Underwater detection and remediation of unexploded ordnance (UXO) remains a challenging problem for many reasons including the dynamical nature of the environment, limited visibility, mobility of Targets of Interest (ToIs), and the absence of global positioning system positioning. Even after ToIs have been identified through Wide Area Assessment techniques, the targets have to be accurately reacquired for precise positioning the instrumentation to classify the ToI. The objective of this project was to investigate the feasibility of a magnetic-gradient-based detailed survey technique for real-time ToI localization and potentially for UXO classification enhancement. The localization method required the development of an ultra-sensitive short-baseline scalar magnetic gradiometer with the combined sensitivity and baseline requirements far beyond the capability of any gradiometer composed of individual magnetometers. A proof-of-concept gradiometer was constructed to explore the possibility of meeting the sensitivity and base-line requirements and accuracy of the localization method is studied theoretically based on sensor performance.

Technical Approach

The gradient from a magnetic dipole moment, which was a good model for UXOs, decayed as a function of R-4, where R is the distance between the dipole moment and the sensor. If gradients at four different points are measured, the location of the dipole moment can be calculated in the coordinate system defined by the four points. This simple calculation can provide the project team with a good initial estimate of the dipole location, which can greatly facilitate the full inversion method for achieving better and more complete dipole parameters. The method poses a combined requirement of low noise and short baseline on the gradient measurement, which cannot be met with commercially available magnetometers. The project team investigated a new intrinsic scalar magnetic gradiometer sensor to address this requirement. The intrinsic gradiometer uses a single far detuned laser beam to probe two atomic ensembles. The intrinsic gradient measurement has the advantage of cancelling out many common-mode noises and hence achieves much better sensitivity.

Interim Results

A magnetic-gradient-based inversion algorithm was developed for real-time localization of a magnetic dipole moment. The project team's theoretical simulation indicated that accurate real-time localization is possible if the peak-to-peak gradient noise is about 4 pT at 5 cm baseline. For a 100 Hz sample rate, 4 pT peak-to-peak noise is equivalent to about 150 fT/√Hz noise density, which is demonstrated in a magnetically unshielded environment by a prototype portable intrinsic gradiometer constructed using vertical-cavity surface-emitting lasers. Inside a magnetically shielded environment, better than 90 fT/√Hz sensitivity was achieved.


This research work eliminates the fundamental scientific risk of obtaining ultra-sensitive short-baseline magnetic gradient data in practical applications. The gradient measurement can be used for real-time UXO localization, greatly reducing the cost of UXO disposal by shortening the time-consuming localization procedure. Current classification methods using time domain electromagnetic (TDEM) require very high accuracy in real-time target localization to position the instrument over the target; this method enhances reacquisition efficiency and accuracy and may be integrated with TDEM instruments to reduce location-based uncertainty in classification results.

  • Magnetometer ,

  • Sensors