Preliminary Study of Information Extraction of LANDSAT TM Data for a Suburban/regional Test Site

A substantial amount of spectral information is available from TM (as compared to MSS) data for a 14.25 square km area between Beltsville and Laurel, Maryland. Large buildings and street patterns were resolved in the TM imagery. While there was added information content in TM data for discriminating surburban/regional land cover, characteristics of MSS can improve land cover discrimination over TM when conventional classification procedures are used on digital data. The improved qualitization of TM is likely valuable in situations where there are spectral similarities between classes. The spatial resolution in TM decreased land cover discrimination as a result of increased within class variability. For many general digital evaluations, inclusion of four bands representing the four spectral regions can provide much useful land cover discrimination. Inclusion of TM 6 indicates an improvement in spectral class discrimination. Of primary spectral importance is the discrimination between water, vegetative surfaces, and impervious surfaces due to differences in thermal properties. Results from the principle component transformed data clearly indicates additional information content in TM over MSS.