAIP imagery was found to be more accurate than from drone imagery (91% versus 86% overall accuracy, respectively) in Trinidad. Findings from Chapter 2 indicate that the kelp detection accuracy may be improved with higher altitude UAV surveys, but overall, the results showed success in the usage of UAV and NAIP imagery to monitor the presence and distribution of bull kelp