Examination of multivariate dependency structure in soil parameters

Abstract

A typical site investigation involves sampling for laboratory tests and a variety of field tests. In other words, it is common to measure more than two soil parameters. The literature is replete with correlation equations between two soil parameters. Consistent synthesis of more than two soil parameters through construction of a multivariate probability distribution function is rare, despite obvious practical usefulness of such an approach. This paper examines the feasibility of modifying the well known multivariate normal distribution to model the dependency structure of soil parameters in two databases. © 2012 American Society of Civil Engineers.