Contamination of reservoir sediment with radiocesium has been thoroughly investigated, particularly since the accident at the Chernobyl nuclear power plant. Temporal changes in 137Cs concentration, inventory1,2,3,4, spatial/vertical distribution5,6,7,8,9 and physicochemical properties10,11,12 have been well documented in reservoirs. However, few studies have documented the initial accumulation and flushing of 137Cs from reservoirs exposed to the fallout of radioactive materials from nuclear power plant accidents.

Reservoir inputs of 137Cs are removed from the water column mainly by direct adsorption to sediment1,13,14, sedimentation with suspended particles12,15 and hydraulic flushing1. Davison et al.1 investigated the concentration of 137Cs in the water and sediments of English lakes immediately following the Chernobyl nuclear power plant accident. They estimated, from the flow rate and concentration of radiocesium in the water, that 32–41% of the 137Cs that was initially deposited had been hydraulically flushed from the lakes and the remaining 137Cs had accumulated in sediments. However, it is difficult to collect such data from many reservoirs immediately following a nuclear accident.

The concentration of 137Cs in reservoir sediment can be affected by many factors, such as the initial amount of radiocesium deposition, organic matter content of the sediment, water depth, catchment area, pond volume and hydraulic residence time2,3. Additionally, it has been well established that sediment particle size can affect the concentration of 137Cs in sediment10,16,17,18,19. He and Walling10 demonstrated, by laboratory experiments and empirical observation, that the increase in 137Cs concentration in soil and sediment was associated with a decrease in particle size. Radiocesium in the surface water of ponds, which are shallower than lakes and usually only a few meters deep, is inferred to be directly associated with resuspended solids, particularly fine particles. Since hydraulic flushing of fine particles is reflected in the particle size distribution of the sediment, assessment of sediment particle size is important in determining the amount of 137Cs removed from a reservoir by hydraulic flushing.

On March 12, 2011, an accident at the Fukushima Dai–ichi nuclear power plant (DNPP) released ~1–2 × 1016 Bq of 137Cs into the atmosphere20, contaminating a large proportion of the surrounding area21,22,23,24, which included a forested area25. Fukushima Prefecture, the area most affected by the 137Cs fallout, is a major rice-farming region where there are many artificial ponds used for irrigation, which are surrounded by forest. Since reservoirs accumulate soil eroded from the catchment and decrease the flux of sediment to downstream systems26, it is also likely that they have an impact on the movement of sediment-sorbed radiocesium to downstream systems. It has been reported that the initial inventory of 137Cs in lakes is related to the total 137Cs directly deposited in/around the lakes, with a small contribution of 137Cs from the catchment1,13. Thus, trap efficiency can be calculated as the proportion of 137Cs inventory in pond sediment to the 137Cs inventory in the soil surrounding the pond (i.e., total 137Cs inventory). Assessing the accumulation and flushing of 137Cs in ponds is important in understanding the initial contamination of sediment and the discharge of radiocesium to downstream systems. To our knowledge, no studies have investigated the initial dynamics of 137Cs in irrigation pond sediment following the Fukushima DNPP accident.

In this study, we assessed the concentrations of 137Cs in the sediments of four irrigation ponds (Fig. 1) 4–5 months after the Fukushima DNPP accident. We used sediment particle size analysis as a simplified estimation of 137Cs dynamics, to determine initial accumulation and hydraulic flushing of 137Cs in ponds exposed to nuclear fallout.

Figure 1
figure 1

Location and plain views of irrigation ponds.

Cross in plain view indicates sampling point. These maps were generated by ArcGIS 10 software and combined with the 137Cs distribution obtained from the Third Airborne Monitoring Survey (MEXT21), provided in the Database on the Research of Radioactive Substances Distribution34.


The description of each pond, sampling dates and 137Cs inventory in the soil surrounding the ponds, which was measured as part of the Third Airborne Monitoring Survey by MEXT on 2 July 2011 (calibrated with 2200 points of soil core samples21), are provided in Table 1. The inventory of 137Cs in the surrounding soil was estimated to be 206–360 kBq m−2. While heterogeneity in the local distribution of radioactivity has been observed27, no hot particles derived from the fuel of nuclear reactor have been detected around our sites28.

Table 1 Descriptions of ponds, sampling dates and inventories of 137Cs in soils surrounding ponds estimated from the Third Airborne Monitoring Survey, by MEXT21. Total inventories of 137Cs were estimated as inventory on ground surrounding the ponds within 25 m distance and the decay is corrected on 2 Jul. 2011

The mass-depth distributions and inventories of 134Cs and 137Cs in the pond sediments are presented in Fig. 2. In all ponds, the radiocesium concentrations were highest in the surface layer of sediment and quickly decreased with depth. Concentrations of 137Cs in the surface layer were 17.9, 34.9, 6.4 and 10.9 kBq kg−1 at Oyado, Takayashiki, Netsupami–ike and Matsuzawakami–ike, respectively. Radiocesium was detected as far down as the fifth layer at Takayashiki and the fourth layer in the other ponds. However, >80% of the total 134Cs and 137Cs inventories were in the top two surface layers (<16 kg m−2). The accumulation of 137Cs in surface sediment (<3.5 cm) was simulated in laboratory experiments, using the kinetics of 137Cs sorption to and desorption from sediments29,30 and the vertical distribution of 137Cs in our sediment samples. The results indicate that downward migration or agitation of radiocesium in sediment was minimal in the ponds studied and that most of the radiocesium was trapped in surface sediment.

Figure 2
figure 2

Mass depth distributions of concentrations of radiocesium in sediment (gray line for 134Cs and black line for 137Cs) and D50 of sediment (broken line).

Inventories are also shown.

Matsuzawakami-ike has a relatively large surface area compared to the other ponds (Fig. 1; Table 1), so there could be spatial variation in the inventory of radiocesium in sediment. Another investigation collected sediment core samples at three locations (inlet-side, middle and outlet-side) in Matsuzawakami-ike at 30–40 m intervals in Apr. 2013 (not published). The coefficient of variation for 137Cs-inventory was 7%; thus, the spatial variation in inventory was inferred to be small.

Radiocesium inventory was related to the concentrations of radiocesium in sediment; the highest radiocesium inventory was observed in Takayashiki, which showed the highest concentration of radiocesium. Radiocesium inventory in Oyado, which had the highest amount of 137Cs deposition among the study ponds, was lower than that in Takayashiki, indicating a difference in radiocesium accumulation efficiency. To evaluate the efficiency of sediment in accumulating radiocesium, we compared the inventory of 137Cs in pond sediment to the total 137Cs inventory (Table 1), referred to hereafter as trap efficiency. Based on the findings of previous studies1,13, we assumed that the amount of 137Cs deposited in the surrounding soils represented the total amount of 137Cs deposited into the ponds. There was marked variation in trap efficiency among ponds: 23, 133, 19 and 37% for Oyado, Takayashi, Netsupami–ike and Matsuzawakami–ike, respectively.

Reservoir characteristics, such as water depth, catchment area and pond volume, can affect the hydraulic residence time and the associated removal of 137Cs through hydraulic flushing. The ratio between catchment area and pond volume was calculated as an estimate of hydraulic residence time, which was compared to the trap efficiency. There was no significant correlation between our estimate of hydraulic residence time and the trap efficiency. We also investigated the elemental composition, geology and organic content of sediment, which are factors that can affect the adsorption of radiocesium by sediment particles. The elemental composition is presented in Table 2. There was no unusual composition and there was no clear relationship between content and trap efficiency for any sediment element. As regards the geology of the ponds studied, they are all granite bedrock31. The average organic matter contents in the three surface layers of sediment were 16.7 ± 0.6%, 12.6 ± 0.1%, 12.0 ± 0.2% and 12.3 ± 1.0% for Takayashiki, Oyado, Netsupami-ike and Matsuzawa-ike, respectively and there was no significant correlation between organic matter content and trap efficiency. Water quality factors such as pH can also affect the adsorption/desorption dynamics of radiocesium in sediment. The pH in each pond was measured in April and May 2013. The pH values were 7.5–7.7 for Takayashiki, 7.5–7.6 for Oyado, 7.6–7.9 for Netsupami-ike and 7.9–8.0 for Matsuzawakami-ike (personal data); there was no correlation between pH and trap efficiency. Based on these results, the variation in trap efficiency was not the result of differences in the reservoir characteristics, the sediment characteristics, or pH among study ponds.

Table 2 Elemental composition of surface (0–2 cm layer) sediments (%). The value was obtained as weight composition

The inventories of radiocesium reflected the concentration of radiocesium in our study ponds. It has been established that sediment particle size can affect the concentration of radiocesium in sediment10,16,17,18,19. Therefore, we assessed the effect of particle size by comparing 137Cs-based weighted average values of D50 and specific surface area to trap efficiency (Fig. 3). The D50 and specific surface area of sediment in Takayashiki, which had the highest trap efficiency among our study ponds, were 15.1 μm and 0.41 m2 g−1, respectively, representing the smallest particle size among ponds. Trap efficiency was correlated with sediment particle size, showing a negative correlation with D50 (r2 = 0.69; Fig. 3a) and a positive correlation with specific surface area (r2 = 0.66; Fig. 3b).

Figure 3
figure 3

Variation in trap efficiency of 137Cs into sediments associating with weighted averages of (a) D50 and (b) specific surface area.

Solid curves indicate regression lines.


Davison et al.1 reported trap efficiencies between 59–68% for two English lakes, over an 18-month period after the Chernobyl nuclear power plant accident and that 32–41% of the total 137Cs was hydraulically flushed from the system. It is likely then, for the ponds in this study that had trap efficiencies <100%, that some of the initial radiocesium inputs were removed through hydraulic flushing. Hydraulic flushing of 137Cs from a reservoir can be affected by precipitation, as large rain events will lead to shorter hydraulic residence times. Oyado and Takayashiki are very close to each other (<1 km apart) and likely experience similar rates of precipitation, but they had very different trap efficiencies (23 vs. 133%). While reservoir characteristics, such as water depth, catchment area, pond volume and hydraulic residence time can also affect hydraulic flushing2,3 of 137Cs from ponds, we observed no correlation between these factors and trap efficiency. However, trap efficiency was correlated with sediment particle size, indicating that particle size can be an important factor in determining trap efficiency. Accumulation of 137Cs in sediment occurs through direct adsorption to sediment1,13,14 and sedimentation with suspended particles12,15. It is likely that the finer sediment in Takayashiki facilitated the accumulation of 137Cs, resulting in the highest trap efficiency. In contrast, ponds with coarser sediment and lower trap efficiencies may have lost 137Cs through hydraulic flushing of finer sediment. Thus, sediment particle size is one of important factors in estimating the amount of 137Cs lost through hydraulic flushing, in addition to its accumulation in pond sediment. More research and data are needed to determine the strength of the correlation between radiocesium trap efficiency and sediment particle size.


Sample collection

We selected four irrigation ponds, located ~40–50 km from the Fukushima DNPP (Fig. 1), to investigate initial 137Cs accumulation from the fallout of the Fukushima DNPP accident. Oyado and Takayashiki are located in the city of Nihonmatsu and Netsupami–ike and Matsuzawakami–ike are located in the town of Kawamata. The ponds were artificially constructed (i.e., reservoirs) and receive water inputs from precipitation and streams within the catchment. As the ponds all have relatively flat bottoms, water depths were similar for large areas in all of the ponds.

Sediments were collected at Oyado and Takayashiki on 8 July 2011 and at Netsupami–ike and Matsuzawakami–ike on 4 August 2011. Four sediment cores (core depth: >12 cm) were taken from approximately the center of each pond using a gravity corer (i.d.: 4 cm). Each core was sampled at a distance of 1–3 m from other core-sampling points to avoid the influence of sediment agitation on samples. Core samples were sliced into 2-cm layers, for a total of five layers for each core. Across replicate cores, sliced samples were combined at each depth. The samples were then dried at 105°C for 24 h. The dried samples were homogenized, using a mortar and pestle and used for measurements.

Sample analysis

Concentrations of 137Cs and 134Cs were measured using a high-purity n–type germanium coaxial gamma–ray detector (EGC25–195–R, Canberra–Eurisys, Meriden, U.S.A.), equipped with an amplifier (PSC822, Canberra, Meriden, U.S.A.) and multichannel analyzer (DSA1000, Canberra, Meriden, U.S.A.). Instruments were calibrated using standard gamma sources: 210Pb and 137Cs (EG–CUSTOM; Isotope Products Laboratories, USA), 241Am, 109Cd and 152Eu (EG–CUSTOM; Eckert and Ziegler Isotope Products, USA). The samples were placed on top of the detector head and radioactivity was measured for 5 min to 12 h, depending on the activity of the sample. The activities of 137Cs and 134Cs were corrected for decay, based on when the samples were collected. The counting error was less than 10% and the measurement accuracy satisfied the IAEA–CU–2006–03 World–Wide Proficiency Test on the Determination of Gamma Emitting Radionuclides32. Further details can be found in Kato et al.33.

Samples were then analyzed using a laser diffraction particle size analyzer (SALD–3100, Shimadzu Co., Ltd., Kyoto, Japan) to determine the median-diameters (D50) and specific surface areas of sediment particles, for each layer of sediment. To evaluate the effect of particle size on the accumulation of 137Cs in sediment, 137Cs-based weighted average values of D50 and specific surface area were calculated according to the following equation:

where Vi is D50 or specific surface area, Ii is the inventory of 137Cs in layer i and Itotal is the total inventory of 137Cs in a given sediment core.

After radiocesium analysis, the elemental composition of each surface layer of sediment (0–2 cm) was determined using an X-ray fluorescence analyzer (Niton XL3t, Thermo Scientific) with a helium gas-purging system. To measure the organic content, ~1 g of sample was subdivided from three surface layers (0–2-, 2–4- and 4–6-cm layers). The subdivided samples were weighed and then combusted at 500°C for 24 h. The mass lost upon ignition represented the organic content.