Application of Kompsat-2 image to estimate distribution of seagrass beds at Deukryang Bay, Korea

Abstract

Despite the ecological importance of segrass beds, their distributional data in Koreancoastal waters is still limited. Here, we used hydroacoustic system for collecting accurateground truth data, and Kompsat-2 (4 m spatial resolution) image for detection of seagrassbeds. The area of seagrass beds from satellite image classification was underestimatedcompared to hydroacoustic survey result. Nonetheless, the accuracy of Kompsat-2 imageclassification based on the field reference data showing 90% (Kappa = 0.85) for the threeclass map (seagrass, unvegetated seawater and aquaculture). The agreement between thesatellite image classification and the hydroacoustic result was 77.1% for the seagrasspresence/absence map. Our results showed that the Kompsat-2 image is desirable forseagrass mapping with resonable accuracy.mage for detection of seagrassbeds. The area of seagrass beds from satellite image classification was underestimatedcompared to hydroacoustic survey result. Nonetheless, the accuracy of Kompsat-2 imageclassification based on the field reference data showing 90% (Kappa = 0.85) for the threeclass map (seagrass, unvegetated seawater and aquaculture). The agreement between thesatellite image classification and the hydroacoustic result was 77.1% for the seagrasspresence/absence map. Our results showed that the Kompsat-2 image is desirable forseagrass mapping with resonable accuracy.