Unexploded ordnance (UXO) remediation at former Department of Defense (DoD) sites and return of the land to public use is a central U.S. government environmental policy goal. Successful remediation requires essentially 100% probability of detection (Pd), but economically practical remediation requires dramatic reduction of the high false alarm rates (Pfa) that lead to large numbers of clutter items needlessly excavated at huge expense (typically more than 70% of site costs). A higher performance integrated solution to this problem must combine a high Pd sensor technology with data of sufficient quality and diversity that robust target discrimination and classification may be performed, even under challenging field conditions (e.g., difficult topography, heterogeneous soil, dense clutter, and overlapping or deep targets). The newest electromagnetic induction (EMI) sensors have not only the required Pd but also the vastly improved dynamic range and spatial diversity necessary to acquire data with greatly improved discrimination information. Taking proper advantage of this data, however, requires physics-based forward models that provide a rigorous connection between target physical characteristics (geometry, material properties) and the measured signal.

The objective of this project was to validate the physics-based “mean field” and “early time” approaches to modeling of time-domain electromagnetic (TDEM) responses of compact, highly conducting targets.

Technical Approach

This project applied methods to the analysis of laboratory-style data collected by the Naval Research Laboratory TEMTADS system using artificial spheroidal targets. The models used the detailed system parameters (transmitter and receiver coil position, orientation, and geometry; transmitted pulse waveform; target position, orientation, geometry; target conductivity and permeability) to generate first principles predictions for the measured time-domain voltages. The models were designed to be essentially exact for spheroidal targets and the agreement between measurements and predictions supports this conclusion.


The results demonstrate the accuracy available from this project’s first principles – physics-based models covering the entire measurement window – from the early time multi-power law regime, all the way through the multi-exponential regime to the late time mono-exponential regime. Prior to the mean field code’s current upgrade, the number of accurately computed modes used to describe the multi-exponential regime was limited to perhaps a few dozen. The validation results showed this upgrade is critical to the success of the predictions, by generating the required overlap of the early time and multi-exponential regimes.


Reduced false alarm rates achievable using the new target discrimination technologies will lead to reductions in UXO remediation costs. The novel physics- and data-based signal processing algorithms may be tailored to a broad array of target discrimination and classification problems.