Data Series 901
 Marquardt InversionsA Marquardt inversion is a damped, leastsquares ridgeregression algorithm used to obtain conductivity versus depth from electromagnetic sounding data (see fig. 5). It requires a starting model that has equal to or less than the number of layers that can reasonably be observed in a Fischer inversion. The algorithm modifies the resistivities and thicknesses of the specified number of layers until it minimizes the leastsquares error between the model response and the observed data. It is very sensitive to the starting (initial) model and commonly will not work if one attempts to invert for more layers than are resolved by the data (Marquardt, 1963; Pujol, 2007). All Marquardt Inversions were done for the TE = XY orientation. If there was apparent anisotropy, the inversion was also done for TE = YX.
