| dc.description.abstract | A wide variety of data and information on forest cover in Indonesia may be due to the variety of source of data, date of acquisition, and methods applied. For a wide area, terrestrial inventory methods are usually costly and time consuming. One alternative that may be used to minimize the cost and time is satellite based remote sensing technology. In the tropical country such Indonesia, cloud, fog, and smoke mainly limit the use of optical remote sensing during identification process and object monitoring on earth surface. Objects under the cloud, fog, and smoke could be identified using using Radar images. ALOS is remote sensing satellite which launched by Japan in 2006. One of its censor is PALSAR (Phased-Array type L-band Synthetic Aperture Radar). PALSAR is an advanced development from SAR which carried by the former satellite JERS-1. This is the microactive wave censor which can observe day and night without weather influence. Through one observation mode that is Scan SAR, this censor can observe earth surface in wide area than the conventional SAR. The objective of this study is to evaluate the ability of low resolution ALOS PALSAR image to classify regional scale land cover in Kalimantan Island. ALOS PALSAR image have 200 x 200 m resolution acquired in 2007. Other supporting data used are Land Cover Map year 2003, Administration Border Map, Forest Area Map, and the Base Thematic Forestry Map. The data were analyzed using GIS 3.2 and Erdas Imagine 9.1. The method are consisted of image preprocessing, image processing, separability accuracy evaluation and spatial analysis The study shows that low resolution ALOS PALSAR image could classify land cover into six classes. There are water body, rice field, shrub/bush, estate crop, and forest. Separability analysis for these classes show that there are 2 unseparable class pairs. These classes were then reclassified into four classes. The new classes are water body, sparse vegetation, medium density vegetation, and high density vegetation. The result of separability analysis shows that the these class separabilities are good (well separated). The accuracy of the classification are 88,21% for Overall Accuracy and 85,26% for Kappa Accuracy. Based on ALOS low resolution images (200 m x 200 m spatial resolution, the acreages of each land cover are 11.459.400 hectares (21,33%) for sparse vegetation, 5.070.008 hectares (9,44%) for medium density vegetation, and 36.806.058 hectares (68,52%) for high density vegetation. While the acreages of each land cover based on Land Cover Map year 2003 are 802.233 hectares (1,51%) for sparse vegetation, 20.841.843 hectares (39,32%) for medium density vegetation, 27.583.553 hectares (52,04%) for high density vegetation and 2.457.825 hectares (4,64%) for smoke. Key words : ALOS PALSAR Image, Land cover, Separability, Radar images | id |
