Abstract:
The purpose of this study is for subpixel classification with Landsat 7 imagery for small opium field study. Nowadays there is still a continuous trend of illegal opium cultivation and trading in Thailand. Office of the Narcotics Control Board assigns its Northern branch to survey opium field for crop eradication. Satellite imagery has been employed to assist in increasing opium cultivation survey efficiency. This study covers the area around Amphur Maejam, Chiangmai province, one of the northern provinces, that have several scatter opium cultivation fields. Currently the cultivators have employed up-to-date technology for cultivation, for example, there is a method of divide opium field into small area or planting small opium fields to blend in with other vegetations, or planting late season opium cultivation by using suitable watering technique to avoid the eradication from the authority. This study employs Landsat 7 satellite imagery recorded on February 10, 2003 Path-Row 141-047. Classifying opium field with subpixel classification by creating opium field pure signature to represent the material of interest. Additionally there is a background removal process, environmental correction, signature derivation and MOI classification. Finally all the above mentioned variables for opium field classification must be employed. The ending result will be shown as interested material ratio together with attribute table. This study is also distinguished the supervised classification result from the subpixel classification which in turn shown that the subpixel classification for the opium field is capable of more accuracy detection, which is about 89%, than the supervise classification which is about 72%. However, subpixel classification can practically be deployed to gain efficiency in small opium field detection, there is also a limitation of this method about facing the commission error due to mixed data up to 48%.