Integration of gis technology, remote sensing and multivariate analysis in the delimitation of physiographic units for pedological mapping

Abstract

This study was based on an association of techniques and equipment that could lead to the delimitation of physiographic units not merely restricted to the photointerpreter's criteria. A data matrix was evaluated which was obtained from the crossing, by means of Geographic Information System (GIS), of a regular grid with cells 300 x 300 meters and a drainage base map of the study area and bands 3, 4, 5 and 7 from TM-LANDSAT and a slope map generated in a GIS. The variables collected in each cell of the regular grid were: long, medium and short channels from the first to the seventh order, weighted slope and the grey levels (DNs) in bands, resulting in a matrix with 1167 individuals with 27 variables, whose data were subjected to cluster analysis. This analysis provided dendrograms which enabled the separation of groups of individuals represented by cells from the regular grid, aiming at creating pictograms to serve as guides. These guides overlaid upon properly treated and handled orbital HRV-SPOT images oriented the discrimination of physiographic units on the color monitor of the image processing system. Among the variables used, those referring to drainage network were best for cluster analysis of regular grid cells. The variables relief and grey levels of TM-LANDSAT image also presented satisfactory results with the methodology applied.