Impact of typhoons on particulate and dissolved 137Cs activities in seawater off the Fukushima Prefecture: results from the SOSO 5 Rivers cruise (October 2014)

Cruise SoSo 5 Rivers took place during October 2014 off the coast of Fukushima Prefecture shortly after the passage 10 of two typhoons. Detection of dissolved 134Cs and 137Cs in all samples reflected contamination caused by accidental releases of radiocaesium from the Fukushima Dai-ichi Nuclear power plant (FNPP1) accident. The dissolved activities were generally higher at coastal sites and decreased with distance from shore, and they were higher in the surface than in the bottom water. The tendency of 137Cs activities to decrease with distance from the coast reflected mixing of coastal water and open-ocean water of which 137Cs activity concentration was ~1.5 Bq m−3. At stations very close to the coast, we observed high particulate 15 137Cs activity concentration that exceeded dissolved 137Cs activity concentration. 137Cs activities were generally 1–2 orders of magnitudes lower in organic particles than in dissolved form, and the ratios of 137Cs activity concentration in organic particles to 137Cs activity concentration in dissolved form ranged from 0.01 ± 0.00 to 0.12 ± 0.01. The ratio of 137Cs to 134Cs activity concentrations in organic particles did not change with distance from shore or with 137Cs activity concentration and generally remained around 1, even in samples collected far from the coast. This pattern indicated that the organic particles had come 20 from rivers or a source very close to the coast. The 137Cs/134Cs activity ratio in dissolved form north of FNPP1 region was estimated to be 1.074 ± 0.015, a ratio that is in good agreement with the 137Cs/134Cs activity ratio in the core of Unit 1 of the FNPP1 while the 137Cs/134Cs activity ratio at Tomioka port which located south of FNPP1 was 0.998 ± 0.017. Therefore we can conclude the source of radiocaesium in seawater in the coastal region north of FNPP1 was deposited radiocaesium released from the core of Unit 1 of FNPP1, while the source of radiocaesium observed in the coastal region south of FNPP1 was a 25 mixture of deposited radiocaesium released from the core of Unit 2 and the core of Unit 1 of FNPP1. During September– October of each year, the typhoon season in Japan, the 137Cs activity concentration generally increased at Ukedo port, Tomioka port, FNPP1, and Iwasawa beach, and showed a good relationship with the 7-day modified antecedent precipitation index (API) while there is less correlation between the modified API and 137Cs activity concentration near the outlet of canal from unit 5 and 6 of FNPP1 to the sea. 30 https://doi.org/10.5194/bg-2020-491 Preprint. Discussion started: 8 February 2021 c © Author(s) 2021. CC BY 4.0 License.


Introduction
The Fukushima Dai-ichi Nuclear Power Plant (hereafter FNPP1) accident, which occurred in northeastern Japan on 11 March 2011, resulted in releases of large amounts of various radionuclides, especially 137 Cs and 134 Cs, which have half-lives of 30.2 y and 2.06 y, respectively. Estimates tend to converge on 15-20 PBq for the combined inputs of 137 Cs from atmospheric fallout 35 and direct discharge to the North Pacific (Aoyama et al., 2019;Aoyama et al., 2016). According to Nishihara et al. (Nishihara et al., 2012), the 137 Cs to 134 Cs activity concentration ratios in reactor cores 1, 2, and 3 were 1. 06, 0.92, and 0.96, respectively, at the time of the accident. In the coastal region, however, the 137 Cs to 134 Cs activity concentration ratio in dissolved radiocaesium was reported to be uniform and very close to 1 (Buesseler et al., 2012;Buesseler et al., 2011), probably because of inaccuracies in the measurements made during monitoring by the Japanese Government and Tokyo Electric Power Company 40 Holdings, hereafter TEPCO. Small but significant differences have already been reported in the 137 Cs to 134 Cs activity concentration ratios of the environmental samples collected on land. Those differences depended on whether the radiocaesium came from the unit 1, 2, or 3 reactor core. Miura et al. (Miura et al., 2020) have reported two types of caesium-bearing microparticles emitted from the FNPP1 accident that were separated from road dust and non-woven fabric cloth. Type-A particles, which were spherical and ~0.1-10 µm in diameter, contained ~10 -2 to 10 2 Bq of 137 Cs radioactivity concentration. 45 Type-B particles, which had various shapes and were 50-400 µm in diameter, contained 10 1 -10 4 Bq of 137 Cs radioactivity concentration. The 137 Cs to 134 Cs activity concentration ratios in Type A and Type B particles were in excellent agreement with the corresponding ratios in cores 2 and 3 in the former case and in core 1 in the latter case. The evidence therefore indicated that the type A particles came from Units 2 and 3, and the Type B particles came from Unit 1. There has not been a similar study of the dissolved and particulate ratios of 137 Cs to 134 Cs activity concentration in seawater. 50 Although these major releases from the FNPP1 accident occurred in March and April 2011 (Tsumune et al., 2012;Tsumune et al., 2013), small amounts of radiocaesium have continued to be released, and the rate of release of 137 Cs was estimated to be 10 GBq day −1 or 3.6 TBq year −1 in 2014 (Tsumune et al., 2020). Also, riverine inputs due to runoff from contaminated watersheds, though of lesser importance, are expected to continue for a long time (Adhiraga Pratama et al., 2015). Movement of primarily particulate radiocaesium in fresh water into the Pacific Ocean appears to have occurred during extreme weather 55 events because of the high affinity of Cs for particles. The result was the transfer of up to 10-12 TBq of radiocaesium from the land to the ocean during the first year after the accident (Evrard et al., 2015). (Nagao et al., 2014) found that about 50% of the radiocaesium in the Niida River was in dissolved form in the first few months after the accident, whereas after September 2011 more than 70% of the radiocaesium was associated with particles, and that percentage was even higher after storms.
Recently, inputs of particulate and dissolved radiocaesium by several rivers in the Maeda Basin following heavy rain events 60 including typhoons were estimated (Sakuma et al., 2019). They have developed a simple model to estimate the 137 Cs discharge from catchments. They used this model to estimate the discharge of 137 Cs and the ratio of discharged 137 Cs to the inventory of 137 Cs deposited in the catchments of the Abukuma River and 13 other rivers in the Fukushima coastal region from the time immediately after the Fukushima accident until December 2017. The discharge of 137 Cs and the ratio of discharged 137 Cs to the inventory of 137 Cs during the initial six months after the accident were estimated to be 18 TBq (3.1%) for the Abukuma River and 11 TBq (0.79%) for the 13 other rivers. These 137 Cs discharge ratios were 1-2 orders of magnitude higher than those observed after June 2011 in previous studies (Ueda et al., 2013;Tsuji et al., 2016;Iwagami et al., 2017). The impact on the ocean from the initial discharge of 137 Cs through the rivers was limited because the discharge of 29 TBq (18 TBq + 11 TBq) of 137 Cs from the Abukuma River and the 13 other rivers in the Fukushima coastal region was two orders of magnitude smaller than the direct release into the ocean of 3.5 ± 0.7 PBq from the FNPP1 (Tsumune et al., 2012) and 7.6 PBq from atmospheric 70 deposition (Kobayashi et al., 2013). During the period from October 2012 to December 2017, the total discharge from 13 rivers in the Maeda basin has been estimated to have been 12 TBq (23 TBq − 11 TBq) (Sakuma et al., 2019). Direct discharge from the FNPP1 site decreased dramatically during that time, and the direct discharge of 137 Cs to the ocean was estimated to be only 0.73-1.0 TBq year −1 during 2016-2018 (Aoyama et al., 2020d). This pattern indicates that discharges from the 13 rivers in the Maeda basin and direct discharges might have become similar in magnitude recently. Among the 13 rivers in the Fukushima 75 coastal region, the Ukedo River was the major contributor of 137 Cs discharged to the ocean during the study periods of 11 March 2011 to 27 September 2012 and 11 March 2011 to 31 December 2017 because the total inventory of 137 Cs in the catchment of the Ukedo River derived from atmospheric fallout was the largest component (536 PBq) of the total amount of 137 Cs in the catchments of the 13 rivers in the Fukushima coastal region, 1282 PBq (Sakuma et al., 2019). The results of Sakuma et al. (2019) showed that the particulate 137 Cs discharge from the Abukuma River was approximately 1-2 orders of 80 magnitude larger than the dissolved 137 Cs discharge under both base flow and storm flow conditions. The particulate 137 Cs discharge from the Maeda River was close to the dissolved 137 Cs discharge under base-flow conditions, but the discharge of particulate 137 Cs was approximately 1-3 orders of magnitude larger than the discharge of dissolved 137 Cs under storm flow conditions. We therefore expected that the discharge of particulate 137 Cs might account for most of the discharge of 137 Cs from rivers in our study region after heavy rains associated with typhoons. 85 The goal of this study, which was conducted in October 2014, was to analyse coastal waters close to small coastal rivers that flowed through highly contaminated watersheds close to the damaged FNNP1 after the passage of two typhoons. We examined dissolved 137 Cs activities and all forms of particulate 137 Cs activity concentration in seawater, as well as the ratios of 137 Cs to 134 Cs activities. We also analyzed the long-term trend of radiocaesium data in surface water at Tomioka together with precipitation records and monitoring data obtained by TEPCO to understand riverine fluxes in the coastal regions of 90 Fukushima.

Material and methods
In October 2014, the cruise SOSO 5 Rivers took place off the coast of Fukushima prefecture. The sampling targeted the areas off the mouths of five rivers (i.e., the Mano, Nitta, Ota, Odaka, and Ukedo rivers) located north of the FNPP1, the watersheds of which were highly contaminated by fallout from the FNPP1 accident. Seawater was sampled at the surface and from 1 m 95 above the sea bottom at 5 stations along 5 radial transects from the mouths of each of these five rivers as total 25 stations as shown black dots in Fig. 1. We also collect surface seawater at 12 locations, S1 to S12, along the coast from north of the mouth https://doi.org/10.5194/bg-2020-491 Preprint. Discussion started: 8 February 2021 c Author(s) 2021. CC BY 4.0 License. of the Mano River at station S1 to south of the mouth of the Ukedo River at the stations S11 and S12 where the water depths were 8-10 m as shown red solid circles in Fig. 1. In addition, one water sample was collected from four of the five estuaries (the Ukedo estuary was not accessible at the time of sampling, locations are not shown in Fig.1). All samples were filtered 100 through 0.45-µm filters.
Before the cruise, typhoon #18 (Phanfone) caused heavy rain in the area of interest on 6 October 2014. We collected samples along the Mano River transect on 9 October and along the Ota River transect on 10 October. Because typhoon #19 (Vongfong) made landfall on 13 to 14 October, sampling was postponed until 17 October for the Ukedo River transect, 18 October for the Odaka River transect, and 19 October for the Nitta River transect. Sampling at all stations S1-S12 was carried out between 16 105 and 19 October.
An improved ammonium phosphomolybdate (AMP) procedure ) was used to extract radiocaesium from the seawater samples as dissolved form which were obtained by filtration through 0.45-µm filters. With this procedure, the weight yields of the AMP/Cs compound as well as the radiochemical yields of radiocaesium generally exceeded 99% for 2-liter samples. The activities of AMP/Cs compounds were measured at the Ogoya Underground Facility of the Low Level 110 Radioactivity Laboratory of Kanazawa University using high-efficiency, well-type, ultra-low-background Ge-detectors (Lutter et al., 2015;Aoyama et al., 2009;Hamajima and Komura, 2004). Organic form of radiocaesium of the samples were obtained by disillusion of organic portion on the filter using concentrated nitric acid and concentrated hydrogen peroxide at all stations, then filled and dried up in a Teflon tube. The activities of organic portion of radiocaesium in Teflon tube were also measured at the Ogoya Underground Facility of the Low Level Radioactivity Laboratory At some stations, we were able to recover a 115 sufficient amount of suspended matter (by filtration through 0.45-µm filters) to measure radiocaesium activities by gamma spectrometry at the Modane underground laboratory in France using ultra-low-level, well-type, high-purity Ge γ-detectors (Canberra Industries). The counting time ranged between 240,000 and 320,000 s. Because these samples were collected separately from other samples and the suspended matter was heterogeneous, the mass concentrations of samples for all particles and for organic particles were not the same. However, the 137 Cs and 134 Cs activities in the organic particles and all particles 120 did not depend on the mass concentrations. We have therefore reported the activities of radiocaesium in both kinds of particles in this article. Fig. 1), we measured the dissolved radiocaesium activity concentration in all samples and the particulate radiocaesium activity concentration for selected samples during the period from June 2014 to April 2019 using the methodology described above. 125

Results
Activity concentrations of 134 Cs and 137 Cs in dissolved form, all particles, and organic particles collected during the SoSo 5 Rivers cruise activities used in this study are in a published dataset entitled "Dataset of 134Cs and 137Cs activity concentration We also used TEPCO time series monitoring data of 137 Cs activity concentration in surface water at Ukedo port, near the outlet of canal from unit 5 and 6 of FNPP1 to the sea, hereafter 56N of FNPP1, and Iwasawa beach (open circles in Fig. 1) and at the 140 Fukushima Dai-ni Nuclear Power Plant, hereafter FNPP2 (solid square in Fig. 1). TEPCO time series data used in this study are available at https://emdb.jaea.go.jp/emdb/portals/1060113000/.

Dissolved radiocaesium
The detection of both 134 Cs and 137 Cs in all samples confirmed contamination from the FNPP1 accident. Generally, the 145 dissolved concentrations were higher at coastal sites and decreased with distance from the coast (Fig. 2), and they were higher in the surface layer compared to the bottom layer ( Fig. 3).
At the stations very close to the coast within 1 km from the coast, 137 Cs activity concentration ranged from 10.6 ± 0.6 Bq m −3 at the bottom of Niida 1 station to 62.5 ± 3.5 Bq m −3 at the surface of station S12 as shown in Fig.3 and in a dataset doi: 10.34355/CRiED.U.TSUKUBA.00030 (Aoyama et al., 2020a). They then decreased to less than 10 Bq m −3 at the stations 150 about 2 km from the coast for bottom samples and at the stations about 4 km from the coast for surface samples along each radial transect (Fig. 3). In general, the lowest 137 Cs activity concentration was observed at the bottom at the stations furthest from the coast (Fig. 3). The lowest 137 Cs activity concentration was therefore observed at the bottom at the Nitta 5 station (2.0 ± 0.1 Bq m −3 ), whereas the highest 137 Cs activity concentration was observed at the surface at station S12 (62.5 ± 3.5 Bq m −3 ) as shown in the dataset doi: 10.34355/CRiED.U.TSUKUBA.00030 (Aoyama et al., 2020a). 155 The decreases of 137 Cs activities in surface water with distance from shore were similar along four of the radial transects (the Mano, Niida, Odata, and Ota transects). The activity concentration at a distance of 1 km decreased to 90-96% of that at the first station along each transect and 36-68% of the corresponding activity concentration at a distance of 10 km. The rate of decrease was higher along the Ukedo transect: 85% at 1 km and 19% at 10 km. These differences might reflect differences in the magnitudes of the river fluxes associated with the heavy rains caused by the typhoons that passed over the study area and 160 the initial 137 Cs activities in the river water. The entire dataset showed clearly higher activities at all the locations nearest the coast, regardless of salinity. In general, 137 Cs activities tended to decrease with distance from the coast. The 137 Cs activities also tended to decrease with increasing salinity (Fig. 4) because of mixing of coastal water and open-ocean water in which the 137 Cs activity concentration was ~1.5 Bq m −3 (see section 4.2). 165

Particle-associated radiocaesium
The 137 Cs activity concentrations in all particles observed in this study ranged from 2.25 ± 0.33 Bq m −3 in surface water at Odaka 1 to 704 ± 88 Bq m −3 in surface water at Ukedo 1. At the stations very close to the coast, Ukedo 1, S4, S10, S11, and S12, we observed 137 Cs activity concentration in all particles exceeded the dissolved 137 Cs activity concentration (Fig. 5). The ratio of particulate 137 Cs activity concentration to the sum of particulate and dissolved 137 Cs activity concentration ranged from 170 0.13 ± 0.02 to 0.96 ± 0.13 at all stations. Ratios at the lower end of this range are comparable to the ratios observed at Tomioka in August 2014. Those ratios ranged from 0.06 ± 0.01 to 0.18 ± 0.01 during a time of not much rain (Aoyama et al., 2020b).
Particulate 137 Cs activities in organic particles were generally one or two orders of magnitude lower than dissolved 137 Cs activities, and the ratio of 137 Cs activity concentration in organic particles to dissolved 137 Cs activity concentration ranged from 0.01 ± 0.00 to 0.12 ± 0.01, except at a few stations. The ratios at those exceptional stations were 1.18 ± 0.09 at the bottom at 175 Odaka 4, 0.15 ± 0.01 at the surface of Ota 3, and 0.32 ± 0.03 at the surface of station S4.

Activity ratios of 137 Cs to 134 Cs in dissolved form, all particles, and organic particles
The ratio of dissolved 137 Cs to dissolved 134 Cs activity concentration changed because of mixing with open-ocean water of which 137 Cs activity concentration is ~1.5 Bq m −3 and 134 Cs activity concentration can be assumed zero due to relatively short 180 half-life (Figs. 6 and 7). Along a radial transect, the ratio of dissolved 137 Cs to dissolved 134 Cs activity concentration tended to increase with distance from the coast (Fig. 6). Mixing with open-ocean water tended to increase the ratio of 137 Cs to 134 Cs activity concentration because the open-ocean seawater contained virtually no 134 Cs. It is very clear from Fig. 7 that the ratio of dissolved 137 Cs to dissolved 134 Cs activity concentration increased as the dissolved 137 Cs activity concentration decreased because of mixing between open-ocean seawater and freshwater (see section 4.2). In contrast, the ratio of 137 Cs to 134 Cs activity 185 concentration in organic particles did not change with distance and remained at around 1 along 5 radial transects of SoSo 5 river cruises. This observation indicates that the source of the organic particles was very close to the coast, or that the particles were present in the rivers, indeed Naulier et. al. (Naulier et al., 2017) showed that organic particulate matter from contaminated watersheds could be a significant radiocaesium carrier towards the sea.

An estimation of initial activity concentration of freshwater from land
There was no clear relationship between salinity and dissolved radiocaesium activity concentration (Fig. 4), except off the mouth of the Ukedo River where dissolved radiocaesium activity concentrations clearly decreased with increasing salinity https://doi.org/10.5194/bg-2020-491 Preprint. Discussion started: 8 February 2021 c Author(s) 2021. CC BY 4.0 License. (Fig. 8). This relationship could be the signature of riverine inputs to the coastal areas because the Ukedo River flows through 195 a highly contaminated watershed. However, several authors have emphasized that rivers discharge small amounts of dissolved radiocaesium (Sakuma et al., 2019;Nagao et al., 2014). As mentioned in the Introduction, during the period of our sampling, typhoons #18 (Phanfone) and #19 (Vongfong) made landfall on Japan on 6 and 13 October, respectively. These typhoons were associated with heavy rainfall and runoff that lowered the salinity of coastal waters and increased the fluxes of radionuclides from the FNPP1 accident. Because the Ukedo River is close to the FNPP1, this relationship between salinity and dissolved 200 radiocaesium activity concentration off the mouth of the Ukedo River can be linked to runoff of contaminated water because the highest 137 Cs activity concentration during the typhoons was observed in the 56N canal of the FNPP1 (Fig. 9)). A simple mixing model that explained the 137 Cs data along the Ukedo radial transect on 17 October 2014 showed that the 137 Cs activity concentration at a salinity of zero (i.e., the 137 Cs activity concentration in the Ukedo River) should be around 260 Bq m −3 (Fig. 8). This 137 Cs activity concentration is in good agreement with the activity concentration of 230 Bq m −3 reported in 205 the TEPCO monitoring data at the Ukedo port on 16 October 2014 (Fig. 9). In contrast, off the mouths of the other four rivers, a negative correlation or weak relationship (data not shown) was found between salinity and radiocaesium activity concentration, though the range of salinity was relatively small. In the case of the Ota River, where the correlation was positive, the most-saline water was found near the coast, and lower salinity water was found further from the river mouth (figure not shown; data are in Aoyama et al., 2020a). This pattern might reflect complex physical processes associated with southward 210 advection and a small amount of eddy mixing along the coast.

Source term estimation of observed radiocaesium based on 137 Cs/ 134 Cs activity ratio in seawater
We carried out standardized major axis regression which accounts for the uncertainty in both x and y by minimizing the errors in both directions on decay-corrected (to 11 March 2011) 134 Cs and 137 Cs activity concentrations in the samples collected by 215 the SoSo 5 rivers project (Aoyama et al., 2020a) and the time series of observations at Tomioka (Aoyama et al., 2020c). Then we obtained the 137 Cs/ 134 Cs activity ratio at the time of the accident as a slope of standardized major axis regression line and the pre-Fukushima 137 Cs activity concentration due to global fallout from atmospheric weapons tests as an intercept of regression line. Results of a standardized major axis regression are shown in Table 1 and Figures 10 and 11. The 137 Cs/ 134 Cs activity ratios in the cores of the FNPP1 at the time of the accident has been estimated (Nishihara et al., 2012) and are also 220 shown in Table 1.
The relationship between the decay-corrected 137 Cs and 134 Cs activities at the time of the accident (Fig. 10) showed that the pre-Fukushima 137 Cs activities due to global fallout from atmospheric weapons tests were 1.3 ± 0.2 Bq m −3 , in agreement with the pre-Fukushima 137 Cs activity concentration reported . Tsuruta et al., (Tsuruta et al., 2014) have shown that the first plume from Unit 1 of the FNPP1 contaminated the catchment area of the Ukedo River. The 137 Cs/ 134 Cs activity 225 ratio in dissolved form in the SoSo project samples associated with the FNPP1 accident was estimated to be 1.074 ± 0.015. This 137 Cs/ 134 Cs activity ratio is in good agreement with the 137 Cs/ 134 Cs activity ratio of 1.06 (+-10%) of the radiocaesium in the core of Unit 1 of the FNPP1 at the time of the accident (Nishihara et al., 2012). The pre-Fukushima 137 Cs activity concentration originated from atmospheric weapons tests in the seawater samples collected at Tomioka port was estimated to be 1.2 ± 0.7 Bq m −3 (Fig. 11), again in good agreement with the activity concentration in the 230 SoSo project samples and previous reported pre-Fukushima 137 Cs activity concentration . The 137 Cs/ 134 Cs activity ratio in the dissolved radiocaesium in surface water at Tomioka port was 0.998± 0.017. The Tomioka port located 10 km south of the FNPP1 (Fig. 1) and also locates at the mouth of the Tomioka river. In the Tomioka River catchment, the source of radiocaesium might be the core of Unit 2 of FNPP1 (Nakajima et al., 2017), then the 137 Cs/ 134 Cs activity concentration ratio might be 0.92 (+-10%) (Nishihara et al., 2012). The 137 Cs/ 134 Cs activity ratio of 0.998 ± 0.017 observed at Tomioka port was, 235 however, slightly higher rather than that in the core of Unit 2 of the FNPP1. Possible explanation of this finding are that the radiocaesium in the coastal seawater at the Tomioka port might be a mixture of radiocaesium from the core of the Unit 1 (ratio = 1.06) and the Unit 2 (ratio = 0.92) of the FNPP1 because the flow direction of the coastal current is primarily southward in this region.
Therefore we can conclude the source of radiocaesium in seawater in the coastal region north of FNPP1 was deposited 240 radiocaesium released from the core of Unit 1 of FNPP1, while the source of radiocaesium observed in the coastal region south of FNPP1 was a mixture of deposited radiocaesium released from the core of Unit 2 and the core of Unit 1 of FNPP1.

Analysis of relationship between 137 Cs activity concentration in surface water and the antecedent precipitation index
During our concentrated study period in October 2014, we observed increase of 137 Cs activity concentration in seawater which 245 might reflect increased riverine flux of 137 Cs due to heavy rain as discussed in the previous sections. We therefore investigated the relationship between 137 Cs activity concentration in seawater observed at several stations along the Fukushima coast and precipitation observed by Automated Meteorological Data Acquisition System located at Tomioka Town (Fig. 1). We adapted the antecedent precipitation index, hereafter API, which is a weighted summation of daily amounts of precipitation and is used as an index of soil moisture (https://glossary.ametsoc.org/wiki/Antecedent_precipitation_index, accessed on 6 October 2020) 250 as an index of the fluvial flux of radiocaesium in this study. The API is usually calculated with equation (1) . API = P0 + k 1 P1 + k 2 P2 + , , , + k n Pn (1) where P0 is total amount of precipitation on the day of radiocaesium measurement, Pn is total amount of precipitation on day 255 n before radiocaesium measurement, k is constant that depends on region and season and n is period of calculation.
In this study, we did not have information about k and n. We set k equal to 1 and let n vary between 1 and 9. We also did not include P0 because sampling was generally conducted in the morning, and there was a time lag between a rainfall event in the catchment of each river and the time the associated runoff reached the coast. We therefore modified the API in this study. 260 Because of the tsunami and radioactive contamination from the FNPP1, the precipitation-observation system was not available until April 2014, the modified API could be calculated only after April 2014. We calculated the correlation coefficients between the 1-9-day modified APIs and the 137 Cs activity concentration in surface water at Ukedo port, the 56N of the FNPP1, Tomioka port, and Iwasawa beach. In Figs. 12a to 12e, we also show examples of relationships between modified API and the 137 Cs activity concentration in surface water at Tomioka port for 1 day, 3 days, 265 5 days, 7 days and 9 days APIs. As shown in Figs. 12 to 16 and Fig. 17, the correlation became better with increase of days of integration from 1 day to 5 days to calculate API. The correlation coefficient reached a maximum for 5-7-days modified APIs at Tomioka port, Ukedo port, and Iwasawa beach (Fig. 17) and since most of the seawater sampling was conducted for ~7 days interval, we chose the 7-day modified API as an appropriate API index in this study hereafter.
To show general figure on the relationship between 7 days modified API and the 137 Cs activity concentrations, a Hovmöller 270 diagram of the 137 Cs activity concentrations at Ukedo port, the 56N of the FNPP1, Tomioka port, at FNPP2 port and Iwasawa beach is presented with the 7-day modified API with a reversed ordinate in this Figure (Fig. 18). It is clear that in September-October of each year, the typhoon season in Japan, the 7-day modified API exceeded ~150 mm and then the 137 Cs activity concentration increased at Ukedo port, Tomioka port, FNPP2, and Iwasawa beach, but the relationship with heavy rainfall was weak in the 56N of the FNPP1. The reason for the lower correlation coefficient between 137 Cs activity concentration at the 275 56N of the FNPP1 and the 7-day modified API might reflect occurring extreme increases of 137 Cs activity concentration without heavy rain events within the FNPP1 site as show in Fig. 19.

Conclusions
The detection of dissolved 134 Cs and 137 Cs activity concentrations in all samples demonstrated contamination from the FNPP1 280 accident. The dissolved 137 Cs activity concentrations were generally higher at coastal sites and decreased with distance from the coast, and they were higher in the surface layer compared to the bottom layer. The decreases of the rates of 137 Cs activity concentration with distance from shore in surface water were similar along radial transects at Mano, Niida, Odata, and Ota.
The decrease of 137 Cs activity concentration with increasing salinity reflected mixing of coastal water with open-ocean water, the 137 Cs activity concentration of which was only ~1.5 Bq m −3 . 285 137 Cs activity concentration in all particles observed in this study ranged from 2.25 ± 0.33 Bq m −3 at the surface at Odaka 1 to 704 ± 88 Bq m −3 at the surface at Ukedo 1. At the stations very close to the coast, Ukedo 1, S4, S10, S11, and S12, we observed relatively high 137 Cs activities in all particles, and those activities exceeded the dissolved 137 Cs activities. The ratio of 137 Cs activity concentration in all particles to the sum of dissolved and particulate 137 Cs activity concentration ranged from 0.13 ± 0.02 to 0.96 ± 0.13 at all stations and the lower range of this ratio was consistent with the analogous ratios at Tomioka port in 290 August 2014 when there was no heavy rainfall. 137 Cs activities were generally one or two orders of magnitudes lower in organic particles than in dissolved form; the ratio of 137 Cs activity concentration in organic particles to dissolved 137 Cs activity concentration ranged from 0.01 ± 0.00 to 0.12 ± 0.01, except at a few stations. The ratio of dissolved 137 Cs to dissolved 134 Cs activity concentration changed dramatically because of mixing with open-ocean water; it tended to increase with increasing distance from shore and to increase as the 295 activity concentration of dissolved 137 Cs decreased. In contrast, the ratio of 137 Cs to 134 Cs activity concentration in organic https://doi.org/10.5194/bg-2020-491 Preprint. Discussion started: 8 February 2021 c Author(s) 2021. CC BY 4.0 License. particles did not change with distance from shore or with 137 Cs activity concentration and generally remained at ~1, even at locations far from the coast. This pattern indicated that the source of the organic particles was the rivers or another source very close to the coast. A simple mixing model for the 137 Cs data obtained along the Ukedo transect on 17 October 2014 indicated that the 137 Cs 300 activity concentration at the mouth of the Ukedo River should be ~260 Bq m −3 , which is in good agreement with the activity concentration of 230 Bq m −3 recorded in the TEPCO monitoring data at the Ukedo port on 16 October 2014.
The relationship between the decay-corrected 137 Cs and 134 Cs activity concentration corrected to the time of the accident showed that the pre-Fukushima 137 Cs activity concentration due to global fallout were 1.3 ± 0.2 Bq m −3 at sea area north of FNPP1 region and 1.2 ± 0.7 Bq m −3 at Tomioka port, respectively. These 137 Cs activity concentrations are in agreement with 305 the pre-Fukushima 137 Cs activity concentration . The 137 Cs/ 134 Cs activity ratio in dissolved form in the SoSo project samples collected north of FNPP1 was estimated to be 1.074 ± 0.015 which is in good agreement with the 137 Cs/ 134 Cs activity ratio of 1.06 (+-10%) of the radiocaesium in the core of Unit 1 of the FNPP1, while it at Tomioka port was 0.998± 0.017. In the Tomioka River catchment, the source of radiocaesium might be the core of Unit 2 of FNPP1 (Nakajima et al., 2017), then the 137 Cs/ 134 Cs activity concentration ratio might be 0.92 (+-10%) (Nishihara et al., 2012). Therefore we 310 concluded that the source of radiocaesium in seawater in the coastal region north of FNPP1 was deposited radiocaesium released from the core of Unit 1 of FNPP1, while the source of radiocaesium observed in the coastal region south of FNPP1 was a mixture of deposited radiocaesium released from the core of Unit 2 and the core of Unit 1 of FNPP1. 137 Cs activities at Fukushima coast and 7-day modified API showed a good positive relationship with the exception at the 56N canal of the FNPP1. The reason for the lower correlation coefficient between 137 Cs activity concentration in the 56N canal of 315 the FNPP1 and the 5-7 day modified API might reflect some very high 137 Cs activities with not heavy rain event.

Data availability
Activity concentrations of 134 Cs and 137 Cs in dissolved form, all particles, and organic particles collected during the SoSo 5 Rivers cruise activities used in this study are in a published dataset entitled "Dataset of 134Cs and 137Cs activity concentration 320 concentrations in dissolved for, all particles and organic form of particles obtained by SoSo 5 rivers cruise in 2014" as doi: 10.34355/CRiED.U.TSUKUBA.00030 (Aoyama et al., 2020a).   Fig. 5. Relationship between the distance from the coast and 137 Cs activityconcentrations in dissolved form (blue solid circle), all particle (brown solid square) and organic form of particle (red solid star).