“…While several stabilizing functions have been successfully used (least squares, minimum norm of the gradient, distance from an a priori model, etc.) in various geophysical applications (Smith et al, 1999, Mehanee and Zhdanov, 2002, de Groot-Hedlin and Constable, 2004, many magnetotelluric algorithms employ the minimum norm of the Laplacian, which will seek the smoothest model possible that fits the data (for a given misfit). The argument is that by seeking the smoothest model (again, for a sufficiently large misfit), only structure explicitly required by the data will be resolved (de Groot-Hedlin and Constable, 1990).…”