Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


One Landsat Thematic Mapper (TM), and two Multispectral Scanner (MSS) data sets were digitally analyzed for forest stand and type mapping of the Kisatchie Ranger District, Kisatchie National Forest, Louisiana. Detailed ground-verification maps were produced from interpretation of 1:12,000 and 1:58,000 color-infrared (CIR) aerial photography of nine test compartments in the study area. Stand boundary and soils maps of the study area were obtained and input to the digital geographic information system along with the satellite and ground-verification data. Aerial uniformity and volume information were interpreted and combined with the cover-type data, which included the following categories: open areas, longleaf-slash pines, loblolly-shortleaf pines, and hardwoods. Unsupervised classifications of the three Landsat data sets resulted in poor results for identification of the above classes. Supervised classifications were tested by both training fields and stand agreement to the ground-verification. The highest four-class overall agreement (by stand) was obtained for the TM classification (76 percent). Three-class stand agreements (open, pine, and hardwoods) for the supervised classifications of the three data sets were insignificantly different as tested by ANOV (alpha level 0.1). The three-class agreement for all classifications ranged from 81 to 85 percent. Soils data were determined to be important for modification of single-date classification results. The main conclusions were that TM data were superior to the single- and two-date MSS classifications for detailed forest mapping. However, the supervised TM classification was not significantly different from the MSS classifications for identification of pines and hardwoods. Two-date TM analysis and additional data processing cost analysis were suggested for future research.