AUTHORS: Tyler Hessler, Missouri Cooperative Fish and Wildlife Research Unit, School of Natural Resources, University of Missouri; Craig Paukert, U.S. Geological Survey, Missouri Cooperative Fish and Wildlife Research Unit, School of Natural Resources, University of Missouri; Yi Shang, Department of Electrical Engineering and Computer Science, University of Missouri; Zhenduo Zhai, Department of Electrical Engineering and Computer Science, University of Missouri; Shiqi Wang, Department of Electrical Engineering and Computer Science, University of Missouri; Jason Persinger, Missouri Department of Conservation
ABSTRACT: The use of side-scan sonar has seen a dramatic increase in use in a variety of studies including those with the aim to characterize fish habitat. Side-scan sonar has been used to quantify habitat for well over a decade now and has seen promise in a variety of aquatic systems from coastal reefs to backwater bayous. The use of UAS (also known as aerial drones) has also seen more frequent use in mapping features in aquatic systems. In our study, we propose the use of side-scan sonar and UAS to better standardize and streamline the collection of fish-relevant habitat data in non-wadeable, mid-sized rivers. Traditionally, identifying habitat from side-scan sonar and aerial imagery was a manual process that could take several hours to days to complete a site. The amount of data that can be quickly accumulated in riverine systems can be daunting, but the use of AI and other software may streamline the implementation of these data and provide another option for a manager’s toolbelt that can increase confidence associated with making informed decisions that rely on accurate habitat data. To date, we have conducted 25 side-scan surveys across 8 sites across the state of Missouri totaling over 80 river kilometers. In addition, we have conducted 30 UAS surveys across these 8 sites as well as another site just used for UAS imagery. Developing a standardized protocol using side-scan sonar and UAS with the power of AI modeling could significantly increase the quality of data collected in mid-sized rivers and also decrease the costs associated with collecting habitat data and delineating imagery manually.
