On: “An Automatic Least‐Squares Multimodel Method for Magnetic Interpretation” by Peter H. McGrath and Peter J. Hood (GEOPHYSICS, April 1973, p. 349–358)

Abstract

McGrath and Hood present a magnetic interpretation method whereby the search for a solution is carried out in the (hyper) space of n parameters defining the shape and position of an assumed model. The problem is an optimization problem and should be viewed within the general context of nonlinear optimization techniques. McGrath and Hood simply present one optimization method. The usefulness of individual methods is limited. One could similarly propose the use of the method of rotating coordinates (Rosenbrock, 1960), the “complex” method (Box, 1965), Davidon’s methods (Fletcher and Powell, 1963; Stewart, 1967; Davidon, 1969), etc. We currently have a wealth of these methods at our disposal. In fact, the use of these methods for magnetic interpretation has already been presented (Al‐Chalabi, 1970). As this and subsequent work indicated (Al‐Chalabi, 1972), these methods should be used as an integral group for interpreting magnetic and gravity anomalies. The exclusive use of individual methods is inefficient. Studies performed on objective functions used in magnetic and gravity interpretation have shown that the behavior of these functions in the parameter hyperspace is extremely complicated. Consequently, the search for a solution requires different strategies at different stages between the initial estimate and the ultimate solution (Al‐Chalabi, 1970, 1972).