Forest cover for 7.25 million acres (2.93 million hectares) in southeastern Georgia was characterized for the years 1988 and 1994 with the intent of assessing the efficacy of remote sensing procedures for broad scale forest inventory. Landsat-5 Thematic Mapper digital satellite scenes of seven spectral bands were obtained for winter and summer of each year and were analyzed two separate 14-band multi-temporal images. Images were geo-referenced to the universal transverse mercator (UTM) coordinate system prior to classification. Spectral classification with the 1SOCLUSTER algorithm produced 250 categories. Color infrared aerial photographs were mapped to the digital imagery and were used to convert spectral categories to land cover features. For this study, land features of interest were limited to water, marsh, pine forest, hardwood forest, mixed pine/hardwood forest, urban, and where distinguishable, clearcut and agriculture. Accuracy assessment techniques indicated very good consistency.
Bennett, Chris W. and Weih, Robert C. Jr.
"Methodology for Integrating Aerial Photography and LANDSAT TM Imagery for Inventory of Forest Land Cover,"
Journal of the Arkansas Academy of Science: Vol. 51
, Article 6.
Available at: http://scholarworks.uark.edu/jaas/vol51/iss1/6