International Network of Engineers and Scientists Against Proliferation

Global Modeling of Atmospheric Krypton-85 Concentrations

Daeok Youn, Donald J. Wuebbles, Martin B. Kalinowski

This study documents the derived atmospheric distribution of ⁸⁵Kr from using a recent new compilation of the annual emissions from all known sources (e.g., reprocessing facilities) for 1945-2000[1] for this radioactive noble gas in a state-of-the-art three-dimensional (3D) model of global atmospheric physical processes. Such an atmospheric inventory could be useful in future studies aimed at detecting the clandestine separation of plutonium possibly related to nuclear weapons production.[2] After a brief description of the global atmospheric model, results from this study are presented followed by a discussion of further work needed for better resolving the source attribution in future studies.

Global Modeling Tools

Both Eulerian (grid-based) models and Lagrangian (parcel-following) models have been used in past studies of radioactive isotopes in the atmosphere. Eulerian atmospheric models, such as the model used here, have the advantage that they take into account fully 3D descriptions of the meteorological fields, including many vertical levels, rather than only considering single trajectories. Moreover, they can incorporate complete 3D transport and other relevant physical processes in the atmosphere throughout the spatial and temporal domain of interest. However, when used with fixed meshes, Eulerian models show difficulty in resolving steep gradients from a single point source. If a coarse Eulerian mesh is used then the release is immediately averaged into a large area, which smears out the steep gradients and creates a large amount of numerical diffusion. The result will be to underpredict the maximum concentrations within the near-field plume and to overestimate the plume width. This problem, however, can be addressed, as it is here, by enhancing the model resolution or by nesting a finer resolution grid to better resolve steep gradients close to the source.

On the other hand, the Lagrangian particle models (that follow the movement of air masses) mostly focus on atmospheric processes at a single location close to a single point source. While quite accurate in such situations, the Lagrangian model results tend to be less reliable when long-range atmospheric transport of the emitted gas must be considered. They cannot readily treat vertical structure including changes in wind speed and direction with altitude, and pollutant transport in layers aloft. The Lagrangian particle models, however, can be easily adapted for assessment of all possible emission sources on the given receptor. Global-scale Lagrangian parcel models are able to compute complete fully 3D atmospheric physical processes, but have the problem of reduced mixing between parcels in the upper levels. They need an arbitrary parameter to compensate for this problem. Also, the global Lagrangian models demand more computational resources with an increasing number of emission sources.

3D chemistry transport models (CTMs) in the Eulerian framework have proven to be accurate research tools to explain many of airborne and satellite measurements of chemical tracers in a global or regional sense. These models have undergone extensive improvements in their capabilities over the last decade, to the point where these tools are now ready to truly aid the determination of atmospheric radioactive isotopes background and the discrimination of suspicious/unknown signals. Unlike the Lagrangian particle models, the Eulerian models are capable of easily treating a large number of emission sources with finer resolution using nested grids. A well-established 3D model with state-of-the-art representation of atmospheric processes is used in this study.

Evaluation of the Global Atmospheric ⁸⁵Kr Distributions

Our evaluation of the atmospheric distributions of ⁸⁵Kr has been done by employing the Model of Ozone and Related Tracers (MOZART)[3] in combination with the emissions inventory mentioned above.[4] The spatial and temporal variations of the atmospheric background ⁸⁵Kr concentration as well as the dependence of this distribution on seasonal weather patterns were determined. Two 30-year (1971-2000) simulations of ⁸⁵Kr transport have been done with the latest version of MOZART (which is called MOZART-4). The simulations used two different meteorological data, one based on the assimilation of available meteorological observations, the U.S. National Centers for Environmental Prediction Reanalysis-II (NCEP R-2), on a 1.875°×1.875° grid. The other comes from a global climate model, the National Center for Atmospheric Research’s Whole Atmosphere Community Climate Model (WACCM), on a horizontal resolution of 2.8°×2.8°.

The simulated near-surface distributions of monthly ⁸⁵Kr in 1998 are shown in Figures 1 and 2. The global distribution of ⁸⁵Kr radioactivity concentration shows more detailed features than that determined in the recent annual emissions compilation.[5] The two sets of meteorology fields used here show similar general patterns, although there are some significant differences especially over Europe and Russia. The seasonal difference following surface wind patterns related to monsoon circulations is apparent in both January and July ⁸⁵Kr surface distributions from both model runs. The model-derived concentration with the NCEP R-2 (WACCM) run averaged over 1995-2000 and over the Northern Hemisphere is 1.245 Bq m^-3 (1.414 Bq m^-3) which is increasing at 33 mBq m^-3 yr-1 (37mBq m^-3 yr-1). The MOZART simulation with the WACCM shows higher values than with the NCEP R-2.

However, both of the simulated values are comparable to observed background concentrations of ~1.3 Bq m^-3 increasing at ~30 mBq m^-3 yr-1.[6]

The difference in ⁸⁵Kr concentrations and rates between these model simulations suggests that there might be a resolution dependency. However, the difference in dynamics between those two meteorological fields (one based on past observations while the other is climate-model-based) makes definitive conclusions unclear. Therefore, we will use the same dynamics core (representing all of the forces in the meteorological models) with different resolutions to better understand resolution dependency and we will use a different dynamics core with similar resolutions to better understand the dynamics dependency. The global MOZART transport model we are using can ingest any global meteorological reanalysis data. The use of the NCEP global Final Analyses (FNL) in future studies relative to the default NCEP R-2 can be used to show the dependence of model transport on the resolution. The NCEP provides the FNL data on regular 1.0°×1.0° longitude/ latitude grids covering the entire globe every six hours (4 times a day).

We also plan to compare both of these to the simulation using the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis data on 1.8°×1.8° or 1.125°×1.125° grids to test dynamics dependency. As the ECMWF dataset is generally considered to be better than the NCEP reanalysis dataset, it is worth comparing the simulations with different measurement- based reanalysis data. The global distributions of simulated radionuclides could thus be analyzed and discussed relative to the choice of model resolution and meteorological reanalysis.

Multi-Scale Modeling Approach

The comparisons of the simulated temporal variations of ⁸⁵Kr background concentrations with the observed variations at observational sites will help determine if the higher deviations (outliers) of measurement data at the sampling stations from background value are caused by unknown/accidental sources closer to the receptor sites. If the comparison shows higher deviations, the suspect signals, at a receptor site, regional- scale simulations on much finer resolution are needed for the specific focus studies requiring better representation of estimated point sources and its regional-scale transport. The regional models with appropriate resolution for the regions of interest are inevitably needed to represent temporal and spatial variation of radionuclides concentrations in the vicinity, along with global background concentrations from the global transport model with relatively coarse resolution. The regional CTM can be run in 1-way or 2-way modes interactively with the global transport model for the purpose of better understanding the influence of long distant sources on atmospheric concentration measured at a certain location. Specific regional modeling capabilities to be used will depend on the geographic region for which the INESAP Information Bulletin No. 27, December 2006 14 Figure 1: Near-surface distribution of ⁸⁵Kr concentration in January 1998 simulated by MOZART-4 using (a) WACCM and (b) NCEP R-2 meteorological data Daeok Youn, Donald J. Wuebbles, and Martin B. Kalinowski model was developed and has been successfully applied.

Atmospheric release and measurement data of ⁸⁵Kr from the same region are available for Germany and Japan.[7] The regional data for ⁸⁵Kr provide a good chance to validate the capabilities of global- to regional-scale models in transporting nuclear tracers. In addition, scenarios could examine the effects on concentrations expected if there were hypothetical releases from sensitive countries. The multiscale modeling approach will show significant improvement in diagnosis close to the plume release due to the inclusion of higher resolution model analysis.

Source-Receptor Relationship

Once the global and regional simulations of ⁸⁵Kr are successfully tested to give reasonable agreement with the atmospheric observations, the model could be used as a tool for the assessment of the source-receptor relationship. The model test of event scenarios, like hypothetical plutonium separation (or nuclear explosions for other radioactive isotopes), will likely highlight the increased modeling potential of source attribution through the global and regional transport modeling. In either Eulerian or Lagrangian models, the direct simulation of source emissions based on hypothetical source scenarios will give source-receptor relationships.

When a measurement (receptor) shows a suspicious signal, it is required to backtrack the signal for the identification of sources. The classical adjoint equation used for the calculation of back-trajectories in Lagrangian trajectory models is too simple in atmospheric transport-diffusion processes and is valid only during short time integration. A few years ago, a new approach of backtracking by means of extended adjoint transport equations considering turbulent diffusion and radioactive decay in the Eulerian approach has been discussed.[8] The application of this new inverse transport modeling to the global model using an accurate transport scheme as well as the nested regional chemical transport model will be superior to the classical back-trajectories. We expect that the extended adjoint technique applied to Eulerian 3D transport models inherently includes more of the transport mechanisms and allows longer response time at the receptor when compared with the deficiencies of Lagrangian approaches. The response at certain points or receptors could more properly mimic the sampled time-period of the measurements.

Concluding Remarks

By using the global atmospheric model, MOZART, we were able to determine temporal variations of ⁸⁵Kr concentrations as well as background values that compare well with available observations. The enhanced modeling capability through the multi-scale modeling approach described here would help clearly identify pulsed emissions of ⁸⁵Kr near reprocessing facilities. The multi-scale approach will cover finer resolutions of time and space, up to several hours and kilometers. For this purpose, further studies are needed, including:

• Analyze the effects of model uncertainties in treatment of point source, parameterization of atmospheric processes, and usage of meteorological reanalyses data on the ⁸⁵Kr distribution.
• Implement regional models with a sufficiently high spatial resolution into the global model. The regional model will be nested in suspicious areas.
• Perform studies simulating additional point releases from hypothetical sources along with hypothetical release scenarios from sensitive countries contributing to the assessment of the characteristic higher deviations under mixing with ambient concentrations from multiple reactor releases.
• Apply a method of determining the source location probability based on Lagrangian models for the identification of source regions.
• Provide additional information of source regions by means of adjoint equations applied to global and regional 3D Eulerian models.
• Compare the discrimination of source areas obtained by both backtrajectories and adjoint technique in order to get the best source-receptor relationship.

The knowledge accumulated through the atmospheric transport modeling experiments of radioactive noble gas isotope ⁸⁵Kr will greatly help us to evaluate other important radioactive isotopes such as the four radioxenon isotopes, ¹³⁵Xe, ¹³³mXe, ¹³³Xe and ¹³¹mXe, that are used for global monitoring to verify the Comprehensive Nuclear-Test-Ban Treaty (CTBT). We thus expect to enhance the ability to discriminate sources, i.e., to determine the timing of emissions relative to the source from either nuclear test explosions or nuclear reprocessing plants.

---

1.   K. Winger, F. Feichter, M. B. Kalinowski, H. Sartorius, and C. Schlosser, A new compilation of the atmospheric krypton-85 inventories from 1945 to 2000 and its evaluation in a global transport model, Journal of Environmental Radioactivity, Vol. 80, 2005, pp. 183-215.
2.   M.B. Kalinowski, H. Sartorius, S. Uhl, and W. Weiss, Conclusions on Plutonium Separation from Atmospheric Krypton-85 Measured at Various Distances from the Karlsruhe Reprocessing Plant, Journal of Environmental Radioactivity,Vol. 73, Issue 2, 2004, pp. 203-222.
3.   L. Horowitz et al., A global simulation of tropospheric ozone and related tracers: Description and evaluation of MOZART, version 2, Journal of Geophysical Research, Vol. 108, No. D24, p. 4784, doi:10.1029/2002JD002853, 24 December 2003.
4.   K. Winger et.al., op.cit.
5.   K. Winger et.al., op.cit.
6.   Y. Igarashi, M. Aoyama, K. Nemoto, K. Hirose, T. Miyao, K. Fushimi, M. Suzuki, S. Yasui, Y. Asai, I. Aoki, K. Fujii, S. Yamamoto, H. Sartorius, and W. Weiss, ⁸⁵Kr measurement system for continuous monitoring at Meteorological Research Institute, Japan, Journal of Environmental Monitoring., Vol. 3, Issue 6, 2001, pp. 688-696.
7.   M.B. Kalinowski, op.cit. Y. Igarashi, H. Sartorius, T. Miyao, W. Weiss, K. Fushimi, M. Aoyama, K. Hirose, H.Y. Inoue, ⁸⁵Kr and ¹³³Xe monitoring at MRI, Tsukuba and its importance, Journal of Environmental Radioactivity, Vol. 48, Issue 2, 2000, pp. 191–202.
8.   J.-P. Issartel and J. Baverel, Inverse transport for the verification of the Comprehensive Nuclear Test Ban Treaty, Atmospheric Chemistry and Physics, Volume 3, Issue 3, 2003, pp. 475-486.