Abstract
By use of a simple function to be known as a Natural Classification Function, multispectral signatures of 13910 LANDSAT pixels were classified into categories each of which when plotted on a character map, was found to correspond with a definable ground feature. Classification of the multispectral signatures was done in two stages. In the first stage, 65 categories were generated on the basis of the Natural Classification Function in which the independent variables are the reflectance values of multispectral signatures of pixels. In the second stage the 65 categories were merged into seven realistic categories on the basis of the variance of pixels within each category. The ground features represented on the character map are six plant associations and open bodies of water.
At a cost of less than $50 for analysis of data and generating character maps for about 20 square miles, the method is considered to be a good tool for rapid exploration using LANDSAT Multispectral Data.
Date of this Version
1976