Classification of Land Cover from Sentinel-2 Imagery Using Supervised Classification Technique (Preliminary Study)

This paper intended to classify land cover of high-resolution satellite image using supervised classification method. The object of this research was the land cover image of the central Java area in Indonesia which chosen as an area observed with the cloud cover consideration. The Satellite image was obtained from Sentinel-2 through https://earthexplorer.usgs.gov. Supervised Classification and ArcMap 10.5 was used to classified image object. This research classified the land cover into four classifications class namely Water, Forest, Urban and Bare Land due to show the main classes in the Land Cover Classification of RSNI-l National Standardization Body of Indonesia, RSNI-l was used as reference for classification process. The confusion matrix used to calculate classification accuracy value. Subsequently, confusion matrix results compared to the truth information. The truth information was derived from the actual value of RASTERVALUE obtained from Google Earth. This research shown supervised classification maximum-likelihood classification has been classified land cover into four class (forest, water body, urban and bare land) with overall accuracy = 1 and kappa value = 0.4896. This result showed this research got moderate kappa accuracy value but high overall accuracy value. High accuracy value reached due to fully supervised experiment during classification process.