| dc.description.abstract | The availability of medium resolution satellite imagery (i.e. Sentinel-2A) provides
the rapid, low-cost and more accurate mapping. This report presents the use of satellite
imagery (Sentinel-2A) for mapping of marginal Agricultural Land in the eastern part of
Situbondo Regency. The study area covers three (3) districts, i.e., Arjasa, Jangkar, and
Asembagus. This study uses two methods of image classifications (i.e., unsupervised and
supervised). Sentinel-2A images for dry seasons of 2018 use for this study. The dry season of
this region usually occurs from April to November. Then, 450 ground control point for
training areas collected during the fields surveys between June until Octobre 2019. This study
also uses multi-band (i.e., 2,3,4,5 and 8A) of the sentinel 2a image. Image treatments use “
Multispect” and SNAP, two open-source image processing software. The procedures include
image enhancement, registration, clipping, and classification. The classification consists of preprocessing,
processing and post-processing tasks. Then, classification results evaluated by
confusion-matrix (overall and kappa accuracy). Furthermore, the thematic maps produce from
both unsupervised and supervised classification are then compared to existing thematics maps
and statistics data. The unsupervised method use iso-data algorithm and produce five (5) class
of land uses, i.e., (1) forestry and plantation; (2) build-up area, (3) irrigated paddy field, (4)
non-irrigated rural areas (ladang/tegalan). The unsupervised method did the overall accuracy =
79 % and kappa accuracy = 72%. The supervised methods use maximum-likelihood algorithms
and produce six (6) class, i.e., (1) forestry - plantation; (2) urban or build area, (3) irrigated
paddy field, (4) non-irrigated rural areas, (5) dry-marginal land and (6) water body. Supervised
method provide overall accuracy = 95,8% and kappa accuracy = 93,2%. The result shows the
potential use of Sentinel 2A to map dry-marginal agricultural land in the study area. | en_US |
