Parametric inversion reveals the different sensitivities with which source depths can be recovered from TMI anomalies. Depths can be estimated with higher sensitivity for sources of large depth-extent and with lower sensitivity for sources of limited depth-extent. The highest sensitivities for narrow ‘dykes’ are about 5% and basement blocks beneath sedimentary basins provide depth sensitivities in the range of 10-15%. Inversion is a convenient and versatile procedure to estimate source depth and is able to provide sensitivities to accompany those estimates.
Magnetic field interpretation is a remote sensing technique commonly used to estimate depth to the top of buried source bodies. Many methods of automated batchprocessing have been developed (eg. Thompson, 1982, Reid et al, 1990, Thurston and Smith, 1997) but with increases in computing speed and improvements in graphical user interfaces it is now feasible to obtain depth estimates directly from inversion. Any one survey area may contain few anomalies from which reliable depth estimates can be derived and inversion recovers maximum information from this limited resource. Inversions are selfadaptive to each anomaly, readily permit the introduction of independent constraints and are able to supply sensitivity estimates. Furthermore the interpreter gains valuable insight from pro-active involvement in the inversion process.
All magnetic source depth estimators require assumption or estimation of the source geometry and in general all methods perform well or poorly according to the suitability of this shape factor in representing the geological sources. Inversion of the fixed depth-extent dyke has the largest misfit statistics and therefore the greatest sensitivity to depth. Misfits for inversion of the fixed depth-extent plate are almost 90% of those for the dyke but a much greater difference between these bodies occurs when we set depth-extent free in the inversions. For the dyke freedom of depth-extent reduces sensitivity to depth by 15-30%. For the plate however the reduction in sensitivity to depth is much more extreme. If the reference model was a geological body and the input field a set of measurements we would have no confidence in resolving these field differences due to 30% depth variations. Other than for shallow thin sheets it is difficult to estimate the depth-extent of a source by visual inspection of anomalies. However once we have recovered estimates of depth-extent from inversion (or structural index from extended Euler deconvolution) we should weight depth values from bodies of limited depth-extent with lower confidence.
Depth sensitivity for bodies of large depth-extent is reduced by about 50% if the sides are allowed to taper. However sensitivity to depth is still significantly higher than for bodies of limited depth-extent. In these studies we are estimating sensitivity within the model assumptions and nonuniqueness allows that sources could be at any shallower depth once these assumptions are relaxed. These model sensitivities do however provide minimum error estimates and because they refer to relationships between the source and field they are relevant to all techniques which depend on similar assumptions.
