This study presents change detection analysis using object-based image analysis in Dong Ra Nang national forest, Kalasin province. This national forest was previously cleared and used for agricultural purposes for an extended period of time according to media reports published between 2007-2015. In order to perform a LU/LC examination, a time series of LANDSAT 7 ETM+ images acquired from 3 dates (25/11/1999, 28/11/2007 and 21/11/2015) were used for image segmentation and LU/LC classification. Furthermore, a CART algorithm and crucial image band ratios, such as MSAVI and NDVI, including mean of image layer bands, were used to improve image classification of degradation of the forest. The information from three thematic layers sampling points that had been derived from visual interpretations was used for CART training, applying and classifying the satellite images into 6 LU/LC classes, namely, (1) dense forest, (2) light forest, (3) bare land, (4) agricultural area, (5) plantation area, and (6) bodies of water on hierarchical image networks. Prior to the deforestation detection analysis, each image scene was classified individually using a CART algorithm. Then, the classified images were synchronized with the main map for performing land use/ land cover changes analysis focused on deforestation using image hierarchical image network by relation to image objects in vertical and horizontal aspects. The results indicated that the forest areas decreased dramatically by 50% from 1999-2007. On the other hand, there was a slight increase in bare land by an area of 38.68 sq.km. The majority of the area was used for farm land according to the report of the Forest Management Office, Khon Kaen province. The vegetation area emerged in the central area surrounding by bare land and agricultural area. In 2007-2015, the vegetation rapidly decreased by 30.89 sq.km., and the area tended to be bare land and agricultural area.
Keywords: OBIA; remote sensing; classification and regression tree; CART
*Corresponding author: Tel.: (+66) 85-5503-666 Fax: (+66) 43-3547-59
E-mail: sopholwit.c@msu.ac.th
