Satellite-based seagrass mapping in Korean coastal waters

Abstract

Seagrass beds provide habitat for invertebrate and fish species, many of which are economically important. In addition, they perform important physical functions such as trapping sediment particulates associated with dissipating wave energy, thus are helpful to maintain clear waters. We, here, generated the map of seagrass distribution using remotely sensed images to which atmospheric corrections and water column corrections had been applied. Then, the seagrass habitat distribution changes were calculated by seagrass habitat map. For this study, we selected Deukryang Bay located on the southern coast of the Korean peninsula. It is surrounded by small villages like Jinmok-ri and Ongam-ri. Zostera marina dominated at the bay, small amounts of Z. caulescens and Halophila nipponica are also distributed in this area. The results showed that image classifications to which the water column correction had been applied produced improved accuracies in all the classification algorithms we had employed. The object-based classification algorithm showed the highest accuracy, but it is effective method for the high spatial resolution remotely sensed images, consequently not suitable for monitoring changes of the long-term base. Thus, we applied the Mahalanobis distance method which had been known to suitable for medium spatial resolution images like Landsat. This study revealed that seagrass beds in the study area showed similar pa, thus are helpful to maintain clear waters. We, here, generated the map of seagrass distribution using remotely sensed images to which atmospheric corrections and water column corrections had been applied. Then, the seagrass habitat distribution changes were calculated by seagrass habitat map. For this study, we selected Deukryang Bay located on the southern coast of the Korean peninsula. It is surrounded by small villages like Jinmok-ri and Ongam-ri. Zostera marina dominated at the bay, small amounts of Z. caulescens and Halophila nipponica are also distributed in this area. The results showed that image classifications to which the water column correction had been applied produced improved accuracies in all the classification algorithms we had employed. The object-based classification algorithm showed the highest accuracy, but it is effective method for the high spatial resolution remotely sensed images, consequently not suitable for monitoring changes of the long-term base. Thus, we applied the Mahalanobis distance method which had been known to suitable for medium spatial resolution images like Landsat. This study revealed that seagrass beds in the study area showed similar pa