Emission and transport of cesium-137 from boreal biomass burning in the summer of 2010



[1] While atmospheric concentrations of cesium-137 (137Cs) have decreased since the nuclear testing era, resuspension of 137Cs during biomass burning provides an ongoing emission source. The summer of 2010 was an intense biomass burning season in western Russia, with high levels of particulate matter impacting air quality and visibility. A radionuclide monitoring station in western Russia shows enhanced airborne 137Cs concentrations during the wildfire period. Since 137Cs binds to aerosols, satellite observations of aerosols and fire occurrences can provide a global-scale context for 137Cs emissions and transport during biomass burning events. We demonstrate that high values of the Moderate Resolution Imaging Spectroradiometer aerosol optical depth coincide with detections of 137Cs, and use the relationship between 137Cs and aerosols to model 137Cs based on organic carbon emissions and transport with the Goddard Earth Observing System, version 5 model. The model's boreal biomass burning tracer explains approximately half of the daily variability in detected 137Cs concentrations at a monitoring station in western Russia. Constraining the model with the station observations, we calculate 137Cs emissions of 1.5 × 1012 Bq from biomass burning north of 40° in July and August 2010. The emissions and subsequent deposition lead to a small northward redistribution of 137Cs.

1. Introduction

[2] Cesium-137 (137Cs) is an anthropogenic radionuclide with a half-life of 30 years. This lifetime, along with its bioavailability, makes it a concern for human health. The bulk of 137Cs entered the environment from nuclear weapons testing, with the peak atmospheric fallout occurring in the 1960s [United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR), 2000a]. 137Cs was deposited throughout the northern hemisphere, with the maximum deposition occurring in the latitudinal band around 45°N [Aoyama et al., 2006]. The Chernobyl accident in 1986 introduced an additional pulse of 137Cs to the environment [UNSCEAR, 2000b], especially in Europe. Although our data precede the 2011 Fukushima-Daiichi accident, this event also liberated a large amount of 137Cs into the environment.

[3] Atmospheric concentrations of 137Cs decreased following the nuclear-testing period, and by 2010 were very low. However, 137Cs is still present in soil and vegetation, and remobilization from these reservoirs back to the atmosphere can provide a small ongoing source.

[4] Knowledge of sources of atmospheric 137Cs is important to the nuclear explosion monitoring community since remote detection of 137Cs and other radionuclides can be indicative of a nuclear explosion [Preparatory Commission for the Comprehensive Nuclear-Test-Ban Treaty Organization, 2011]. An International Monitoring System (IMS) for the detection of airborne radioactivity was designed and is being implemented for the Comprehensive Nuclear-Test-Ban Treaty (CTBT).

[5] Studies show increased 137Cs associated with African dust events over the Canary Islands [Hernandez et al., 2005; Karlsson et al., 2008] and France [Menut et al., 2009], and with Asian dust over Japan [Fukuyama and Fujiwara, 2008]. The ratio of 137Cs to 90Sr can provide information on the origin of observed dust [Igarashi et al., 2001, 2005].

[6] Biomass burning provides another mechanism for the remobilization of 137Cs [Paliouris et al., 1995]. Amiro et al. [1996] determined that 40–70% of the 137Cs in the fuel is released to the atmosphere during a typical field fire, and release increases with temperature. The higher levels of 137Cs still present in northern ecosystems [Paliouris et al., 1995] make boreal fires a particular concern for 137Cs resuspension. Several studies have examined radionuclide resuspension and the potential exposure of firefighters and other populations from wildfires in areas near Chernobyl [Kashparov et al., 2000; Yoschenko et al., 2006; Hao et al., 2009]. The modeling study of Wotawa et al. [2006] shows that biomass burning emissions in boreal North America and Asia can explain observed summertime increases in 137Cs concentrations at an IMS station located in Yellowknife, Canada in 2003 and 2004, and demonstrates the impact of intercontinental transport on atmospheric 137Cs. Bourcier et al. [2010] examined the correlation between 137Cs and levoglucosan, a product of cellulose pyrolysis used as a specific tracer of biomass burning. They found significant covariation between 137Cs and levoglucosan, indicating a common source from biomass burning.

[7] The summer of 2010 was an intense biomass burning season in western Russia, with high levels of air pollutants from wildfires reaching Moscow. The fires were associated with an intense heat wave that began in June, intensified around 18 July and ended with a cold front passage on 18 August [Lau and Kim, 2012]. The heat wave was caused by atmospheric blocking [Dole et al., 2011; Matsueda, 2011], and, during the period of intense fires, anticyclonic flow brought smoke into Moscow [Witte et al., 2011].

[8] Fires in the Bryansk region raised concerns about whether radioactive particles could be released to the atmosphere (N. Gilbert, Russia counts environmental cost of wildfire, 2010, http://www.nature.com/news/2010/120810/full/news.2010.404.html). This study focuses on 137Cs emissions from Russian biomass burning in July and August 2010. We compare global atmospheric model output to ground-based measurements of 137Cs from IMS stations and to satellite measurements of aerosol optical depth (AOD) to constrain the boreal biomass burning source of 137Cs during this intense fire season. We also examine the subsequent particulate transport and deposition of 137Cs.

2. Observations of 137Cs and Aerosol Optical Depth

[9] The Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) is constructing and testing radionuclide monitoring stations as part of its International Monitoring System [Hoffmann et al., 1999]. Atmospheric 137Cs is expected to be in aerosol form [Yoschenko et al., 2006]. IMS stations contain equipment to measure radionuclides bound to aerosols [Schulze et al., 2000]. They provide daily data on 137Cs and other radionuclides, through the use of high-purity germanium detectors which measure the radionuclides deposited on filters that collect aerosols from large air volume samplers. For most of the summer of 2010, 137Cs concentrations were below the detection limit (typically a few μBq m−3, depending on the station) at many of the IMS stations. This study focuses on two stations located at high latitudes that detect a signal from biomass burning. These are the Yellowknife, Canada station (114.5°W, 62.4°N) previously described by Wotawa et al. [2006], and the Dubna site (56.7°N, 37.3°E) in the Russian Federation, which is located near the region of biomass burning in western Russia.

[10] Satellite measurements of aerosols can provide additional information related to biomass burning emissions and particulate transport of 137Cs. This study used the 550 nm aerosol optical depth from the Moderate Resolution Imaging Spectroradiometer (MODIS) on the EOS Aqua satellite, which is in a sun-synchronous orbit with a 13:30 equator-crossing time. MODIS provides nearly global daily coverage [Levy et al., 2009]. We obtained daily gridded data from the Giovanni online data system [Acker and Leptoukh, 2007].

[11] Figure 1 shows the relationship between MODIS AOD and 137Cs at Dubna and Yellowknife during the summer of 2010. Many of the days on which 137Cs is detected at Yellowknife also show enhanced AOD. In Dubna, AOD increases dramatically in late July and early August due to the intense biomass burning in the region during that period. While AOD values were typically below 0.5 in June and early July, they reach values up to three in early August. A corresponding increase is also evident in both the frequency of detection and the concentration of 137Cs during the fire period, suggesting that resuspended 137Cs is present in the biomass burning aerosol. Considering both sites together, the mean AOD on days when 137Cs was detected is 1.2, while the mean on days without a 137Cs detection is 0.18. This difference is statistically significant at the 95% level based on a Student's t test.

Figure 1.

Comparison of MODIS AOD (black lines, left axis) at (a) Dubna and (b) Yellowknife with 137Cs concentrations (gray circles, right axis) during the summer of 2010. Error bars represent the uncertainty in the 137Cs measurements.

[12] The peak value of 137Cs at the Dubna station, 32.83 μBq m−3 on 6–7 August, is a particularly prominent feature in the time series (Figure 1a). The MODIS aerosol optical thickness and the aerosol index (AI) from the Ozone Monitoring Instrument (OMI) reached peak values over Moscow on 7 August [Witte et al., 2011], and high AOD values were observed during early August at the Moscow-MSU-MU AERONET site as well [Mei et al., 2011]. Yurganov et al. [2011] also observed a peak in carbon monoxide (CO) on this day.

3. Model Simulation

[13] We use the Goddard Earth Observing System, version 5 (GEOS-5) global atmospheric general circulation model (AGCM) to examine the sources of the observed aerosol and 137Cs concentrations, and to test the consistency of our emission estimates with the observations. The model is constrained by the Modern Era Retrospective-Analysis for Research and Applications (MERRA) meteorological fields [Rienecker et al., 2011] to reproduce the winds and temperatures of the period of study. The horizontal resolution is 1.25° longitude by 1° latitude. Aerosols are simulated online within the GEOS-5 AGCM using the Goddard Chemistry, Aerosol, Radiation, and Transport (GOCART) model [Chin et al., 2002; Colarco et al., 2010]. The aerosol simulation includes black and organic carbon, sulfate, dust, and sea salt aerosols. We use a factor of 1.4 to convert organic carbon to particulate organic matter (POM).

[14] Biomass burning emissions of organic carbon and other aerosols come from the Global Fire Emissions Database version 3 (GFED3) [van der Werf et al., 2010] with daily scaling factors [Mu et al., 2011]. An initial simulation indicated that the emissions were too low for the model to reproduce the satellite-observed AOD over western Russia during the summer of 2010. As a result, we modified daily emissions for North American, eastern and western Russian regions to reduce the model bias. The modification applied a daily varying, region-specific scaling factor to each of the three regions based on the ratio of MODIS to modeled AOD in the initial simulation over the region. The regions were chosen so that biomass burning would be the main contributor to the AOD of the region. Within each region, the scaling factor does not vary spatially, since the AOD of a single grid box is influenced by transport from surrounding areas as well as by emission into the grid box. While the resulting emissions are imperfect, they do reduce the bias in our simulation compared to the simulation with the original emissions. The improved simulation is used for the rest of this study.

[15] Figure 2 compares MODIS and modeled optical depth on 7 August, a day of strong biomass burning in western Russia. Both MODIS and the model show a strong AOD enhancement over western Russia, as well as a smaller enhancement over northern Canada. However, the model is biased low on this day and does not capture the strength of the observed enhancement in western Russia (Figures 2a and 2c). The AOD contribution from POM released by biomass burning (Figure 2d) shows the same spatial pattern as the total AOD, confirming that these enhancements are due to biomass burning.

Figure 2.

(a) Aerosol optical depth on 7 August 2010 from MODIS overplotted with the locations of the Dubna (D) and Yellowknife (YK) stations in blue. Areas without a valid retrieval are shown in white. (b) Model AOD overplotted with the locations of the tagged tracer regions east and west of Moscow (black boxes). (c) Model AOD shown only where valid MODIS data are present for easier comparison with Figure 2a. Areas without a valid MODIS retrieval are shown in white. (d) The boreal biomass burning POM contribution to AOD with colors corresponding to green color bar. Model results are sampled in the early afternoon in each region for consistency with the Aqua overpass.

[16] We simulate the concentration and deposition of 137Cs by multiplying the modeled POM tracers by the ratio of 137Cs emission to POM emission from biomass burning, since we assume that particles containing 137Cs undergo the same transport and deposition processes. We neglect the radioactive decay of 137Cs since the 30 year half-life is long compared with the time scale of atmospheric deposition and our period of study. Regressing the modeled POM against the Dubna 137Cs data, we obtain a 137Cs/POM ratio of 0.23 kBq kg−1, and validate the resulting 137Cs concentrations against the observations from Yellowknife. This method allows us to utilize the global coverage of MODIS AOD as well as the 137Cs station data for model validation and to optimize our estimate of 137Cs emissions.

4. Results

[17] Figure 3 compares the modeled 137Cs to observations at Yellowknife and Dubna. For Dubna, we sample the model one grid box north of the station coordinates where the correlation is stronger. This may reflect uncertainty in model transport or in the fire emissions. 137Cs was undetected on many days at Yellowknife, so no observations are shown on those days. The model reproduces the timing of many of the observed peaks at both sites, suggesting that the model captures much of the impact of biomass burning on 137Cs concentrations (Figures 3a and 3b). However, there are errors in the magnitude of the enhancement on individual days, which could be due to inhomogeneous 137Cs distributions between different burn areas, or to errors in the strength or location of our biomass burning aerosol emissions. The true value of 137Cs/POM is expected to vary regionally due to the variable levels of 137Cs in soil and vegetation. Furthermore, since AOD is a column measurement, it does not fully constrain the surface concentration of aerosol, and model error in the vertical distribution of the biomass burning plumes could contribute to the mismatch with observed 137Cs on some days.

Figure 3.

Observations (asterisks) and model total tracer (black line) of 137Cs concentrations at (a) Dubna and (b) Yellowknife in July and August 2010. Model tagged tracers for boreal biomass burning (bbbo, red), biomass burning from 40°N to 50°N (bb30–50, purple), western Russian biomass burning (bbru, cyan), biomass burning in the southwestern (SW box, blue) and southeastern (SE box, pink) boxes, and North American biomass burning (bbna, green) are shown for (c) Dubna and (d) Yellowknife. Some tracers are scaled as described in the legend to fit on a single axis. The scaling factor is shown in the legend.

[18] We examine the origin of the 137Cs at Dubna using tagged tracers for biomass burning in different regions (Figure 3c). The tracer for western Russian BB (cyan dashed line) dominates the boreal biomass burning (red line) and total (black line) tracers at Dubna (Figures 3a and 3c). These tracers have an r2 correlation with the detected 137Cs observations of 0.49, indicating that they explain approximately half the variance in observed concentrations. Given the heterogeneity of 137Cs deposits, we examine the possibility that emissions from particular regions might make greater contributions to the observed concentrations.

[19] Figure 3c shows tagged tracers for the 40°N–50°N latitude band (purple line), where zonal mean 137Cs fallout was largest, for a box southwest of Moscow that includes Chernobyl (blue line), and for a box southeast of Moscow (pink line). Figure 2b shows the location of these boxes. The southeastern box has the highest correlation with the observations, r2 = 0.53, while the southwest box does not correlate well with the observations. The 40°N−50°N tracer has a weaker correlation of r2 = 0.34. Since the range of 137Cs concentrations is truncated by the detection limit, the data cannot be considered normally distributed. We therefore apply a nonparametric test, Spearman's rho, to determine the statistical significance of the tagged tracers. Based on this test, all of the tracers except for the southwestern box and the North American tracer have a statistically significant correlation with the Dubna observations. The southeastern box has the highest rho value, consistent with the results of the t test.

[20] Witte et al. [2011] used back trajectories to show that during the peak fire period southeasterly flow brought air from regions impacted by biomass burning into Moscow. Our strong correlation for the southeastern box is consistent with that result. Although the southwestern box would be expected to have more 137Cs available for resuspension, it is important to note that the southwestern box tracer in Figure 3c is scaled by a factor of 10 to be visible on the plot. This shows that the contribution of the southwestern box to the POM over Dubna is much smaller than the contribution from the southeastern box. Figure 4 shows that the 2010 burned area and organic carbon emissions from the southeastern box greatly exceed those of the southwestern box, leading to the dominant contribution of the southeastern box.

Figure 4.

The 1997–2010 (a) burned area and (b) organic carbon emissions from GFED3 for the southeastern (black bars) and southwestern (gray bars) boxes. The box definitions are the same as in Figure 3.

[21] At Yellowknife, the total tracer is dominated by North American biomass burning (green dashed line) and shows peaks corresponding to the 137Cs detections (Figures 3b and 3d). The 137Cs/POM ratio determined for Dubna yields reasonable magnitudes for the model compared to observations at Yellowknife as well. However, the model also shows peaks at Yellowknife on days with no 137Cs detection and overestimates the magnitude of the observed peak.

[22] The 137Cs/POM ratio leads to an estimate of 1.5 × 1012 Bq 137Cs emitted from biomass burning north of 40° in July and August 2010, of which 1.2 × 1012 Bq was in Eurasia and the rest in North America. For comparison, Wotawa et al. [2006] estimated that 4.3 × 1012 Bq and 2.2 × 1012 Bq were emitted by boreal biomass burning from May–September in 2003 and 2004, respectively. Using the monthly GFED3 burned area data set [Giglio et al., 2010], we determine 137Cs emissions per area burned for Canada and Russia of 2.1 × 105 and 2.6 × 105 Bq ha−1, respectively. We note, however, that the ratio of organic carbon emission to burned area can vary from year to year (Figure 4), so our ratio of 137Cs to burned are would also differ if a different year had been considered. None the less, our estimated values lie within the uncertainty range of 1.2 × 105–2.3 × 105 Bq ha−1 for North America and 0–5 × 105 Bq ha−1 for Siberia determined by Wotawa et al. [2006] for the 2003 and 2004 fires.

[23] A comparison of model deposition and emissions (Figure 5) shows the net effect of July and August boreal biomass burning on the surface distribution of 137Cs. Atmospheric transport acts to spread the emissions from confined biomass burning over broader regions where it is deposited (Figures 5a and 5b). The net effect is a northward shift in deposition compared to emissions (Figure 6). Average surface concentrations of 137Cs in the Northern Hemisphere midlatitudes are around 2 kBq m−2 [Wotawa et al., 2006]. Thus, the modeled redistribution of 137Cs from July and August 2010 is small compared to the existing land concentration. However, on regional scales our peak modeled 137Cs deposition from boreal biomass burning (Figure 7) reaches values comparable to those measured during dust events. Fukuyama and Fujiwara [2008] measured 137Cs deposition of 62.3 mBq m−2 week−1 during an Asian dust event over Japan, while the peak weekly total boreal biomass burning deposition values in this study reach hundreds of mBq m−2 week−1. Biomass burning occurs regularly in the southwestern and southeastern boxes discussed above (Figure 4). However, the 2010 emissions were anomalously high (Figure 4b). In the future, a multiyear study would be useful for assessing the long-term impact of biomass burning on the 137Cs distribution.

Figure 5.

(a) Total boreal biomass burning emissions (red contours) for July and August 2010 overplotted on the maximum 137Cs column density (gray-filled contours) at each location for the period. (b) Net surface flux of boreal biomass burning 137Cs tracer, defined as deposition minus emission.

Figure 6.

Latitudinal distribution of zonal mean 137Cs emission (black) and deposition (gray) fluxes for the boreal biomass burning tracer in July and August 2010.

Figure 7.

Maximum modeled weekly 137Cs deposition value reached in July–August at each location.

[24] The accident at the Fukushima power plant in 2011 introduced additional 137Cs to the atmosphere. Although concentrations decreased by orders of magnitude with distance from accident, 137Cs was detected at sites around the northern hemisphere [Hsu et al., 2012]. In Europe, far downwind of the accident, observations show 137Cs concentrations reaching several hundred μBq/m3 in Stockholm [Stohl et al., 2011]. These values are much higher than the observations from the Russian fires. Deposition from the Fukushima accident has the potential to increase the resuspension of 137Cs during future wildfires.

5. Conclusions

[25] Biomass burning during the summer of 2010 led to high concentrations of aerosols and other atmospheric pollutants. We find that increases in aerosol optical depth in regions affected by burning are also related to increased 137Cs detections. Satellite measurements of fire activity and AOD can thus provide additional information on the emission and transport of 137Cs from biomass burning. Observations of high 137Cs concentrations at the Dubna station coincide with high aerosol optical depth detected by the MODIS satellite during a period of intense wildfires.

[26] We use the GEOS-5 AGCM with an embedded GOCART aerosol model to simulate aerosol concentrations during July and August 2010, and relate the modeled organic carbon to 137Cs based on an emission ratio. This approach allows us to use MODIS AOD as an additional tool to validate the model simulation, providing far greater spatial coverage than would be available from 137Cs concentration data alone. The model's boreal biomass burning tracer has a statistically significant correlation with detected 137Cs concentrations observed at a site near the wildfires in western Russia during July and August 2010. Since 137Cs has a limited number of sources to the atmosphere, it can be a useful tracer of the transport of biomass burning emissions.

[27] Constraining the model with 137Cs observations yields a best guess estimate of 0.23 kBq 137Cs kg−1 POM, leading to total boreal biomass burning 137Cs emissions of 1.5 × 1012 Bq for July and August 2010. This corresponds to emissions of 2.1 × 105 and 2.6 × 105 Bq ha−1 for Canada and Russia, respectively, consistent with the 2 × 105 Bq ha−1 estimate of Wotawa et al. [2006].

[28] Satellite-based information on AOD, ground-based measurements of radioactivity, and other information could be combined with atmospheric transport modeling to better constrain 137Cs emission sources. The correlation between observed measurements using AOD and the ground-based IMS radionuclide sensors imply that resuspended 137Cs can be strongly tied to phenomena such as wildfires. The net effect of the biomass burning emissions and subsequent deposition in our period of study is a small northward redistribution of 137Cs. Boreal forest fires are a recurring phenomenon, and the methods described in this paper could be extended to help predict large-scale impacts of future fire scenarios on the 137Cs distribution.


[29] We are grateful to NASA for funding this work through the Modeling, Analysis and Prediction program and for providing high-performance computing resources on “Discover” at the NCCS. The Giovanni online data system is maintained by the NASA GES DISC. We thank Arlindo da Silva and Anton Darmenov for helpful discussions.