Visual interpretation of vegetation through digitally enhanced LANDSAT‐MSS images

Publisher Website
Google Scholar
Abstract
Forest or wildland vegetation classification is one of the most challenging applications of LANDSAT‐MSS data, because of the common heterogeneity of this cover type and the numerous factors affecting its spectral response. Vegetation classes are difficult to define on the basis of their spectral characteristics alone, they are better characterized by a spatially distributed pattern of spectral responses. Studies having indicated that spectral data analysis is a limited tool for classifying vegetation, in 1978 the Laurentian Forest Research Centre (LFRC), facing the problem of choosing an operational methodology to produce vegetation mapping over large areas, adopted a method that takes advantage of human interpretation and computer capabilities: the interpretation of digital enhancements based on principal component analysis. The main steps of the enhancement methodology are given, from the delineation of training areas for statistic generation to various histogram stretches of the output components mapped into a brightness, red‐green, blue‐yellow colour space. After reviewing applications of LANDSAT‐MSS imagery through digital classification, interpretation of photographic products and various digital enhancements, the paper describes vegetation mapping projects of boreal forests and subartic environments in Quebec. Those projects illustrate the operational aspect of the methodology for mapping forest vegetation over broad territories. In forested land, the output maps located hardwoods, mixedwoods, three density classes in softwood stands and various types of regeneration after cut‐overs and burns. It is often possible to distinguish two stand age classes on the basis of patterns originating from perturbated areas. Visual interpretation of principle component enhanced LANDSAT‐MSS imagery has proven to be an effective way to map wildland vegetation in Quebec at scales 1:125 000 and smaller. Taking advantage of available satellite images is considered to have been a good opportunity to rapidly acquire information on remote areas. At present, the principal limitation to a more generalized application of satellite remote sensing remains the restricted number of qualitatively acceptable images taken at appropriate dates, particularly for multidate studies. Nevertheless the future of this new technology is promising. However, to be cost‐effective, it should be able to replace traditional aerial photographs in many applications; it should become more than the well‐known “additional tool.”

This publication has 8 references indexed in Scilit: