Distinguishing open land areas from forest using single ERS SAR images

Abstract

Research of the usefulness of SAR images for forest applications has only recently received more attention. Good results have been obtained both with multitemporal analysis and interferometry. If the area of interest is very large, the double number of images needed for multitemporal approach or interferometry may become a problem. Also the time interval needed to obtain two suitable images may be a drawback. Therefore a forest/non-forest classification method using single SAR images has also potential, although it is clear that additional information brought by double imagery should increase reliability. Single ERS SAR images can be used for forest/non-forest classification of boreal forests in winter and up to the beginning of the growth of vegetation in the agricultural fields. Very wet weather conditions will probably disable the usage of a SAR image for classification.