Value of Multi-component TEM Data for Estimation of UXO Target Parameters
Donald D. Snyder, Snyder Geoscience, Inc., Grand Junction, Colorado
Scott C. MacInnes*, Zonge Engineering & Research Org., Inc, Tucson, Arizona
Jennifer L. Hare*, Zonge Engineering & Research Org., Inc, Tucson, Arizona
Robert E. Grimm, Southwest Research Institute, Boulder, Colorado
Mary Poulton, University of Arizona, Tucson, Arizona
Anna Szidarovszky*, University of Arizona, Tucson, Arizona
Paper – [pdf] UXO_Value of Multi-ComponentTEM Data
Two multi-component multi-gate data sets from the Naval Research Laboratory’s Baseline Ordnance Classification Test Site at Blossom Point, one acquired statically with a Geonics EM61-3D- 3C and the other acquired dynamically with a Zonge NanoTEM system (DNT), are analyzed to determine the relative classification performance of the two systems. Not surprisingly, our classification performance is better with 3-component static data than it is with the 3-component dynamic data. Confirming published work by Grimm , classification is significantly improved when it is applied to the 3-component static data than when it is applied to a decimated data set consisting of only a single (z) component. However, early analyses of the dynamic data indicated that horizontal components provide marginal, if any, improvements in classification. Noise analyses of data from the two systems show that noise levels in the EM61-3D data set are approximately 40dB lower than those in the DNT system and that noise levels in the horizontal components at late times are 2-5 times higher in the vertical component. Noise reduction in statically acquired data can be
attributed to stacking (~20dB) and the elimination of microphonic noise from antenna cart movements. With dynamically acquired data, the higher noise levels in the horizontal components together with uncertainties in antenna position and attitude are most likely the reason that the horizontal components do not unequivocally improve classification performance in the dynamic data.
Recently published numerical model studies [1-3] and precision static measurements  conclusively demonstrate that target classification noticeably improves when physics-based target analysis is applied to multi-component multi-gate TEM data. The obvious explanation for the improvement is that the added independent information provided by the addition of the transients from the horizontal receiver antennas produces a more robust solution to the anisotropic dipole model that is widely used for target parameterization. But these conclusions are not as clear when using dynamic multi-axis, multi-gate data from the same target set .