| dc.description.abstract | ASTER with its VNIR sensor have the capability for mapping vegetation. With the resolution of 15m x 15m, this image should be able to classify mangrove is more details than Landsat TM with resolution of 30m x 30m. Based on this, algorithms was applied to explore the capability of this image for classifying mangrove. These are Standard Back Propagation Neural Network and some vegetation indices (DVI, NDVI, SR, and the modified of these algorithm by using Natural Color Composite : NCC-DVI, NCC-NDVI, NCC-SR). While, as study area, Berau Delta, East Kalimantan is chosen because there are still may found many species of mangroves. Each algorithm show different results, but almost all of them capable to classify mangrove into zone level, except modification of vegetation index algorithms. Standard Back Propagation Neural Network is not quite good for classifying mangrove, because it can generate pixels into lower number of classes. While, DVI, NDVI, and SR can classify mangrove into 4 class. Its rather low than Maximum Likelihood Supervised classification and ISODATA Unsupervised classification which produced 5 class. But, when modified vegetation index applied, it a different results. They can detect Nypa fruticans class clearly than other species. Even, by using NCC-NDVI and NCC-SR class Nypa still can differentiate into detailer condition. The specific condition of the area of study where high frequency of rain, may cause the amount of water content in the vegetation become increase. Since gives influence in low value of NIR reflectance and high reflectance of RED channel, the modified algorithm of vegetation indices using NCC applied, and show result better than standard algorithm of vegetation indices. | id |
