The people, science and technology behind discovery

UXO Lands Restoration and Release: Mitigating Residual Risk

By Graham Chandler

When it comes to finding hand grenades, artillery shells or other UXO (unexploded ordnance) buried in the soil, Tamir Klaff knows that you can never be 100% sure you've found them all.

"We'll never go onto a property that has had munitions on it and do a geophysical survey, intrusive excavation and come out with a report that says you have a clean piece of land," says the Munitions Response Geophysicist at CH2M HILL who's responsible for the worldwide engineering conglomerate's operations related to munitions response geophysics. "The only way to absolutely ensure that the land is clean is to sift the soil."

Sifting every cubic centimetre of soil is of course prohibitively expensive if not totally impractical. As a result consultants and contractors that are hired to clean up properties previously used for gunnery and bombing practice, testing or disposal around the world have to rely on good research, impeccable planning, specialized geophysical survey equipment, efficient software and statistics to keep costs down while attaining acceptable margins for land clearance. Trading one facet against another is a constant challenge.  How does one determine when a piece of land is considered safe enough to be restored to public use?  "Nobody has answered the question: how clean is clean?" says Klaff.

Tens of millions of acres of such land have been deemed unessential by governments which want them released for civilian uses such as housing and business developments, and the establishment of recreational land or wildlife parks. Furthermore, the military are frequently faced with re-designation of still-active military lands.  Ranges once used for grenade or mortar practice are now needed for uses like new barracks or a magazine area. Faced with the limits of technology, it all amounts to a massively expensive undertaking.

In the US alone, as of September 2008, the Department of Defense's inventory of Military Munitions Response Program (MMRP) lands—those needing restoration—under the US Army Environmental Command totalled 3,662 sites, according to Mary Ellen Maly, MMRP Program Manager for Active Army Installations. "Of which 1,661 were FUDS [Formerly Used Defense Sites], 1,658 were attributed to active Army, Navy or Air Force installations, and 343 sites were located at BRAC [Base Realignment and Closure] installations," she says. "There are 896 MMRP sites located on active Army installations, so the Army controls approximately 54% of the MMRP sites in the Department of Defense inventory of active installations."

Land that has UXO rarely (if ever) has landmines. These are typically terrestrial maneuver areas and as such, real landmines would not be used in these areas. Land mines are primarily made out of a type of plastic, with the only metal in them being the spring and/or firing pin, and these items are so small that they can't be easily located, explains Elizabeth Baranyi, Geophysics Specialist with Geosoft, a company that develops software solutions for UXO detection, data and project management. "They are within the noise levels of other things like small nails or pieces of fragmentation." Baranyi says that in the US, Canada and Europe there are few sites involving land mines, but others like Laos, Vietnam, Afghanistan and Cambodia are still littered with them.

For US Army lands needing restoration, the Army starts by developing a performance work statement (PWS) and qualified contractors are given the opportunity to submit a proposal. "After careful evaluation of each proposal, the Army selects the contractor with the best proposal for conducting the required work," says Maly. In some cases the extent of UXO is so uncertain that the PWS may not be to clean the area up but rather to complete a remedial investigation (RI) that is then used to determine the extent of contamination. This RI is completed before any remediation action efforts are even considered.  

From there, it's a phased process. Klaff's role is to develop the geophysical approach for the project team. "We're a management integrator company," he explains. "We bring all aspects of the project together into a uniform approach. We contract directly with, for example, the US Army and manage the entire project including Quality Control (QC), safety aspects and technical project planning." Upfront he says they sit down with all stakeholders to agree on approaches, sampling methodologies—the whole process, to make sure it will meet everyone's objectives and goals.

The process can't be hurried. "It is of particular importance that the entity responsible for the work (such as the Navy, Corps of Engineers, AFCEE, etc.) and the regulatory community (State of California, EPA, etc.) take the time to agree to a sound decision tree that outlines the way ahead," says Jeff Sabol, Sr. Geophysical Manager for the Ordnance Response Division for ECC, a Colorado firm handling munitions response activities around the world. It is critical that initial decisions include an all-party agreement on an acceptable end state, he adds.

The next phase is background research: what was the land used for? What type of ordnance is expected? Where would most projectiles have impacted? These questions and more will lead to an initial site investigation (SI) involving aerial photographs, historical research and some onsite reconnaissance—visual and geophysical.

In order to keep the field costs down, this research needs to be as in-depth as possible. "We'll look at the old range maps that show the historical firing points and firing fans to help focus the investigation for example," says John Breznick, General Manager at NAEVA Geophysics based in Charlottesville Virginia. In most cases the next step is a remedial investigation feasibility study (RIFS). "At that point you get a little more aggressive with your site characterization methods, utilizing geophysical investigation techniques which may include a combination of transects and grids that are commonly used for digital geophysical mapping over the property," he says. The number of transects and grids are often chosen based on the findings of the SI. "You would also try and define the higher impact areas where concentrations of MEC would be expected to be the highest and also the medium to light areas." From all this he says you develop the RIFS, "and that gives you guidance as to what it will take to reclaim this property." Importantly too, an estimated real cost comes out at this point.

Klaff says they often use a software package called Visual Sampling Plan (VSP) to help define the geophysical plan. According to the VSP website, this software supports the development of a defensible sampling plan based on statistical sampling theory and the statistical analysis of sample results to support confident decision making.  "It has statistical tools that can be used to make statements, for example, ‘here's a certain size area and if I want 98% confidence that I have crossed an impact area of X size, then here's how many transects and their spacing that would need to be placed'," he says. Still, he notes that even after using this software, or any other sampling/confirmation approach, there can never be a guarantee of finding 100% unless you physically sift the soil.

Before the field detection and excavation phase begins, ECC's Jeff Sabol highly recommends that "seed items be emplaced before the operation starts." These, he explains, provide critical, real-time tests over the entire "investigation / remediation system" that can then be used for QC and Quality Assurance (QA) purposes.

Although the overall team is a cohesive unit, typically one contractor is used to complete the geophysical mapping and another for the specialized task of excavation and removal of anomalies. "We're typically involved with another contractor who has an explosive ordnance capability," says Breznick. "Our responsibility is the geophysical mapping for investigating the property. Once we've identified targets in the subsurface to be considered for excavation we'll often go back to the field and flag them for subsequent removal."

At times, individual target identification becomes impossible says Breznick. "Some areas are so concentrated with metal it doesn't allow you to identify individual targets." In those cases he says they put a polygon around the entire area and the remediation group does one of two things—go in with their hand-held metal detectors and keep removing metal until the upper layers of the area are relatively clean, or use heavy equipment to spread the soil and go through it by hand or sieve it.

Instrument depth of detection capabilities have their limitations as well. For instance a 20mm projectile is about the size of a roll of nickels. If this 20mm projectile is on the ground surface it is easily detected by the geophysical instrument, however, if it's 12 inches below the ground surface, it will probably not be detected and therefore not removed. Sometimes natural forces will shift ordnance or bring it closer to the surface. Klaff says he has seen instances where freeze-thaw cycles have pushed items to the surface over the years. "So things can start popping out that weren't there before," he says.

Most geophysical surveys for UXO use magnetic and/or electromagnetic detection equipment and are completed on the ground using vehicles or walking. Aerial surveys can be conducted however this type of survey is typically used for wide area assessment or "footprint reduction".  Because aerial surveys are a greater distance from the ground, their detection capability is diminished exponentially as the distance from the object increases. Furthermore the resolution of the anomaly is also degraded as distance increases. Ideally you want some definition of the shape of the anomaly. But, according to Baranyi, the higher you are, the greater the distance between your instrument and the anomaly and the less likely you are to detect it. Furthermore, the higher the detection instrument, the greater the area that is being averaged for that reading. Thus, you may be able to detect an anomaly, however the resolution may be so low that multiple anomalies look like a single anomaly. The resolution of an aerial survey may not be good enough to pinpoint exact digging location or even to detect individual items of interest, but may be used at the pre-planning stage, she adds.

Often it is not only land that needs clearing, but waterways, lakes and oceans have also been used for weapons testing, practice or dumping. These areas have their own sets of considerations. According to Sabol, oceans were used as disposal areas at the end of World War II for both conventional and chemical weapons. Furthermore, depending on the site, some ordnance items can be transported due to current, wave or storm activity. This is a problem unique to underwater UXO in that, after the UXO item has been detected, by the time the recovery crew arrives to remove the item it may have moved. 

QC and QA are continuous throughout the operation. Contracting agencies such as the US Army maintain their own quality control as recovery progresses. "The bulk of the Army's field oversight of UXO detection and response activities is performed by the Corps, which has ordnance specialists trained to supervise or monitor field UXO work," says the Army's Maly.

Most UXO clearance projects generate gigabytes of data and often tens of thousands of targets that then need to be interpreted. "The big question in our part of the industry is ‘can we tell the difference between UXO and a pop can?" asks Darren Mortimer, Technical Geophysical Analyst for Geosoft. "Or between UXO and a 45-gallon drum. You're just looking for metallic objects, however the trick is deciding whether that object is the target you're looking for." The statistics often astound: for example, out of 49,521 anomalies excavated at Kaho'olawe, Hawaii, only 3% turned out to be UXO, according to the University of British Columbia's Unexploded Ordnance and Landmine Research Group. Incredibly, this 3% is high compared to typical ordnance remediation sites which have UXO percentages of ~ 1% or less.

For UXO target detection, Klaff finds Geosoft's montaj UX-detect software indispensable. "It's a very comprehensive data processing and earth mapping system that has specialized workflows for the types of data we deal with on munitions response projects," he says. The fact that Geosoft works with the Army he says gives them probably the most experience in the US. "The UX-detect module has been designed specifically for our industry so it is extremely helpful in conducting these projects."

NAEVA's Breznick also uses UX-detect for analysis and project management. "Geosoft provides a complete solution for UXO projects, in fact it's the only one on the market," he says. "It's very robust and allows you to manage large volumes of data. That's critical to our ability to operate with limited budgets." He adds that Geosoft have been responsive to user feedback, and has made continual improvements, over the years, to simplify the workflow.

Once the geophysical anomalies have been identified and excavated, frequently QC protocol dictates that a second geophysical survey and excavation be completed. "If there's a high density of anomalies, going back and removing what you can and then going back over it as a second run, inevitably you will find more," says Klaff. "Sometimes there are just too many anomalies that blend into other adjacent anomalies. When this is the case a procedure called "mag, flag and dig" may be used." The mag, flag and dig operation entails having the UXO technicians (excavation team) use hand held metal detectors that they then use to "sweep" the area and remove metal both above and below the surface. Once this is completed then a geophysical survey will be done to see what is left. According to Klaff, "There can be multiple iterations of this."

Then, at the end of the excavation process, a final geophysical survey may be required.  According to Sabol, "A statistically sound number of geophysical transects, or grids, may be required to be collected so as to establish confidence in the entire geophysical and excavation process. Sites with special end land uses, such as a future school ground, might even have 100% of the area geophysically surveyed as part of the required, predetermined and agreed upon QC or QA process."

Even when a 100% QC or QA post excavation geophysical survey has been completed, questions of confidence linger. The question then becomes: when is land previously contaminated with UXO deemed acceptable for use? The answer is dependent upon the eventual use slated for the property. For example, if the area is going to be a school there is obviously going to be a high degree of confidence that all UXO items have been removed because of the required safety factor that is necessary when kids are involved. If it the area is going to be a wildlife refuge where there is a much lower possibility of interaction between humans and potential UXO then the degree of confidence that is required might be less. "These kinds of risk decisions have to be made by the stakeholders and communicated to the land users, says Klaff. "And it is particularly difficult proving the negative: if you go out to a site and find nothing, can you really say there's nothing out there?"

Geosoft's Mortimer says what's critical in an acceptability decision is to look at what the future activities for the site are and how are they going to affect the ground surface. "For instance a hunter is just going to walk across the ground, he's not going to dig any holes or chop down any trees; he won't be disturbing the ground in any significant way," he says. On the other end of the spectrum, in an urban setting with the construction and use of densely spaced houses, "there would be a much greater chance of encountering munitions." Mortimer says similar considerations are in order for underwater restorations. "What will people be using the area for, a wildlife habitat for example, or for human recreation? Will people be throwing anchors into the water or bordering on a marine transit way where they might employ dredging etc?"

To improve end state confidence levels, there's a pressing need to reduce costs and become more efficient at identifying targets that are specifically UXO or munitions related. "One way is to stop digging up non-threatening metallic objects—the big push going forward is the ability to identify those anomalies that are ordnance and not a pop can," says Breznick. He says both sensors and software need more research and development, working in parallel so that new software can work with new information sources. "Geosoft is working in that direction right now," he says. "They've incorporated software developed through the ESTCP program by SAIC called UX-analyze. It allows you to start to discriminate these targets." Research into better discrimination has been working reasonably well in laboratories and controlled environments but has yet to demonstrate confidence in the field, he adds.

Better success would go a long way to easing beleaguered contractors faced with the trend towards fixed-price performance-based contracts in the US that tend to put all the risk on the contractor. "There's very little room for change orders or adding on," says Klaff. "Companies are having to take on a lot more risk than the government or the party responsible for the cleanup." For them, vastly improved discrimination is certainly the desired future for keeping costs down and reducing their financial and business risk.

A unique firm called Neptune and Company think they have a new tack to not only address the formidable cost the US federal government faces with the cleanup campaign but also those risk-facing contractors. Because so much of the UXO business is based on probability, they apply statistics in a much larger way.

Before conducting the remediation, "we build a probabilistic model based on available information," explains Paul Black, Neptune's CEO, who holds a PhD in statistics. It's a comprehensive and highly detailed one he says. "Firing points, target areas, ballistics of the munitions used, digital elevation maps where the ballistics match. We start working from a modeling perspective where the UXO are likely to be." Then, once the data are collected they're used to update or calibrate that model. The company has demonstrated success in a recent project in Montana.

The approach is a fresh one, and needs some outdated concepts renewed, says Black. Like the idea that the probability of UXO detonation statistically is always 1. "That's based on some work done ten years ago and is still leading us down the path," he says. "But a few years ago a researcher named Jackie McDonald did some work and found she could put a probability of between 0 and 1 on a UXO—with a uniform distribution between 0 and 1." He says it can be made a function of inputs like age of ordnance, type, fuze etc. Incorporating statistical considerations like these can go a long way towards cost reductions. But funding for the concept has been an uphill battle, says Black.

"It has been a hard sell to the government," he says. "The government's focus has been on instrumentation and statistics for geophysical survey data—there is a lot of interesting research going on there." Discrimination is another important focus, however there is virtually no effort put forth on statistical modeling of the overall problem like Neptune's approach, which has already shown financial benefits. "In the Montana project we were told it saved $10 million dollars."

In the end, even with all these improvements UXO will remain a probability game. But as the odds improve, contractors will edge closer and closer towards that 100% confidence figure which will ultimately reduce the massive costs of returning these properties to safe civilian uses.