Most airborne electromagnetic (AEM) data are processed using successive 1D approximations to produce stitched conductivity-depth sections. Because the current induced in the near surface by an AEM system preferentially circulates at some radial distance from a horizontal loop transmitter (sometimes called the footprint), the section plotted directly below a concentric transmitter-receiver system actually arises from currents induced in the vicinity rather than directly underneath. Detection of paleochannels as conduits for groundwater flow is a common geophysical exploration goal, where locally 2D approximations may be valid for an extinct riverbed or filled valley. Separate from effects of salinity, these paleochannels may be conductive if clay filled or resistive if sand filled and incised into a clay host. Because of the wide system footprint, using stitched 1D approximations or inversions may lead to misleading conductivity-depth images or sections. Near abrupt edges of an extensive conductive layer, the lateral falloff in AEM amplitudes tends to produce a drooping tail in a conductivity section, sometimes coupled with alocal peak where the AEM system is maximally coupled to currents constrained to flow near the conductor edge. Once the width of a conductive ribbon model is less than the system footprint, small amplitudes result, and the source is imaged too deeply in the stitched 1D section. On the other hand, a narrow resistive gap in a conductive layer is incorrectly imaged as a drooping region within the layered conductor; below, the image falsely contains a blocklike poor conductor extending to depth. Additionally, edge-effect responses often are imaged as deep conductors with an inverted horseshoe shape. Incorporating lateral constraints in 1D AEM inversion (LCI) software, designed to improve resolution of continuous layers, more accurately recovers the depth to extensive conductors. The LCI, however, as with any AEM modeling methodology based on 1D forward responses, has limitations in detecting and imaging in the presence of strong 3D lateral discontinuities of dimensions smaller than the annulus of resolution. The isotropic, horizontally slowly varying layered-earth assumption devalues and limits AEM’s 3D detection capabilities. The need for smart, fast algorithms that account for 3D varying electrical properties remains.
Design of an SVM Classifier Assisted Intelligent Receiver for Reliable Optical Camera Communication
