The project team is currently developing a computer-based expert system that synthesizes recent research to model munitions burial and mobility in a range of underwater environments. There are numerous inland water and coastal sites contaminated with munitions and explosives of concern (MEC). Given the constrained resources available for underwater munitions detection, it is advantageous to have the ability to predict areas of munitions aggregation, burial versus exposure, and temporal variability to support planning for efficient site remediation. Towards this end, the project team will continue development of the Underwater Munitions Expert System (UnMES) in this project. The UnMES is a probabilistic Bayesian expert system which synthesizes databases of environmental conditions and models of MEC behavior in response to environmental forcing, to predict the location of munitions and their degree of burial at underwater sites. This effort will be performed in collaboration with other SERDP Munitions Response (MR) studies that are focused on MEC response to underwater forcing. As a result, the UnMES will provide improved guidance for underwater munitions site assessment and remediation efforts.

Technical Approach

The UnMES is a probabilistic Bayesian expert system for predicting the location of munitions and their degree of burial at underwater sites. This expert system predicts the contamination density (relative abundance) of MEC across specific sites of interest. Initial development of the UnMES was carried under SERDP MR-2227 and incorporates process models of MEC behavior developed by prior and ongoing SERDP and ESTCP mobility and burial studies. This project will continue development of the UnMES and focus on incorporation of laboratory and field data on mobility and burial currently being obtained by other investigators in the SERDP MR Program area, and improved modeling of burial, migration, impact penetration and re-exposure. Also, the UnMES architecture will be extended to incorporate conditions appropriate for the numerous aquatic environments of interest. A “proof of concept” UnMES validation study will be undertaken utilizing the compiled databases.


The probabilistic construct of UnMES feeds naturally into risk-assessment models used by site managers for remedial investigation decisions. Guidance regarding the timing, location, and operational choices for wide area assessments and subsequent clean-up efforts can be more efficiently planned and executed. Prediction of processes that affect detection and classification performance by geophysical, acoustic, and optical sensors will guide optimal selection of sensor technologies for use within the characterized sub-regions. (Anticipated Project Completion - 2019)