Physical and operational challenges in the marine electromagnetic induction (EMI) sensing involve reduced signal levels due to standoff distance to seafloor, obscuring of target signals by conductive background response, reduced positional accuracy and non-uniform data coverage due to navigational difficulties. In the SERDP MR19-1261 project, the project team addressed aspects of these technical difficulties resulting in several key developments. These include an integral equation technique of characterizing background, an optimal stacking response scheme of enhancing target detectability, and a data inversion method that accounts for sensor positioning errors. Nevertheless, it has been observed that low signal-to-noise ratio coupled with low signal-to-background ratio remains an issue when inverting for small targets of interest and/or at larger standoff distance. The methods developed to account for sensor positioning errors are promising, but introduce a trade-off for finding accurate extrinsic source parameters (locations and orientations) due to the use of multiple sets of intermediate sources. The work performed in SERDP MR19-1261 used data collected under ideal conditions (e.g., consistent water depths of approximately 20 m, minimal swell and minimal current). To achieve the full potential of marine EMI performance in more realistic environments, this project aims at 1) improving signal-to-noise ratio of measured target responses; 2) exploring methods of predicting modeling sensor positions; 3) developing alternative methods for dealing with sensor positional uncertainty; 4) developing methods to mitigate the influence of dynamic environments with significant current and tidal variations.

To accomplish the objectives, the project team is proposing to investigate the following approaches:

1. Noise suppression. Two data-driven methods will be studied: Singular Spectrum Analysis Method and Autoencoder, to understand and characterize time-domain electromagnetic signal of interest and noise contained in underwater measurements and ultimately to suppress the impact of any interfering signals.
2. Estimation of survey position offset. For a multiple pass survey, absolute sensor positions may be transformed into relative positions with respect to a given survey line. A kinematic model will be explored that can be used to make predictions of offset between lines in a retrospective process.
3. Robust and accurate inversion under sensor positional uncertainty. Two formulations that explicitly account for the sensor positioning errors by introducing either response corrections or corrections in sensing locations will be pursued.
4. The project team will try to find ways to understand how the dynamic processes (e.g., wave action, tidal variations etc.) affect marine advanced geophysical classification (AGC) data and develop procedures, coupled with (2)-(3), for mitigating their impact.

Marine EMI sensing has considerable potential to be deployed as a practical and effective AGC tool. Extracting accurate target information from underwater EMI data can be achieved given that the impacts of dynamic marine environment, sensor positioning errors, and stochastic noise can be properly accounted for. This project will attempt to further improve and enhance the EMI processing and interpretation capabilities in realistic underwater environments. The ultimate objective is to provide stakeholders with efficient tools to rapidly and effectively clear areas contaminated with underwater ordnance.