Overcoming Overlapped Fields and Geolocation Errors in Point-Dipole Magnetic Field Inversions

In magnetic field surveys for unexploded ordnance (UXO), point-dipole inversions are generally used. Ground, aerial, and marine surveys often use magnetometer arrays that sweep out one or more curviplanar surfaces. Where such swaths are adjacent or overlap, geolocation errors, differences in swath elevations, artifacts of demedian or other filtering, and imperfect corrections of heading noise, maneuvering noise, and geomagnetic variations, can yield herringbone noise. Leveling or microleveling can reduce herringbone, but fields at actual measurement locations are desirable for optimal inversions. An automatic geolocation error correction (AGEC) was therefore developed to simultaneously estimate dipole parameters and optimal 1-, 2-, or 3-d translations of one or more swaths relative to a reference swath where a point-dipole anomaly spans multiple swaths. Intra-inversion filtering (IIF) can now be directly applied to ungridded data using linear, pointby- point adaptive, “stabilized” gradient nulling (SGN) filters, F, which null functions with constant horizontal gradients. That is, F{a+bxi+cyi} = 0 for a, b, and c constants and (xi,yi) horizontal positions scattered on the measurement surface z(x,y). IIF ameliorates effects of complex background fields, overlapping fields of nearby dipole-like sources, and residual low frequency noise components due to imperfect pre-inversion processing. These new SGN filters can be applied to individual swaths without amplifying effects of herringbone noise. Because they are applied to the input data and to components of fields being modeled in the mixed gradient-regression inversion algorithm, SGN filters yield perfect inversions for point-dipole anomalies superposed on a constant-gradient background field. Multi-window, multi-dipole inversions and AGEC with and without IIF are shown to be effective. AGEC and IIF can be usefully applied to total-field or vertical-gradient data. Measures of fit using post-inversion bias estimates and polynomials fit to residual fields are useful for IIF inversions. Single-window IIF inversion with AGEC takes a few seconds on a laptop computer.