2.1. Sampling sites and sampling preparation of flood sediments and decontaminated soils

FS and DS samples were collected along Niida River and Mano River within Iitate Village, Fukushima, Japan, which is located about 40 km northwest from the FDNPP. These rivers overflowed at many locations following Typhoon Hagibis (typhoon 19) in October 2019. Their catchments include upstream mountainous plateaus at an altitude of 700–900 m above sea level connected to a flat coastal plain by deep and strongly incised rivers [Chartin et al. 2017]. About two-thirds of the land in these catchments is occupied by forested areas. Most of the agricultural fields along these rivers were used as paddy fields before the 2011 nuclear accident. Geologically, granite and granodiorite are dominant in the area, with a minor occurrence of sedimentary rocks and basalts [AIST Geological Survey of Japan website 2022].

DS samples were collected using a shovel from the upper 0–15 cm layer of surface soil from 18 decontaminated fields in March 2019. In each field, subsamples of soil were collected from five points (at the center and the four vertices of a square with a side length of 15 m) and mixed together well. Also, FS samples were collected from 19 points where river flooding occurred in the close vicinity of the DS sampling sites after the Hagibis typhoon in October 2019 (Figure 1). The FS samples correspond to fresh sediments deposited as a fine material draped on the upper part of the embankment along the Mano River (n = 6) and Niida River (n = 13). FS samples were collected from the uppermost layer of about 0–1 cm of fresh sediments using a shovel. The DS and FS samples were dried at room temperature and sieved to <2.0 mm.

2.2. Radiocesium analyses

Total and exchangeable ¹³⁷Cs contents in FS or DS were determined following the method of Kurokawa et al. [2019]. Approximately 20–40 g of the soil sample was filled in a polyethylene vial (20 cm³). The vial was set in an auto gamma counter with a NaI (Tl) detector (2480WIZARD2, PerkinElmer, Waltham, Massachusetts, USA, Radioisotope Research Center of Kyoto Prefectural University) to determine the gamma activity (between 589 and 735 keV). To determine the exchangeable ¹³⁷Cs content, duplicates of 50 g of sample were shaken with 250 mL of 1 mol⋅L⁻¹ ammonium acetate (CH₃COONH₄) for 24 h at 20 °C. The extracts were heated on a hot plate and concentrated to about 20 mL. The concentrated solution was measured with NaI (TI) detector in the same way as for total ¹³⁷Cs content. The data were calibrated by comparing the measured values with those obtained using a Ge semiconductor detector connected to a multichannel analyzer system (GC-4018, Canberra, Japan, Radioisotope Research Center of Fukushima University). Details of this procedure are provided by Kurokawa et al. [2019].

The ¹³⁷Cs exchangeable fraction (%), i.e., the percentage of exchangeable ¹³⁷Cs content compared to the total ¹³⁷Cs content, was calculated using the following equation:

[¹³⁷Cs exchangeable fraction (%)] = [Exchangeable ¹³⁷Cs content in FS and DS (Bq⋅kg⁻¹)]/[Total ¹³⁷Cs content in FS and DS (Bq⋅kg⁻¹)] × 100.

2.3. Particle size distribution

Particle size distribution, i.e., respective content in coarse sand (2000–200 μm), fine sand (200–50 μm), and silt and clay ( < 50 μm) in FS and DS, was assessed by pipetting and sieving after removing soil organic matter with oxidizing reagents.

2.4. Total carbon content

The total carbon (TC) content in sample was measured, as an indicator of total organic matter content, determined by the dry combustion method with an NC analyzer (NC-95A, Smika Chem. Anal. Service, Tokyo, Japan) and gas chromatography (GC-8A, Shimadzu, Kyoto, Japan) after fine grinding with agate mortar.

2.5. X-ray diffraction (XRD) analysis

XRD diffractogram was determined using the matrix flushing technique [Chung 1974]. 1.0-g soil sample was mixed with 7-mL methanol and 0.25-g 𝛼-Al₂O₃ (Baikalox 3.0CR, Bakowski, France) as internal standard and finely-ground to a powder using a grinding machine (XRD-Mill Mcrone, Retsch, German) for 10 min. The mixed powder was packed into a holder to ensure a random sample orientation and then irradiated from 5° to 65° two-theta using CuK𝛼-radiation, with 0.02° steps and a counting rate of 10° per min by XRD (MiliFlex 600, Rigaku, Tokyo, Japan). The diffraction peak at 8.8° two-theta corresponded to the 1.0 nm lattice spacing associated with the presence of non-expanding mica layers [Kitayama et al. 2020, Ogasawara et al. 2019]. As K in these non-expanding layers is the primary source of Nex-K, XRD_micas provided a rough index of the amount of Nex-K present. Therefore, XRD_micas was calculated by the following equation:

2.6. Non-exchangeable K as an indicator of acid-extractable K

Nex-K could be used as a good proxy of K phytoavailable fraction especially in mica-rich soil [Mengel and Rahmatullah 1994, Surapaneni et al. 2002]. Kurokawa et al. [2020] and Wakabayashi et al. [2022] reported that the Nex-K reduced RCs crop migration as well as that of exchangeable K (Ex-K). The Nex-K content of the FS and DS samples was determined by subtracting Ex-K from the estimation of K concentration determined as the elemental content extracted with the hot nitric acid method [Helmke & Sparks 1996]. The Nex-K is often used as an indicator of the phytoavailable K fraction, which is not extractable with ion-exchange reaction. A 2.50-g soil sample was gently heated using a hot plate (EA-DC10, ZOJIRUSHI, Osaka, Japan) at nearly 100 °C for 15 min with 25.0 mL of 1 mol⋅L⁻¹ HNO₃ after boiling started. After cooling, the extract was filtered and filled up to 100 mL with 0.1 mol⋅L⁻¹ HNO₃. The K content of the extract solution was determined via Atomic Absorption Spectroscopy (AAS) (AA -6200, Shimadzu, Kyoto, Japan). To determine the Ex-K content, the DS and FS samples were shaken with 1 mol⋅L⁻¹ CH₃COONH₄ at pH 7.0 in a soil:solution ratio of 1:5 for 30 min at 20 °C. The suspension was centrifuged, and the supernatant was collected by filtration. These processes were repeated three times. Approximately 75 mL of supernatant was brought up to a total volume of 100 mL with 1 mol⋅L⁻¹ CH₃COONH₄, and the K content of this solution was determined using the same instruments.

3.1. Total ¹³⁷Cs content in flood sediment and its controlling factors

The total ¹³⁷Cs content in FS reached on average 4.2 kBq⋅kg⁻¹ (Table 1). This observed value was lower than that obtained in FS collected along the same two rivers in 2011 and 2015 [Chartin et al. 2017]. Previous research indicated that the amount of ¹³⁷Cs had strongly decreased in river sediment and in the soil surface in 2019 compared to 2011 [Evrard et al. 2021, Funaki et al. 2019, Iwagami et al. 2017, Yoshimura et al. 2016], which further supports the results obtained in the current study.

Typhoon Hagibis caused debris flows that apparently excavated soils from contaminated topsoil layers, and much less contaminated subsurface soils as shown for FS in Miyagi Prefecture [Moriguchi et al. 2021]. Highly contaminated surface soils from forested areas were therefore likely mixed with subsurface soils from landslides and channel bank collapse events, which may consequently explain the overall low total ¹³⁷Cs content in FS.

Although total ¹³⁷Cs contents in FS reached an average value of 4.2 kBq⋅kg⁻¹, they showed a wide range of values varying from 0.51 to 16 kBq⋅kg⁻¹ (Table 1). Potential explanation factors relating to this variation were investigated by means of the Pearson correlation analysis, which revealed that the silt + clay content of FS, averaging 18% (range: 4.9 to 36%), was positively correlated with total ¹³⁷Cs content (r = 0.59,p < 0.01) (Figure 2a). This may reflect the strong interaction of ¹³⁷Cs with sediments, i.e., the finer mineral particles have a larger specific surface area and a higher proportion of 2:1 clay minerals, which can strongly adsorb Cs in their frayed-edge sites [Fan et al. 2014, He & Walling 1996, Nakao et al. 2012].

Organically-bound ¹³⁷Cs was not considered as a major fraction of the total ¹³⁷Cs in FS. We observed few macro-organic material (e.g., leaves and litter from forest vegetation) in the field affected by FS migration via flooding events. Since DS fields were not cultivated at that time, crops could not have trapped macro-organic material, which would instead have been carried away by the river water. As previously reported, ¹³⁷Cs in soils and sediments was found to be bound predominantly to the soil mineral phase whereas it was found bound to a lesser extent to organic matter [Koarashi et al. 2019, Tsukada et al. 2008]. The amount of total carbon in the FS was low, with an average of 3.7 g⋅kg⁻¹ (range: 0.47 to 8.1 g⋅kg⁻¹), and it was found not to be significantly correlated with total ¹³⁷Cs content or the ¹³⁷Cs exchangeable fraction (Figure 2b). Based on these results we considered that the amount of organic matter-bound ¹³⁷Cs in the FS should only provide a minor contribution, if any, to the total ¹³⁷Cs content.

3.2. Comparing total ¹³⁷Cs content and ¹³⁷Cs exchangeable fraction between FS and DS

The total ¹³⁷Cs content in DS reached on average 1.2 ± 2.0 kBq⋅kg⁻¹ (Figure 3). The entire range of values (Table 2) was less than one-fifth of the total ¹³⁷Cs content in agricultural land observed before decontamination in Iitate Village [Ministry of Agriculture 2011]. A drastic decrease in total ¹³⁷Cs content after decontamination was observed in the agricultural fields in Tomioka Town [Kurokawa et al. 2019]. Thus, decontamination activities for agricultural fields in Iitate village were thus confirmed to have effectively reduced the total ¹³⁷Cs content in surface soils corresponding to about 80% in a decrease of the initial contamination levels. Although total ¹³⁷Cs was not significantly different between FS and DS, this value range was lower than that of FS (Figure 3). These comparisons suggest that FS deposition on DS may lead to a new increase in their total ¹³⁷Cs content, at least in their surface layer. However, the increase in total ¹³⁷Cs content due to FS deposition, if any, may not raise the transfer risk of ¹³⁷Cs from soil to crops for the following reasons. First, our field observations in Iitate Village confirmed that the thickness of the FS layer deposited on decontaminated fields following the flooding reached at most about 1 cm in depth. Assuming that the thin FS layer contained a total ¹³⁷Cs activity of 4.2 kBq⋅kg⁻¹ mixed with 15 cm of the plowed layer with a total ¹³⁷Cs content of 1.2 kBq⋅kg⁻¹, the total ¹³⁷Cs content of the entire profile would only increase by 15% to an average of about 1.4 kBq⋅kg⁻¹. While this is a rough estimation based on field observation of the thickness of sediments deposited in several agricultural fields, it indicates that the FS incorporation into a realistic larger volume of soil will not lead to a strong increase in the total ¹³⁷Cs content in decontaminated fields. It should be emphasized that the 1 cm thickness was an exceptional case for FS migration to DS. The average trend was unfortunately not quantitatively recorded in this study, but more accurate observation is needed in the future when larger volumes are involved. A comprehensive risk assessment would require a more detailed calculation of the thickness of sediment deposited onto agricultural soils.

Second, the average ¹³⁷Cs exchangeable fraction of FS was 3.7%, only about one-third of that rate obtained from DS (i.e., 9.8%) (Figure 4). Whereas the ¹³⁷Cs exchangeable fraction from DS remained in the medium range of values (range: 6.5 to 16%) previously reported for ¹³⁷Cs contaminated soils in Fukushima, corresponding values from FS remained close to the lowest values previously reported for contaminated soils and sediments [Kurokawa et al. 2019, Ogasawara et al. 2019, Saito et al. 2014]. These data suggested that ¹³⁷Cs was assumed not to be desorbed easily from FS particles. Third, FS showed high levels of Nex-K content, with an average of 162 mg K₂O 100 g⁻¹ (range: 80 to 317 mg K₂O 100 g⁻¹), as high as in DS (i.e., on average 218 mg K₂O 100 g⁻¹). Most of these values were higher than the recommended value (i.e., 50 mg K₂O 100 g⁻¹), above which the transfer rate of ¹³⁷Cs from soil to crops was confirmed to be low [Kurokawa et al. 2020]. Since ¹³⁷Cs exchangeable fraction and Nex-K content in soil control soil-to-plant transfer of ¹³⁷Cs rather than the total ¹³⁷Cs content itself, these results may overall lead to the conclusion that FS deposition in decontaminated fields may not increase the transfer risk of ¹³⁷Cs in the case of Iitate Village in Fukushima.

A missing source of ¹³⁷Cs that might have been incorporated into the DS field was dissolved ¹³⁷Cs in flood water. Takata et al. [2020] indeed showed higher dissolved ¹³⁷Cs activity in nearshore water after the 2019 typhoon. Although we could not measure the amount of ¹³⁷Cs in DS after the typhoon, it is unlikely that dissolved ¹³⁷Cs may be supplied to DS after the deposition of FS, as the ¹³⁷Cs exchangeable fraction was found to be lower than 5% in the current research.

3.3. Factors relating to Nex-K content of FS and DS

The Nex-K content of FS and DS (i.e., on average 162 and 218 mg K₂O 100 g⁻¹) was relatively high, compared to those levels previously found in other areas of Fukushima Prefecture, including a mean value of 66 mg K₂O 100 g⁻¹, obtained on 173 farmland soils in Tomioka Town [Kurokawa et al. 2019]. The regional difference in Nex-K content may reflect to a large extent the type of base rock in each area. For example, Iitate Village is a locality surrounded by Abukuma granite mountains. Therefore, the fluvisols on which cropland was installed in the valleys, and the coarse material extracted from the quarries to replace the contaminated upper soil layer during remediation, both contain minerals, derived from granite or granitoid, which in turn contains biotite as a major rock-forming mineral. To investigate whether the quantity of mica defines Nex-K, we investigated the correlation between Nex-K and XRD_micas, which is an index of mica quantity, and found that Nex-K showed significantly positive correlation with XRD_micas (p < 0.001) (Figure 5). This strongly supports the hypothesis that Nex-K content was primarily controlled by micaceous minerals probably derived from granite. Granite and granodiorite widely cover the surfaces of Abukuma Highlands, which have been weathered to supply biotite to watershed areas via river water transfer. Several studies on the mineral composition of sediments have indicated the occurrence of biotite in this material [Fan et al. 2014, Konoplev et al. 2021]. Accordingly, the K release potential of FS which was found to be comparable to that of DS should be probably ubiquitous across this region. The granitic biotite in both FS and DS may contribute to long-term K resources for plants, which effectively inhibit ¹³⁷Cs migration from soil to crops.