Study On Practicable Methods For The Interpretation Of FAEM Data

With careful studying of different inversion methods of AEM data, some helpfulalgorithms are introduced to frequency-domain AEM data processing, whichmarkedly promotes the interpretation skills and enhance the ability in extractingpotential information from field data. Several examples of inversion results based ontheoretical data and survey data are also discussed.The 1D forward modeling techniques are based on well-known formulae. Sincethe integrating factors of the equations include ether zeroth or first order Besselfunctions, a approximate calculating algorithm called Fast Hankel Transform is used,FHT has good balance in the computing precision and speed, and lays a good base to1D inversing programs.With a phasor diagram of normalized in-phase and quadrature response for thehalf-space model of different conductivity and flying height, a quick approximation ofconductivity inversing algorithm is introduced. Also a inversing method called theApparent Resistivity-Centroid Depth algorithm is offered, and it makes fastapproximate inverse modeling of FAEM data possible.The Marquardt's method is also well discussed in this article, which uses aMarquardt's technique in the singular value decomposition of the Jacobian matrix.This algorithm is both simple and time saving, and mostly suitable for a model of lessthan 3 layers.the Occam's method is a algorithm quite different from the Marquardt's. Itconcerns about not only the RMS misfit but also the smoothness of the model, anduses a line search algorithm with the Lagrange multiplier to finding the target modelthat mostly close to the extremum, so it can steadily converge with wide initial value.The Occam's algorithm is mostly suitable for a model with 10～30 layers.