Hawaii UXO Clearance: Volcanic Soil Defeats Standard Electromagnetic Detection

Technical Summary

More than a century of US military training across the Hawaiian Islands has left Explosive Remnants of War (ERW) and Abandoned Explosive Ordnance (AXO) distributed through volcanic soils, watersheds, reef margins, and former impact areas. Reporting from the 8th Military Police Brigade Explosive Ordnance Disposal (EOD) element at Schofield Barracks, Oahu, identifies two compounding technical challenges: storm-driven migration of munitions through soft volcanic substrate, and a basalt-iron geochemistry that defeats the electromagnetic induction (EMI) sensors most widely deployed for sub-surface UXO survey.

Specialist contractor Quality Metric Solutions reports false-positive rates as high as 80% during EMI sweeps on certain Hawaiian sites, attributed to ferromagnetic basaltic clasts whose magnetic signature can overwhelm that of ferrous-cased ordnance. The operator quote was: "hot rocks will ring louder than metal sometimes." University of Hawaii researchers, working under an Army-funded programme tied to the former Waikoloa Maneuver Area (WMA) on Hawaii island, are evaluating smaller portable detection assemblies alongside robotic platforms and machine-learning discrimination algorithms that separate genuine ordnance signatures from soil-derived clutter.

The Soil–Sensor Physics

The Hawaiian detection problem is a materials problem before it is an instrument problem. Soils developed on basaltic parent rock (andisols and related iron-rich profiles) carry high concentrations of the ferrimagnetic minerals magnetite and titanomagnetite. These minerals give the ground a magnetic susceptibility far above that of the quartz-dominated sedimentary soils for which most commercial UXO sensors were calibrated. Susceptibility on strongly magnetic Hawaiian and similar tropical-basalt sites is reported in the geophysical literature at roughly 0.01 to 0.1 SI, against typical low-magnetic soils nearer 0.0001 to 0.001 SI. (These are indicative ranges; site values vary widely and should be measured, not assumed.)

Two failure mechanisms follow. For time-domain EMI, the soil itself produces a strong magnetic-viscosity response whose decay overlaps in time with the eddy-current decay of a small steel target, so the wanted signal is buried in soil background rather than cleanly resolved. Effective detection depth for small submunition-scale targets can fall below roughly 0.5 to 1 metre on high-susceptibility ground, against 2 to 4 metres in benign soils. For total-field magnetometry the problem is geometric: a basaltic "hot rock" and a buried steel fragment both present as compact dipolar anomalies of similar amplitude and spatial scale, so the operator cannot separate them on the map. That is the physical content of the 80% false-positive figure: it is not sensor failure, it is the soil generating ordnance-like anomalies faster than analysts can dig and clear them.

Detection Performance and Discrimination

The defensible response is to move from detection alone to classification. Advanced time-domain EMI arrays with multi-axis transmit and receive coils, such as the MetalMapper and the TEMTADS (Time-domain Electromagnetic Multi-sensor Towed Array Detection System) platforms demonstrated under the US Environmental Security Technology Certification Program (ESTCP), recover the three principal-axis polarizabilities of a buried object. A manufactured munition behaves as a symmetric, elongated body with a distinctive decay signature; clutter and geology behave differently. That polarizability "fingerprint", not raw signal amplitude, is what allows reliable separation of ordnance from scrap and from hot rocks.

On top of the physics-based models sit statistical classifiers. Random forests, neural networks, and physics-informed models trained on emplaced-target libraries or synthetic data can suppress false positives substantially, which is precisely the gap the University of Hawaii portable-sensor and machine-learning work is aimed at. Two caveats belong in any survey design. First, ground-penetrating radar, often proposed as a complementary sensor, performs poorly in Hawaiian conditions because moist, iron-oxide-rich, electrically conductive soils attenuate the radar signal severely. Second, robotic and drone-borne carriage plus repeat-survey change detection are well suited here: they reduce operator exposure on steep terrain and, by re-flying a grid after each storm, directly address the migration problem rather than assuming a static target field.

Analysis of Effects: ICM and Submunition Legacy

The most operationally significant ERW category in Hawaii is the residue of Improved Conventional Munitions (ICM), including dual-purpose improved conventional munitions (DPICM) and analogous air-delivered submunitions used at sites such as Makua Valley on the Wai‘anae Coast. EOD personnel cite an initial dud rate of approximately 30% for the historical ICM populations encountered in the islands. That figure sits at the high end of, but is consistent with, the wider record: acceptance-test dud rates for legacy DPICM families (M42, M46, M77 and related) are commonly quoted between about 2% and the mid-teens, but field rates after vegetation interception and soft-soil or oblique impact run considerably higher.

This is not an academic point. The US statutory threshold for transfer or export of cluster munitions is a post-arming unexploded-ordnance rate of no more than 1% across intended operational environments, and the US Army ceased procurement of the GMLRS DPICM warhead in 2008 precisely because its submunition dud rate ran up to 5%. The Congressional Research Service notes that delivery technique, soft or muddy ground, and vegetation interception can all push real-world failure well above laboratory figures even where self-deactivation features are fitted. The Hawaiian legacy population predates these controls entirely.

Storm-driven migration defeats any clearance certification tied to a fixed surface grid. Post-storm re-survey is a planning requirement, not a contingency.

Submunition behaviour after impact is the core hazard driver. Failure modes typically leave the M223-family or analogous fuze in an armed or part-armed state, with the shaped-charge liner and explosive fill (commonly Composition A-5 or Composition B in legacy stocks) intact. The March 2026 Kona-low storm events that "washed things down and off the mountains" into a creek bed at Makua are the local illustration of a general rule: on steep, erodible volcanic terrain, the contaminated footprint is not stationary.

Munitions Constituents: Fate and Transport

Detection and disposal address the intact item. A second, slower hazard is chemical. As legacy casings corrode, energetic fills and their breakdown products enter the soil. The compounds of concern are RDX (cyclotrimethylenetrinitramine), HMX (cyclotetramethylenetetranitramine), TNT (trinitrotoluene), and, where present, perchlorate from propellants and certain fuze trains. Hawaiian volcanic soils are frequently acidic, with pH commonly in the 4 to 6 band, and coastal sites carry marine salt influence; both conditions accelerate casing corrosion and the mobilisation of constituents. RDX and perchlorate are relatively mobile and can migrate toward groundwater, while runoff pathways on coastal ranges raise a nearshore reef-ecosystem question. Munitions-constituent fate is therefore a distinct workstream from the explosive-hazard clearance, with its own sampling, modelling, and institutional-control requirements; it should not be folded into the UXO line of effort by default.

Remediation Options Beyond Detonate-in-Place

The default render-safe option of detonation in place (BIP) is constrained at many Hawaiian sites by archaeological, ecological, and wildfire-risk overlays. A fuller option set is worth specifying at the planning stage:

Low-order disposal. Controlled low-order techniques and shaped-charge disruptors can neutralise an item with far less fragmentation and blast than a high-order BIP, reducing collateral effect on cultural and habitat features. They demand a more confident item identification and a higher handling-skill threshold.

Soil management for constituents. Where munitions constituents are the driver rather than intact items, excavation and mechanical screening, thermal desorption, or in-situ bioremediation are established treatments; selection turns on contaminant, concentration, and the sensitivity of the receiving environment.

Risk-based clearance. Because total physical clearance of a 100,000-acre former range to depth is not realistically fundable, a Munitions and Explosives of Concern Hazard Assessment (MEC HA) and probabilistic residual-risk modelling, paired with durable institutional controls, is the governing framework. This is exactly the model the Hawaii Department of Health (HDOH) sets out in its Areawide Environmental Hazard Management Plan for the WMA.

Personnel and Safety Considerations

For EOD personnel and range-remediation contractors operating in similar volcanic-soil environments, the operational lessons are well documented but worth restating. EMI-only sub-surface survey is not, on its own, a defensible clearance methodology where soil iron content is high; multi-sensor approaches combining advanced EMI with magnetometry, selectively applied GPR, and machine-learning signature discrimination should be specified at the survey-design stage. Post-storm re-survey is a planning requirement. The movement-to-demolition-area option is the safer aggregate choice where the munition condition permits, but it increases handling exposure for the team and must be traded against the in-situ constraints case by case.

For ammunition technicians supporting deployed forces, the Hawaii experience reinforces a wider point: dud rates published at acceptance testing routinely under-state field rates once vegetation, soft soil, and oblique impact are factored in. Stockpile-management planning for current-generation munitions retaining submunition payloads, or for their replacement unitary rounds, should account for that delta.

Wider Programme Context

The Army has indicated intent to vacate the state-owned parcel at Makua Valley when the current lease ends in 2029, and federal officials are reportedly weighing relinquishment of further Oahu training land in favour of consolidating activity at the Pohakuloa Training Area on Hawaii island, whose own state lease expires in August 2029. The cost, schedule, and technical envelope of any UXO clearance prior to land transfer remains open and is, by the contractors' own assessment, a multi-generational effort. Official activity continues in parallel: U.S. Army Garrison Hawai‘i announced a UXO clearance operation at Makua for 20 January 2026, with controlled detonations and full public exclusion, explicitly to protect conservation and cultural-access work. The Hawaii case is also instructive for NATO range-remediation planners working legacy ICM-contaminated sites in continental Europe.

Forward Look: Emerging Technology

Several technology lines under SERDP/ESTCP and allied programmes bear on the Hawaiian problem. Drone-borne multi-sensor arrays extend reach over terrain that is unsafe or impassable for foot survey. Quantum and optically-pumped magnetometers offer higher sensitivity and lower noise floors, potentially improving the soil-versus-target contrast that defeats conventional magnetometry. Confirmatory neutron-based interrogation of the PELAN (Pulsed Elemental Analysis with Neutrons) type can identify explosive fill in a suspect item without intrusive access, helping triage what must be treated as live. None of these removes the soil-physics constraint, but each chips at it, and several have moved from demonstration toward field use.

Official References and Further Reading

AI-assisted technical assessment based on open-source material. Not a formal intelligence product. Source reliability assessed B/2 (named EOD personnel, named contractor, named academic, all on-the-record to a regional newspaper of record), corroborated by HDOH, CRS, SERDP/ESTCP, and U.S. Army Garrison Hawai‘i primary sources. Newly added quantitative ranges (magnetic susceptibility, detection depth, soil pH) are indicative figures from the geophysical and environmental literature and require site-specific measurement before operational use. Imagery is U.S. Department of Defense visual information distributed via DVIDS and is in the public domain; its appearance does not imply or constitute DoD endorsement. © 2026 Integrated Synergy Consulting Ltd. Open Source / Unclassified.