The U.S. Department of Defense (DoD) has custody and responsibility for human safety and environmental stewardship for coastal ranges, many of which have underwater sites that are known to contain underwater military munitions (UWMM), such as discarded military munitions (DMM) and unexploded ordnance (UXO), as a result of historic military activities. In addition to explosive blast (safety) considerations, regulators are increasingly concerned about potential ecological impacts of munitions constituents (MCs) on the marine environment, which has resulted in costly risk characterization efforts and could lead to more-resource-intensive remediation efforts. Accurate assessment of MC in underwater environments includes a high level of effort or difficulty required to (1) measure MC at very low (nanogram [ng]/liter [L]) concentrations; (2) identify leaking UWMM, and evaluate the nature of the leakage (e.g., varying levels of corrosion, MC release rates attenuated by currents, dissolution rate, biofouling, MC degradation); (3) measure MC release during episodic events; and (4) measure MC in biota in spite of low bioaccumulation potential.

This demonstration focused on field validation of commercially-available passive sampling devices (PSDs), specifically Polar Organic Chemical Integrative Samplers (POCIS), that had recently been optimized for detection and quantification of MCs under environmentally-relevant conditions in laboratory-based studies.

The technical objectives of the effort included the following Tasks:

Task 1: Conduct a controlled field validation study using a known source (i.e., fragments of the explosive fill material Composition B [Comp B]) placed in a marine environment;

Task 2: Conduct a calibration study to evaluate the performance of POCIS under multiple flow velocities and different levels of source material (e.g., shell) encapsulation, including fully-exposed versus breach-hole scenarios;

Task 3: Use the results from Tasks 1 and 2 to develop a technology user’s guide for POCIS application at UWMM sites; and

Task 4: Conduct a full field validation study at a UWMM site, specifically a bay in the Live Impact Area at the Vieques, Puerto Rico Naval Training Range (VNTR).

Exposure data from the proposed validation efforts were then compared with existing toxicity criteria (Lotufo et al. 2017; SERDP 2017) to assess potential for ecological risk of UWMM associated with the data derived from the field. A technology user’s guide is also appended to the Final Report (Rosen et al. 2017).

The measurement of polar organic compounds in environmental matrices, especially at trace concentrations, represents a significant challenge. In recent years, significant improvements in analytical techniques coupled with the development of PSDs have much to offer towards in situ monitoring of ultra-low concentrations of emerging contaminants by providing a time-integrated sample with low detection limits and in situ extraction. PSDs are fairly well developed for legacy hydrophobic compounds (e.g., low density polyethylene membranes, polymer-coated jars or fibers) as well as for polar organic compounds (e.g., POCIS and Chemcatcher®).

The POCIS technology offers an advantageous alternative to traditional sampling methods (e.g., grab sampling) at sites where very low concentrations (ng/L) or fluctuation in concentrations are expected to occur, such as near underwater munitions. A continuous sampling approach allows detection and quantification of chemicals in an integrated manner, providing time-weighted average (TWA) concentrations, and the detection of chemicals that rapidly dissipate or degrade in the environment following release from the source. Unlike samplers that rapidly achieve equilibrium using very high surface area to sorbent volume, POCIS exhibits negligible loss rates and does not require a lengthy timeframe in order to reach equilibrium, allowing small masses of chemicals from episodic release events to be retained in the device by the end of the deployment period. The POCIS vastly simplifies sampling and preparation steps by elimination of electrical or fuel powering requirements, significantly reduces the numbers of analyses required, and provides protection of analytes against decomposition during transport and storage.

Performance was analyzed using a combination of quantitative and qualitative measurements to achieve the objectives of the project. The extent to which expected performance objectives were achieved was evaluated from the data collected in Tasks 1–3 and are elaborated upon in the report.

Performance Objective 1 was the verification that POCIS could detect MCs in a positive control field study at a clean site. Following permit approvals, 15 grams of Comp B was placed at the site over a 13-day exposure. The performance objective was met, with POCIS-derived TNT and RDX average water concentrations ranging from 9–103 ng/L, with the highest concentrations within 0.3 meters (m) of the source. MC was non-detectable at stations >2 m from the source. Grab water samples collected and oyster tissues deployed at the site were below detection limits for all stations, indicating POCIS was the most sensitive technology for ultra-trace level detection in a controlled field study.

Performance Objective 2 was the verification that POCIS-derived TWA water concentrations and TWA concentrations derived from multiple grab sampling would produce similar results, or better results, for POCIS in a flume study simulating field conditions or in actual field studies. This objective was met for the Comp B flume study, the positive control field study, and the Vieques field validation study. Comp B flume-deployed POCIS estimated TWA water concentrations for TNT and RDX were similar to averaged concentrations generated using multiple grab TWA concentrations. In the positive control field study, MC concentration was successfully determined using POCIS, while the discrete-sampling-derived concentrations as grab water samples resulted only in non-detects. In the Vieques field validation study, 1 of 30 sampling locations resulted in a relatively high water column concentration for TNT and several of its transformation products. The average TNT concentration from the two grab samples at the station was only 11% higher than the POCIS sample. The POCIS-derived average TNT concentration was 19% above the initial grab and 29% below the final grab sample concentration. POCIS-derived average RDX concentrations for 11 stations had detectable concentrations, while at the same locations only 3 stations had detectable concentration from grab samples during the initial period, and all stations were non-detect for the final period. Overall, data from grab samples validated the data obtained using POCIS for all the flume and field studies.

Performance Objective 3 was the demonstration of the effects of varying current velocities, in a series of controlled flume studies with precise velocity control, on the uptake of MC from spiked water to optimize sampling rates based on site-specific flow velocities. The objective was met, with a positive, statistically significant, linear relationship between current velocity and sampling rate for POCIS for multiple MC, providing useful means of applying appropriate sampling rates. From the regression equations derived, simple calculations are able to be used to correct for flow velocity if such measurements are made at the field site. In this project, a Nortek current profiler was used at Vieques to calculate the most accurate sampling rate based on measured flow. Two different explosive fill encapsulation scenarios showed highly comparable TWA concentrations for POCIS and average concentrations from multiple grab samples.

Performance Objective 4 was the demonstration that POCIS would detect MC at levels substantially lower than achievable using typical grab sampling methods. The quantitation limit (QL) for POCIS- derived TWA concentrations were consistently lower than those derived for discrete samples. Lower detection limits are achieved using POCIS sampling because the estimated volumes of water cleared of MC during the deployment time were substantially greater than the volume (1 L) consistently of all grab water samples. For the Comp B positive control study, the concentrations of TNT and RDX in grab samples were reported as non-detects; contrastingly, POCIS-derived TWA concentrations were reported for 12 out of 20 stations. For 12 stations out of 15 in the Vieques field validation study, the concentrations of RDX in grab samples were reported as non-detects; contrastingly, POCIS-derived TWA concentrations were reported for 8 out those 12 stations.

Performance Objective 5 was the demonstration of the success rate in terms of both recovery of POCIS from the field and the determination of useful data. A total of 20, 51, and 30 POCIS canisters (each containing 3 samplers) were deployed in the positive control field study, in the flume studies, and at the Vieques site, respectively. All samplers (100%) were recovered. Data were considered useful whether or not the concentrations were above or below method detection limits (MDLs), as it was expected that many field samples would be non-detect. All flume study data resulted in measurable concentrations, as the flume was spiked at concentrations to ensure detects. The strong correspondence between POCIS and multiple grab-based TWA concentrations in flume studies are a quantitative measure of the value of the POCIS data, showing negligible losses and post-uptake preservation of the parent compounds throughout the exposures.

Performance Objective 6 was the demonstration that all field and laboratory efforts followed experiment-specific quality assurance (QA) objectives and that quality control (QC) criteria were met. All criteria were met for this part of the project. Blanks, including field and laboratory, did not have MCs above the QLs. All spike tests had accuracy and relative precision within 25% of what was expected. In addition, all other sampling handling and instrument criteria were also met.

Performance Objective 7 was the demonstration of the ability to use POCIS TWA data for MC to evaluate ecological risk based on comparison with toxicity benchmarks developed from species sensitivity distributions. Compared to the high incidence of non-detects from grab samples, POCIS reported ≥ low ng/L MC concentrations in all tasks, allowing more quantitative assessment. Measured concentrations indicate negligible ecological risk based on comparison with hazardous concentrations derived from species sensitivity distributions. For Vieques, POCIS-derived TWA concentrations were 10–1,000,000 times lower than hazardous concentration values for 5% of species (HC5) generated from the most up-to-date and comprehensive species sensitivity distributions (SSD) as reported by Lotufo et al. (2017). Despite POCIS having a higher frequency of detection than grab samples at Vieques, detection levels for grab sampling and POCIS were below regulatory screening levels and both sampling methods showed no unacceptable risk. Therefore, the grab samples and POCIS are expected to be of equal value for Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) risk assessment at that site.

Performance Objective 8 was a qualitative objective of ease of operator use, requiring feedback from field and laboratory technicians on the usability of technology, sample prep and extraction, and time requirements. At Vieques, feedback in the field from Navy and contractor personnel was mixed. The deployment and recovery of POCIS went well, but the overall process was highly labor intensive, with dive teams and boat support required for both deployment and recovery of the samplers. The use of munitions response (MR) and scientific divers creates significant safety concerns associated with deployment and retrieval of POCIS. Overall, the level of effort and the associated safety concerns for POCIS are higher than grab sampling, which, if kept at a minimum, can be done in a single field effort without divers. Site managers understood the benefits of integrative sampling and the potential advantages of providing enhanced credibility through lower detection limits and obtaining data representative over extended timeframes, thereby sampling over a larger area. Grab sampling intended to provide temporal trends and TWA concentrations could require substantially more labor, depending on site-specific logistics and study objectives. Similarly, autosampling would require multiple trips to the site to obtain an integrated sample over time and ensure that MC do not degrade (e.g., freezing or extracting samples daily). Laboratory feedback indicated that processing of POCIS as compared to standard solid-phase extraction (SPE) of grab water samples was negligible.

Performance Objective 9 was the demonstration that the relative value of data from POCIS compared well with the cost of measurements from water and sediment porewater. POCIS was the only technology that detected MC at Gulf Breeze, FL, and had a higher frequency of detects compared to grab sampling at Vieques. The costs of using POCIS over more traditional means of water sampling (e.g., grab or composite sampling) are examined using multiple examples in the Cost Analysis section of the report, and suggest that POCIS are less expensive when traditional sampling involves multiple sampling events to develop an integrative sample (as opposed to single grab samples that would be less expensive than POCIS). However, for sites where regulatory requirements are for single grab samples, the costs for a POCIS-based program could be considerably higher. Vieques is a complex site and the demonstration was designed to maximize the likelihood for detecting a leaking munition. It is unlikely that POCIS would be routinely applied in such a manner in a monitoring or regulatory program.

Performance Objective 10 was the qualitative objective of end-user understanding and acceptance of the POCIS technology for potential use at UWMM sites. Site managers and contractors understood the value of integrative samplers for MC and provided a considerable amount of in-kind support to successfully demonstrate the technology at Vieques. The notion that the use of POCIS would help with the criticisms of sampling at the wrong place and at the wrong time was seen as a primary advantage, especially considering the results of the Gulf Breeze study. Site managers on Vieques expressed concerns about the cost, diver safety, and difficulty of implementing POCIS. Site managers also noted that grab samples matched well with the POCIS results and the grab samplers are accepted by the regulators for risk assessment. Although the cost for POCIS is less than grab or composite sampling based on a sampling program that produced similarly integrative samples, the cost of collecting a single grab sample at a site would be less expensive than monitoring with POCIS.

Previous laboratory proof of concept and calibration and work for MC by this project team and the demonstration and validation of POCIS in laboratory and field efforts for this project indicate the technology is highly valuable for assessment of MC exposure at UWMM sites. POCIS-derived TWA concentrations are expected to be more informative about exposure to MC compared to discrete grab samples when MC concentrations are low and MC is released to the water column in a time-varying nature, either from UWMM or from terrestrial-based time varying inputs (e.g., runoff events or tidal pumping of groundwater contaminated with MC). For most applications, the cost associated with POCIS sampling is less than that for multiple grab or composite sampling required to represent a comparably integrated sample. In addition, POCIS sampling is expected to directly address sentiment from those concerned with (1) UWMM as sources of contamination and (2) who perceive grab sampling may take place at the wrong time and in the wrong place, and therefore (3) fail to adequately characterize exposure risk potential. UWMM site characterization using POCIS addresses all three of these concerns, and implementation as part of monitoring programs or for risk assessment should be considered depending on the site-specific objectives. Site characterization using POCIS may be site-wide or spatially focused or may be used to complement traditional sampling approaches to identify or rank sites of potential concern and support leave-in-place (LIP) versus removal decision-making processes.