Delayed signatures of underground nuclear explosions

Abstract

Radionuclide signals from underground nuclear explosions (UNEs) are strongly influenced by the surrounding hydrogeologic regime. One effect of containment is delay of detonation-produced radioxenon reaching the surface as well as lengthening of its period of detectability compared to uncontained explosions. Using a field-scale tracer experiment, we evaluate important transport properties of a former UNE site. We observe the character of signals at the surface due to the migration of gases from the post-detonation chimney under realistic transport conditions. Background radon signals are found to be highly responsive to cavity pressurization suggesting that large local radon anomalies may be an indicator of a clandestine UNE. Computer simulations, using transport properties obtained from the experiment, track radioxenon isotopes in the chimney and their migration to the surface. They show that the chimney surrounded by a fractured containment regime behaves as a leaky chemical reactor regarding its effect on isotopic evolution introducing a dependence on nuclear yield not previously considered. This evolutionary model for radioxenon isotopes is validated by atmospheric observations of radioxenon from a 2013 UNE in the Democratic People's Republic of Korea (DPRK). Our model produces results similar to isotopic observations with nuclear yields being comparable to seismic estimates.

Figure 1. A schematic cross section of physical/simulated system and comparison of measured and modeled chimney pressure.

(a) Model domain (width is horizontally truncated) illustrating region of detonation-produced damage and enhanced permeability. Labeled points A, B, and C in domain indicate where evolution of radioactive xenon daughters was simulated. Point D is an approximate injection point of tracer gases in chimney pressurization experiment. Red Circle represents initial cavity formation by underground explosion, and black box illustrates the chimney formed after collapse of the cavity roof. The chimney consists of a rubblized rock regime resulting from the collapse. Chimney height and damage zone in simulations are scaled to explosion yield. (b) Barometric pressure at ground surface drives pressure changes in underlying chimney region. (c) Propagation of the atmospheric pressure wave through the overburden causes a distinct pressure response in the chimney. Matching the cavity fluctuations with a model, assuming measured surface pressure fluctuations as a time-dependent boundary condition, allows a best-fit estimate for the transport properties of the overlying damaged tuff (Table 1).

Figure 2. Concentrations of Freon 12B2 at SGZ (station #5) normalized by the maximum concentration measured in the time history.

Soil gas samples of 0.1 to 0.5 L were captured beneath a tarpaulin while measuring barometric pressure and background radon concentration. Four other sampling sites produced similar results, but only SGZ data are shown here to reduce the complexity of the plot. The highest concentration occurs during the first 10 days when the chimney is weakly pressurized (∼40 mbar) by pumping air into the borehole with tracer. Sampling was performed either by extracting gases directly from the subsurface at depths of about 3 m or by taking soil gas directly from beneath tarpaulins (∼3 m × 20 m) spread over a crack or similar feature. First detected arrival occurs at SGZ about 2 days after the start of pressurization. Following the 10-day period of pressurization, all sites continue to produce high levels of tracer. The tracking of radon, normalized by highest activity measured, with tracer concentration during chimney pressurization is apparent as measured radon levels are at least 10 ∼15 fold higher during the pressurization period.

Figure 3

Simulated activity concentrations of xenon isotopes for 1 (a) and 10 kilotons (b), with four xenon isotopes arriving above the detection limit at different times. Subplot (c), zoomed plot showing details of isotopic arrivals. In addition to fracture properties as given in Table 1, we simulated heat-pipe transport resulting from the heat of detonation and phase partitioning between pore gas and liquid. The arrival of xenon isotopes occurs much earlier following detonation than for barometric pumping alone.

Figure 4. Comparison of simulated and measured activity-concentration ratios of ^131mXe/133 Xe as a function of time since detonation.

The dashed line is the prediction consistent with England and Rider assuming all parent/daughter isotopes remain in a well-mixed chimney. Red and blue lines correspond to ^131mXe/¹³³ Xe evolutionary paths appropriate for 10-kiloton and 1-kiloton detonations in a volcanic zone, respectively. The geometry of the damage models (e.g., enhanced permeability out to two-cavity radii) is based on field observations from previous underground nuclear explosions and assumes transport-property values consistent with the present study (Table 1). The inset magnifies the region on the Pahute Mesa evolutionary curves containing the Russian and Japanese observations of isotopic ratios. For comparison to the Pahute Mesa results in volcanics, we also include a parameter-sensitivity study involving variations in permeability and nuclear yield in a hard rock zone that is more representative of the DPRK test site assuming a 430 m depth of detonation. The ranges are 1 to 10 × 10⁻¹² m² for permeability, which spans estimated damage permeabilities for explosions in granite, and (a) 1 to 3, (b) 5 to 8, and (c) 9 to 10 kilotons for yield. 1σ measurement errors estimated by Ringbom et al. and given by dotted horizontal lines are found to be comparable to the parametric uncertainties of the simulated DPRK event. The green line is the evolutionary path for gases monitored at tunnel location B (Fig. 1a). Square symbols indicate ratios observed at Russian (two clustered higher-lying squares) and Japanese (3 clustered lower-lying circles) IMS radionuclide stations following the February 2013 event. Cross-over of the evolutionary paths for the 1- and 10-kiloton cases is a result of the sublinear scaling of thermally driven convection with yield-dependent heating while the production of radioxenon isotopes varies linearly with yield.

Figure 5

(a) Experiment site. Site 1 is a small tarp serviced by one sampler. Site 2 is one sampler servicing two tarps (a,b). Site 3 (see also Fig. 5d) is one sampler servicing several tubes inserted in the ground close together by nearby bluff. Site 4 is one sampler servicing tube inserted in fracture. Site 5 (SGZ) is one sampler servicing tarp laid over cement-filled emplacement shaft. The red arrow indicates the location of the drill-back hole. (b) Automated sampler as deployed at Site 2. (c) Tarpaulin installed at Site 2a. (d) Emplacing sample tubes at Site 3. (Fig. 5 photos: Sigmund Drellack).