Klasifikasi Penutupan Lahan Hutan Mangrove di Kecamatan Buduran, Kabupaten Sidoarjo, Propinsi Jawa Timur, dengan Citra TerraSAR-X High Resolution

Abstract

Indonesia actually has used remote sensing imagery in monitoring natural resources especially the optical image since apllied remote sensing technology. Indonesia’s position in the tropics cause the area always cover with clouds in most of the year. This condition affects the ability of an optical image to produce a clear image. Active remote sensing systems (radio detecting and ranging/Radar) has provided an alternative to overcome the limitations of the information than can be taken through an optical image data. TerraSAR-X, an earth observation satellite owned by Germany which was launched on June 15, 2007 at Boikonur Cosmodrome, Kazakhstan, is the latest radar technology for mapping the active wavelength X-band (wavelength 31 mm, frquency 9,6 GHz), which stated able to overcome the cloud cover. Since its launch, satellite TerraSAR-X has been in full operation starting in January 2008. The purpose of this research was to test the ability of high resolution image of TerraSAR-X for land cover classification and identification of constituent species of mangrove forests. Imagery used in this reseach consisted of the main image TerraSAR-X with HH and VV polarization that has made the extraction process to obtain the composition of RGB |HH|-|VV|-|HH-VV|. While the other supporting image are Quickbird image, distribution mangrove map of Sidoarjo regency and Sidoarjo regency administration map. The method used for classification of land cover types to make a class is supervised classification method and unsupervised classification method by reference to the results of the visual image interpretation. Then with the best classification method used for land cover classification process of mangrove forests. By appying the stratified classification method, it is expected to be able to reduce errors during the classification process. This is because the other objects than mangrove forests have been eliminated. The results showed the accuracy of the supervised classification by the method of Support Vector Machine (SVM) in the image of TerraSAR-X is the best method of classification with the overall accuracy of 77,93 % and kappa accuracy by 58,85 %. The value of mangrove forests on the classification accuracy using SVM and applying the two stage classification amounted to 90,74 % and kappa accuracy by 84,30 %. RGB image of the arrangement [HH - VV - (HH-VV)] of the dual polarization HH and VV can produce a 3 class contituent species of mangroves, that is Avicennia alba, Rhizophora mucronata and Xylocarpus molucensis class, Hibiscus tiliaceus class and shrubs and grasses class.