2020 Volume 10 Pages 53-66
We tested the applicability of SAVMA (submerged aquatic vegetation mapping algorithm) method to a WorldView-2 image in Lake Akan, eastern part of Hokkaido, in order to monitor the distribution of submerged aquatic vegetation (SAV) in the lake. Field investigation was carried out at Shurikomabetsu Bay in Lake Akan on August 25, 2010 in order to obtain the actual distribution of the SAV using a side-scan sonar. The in-situ SAV map was then used to validate the estimated SAV distribution by the SAVMA method using a WorldView-2 image on August 18, 2010. Result showed that the use of green band gave the highest accuracy in all bands (κ=0.81). In addition, the additional estimation using the combination between green and yellow bands, and that between green and red bands resulted in the increase of the accuracy (κ=0.84). Although these results indicate that green band is the most suitable band for detecting SAV in lake areas, the accuracy depends on the selection of the DN value for optically-deep water (DNs).