| We have developed and examined numerical methods for fast and accurate forward modeling and inversion of multisource electromagnetic (EM) data based on the localized quasi-linear (LQL) approximation. The objective behind developing this technique is to improve the effectiveness and resolution of EM interpretation methods in structures with complex geology. A rigorous inversion scheme requires at least one forward modeling per frequency, a source, and a predicted model iteratively modified during an inversion process, and it is generally a time-consuming solution. The LQL method provides an efficient solution for this problem.; In contrast to the conventional linearized methods, this approach takes into account the nonlinear effects in the EM data.; The method incorporates both smooth regularized inversion, which generates a smooth image of the physical properties, and focusing regularized inversion, producing a sharper image of the geological targets.; The method is applied to the solution of three important geophysical problems that include sea-bed logging (SBL), cross-well EM tomography, and dipole-dipole array IP surveying.; The practical applications of the method to synthetic data demonstrate that (1) The SBL method has the ability to detect a sea-bottom reservoir in the presence of a salt dome structure. (2) Cross-well EM tomography can recover the resistivity, location, and shape of resistive and conductive rock formations. (3) Dipole-dipole array IP inversion has the ability to recover the conductivity and the chargeability of rock formations as well as the time and relaxation parameters of the corresponding Cole-Cole model parameters.; The results demonstrate that the method can be used as a practical tool for three-dimensional inversion of multisource array frequency domain data. |
