Numerical Designation and Multivariate Discrimination of Drainage Pattern for the Geological Interpretation of Aerial Photoraph

Abstract

Image processing and automatic analysis are the latest trends in geological interpretation of aerial photograph, with the development of remote sensing technique. But the results are not always satisfied, because the quantitative relations connecting the imagery to geology have been left unsolved, which give the computer some numerical standards of judgement.
The use of drainage pattern, one of the most important factors obtained from imagery, should be acceptable at first in order to systematizing these standards.
Some drainage areas of different lithofacies were selected here, and stream-order was distinguished and put down on the topographic map in each area. And each area was divided into small area (about 250×250m in section). Then some stream varialbes such as number, length, area and inclination of stream were measured in the same stream-order in each section on the map, and each lithofacies could be represented numerically as the multidimensional center of grabity with statistical values as mean vector of stream variables.
Discriminant analysis with nonlinear discriminant function in the basis of Maharanobis' generalized distance was used here to distinguish some lithofacies.
As the result, each lithofacies selected here was statistically discriminated by combining some stream variables of the same order, and more over some differences estimated to be affected by weathering and geological structure were recognized in case of similar facies.
Then we could have some prospect to systematize the geological interpretation with the quantitative relation between geology and drainage pattern by accumulating these kind of data and adding the investigation for genetic relation of them