Mapping detailed seagrass habitats using satellite imagery

Essentially, maintenance of adequate seagrass cover is intimately related to coastal ecosystem health and thus monitoring of seagrass habitats is a priority of coastal managers. Remote sensing techniques, especially satellite remote sensing, can provide seagrass habitat information spatially and temporally. In this study, we propose to evaluate and compare the capability of four satellite sensors' (Landsat TM, EO-1 ALI and Hyperion and IKONOS) data for mapping detailed seagrass habitats. After depth-invariant bands were created from the four sensors' data, a maximum likelihood classifier was used to classify the submerged aquatic vegetation (SAV) cover percentage into 3 classes and 5 classes in the study area. The SAV mapping results indicate that Hyperion sensor has produced the best mapping results of seagrass habitats in the two classification schemes: 3-class (Overall accuracy (OAA) = 96%, Kappa = 0.936) and 5-class (OAA = 79%, Kappa = 0.730). ALI outperformed TM for mapping SAV due to its additional blue band.