Regional Dispersal of Fukushima-Derived Fission Nuclides by East-Asian Monsoon: A Synthesis and Review

The WRF/Chem tracer model is employed to simulate the dispersal of radiation plumes from Japan following the 12 March 2011 Fukushima Daiichi Nuclear Power Plant accident. From a direct comparison between the model simulation and the time-series of Fukushima-derived fission nuclides monitored around southeast Asia, we can distinguish between global transport by the Westerlies in the free troposphere and regional transport by the northeast monsoon in the planetary boundary layer. In general, regional (mainly meridional) transport carried more weight than global (mainly zonal) transport in contributing Fukushima-derived radioactivity to the area covered in this review, particularly at the ground-level sites.


INTRODUCTION
The Tohoku-Oki Earthquake on March 11, 2011 resulted in an extremely destructive tsunami that produced waves over 30 m high (Fujii et al., 2011).The earthquake triggered the shut down of the active reactors at the Fukushima Daiichi Nuclear Power Plant (NPP), and the tsunami damaged the plant's backup diesel generators, causing a station blackout.Subsequently, the lack of cooling led to the explosions and reactor cores' meltdowns, and therefore resulting in the worst nuclear accident following the 1986 Chernobyl disaster.
It has been over one year since the Fukushima nuclear accident, and there have been a plethora of publications about the dispersion of radioactive material from the damaged reactors.Most of these works dealt with global transport of Fukushima-derived radionuclides in the northern hemisphere from Japan to the north American and the Eurasian continents (see Stohl et al. 2012 and references therein).Local transport in the vicinity of Fukushima and around Japan has been intensively studied by researchers in Japan (e.g., Chino et al., 2011;Kinishita et al., 2011;Morino et al., 2011;Takemura et al., 2011;Yasunari et al., 2011;Hirose, 2012;Katata et al., 2012;Momoshima et al., 2012;Terada et al., 2012).Besides these studies, either on global or local scales, virtually none of the published papers investigated into dispersal of radiation plumes from Japan to other areas on regional scales.This is because regional dispersal out of Japan in the springtime is most likely dominated by the northeastern monsoon, whereas there are few monitoring stations downwind in the southeastern Asia region.In this respect, we are only aware of the data in Vietnam published by Long et al (2012) in addition to our own data obtained in and around Taiwan (Hsu et al., 2012;Huh et al., 2012).By integrating the data already published in the literature plus those that can be searched from relevant websites, we try to further elucidate the dispersal of Fukushima-derived radiation toward the southeastern Asia region.It should be noted that, in these studies, the aerosol samples were collected by large-volume air samplers using filters similar to those described in Zhao et al. (2012) and Tsai et al. (2012).Therefore, we would address to the validity of our direct comparison of the datasets.

DATA SYNTHESIS AND MODEL SIMULATION
Summarized in Table 1 are the station information and sources of the time series measurement of 131 I, 134 Cs and 137 Cs in Taiwan (NK and MLL), Philippine (Manila), Hong Kong (HK), Vietnam (Hanoi, HCMC and Dalat) and on two small islets in the East China Sea (PCY) and the South China Seas (DS).Fig. 1 shows the geographic distribution of these sites.It should be noted that two of the stations (MLL and Dalat) are located at altitudes above the planetary boundary layer, while the rest are "ground-level" stations.Air transport is highly dependent on altitude as will be discussed later.
The time series data are plotted in Figs. 2 and 3. Due to malfunction of the high-volume TSP sampler, we could not obtain data at DS after March 31 and there was a data gap at NK during April 2-5.Also missing are 137 Cs and 134 Cs for the HK series because radiocesium was detectable only for daily samples (24 hr, noon to noon) collected on April 9 and 13 (http://www.weather.gov.hk/radiation/ermp/rmn/applet/map/KP_Cs137_e.htm).Other than these deficiencies, the time series of Fukushima-derived fission nuclides (FDFN) at the monitoring stations span continuously from the first arrival of these nuclides in late March until late April when nuclide activities went below levels that could be detected by the respective laboratories.
To verify the transport of radioactive materials from Fukushima and relate it to the observed time series, we employed the WRF/Chem (Ver.3.1) tracer model (Grell et al., 2005).The model was run by adopting the Yonsei University (YSU) planetary boundary layer scheme (Hong and Dudhia, 2003;Lin et al., 2009;Bian et al., 2011;Huang et al., in press), with the meteorological initial and boundary conditions obtained from National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) 0.5° × 0.5° analysis data sets at 3-hour intervals (http://www.nco.ncep.noaa.gov/pmb/products/gfs/#GFS).The horizontal resolution for the simulation was 20 km and the grid box has 400 × 400 points in the east-west and south-north directions.There are 35 layers with σ levels set at 1. 000, 0.998, 0.994, 0.990, 0.985, 0.980, 0.970, 0.960, 0.945, 0.930, 0.910, 0.900, 0.890, 0.870, 0.850, 0.820, 0.790, 0.760, 0.720, 0.690, 0.660, 0.630, 0.600, 0.550, 0.500, 0.450, 0.400, 0.350, 0.300, 0.250, 0.200, 0.150, 0.100, 0.050 and 0.000.The lowest level is about 20 m above the surface and the top is around 20 km (50 hPa).To assure the meteorological fields were well simulated, the four-dimensional data assimilation (FDDA) scheme was activated based on the NCEP-GFS analysis data.Our purpose is to make sure the simulation is consistent with the NCEP global simulation results in the long-term run.The tracers in the model originated from the grid of the Fukushima site and the tracer concentration followed the radiation dose rates converted from Chino et al. (2012).
Results of the tracer model animation can be watched from the video (animation 1) in the Supplementary Material.Here, two still images (Fig. 4) are extracted from the video file to show how well the model can predict the first arrival time of air masses laced with Fukushima-derived nuclides.1 for station information.

RESULTS AND DISCUSSION
Shown in Fig. 2 are the time series measurement of 131 I, 134 Cs and 137 Cs activities and the 131 I/ 137 Cs activity ratio at all stations covered in this review.To facilitate the following discussion, the data are separated into nine panels in Fig. 3 for each individual site.

The First Arrival Time of Fukushima-Derived Fission Nuclides
The most important time point in each of the time-series is the first arrival time of FDFN, which spanned a full week (from March 23 to March 29) in the covered area.Since the aerosol samples were collected at a daily resolution, the first arrival time can be unequivocally determined to the date.
During the first several days (March 12-16) of the nuclear disaster, when the water vapor and hydrogen explosions occurred at the NPP, there was a low-pressure system travelling over eastern Japan (Morino et al., 2011;Takemura et al., 2011).Once the vapors and particles were lifted up to the free troposphere, they could be transported easily by the Westerlies.As mentioned earlier, it took only 4 days for Fukushima-derived radioxenon to travel across the Pacific Ocean (Bowyer et al., 2011) 3(c)).The linear transport velocity of this particular plume between these sites is calculated to be about 3 m/s (Huh et al., 2012).
The time series at other more remote ground-level stations (HK, Hanoi and HCMC) show different patterns (Figs. 3(d), 3(f), 3(g)), suggesting different routes followed by different pulses of radiation plumes.Based on a trajectory analysis using the NOAA HYSPLIT model, the initial 131 Ilaced air mass landing at Hong Kong travelled in a counterclockwise direction from Japan toward eastern Russia and then circled down to China before arriving HK on March 26 (http://www.weather.gov.hk/education/edu02rga/radiation/radiation_10_uc.htm).This transport is supported by our model simulation which also suggests that the same plume arrived Vietnam (Hanoi and HCMC) one day later, on March 27.
The remaining two time series (Figs.3(h) and 3(i)) are clearly different from the above.The stations (MLL and Dalat) are located at much higher altitudes where the first arrival times of FDFN are clearly later than those at adjacent ground-level stations.Although the wind speed of the Westerlies is much faster than that of the East Asia monsoon, the around-the-globe transport by the former in the free troposphere is far longer than the regional transport by the latter in the planetary boundary layer.Therefore, it took longer time for the former to arrive in the area covered in this work (Hsu et al., 2012;Huh et al., 2012).It is also noteworthy that the time series at the alpine sites show the lowest activities for individual nuclides but the highest 131 I/ 137 Cs activity ratio (Fig. 2).These characteristics have been ascribed to around-the-globe transport by the Westerlies in the free troposphere (Hsu et al., 2012;Huh et al., 2012).During long-range transport, extended diffusion, mixing with the ambient air, scavenging (by aerosol particles), deposition (both dry and wet), and radioactive decay would conceivably reduce nuclide activities (see the final section).Besides, radioiodine and radiocesium isotopes may be fractionated en route due to their different behaviors.Iodine-131 is more volatile and has a larger fraction in gaseous form.On the other hand, the longer lived 137 Cs is more easily scavenged by aerosol particles.From aerosols released in the Chernobyl accident, Jost et al. (1986) observed size distributions with maximum at 0.35 μm for 131 I and 0.71 μm for 137 Cs; the latter compares favorably with the median aerodynamic diameters of 0.63 μm for radiocesium-enriched aerosol particles released from the Fukushima accident (Kaneyasu et al., 2012).Thus, 131 I in gaseous form or associated with fine particles may travel faster and be lifted to higher altitudes than 137 Cs laden particles can.Alternatively, the earlier arrival of 131 I with respect to 137 Cs at the Dalat and the MLL sites (by 3 days and 4 days, respectively) could be attributed to a combination of factors, including different emission profile and different removal intensities (particularly wet scavenging) during transport (Masson et al., 2011;Stohl et al., 2012) The markedly higher 131 I/ 137 Cs activity ratios observed at these two and other alpine sites (e.g., on the Tibet Plateau; Hsu et al., 2012) than those at ground-level sites indicate the relative enrichment of the more volatile iodine at higher altitudes (Hsu et al., 2012;Long et al., 2012).

The Radiocesium Maximum in the Time Series
Besides the timing of the first detection of FDFN, it is important to note the episodic signals in the time series.During the early stage of the nuclear accident, hardly had one pulse subsided when another rose.Conceivably, the release of fission products from the Fukushima plant must be related to a series of explosions in the nuclear facility which ejected radioactive vapors and particles into the atmosphere at altitudes favorable for long-distance transport.It is also informative to note the relative variation of radioiodine and radiocesium activities released during the nuclear accident.While the 134 Cs/ 137 Cs activity ratio maintained fairly constant (largely between 0.8-1), the 131 I/ 137 Cs activity ratio varied substantially in the course of the accident.At the initial stage of the time series 131 I/ 137 Cs ratios were substantially higher than those monitored later, indicating preferential release of the more volatile 131 I at the early phase of the accident.At PCY, for example, the observed ratio dropped from > 20 during March 26-27 to ~0.3 during April 6-7 (Fig. 2(d)).Such a decrease cannot be ascribed to decay of 131 I because the second pulse of 131 I in early April was even stronger and lasted longer than the first one in late March (Fig. 2(a)).In fact, the 131 I/ 137 Cs activity ratio monitored in late April (~1) was significantly higher than that during April 6-7 (~0.3).The lowest 131 I/ 137 Cs activity ratio is associated with a pronounced radiocesium peak during April 6-7 (Figs.2(b) and 2(c)) which is clearly decoupled from those major 131 I pulses (Fig. 2(a)).The temporal trend of the 131 I/ 137 Cs ratio we observed in northern Taiwan is fairly similar to that monitored near the Fukushima site in showing higher ratios in the early part of the time series, with a fluctuation up to two orders of magnitude for daily measurements in a month-long period (Chino et al., 2011).As mentioned earlier, this ratio may be affected by fractionation during rainfall.It also bears information about nuclear reactions in the damaged reactors and when these fission products were released (Matsui, 2011).
Our model simulation suggests that, when the radiocesium maximum arrived at PCY, the increase of atmospheric radiation above natural background was ~0.1 μSv/hr (see Fig. 4(b)), on the same order as natural radiation in the Taiwan region (http://www.aec.gov.tw/www/gammadetect. php).At other times, Fukushima-derived radioactivity was lower than this (maximum) level by one to two orders of magnitude (or more).Thus, the amount of radioactive iodine and cesium derived from Fukushima was not expected to pose any health risks to the public.
It is important to note here that, Fukushima-derived 137 Cs was detected at the Hong Kong station only on April 9 and 13, with activities reported to be 67 μBq/m 3 and 30 μBq/m 3 , respectively (http://www.weather.gov.hk/radiation/ermp/rmn/applet/map/KP_Cs137_e.htm).Our model simulation indicates that the air mass arriving HK during April 8-9 was the same one passing northern Taiwan earlier (during April 6-7) and Vietnam later (during April 9-12), which caused the radiocesium maximum in the PCY, NK and HCMC time series sequentially (Figs. 3(a), 3(b) and 3(g)).
A mean linear travelling velocity of ~5 m/s can thus be

Decrease of Radioactivity Downwind from Fukushima
As with the decrease of activity concentrations of fission nuclides downwind from Fukushima in the Westerlies (Hsu et al., 2012), similar decrease toward lower latitudes due to dispersal by monsoon winds has been reported (Long et al., 2012).Building on Long et al. (2012), we replotted their Fig. 5 (Fig. 5 in this paper) by removing the data from Dalat and adding the data deduced from the PCY, HK and HCMC time series.As with our data monitored at MLL, the Dalat time series recorded transport of FDFN in the free troposphere, which should be separated from data resulting from monsoon-dominated transport in the planetary boundary layer.
Fig. 5 shows that activity concentrations of both 131 I and 137 Cs integrated over the time series decrease exponentially with distance from Fukushima.Despite its much longer half life, 137 Cs decreases more rapidly with distance (and travel time) than 131 I does.As mentioned earlier, this is due to preferential scavenging and removal of 137 Cs over 131 I from the atmosphere by aerosol particles (Hsu et al., 2012;Kristiansen et al., 2012;Long et al., 2012).

CONCLUSIONS
We have in this work assembled the time series of Fukushima-derived fission nuclides monitored at nine sites in the southeastern Asia region.By running the WRF/Chem tracer model and relate the model result to the time series data, we can distinguish global transport from regional transport and conclude that the latter is more important  et al., 2012).
than the former in contributing to the radioactivity observed in the covered area.An accident like this can be thought of as a non-purposeful, but very valuable tracer experiment that can be used to test aerosol transport models and improve their applications to air quality and climate studies in both global and regional scales.
Fig. 4(a)  shows the advent of the first radiation plume toward Philippine, which led to the increase in radiation peak at the Manila station during March 24-25.Fig.4(b) shows a later plume approaching northern Taiwan, which caused the enhancement in radiation during April 6-7 at PCY and NK.The results are consistent with CTBTO's simulation (CTBTO website) of radiation plumes approaching toward Philippine and Taiwan on March 23 and April 6, respectively.These plumes represent different episodes of north-easterly winds transporting fission nuclides predominately toward the southwest.

Fig. 1 .
Fig. 1.Map showing the location of Fukushima, Japan (red star) and monitoring sites in the southeastern Asia.Groundlevel sites are indicated by pink circles while high-altitude (> 1500 m) sites are indicated by blue circles.Refer to Table1for station information.
. On March 17, 2011, the northeast monsoon wind characteristic of this season in East Asia picked up momentum and switched the radioactive plume from Fukushima toward the southwest.The plume was transported mainly in the marine boundary layer over the open Pacific Ocean, hitting Manila on March 23 without influencing other sites covered in this review.This wave of radiation plume stands out very well in our WRF/Chem tracer model simulation (please see Figs. 3(e) and 4(a), and refer to Animation 1 in the Supplementary Material).The time series at PCY, NK and DS (Figs. 3(a), 3(b) and 3(c), respectively) show a consistent pattern of waxing and waning in nuclide activities with a progressive lapse in time with respect to the distance travelled from the source region.It clearly suggests that these stations were influenced by the same radiation plumes.As revealed by the WRF/Chem tracer model simulation, this wave of radiation plume carried FDFN toward the southwest at somewhat higher altitudes, making landfall and hitting Okinawa on March 24 (CTBTO Preparatory Commission), PCY and NK on March 25 (see Figs. 3(a) and 3(b)), and the Dongsha Islet further downwind in the South China Sea on March 28 (see Fig.

Fig. 2 .
Fig. 2. Time series of (a) 131 I, (b) 137 Cs and (c) 134 Cs activity concentrations and (d) the 131 I/ 137 Cs activity ratio at nine stations discussed in this review.Vertical bars over the data points indicate uncertainty of the measurements and horizontal bars denote sampling intervals (mostly 24 h).

Fig. 4 .
Fig. 4. Still images extracted from the WRF/Chem tracer model simulation showing the first plume approaching Philippine on (a) March 23, 2011 and a later plume approaching northern Taiwan on (b) April 6, 2011.These two plumes led to the radiation maximum in the Manila and the PCY time series, respectively.For more details, please refer to Animation 1 in the Supplementary Information.
calculated for this 137 Cs-enriched air parcel, which is ~60% faster than the first wave of radiation plume entering the South China Sea in late March.

Fig. 5 .
Fig.5.Dependence of integrated activity concentrations (y) of 131 I and 137 Cs on the distance (x) from Fukushima.The yvalues were derived by integrating measured activity concentration (mBq/m 3 ) over the length (in day) of the time series at each station (Redrawn fromLong et al., 2012).

Table 1 .
Station information and data source.