You are here:
EFFECTS OF GREEN MACROALGAE ON CLASSIFICATION OF SEAGRASS IN SIDE SCAN SONAR IMAGERY
Citation:
DEWITT, T. H., P. LATTIN, AND S. STRICKLAND. EFFECTS OF GREEN MACROALGAE ON CLASSIFICATION OF SEAGRASS IN SIDE SCAN SONAR IMAGERY. Presented at Pacific Estuarine Research Society annual meeting, Victoria, BC, CANADA, February 22 - 24, 2007.
Impact/Purpose:
Presentation
Description:
High resolution maps of seagrass beds are useful for monitoring estuarine condition, managing fish habitats, and modeling estuarine processes. Side scan sonar (SSS) is one method for producing spatially accurate seagrass maps, although it has not been used widely. Our team recently developed image analysis methods for automatically classifying SSS imagery for seagrass presence, which produce seagrass maps with high thematic accuracy. However, seafloor objects with high acoustic reflection (such as bubbles trapped under macroalgae) could be mistaken for seagrass in SSS images. In this study, we investigated whether mats of green macroalgae on tide flats and in seagrass beds would interfere with the accuracy of SSS-based seagrass maps. Two 16-ha sites in Yaquina estuary (OR) were each surveyed in April (low macroalgae) and September (high macroalgae) using SSS and underwater video (UV). UV images were classified for presence of seagrass and macroalgae. SSS imagery was processed based on a supervised maximum likelihood classification of SSS focal mean brightness, focal standard deviation of brightness, and depth class, with signature-development training sites determined from georeferenced UV data. Preliminary analysis of the SSS imagery suggests that dense mats of green macroalgae reduced the thematic accuracy of the seagrass maps, but did not affect the estimation of total seagrass area.