Statistical-Dimensional Analysis: An Application to the Assessment of Crater Configuration

Abstract : Definition of crater related environments is influenced by a large number of variables potentially governing the phenomena. The large body of data is derived from many events of different character. A method to exploit this heterogeneous body for prediction is presented. Methods of dimensional analysis are used to reduce the number of independent variables through the introduction of nondimensional pi-groups. More importantly, this allows a common basis for comparing data from grossly dissimilar data bases. These include field events having different soil, stratigraphy and moisture content, and different explosive type, size and configuration, as well as laboratory data in accelerated reference frames. Successive combination of nondimensional variables is then accomplished using a step-wise regression analysis to generate scaling laws which provide relationships among the variables. Because all the various data are compared on a common basis, the statistics resulting from a large base provide higher confidence in predictions as the quantitive uncertainties are less than for each data base considered separately. Correlation of apparent crater volume was used as a pilot model to demonstrate the technique.