Friday, April 13, 2012

The Value of RadNet Ionizing Radiation Detection

In this post, I discuss uses of ionizing radiation measurements made available to the public by the U.S. Environmental Protection Agency's RadNet stations. In particular, I examine the usefulness of RadNet data for the early detection of radioactive fallout from distant radiological emergencies. As example, I use the reactor accidents at the Fukushima Daiichi Nuclear Power Station in the wake of the Mar. 11, 2011, Tōhoku-chihō Taiheiyō Oki Earthquake and Tsunami.

Iodine-131 and cesium-137 are prominent radioisotopes released after severe nuclear reactor accidents with fuel meltdowns. The decay of these isotopes produces a mix of beta and gamma radiation. The gamma radiation is emitted when the atomic nuclei transmute through intermittent energy states at isotope-specific energies.
Near-real time air filter gamma gross count rates measured in nine energy ranges at the RadNet station in Harrisonburg, Virginia (courtesy EPA). Because of the disparate values, the count rates are scaled logarithmically on the ordinate, compressing the peaks at high count rates. 
I used air filter measurements from the RadNet station in Harrisonburg, Virginia. The EPA publishes graphs of hourly gamma radiation gross count rates in nine energy ranges (see graph above).

Gamma spectrum of iodine-131 (Arena, 1971). Counts/channel are plotted on a logarithmic scale versus energy [keV]. Peaks are labelled in MeV.
Iodine-131 with a physical half-life of 8.05 days and an effective half-life in the whole body of 7.6 days possesses its greatest energy peak at 360 keV, a second-ranking peak at 280 keV, a third peak at 638 keV and a fourth peak at 724 keV (Arena, 1971). The lower couple fall into range 3 (200-400 keV) , whereas the higher couple fall into range 5 (600-800 keV) in the RadNet graph.

Gamma spectrum of cesium-137 (Arena, 1971). Counts/channel are plotted on a logarithmic scale versus energy [keV]. Peaks are labelled in MeV
By contrast, cesium-137 with a physical half-life of 30.17 years and an effective half-life in the whole body of 70 days emits gamma radiation at 662 keV, falling into range 5 of the RadNet graph. However, backscatter (Compton effect), that is low-energy detector counts produced by incomplete energy transfer between the ionizing radiation and the detector material, contributes a considerable fraction of the total count rate to range 3. Although the precise shape of the spectral curves shown above depends on the radioactivity of the sources and the instruments used for the measurement, the energy peaks remain invariable and representative. Therefore, the spectra may can be employed for the demonstration of principles.

Partial integration of the areas under the spectral curves taking the logarithmic scale of the counts/channel into account suggests that the iodine-131 decay contributes about 91 percent of the total count rate summed over both ranges to range 3, whereas cesium-137 decay will contribute about one third to this range. Therefore, if both isotopes are present in the sample, the count rate measured in range 3 does not exclusively reflect iodine-131 decay, and iodine-131 will also contribute to the count rate measured in range 5, though to a smaller degree than cesium-137. Despite this cross-contamination, a prominent increase in range 3 suggests the presence of iodine-131 and in range 5 that of cesium-137. Furthermore, the EPA provides offline post-hoc data for identified radioisotopes.

Regardless of its low spectral resolution, the real-time RadNet graph may potentially have its uses for the identification of a radiological incident. An increase above the average counts per minute (CPM) measured in ranges 3 and 5 beyond three standard errors of the mean can be considered statistically significant with 95 percent confidence. Mean count rates measured in the past can be queried in the RadNet database. Sequences of up to 400 measurements can be downloaded in a batch.

To determine mean count rates for the two energy ranges of interest, that is ranges 3 and 5, I chose the period between Feb. 22 and Mar. 10, 2011, and subsequently compared the means averaged over this epoch to the count rates observed in the same epoch this years. Radioactive decay adheres to the Poisson distribution in which the standard error is equal to the square root of the mean count rate. If the current count rate exceeds the mean by three standard errors, the probability of the increase being random is less than five percent.
Graph of gamma gross count rates measured in nine energy ranges taken from air filter samples at the RadNet Station in Harrisonburg, Virginia. The bottom lines of the red boxes indicate three standard errors above the mean count rate of the same epoch a year ago.
The graph above shows that this year's count rates statistically significantly exceeded the mean in the examined time window on roughly a handful of occasions. However, most comprise all energy ranges except 9 (magenta). If increases were only detected in range 3 (blue), we would most likely be confronted with a statistically significant presence of iodine-131. By contrast, a statistically significant increase in ranges 3 (blue) and 5 (yellow) would most likely indicate the presence of cesium-137.

Concomitant increases in gross beta count rates would further affirm the conclusions above, because both radioisotopes also emit beta radiation. Note, however, that the above assertions are purely based on statistical probability and causal inference. Great care must be taken to establish reasonable cause.

Iodine-131 and cesium-137 can be distinguished with time, using the difference in physical half-life between the two isotopes. Iodine-131's half-life is at 8.05 days considerably shorter than that of cesium-137 at 30.17 years. Following a pulse release of both isotopes into the atmosphere, iodine-131 should mainly contribute to the count rates measured early after the release. When the samples are remeasured ten half-lives later, that is after 80 days, the contribution of iodine-131 to the count rate will have diminished to one-thousandth. By contrast, the contribution of cesium-137 with its 30-year half-life is going to persist. Therefore, if repeated measurements of the same sample ascertain a decline in count rate commensurate with iodine-131's half-life, the radioisotope was present in the sample, and the remainder ought to be cesium-137.

Though the RadNet measurements may not immediately lend themselves to the distinction between iodine-131 and cesium-137, they may prove useful for the detection of radioactive fallout from a distant radiological accident. To test this idea, I tapped into the data collected at Harrisonburg in the weeks after the catastrophic nuclear reactor failures at the Fukushima Daiichi Nuclear Power Station situated on the shores of the Pacific Ocean 160 miles north of Tokyo in early March last year, that is between Mar. 12 and Mar. 28, 2011.

The station's reactors incurred loss of cooling in the wake of a magnitude-9 earthquake on Mar. 11, 2011, followed by 15-meter high tsunami waves inundating the structures (TEPCO press release with the title "Analysis and evaluation of the operation record and accident record of Fukushima Daiichi Nuclear Power Station at the time of Tohoku-Chihou-Taiheiyou-Oki-Earthquake," dated May 23, 2011). The nuclear fuel in three operating reactors melted down, producing great amounts of hydrogen. Between Mar. 12 and Mar. 15, hydrogen that had accumulated in the reactor buildings triggered massive explosions devastating the top floors of the structures and releasing vast amounts of radioactive matter into the atmosphere.

The first unit, Unit 1, exploded in the early afternoon of Mar. 12. The second unit, Unit 3, exploded two days later, and the third unit, Unit 4, though it was shutdown for inspection at the time, incurred an explosion in the early morning of Mar. 15. The reactor building of Unit 2, which was operating, suffered minor visible damage. However, since Unit 2's fuel melted, radioactivity was released from this reactor as well.

 Gross count rates from air filter samples collected at Harrisonburg, Virginia, over roughly two weeks after the first radioactive release from Fukushima.
Iodine-131 and cesium-137 were prominent in the airborne releases from the damaged reactors. The graph above depicts the gamma gross count rates collected in nine energy ranges at Harrisonburg over roughly two weeks after the first radioactive release from Fukushima, the statistical significance thresholds (dashed black lines) in ranges 3 (dark blue line) and 5 (yellow line), as well as the gross beta count rate (dashed red line). Peaks of significant gamma gross rate increases are evident in almost all energy ranges. None was confined exclusively to ranges 3 and 5. The light-blue line above the abscissa indicates the period in which the Fukushima Daiichi Nuclear Power Station incurred hydrogen explosions.

March 14 was the first day after the Fukushima Daiichi Nuclear Power Station began to release vast amounts of radioactivity, on which Harrisonburg registered significant peaks in gamma ranges 3 and 5 concomitant with an increase in beta gross count rate (dashed red line). A second such coincidence followed on Mar. 15. Then a triplet of coincident peaks followed on March 17, 19 and 20 (light blue dots above the abscissa). Because the peaks are spaced in about the same time intervals as the three hydrogen explosions at the Fukushima Daiichi Nuclear Power Station, this triplet of peaks represents the most probable harbinger of the arrival of fallout from Fukushima in Virginia, suggesting that radioactive airborne particulate from Fukushima reached Harrisonburg in five days.

The suggested time of arrival is a day ahead of that predicted by the cesium-137 dispersion model of Winiarek and others (2011) at the Centre d'Enseignement et de Recherche en Environnement Atmosphérique (CEREA), Marne la Vallée, France, and two days in advance of that of the iodine-131 dispersion modeled by Wotawa (2011) at the Zentralanstalt für Meteorologie und Geodynamik (ZAMG), Vienna, Austria.

Global iodine-131 dispersion after the Fukushima reactor accidents, 2011, according to the simulation model by Wotawa (2011), ZAMG, Vienna, Austria. 

Taking the above findings together, RadNet air filter count rate measurements seem to provide superb sensitivity for the detection of minute traces of airborne radioactive material, permitting us to identify distant radiological accidents half ways around the globe, if examined in proper factual context. Certainly, RadNet stations represent potent tools capable of alerting us to fallout from radiological accidents closer to home.