The global fallout from the atmospheric nuclear weapons tests during the late 1950s to early 1960s is the main source of artificial radiocesium to the world ocean1,2 and can still be discerned in the surface water, the subsurface, and deep waters (up to 600–1000 m) in the western North Pacific Ocean (NPO)3. After the Fukushima Daiichi nuclear power plant (FDNPP) accident on March 11th, 2011, additional anthropogenic radiocesium with long half-life 137Cs (30.2 year) and short half-life 134Cs (2.06 year) was released into the NPO4,5,6,7,8,9,10,11. Previous studies have estimated that the total amount of 137Cs released by the FDNPP accident ranged between 15 and 27 PBq, of which a very significant portion found its way to the atmosphere and ocean5,12,13. Traceable, FDNPP-derived 137Cs (latitude 37.42° N, longitude 141.03° E) follows the pathway of the North Pacific circulation toward the eastern NPO6,9,14. Thus 60 years after the first nuclear tests in the Pacific region, the fate of long-lives radionuclides on the western NPO and adjacent marginal seas still raises social concerns about potential health risks, e.g. associated with seafood consumption and fishing industry.

Recent studies have observed that 137Cs can disperse through the Subtropical Mode Water (STMW) and Central Mode Water (CMW) in the western NPO9,15. STMW is formed right south of the Kuroshio Current and Extension (between 132° E and near the dateline) by deep vertical convection in winter16,17,18. CMW forms north of the Kuroshio Extension (around 36–41° N, 160° E–165° W), hence is characterised by colder temperatures and a deeper layer (250–500 m)19. These two water masses (i.e., STMW and CMW) circulate clockwise beneath the surface of the western NPO20. When plotted against depth, the vertical profiles of 137Cs activity concentrations recorded at 20° N, 165° E in 2002 exhibited two distinct peaks: the first one at a potential density (σθ) of 25.5 kg m−3 (corresponding to the σθ range of STMW) and the second one at a σθ of 26.0 kg m−3 (corresponding to the σθ range of CMW)3. The subsequent injection of FDNPP-derived radionuclides resulted in a contamination plume spreading eastward and merging with the Oyashio Current; hence the incorporation of radiocesium in the formation of STMW and CMW and its current value as a tracer of ocean circulation4.

Both water masses (i.e., STMW and CMW) extend clockwise to the western boundary of the NPO, where the Kuroshio is most intense. Branches of the Kuroshio Current intrude into the northern South China Sea (SCS), southern Taiwan Strait, and even the southern East China Sea (ECS). Inomata et al.6 estimated that 5.0% of the total amount of FDNPP-derived 137Cs in the STMW entered into the Sea of Japan before 2016 through clockwise spreading in the western NPO, ECS, and the Sea of Japan. Several studies have reported the intrusion of radiocesium into the western NPO but also into the adjacent marginal seas after the FDNPP event4,5,21. However, seawater 137Cs activity concentrations have hardly been investigated in the Kuroshio east of Taiwan and the Taiwan Strait, particularly after the 2011 event.

Subtropical marginal seas of the western NPO are characterized by a monsoon cycle and seasonal coastal currents. For instance, the Taiwan Strait is affected by the northeasterly monsoon in the winter and the southwesterly monsoon in the summer22,23. As a result, the northern half of the Taiwan Strait is characterized by a southward-flowing coastal current in the winter while the southern half of the Taiwan Strait is characterized by a warm, northward-flowing current reinforced by the Kuroshio intrusion branch in the summer24,25,26,27. Furthermore, there are well documented upwelling regions in the study area, such as the waters off northeastern Taiwan28,29, several regions in the Taiwan Strait30,31, and regions in the northern SCS32,33,34. However, to the best of our knowledge, the effects of seasonality and upwelling on the fate of radiocesium in seawater in this study area remain unclear.

This study reports the activity concentration of 134Cs and 137Cs over the shelf break of the western NPO and examines the fates of these radionuclides on the shallow continental shelf. The 134Cs and 137Cs activity concentrations were measured on samples collected in near-surface, subsurface or deeper waters in the Kuroshio region east of Taiwan and the Taiwan Strait between 2018 and 2019. The origin of the 137Cs maximum in the subsurface/deep waters and the factors controlling seasonal variations in 137Cs activity concentration in the near-surface waters are discussed.


From 2018 to 2019, surface (< 5 m), subsurface (5–200 m), and deep seawater (200–1000 m) samples were collected at sites in the Kuroshio east of Taiwan and the Taiwan Strait (Fig. 1). Surface seawater samples (40 or 60 L) were collected mostly from fishing boats by using cleaned 20-L tanks. Subsurface samples were taken by using Niskin bottles mounted on a Conductivity–Temperature–Depth (CTD) rosette, which recorded temperature, salinity (from conductivity), and water depth (from pressure) onboard R/Vs Ocean Researcher I, II, and III. Sampling locations are shown in Fig. 1. Each 20-L sample was acidified using hydrochloric acid (11 M HCl, 100 mL) and was kept at room temperature (~ 15–30 °C) until it was transported to the Radian Monitor Center, Atomic Energy Council, Kaohsiung, Taiwan. Radiocesium was pre-concentrated by adsorption onto ammonium molybdophosphate (AMP)35,36 and counted using a high-purity germanium (HPGe) detector with lead shielding. Each 40-L or 60-L sample was counted for 200,000 s or 120,000 s, respectively. The detection limits of both 134Cs and 137Cs were 0.5 Bq m−3. In this study, we corrected all 137Cs activity concentrations to January 1st, 2020. This date was also applied to data in earlier studies for comparison purposes.

Figure 1
figure 1

Sampling area. The Taiwan Strait and the Kuroshio east of Taiwan are critical regions in the western North Pacific Ocean (NPO). (a) Surface (solid markers) and subsurface and deep-water (open markers) samples were collected in the study area from 2018 to 2019. (b) The two dashed lines indicate the northern and southern boundaries of the Taiwan Strait used in this study. Red dashed circles are potential upwelling sites positioned using the temperature and chlorophyll-a data by Liu et al.31. This image was created by Ocean Data View (Version 4.7.5) (Schlitzer, R., Ocean Data View,, 2016).

Total alkalinity (TA) samples were collected together with nearly 61% of radiocesium samples (92% of 49 subsurface and deep waters, and 55% of 242 near-surface waters). TA water samples were taken into 250-mL borosilicate glass bottles and poisoned immediately with 100-µL saturated HgCl2 solution to eliminate biological activities. We followed the open-cell Gran titration method with a temperature-controlled, semi-automated titrator (AS-ALK2 Apollo Scitech) to determine TA value of each sample. TA measurements were referenced against certified reference materials from A. G. Dickson’s laboratory at Scripps Institution of Oceanography with a precision of 0.1%37. TA is usually treated as a conservative tracer. We used it to determine the water sources and possible mixing processes between fresh water and seawater by normalizing TA values to salinity 35 (NTA = (TA/S) × 35)38.

As variables can be collectively controlled by major oceanographic mechanisms, principal component analysis (PCA)39,40 serves as a multivariate analysis tool to catch the major features of a dataset. PCA can analyze inter-correlations among these variables and reduces the dimension of a dataset. The major factors affecting this dataset is thus obtained. The result is displayed as a subset of new, independent (orthogonal) variables which are referred to as dimensions. The higher the coordinates of a dimension are, the greater the amount of co-variability among the original variables this dimension explains. The results were presented graphically as plots with each dimension and length represented the relationship and weight to the principal components, correspondingly. We applied PCA using R software41 on our data set for which four variables had been determined: salinity, temperature, σθ, and 137Cs activity concentrations from the surface layer to a depth of 400 m. The oceanographic context is then used to interpret the meaning of each dimension.


Surface water properties and the distribution of 134Cs and 137Cs

Surface water samples were obtained from depths of less than 5 m, with temperatures of 9.7–34.9 °C and salinities of 21.8–34.2 psu (Figs. 2 and 3a). Salinity, TA, and σθ displayed large variations in surface waters (Fig. 3). Salinities were comprised between 35 and 21 psu and σθ was usually lower than 24 kg m−3 (Fig. 3a,b), indicating mixing between seawater and fresh water. The average TA in surface waters was 2243 ± 38 μmol kg−1. NTA values associated with low σθ values, i.e., surface waters, deviated significantly from the average value determined in subsurface and deep waters (2309 μmol kg−1) (Fig. 3d), implying that TA values in surface waters were likely to be affected by additional river TA sources.

Figure 2
figure 2

Temperature and salinity diagram. (a) Surface waters showed large variations in temperature and salinity. (b) Deep waters demonstrated salinities higher than 33.5 and represent the σθ range of the Subtropical Mode Water (STMW) and Central Mode Water (CMW) (25.1–26.2 kg m−3). Labels “1” and “2” in (a) represent the σθ range of primary and secondary 137Cs peaks, respectively.

Figure 3
figure 3

Vertical distributions of salinity, TA, NTA, and 137Cs concentration. In the subsurface and deep waters, where (a) salinities varied from 33.7 to 34.9 long the vertical profile, (b) TA values varied from 2284 to 2360 μmol kg−1, (c) 137Cs activity concentrations displayed the primary peak at a depth range of 200–400 m, and where (d) σθ varied from 25.2 to 26.1 kg m−3, (e) NTA was 2309 ± 5 μmol kg−1, (f) the primary peak of 137Cs was observed. (d) NTA with low σθ values deviated from 2309 in the near-surface waters. (f) The secondary peak of 137Cs activity concentration displayed σθ ranging from 19 to 22 kg m−3 in the near-surface waters. All panels share the same legend. Data from NTU2 were noted with triangle markers in corresponding solid circles.

In our samples, 134Cs activity concentrations were under the detection limit (0.5 Bq m−3), and 137Cs activity concentrations ranged from 0.5 to 2.0 Bq m−3 (Fig. 3e,f), with an average of 1.2 ± 0.3 Bq m−3 in the surface water (Supplementary Fig. S1). We also noticed a peak in the surface water 137Cs vertical profile (Fig. 3f), with the maximum value ranging from 1.95 to 1.96 Bq m−3θ = 18.8 to 21.4 kg m−3).

We arbitrarily divided the study area into geographic sectors and listed the average 137Cs values for each sector in Table 1. The average values of 137Cs activity concentrations in each sector were similar, with 137Cs = 1.2 ± 0.3 Bq m−3 in the Taiwan Strait and 1.2 ± 0.2 Bq m−3 in the Kuroshio and its adjacent waters (Table 1). In the surface waters, the 137Cs values binned in one-degree latitude bands were higher between 25 to 26° N than the ones between 21 to 22° N in the Kuroshio and its adjacent waters (approximately to the east of 121° E in our study area) (Fig. 4a). Binned 137Cs activity concentrations were close to each other to the west of 121° E (Fig. 4b).

Table 1 Average 137Cs values for the Taiwan Strait and the Kuroshio Current (KC) east of Taiwan (2018–2019) in Bq m−3.
Figure 4
figure 4

Latitude to 137Cs relationships in the near-surface waters. To the east of 121° E (a), the binned 137Cs activity concentration in one-degree latitude bands from 25 to 26° N was higher than the one from 21 to 22° N (black squares). (b) The binned 137Cs values in each latitude were close to each other to the west of 121° E.

The sea surface temperature (SST) distribution in 2018 displayed a general pattern, where SST values of less than 25 °C were found in the southern ECS and the northern Taiwan Strait while SST values over 27 °C were common for the Kuroshio and the Luzon Strait (Fig. 5). The SST in the northern Taiwan Strait and southern ECS showed strong seasonal variations, displaying SST over 25 °C during the summer-like months (i.e., July–September) and less than 25 °C during winter-like months (i.e., January–March) (Supplementary Fig. S2). The SST in shelf waters of the northern Taiwan Strait displayed stronger seasonal variations than those in the pelagic Kuroshio waters east of Taiwan.

Figure 5
figure 5

Sea surface distributions of seawater temperature. SST is an important factor while understanding the oceanic condition. High SST (MODIS 2018) was detected in the Kuroshio east of Taiwan and its intrusion into the southeastern Taiwan Strait. Low SST was identified along the coastline from the East China Sea (ECS) to the western side of the Taiwan Strait. The white dashed line indicates the contour line of 18 °C obtained from January. Monthly SST data was collected from AQUA-MODIS with a spatial resolution of 1° × 1° ( This data was accessed on July 3, 2021. This image was created by Ocean Data View (Version 4.7.5) (Schlitzer, R., Ocean Data View,, 2016).

In the shelf waters, including the Taiwan Strait and the waters off northern Taiwan, the average 137Cs activity concentration was statistically higher in August than in February (two-tailed t-test, p < 0.05) (Fig. 6a). A statistically significant relationship was observed between monthly variations of 137Cs and temperature (Fig. 6b). Monthly values of 137Cs (Fig. 6a) increased from winter to summer and again started to decrease in fall, implying a seasonal cycle of 137Cs in surface shelf waters.

Figure 6
figure 6

Monthly variations in sea-surface 137Cs activity concentration in the Taiwan Strait and southern ECS. (a) 137Cs activity concentration and corresponding seawater temperature were measured at the study sites on both sides of the Taiwan Strait and the waters off northern Taiwan (refer to Fig. 1b). Both parameters displayed seasonal variations: low in winter and high in summer. (b) The monthly 137Cs activity concentration was statistically correlated to its corresponding seawater temperature.

Subsurface and deep water properties and distributions of 134Cs and 137Cs

Subsurface/deep water samples were taken from depths of 5–1000 m, with temperatures of 4.7–28.2 °C, and salinities over 21.0 to 34.0 psu (Fig. 2). The T–S characteristics of the subsurface and deep waters (Fig. 2b) of the Kuroshio-influenced region of the NPO covered the signals of STMW (σθ: ~ 25.6 kg m−3), Kuroshio Tropical Water (KTW, T = 17.0 °C, S = 34.6 psu), and CMW (σθ ~ 26.1 kg m−3). TA values in the subsurface and deep waters varied in a narrow range (Fig. 3c). The average and standard deviation of NTA value for waters between 5 to 400 m was 2309 ± 5 μmol kg−1 (Fig. 3d).

At station NTU2 in the Kuroshio region, 134Cs activities were below the detection limit (0.5 Bq m−3), and 137Cs activities were lower than 2.5 Bq m−3 from the surface to a depth of 1000 m (Fig. 3e). Moreover, two layers of elevated 137Cs activities were observed: 137Cs activities were mostly higher than 1 Bq m−3 from 0 to 400 m and also higher than 2 Bq m−3 from 200 to 400 m. By contrast, they were lower than 1 Bq m−3 from 600 to 1000 m (Fig. 3f). A synthesis of these results showed that low 137Cs activity concentrations (1.5 to 0.6 Bq m−3) corresponded to the high σθ (> 25 kg m−3) waters between 400 and 1000 m (Fig. 3e,f).


The first three dimensions of the PCA results explained 91% of all variations, including sample temperature, salinity, and σθ, and 137Cs activity concentration above the depth of 400 m. Dimension 1 explained 54% of the variations (Fig. 7a,b) and was dominated by σθ, temperature, and salinity (Table 2). σθ, salinity, and 137Cs were positively correlated with Dimension 1 while the temperature was negatively correlated (Fig. 7a). We suggest that Dimension 1 represents density-induced water layer distributions and explains the primary peak of 137Cs in the deep waters. This suggestion is consistent with the fact that σθ is positively correlated with the variables that have a high loading on Dimension 1 (Fig. 7c) and is a marker for the two layers of elevated 137Cs in the top 400 m of the water column.

Figure 7
figure 7

Results of principal component analysis (PCA) and sample coordinates. (a,b) σθ dominated variations in Dimension 1; increasing temperature and salinity dominated variations in Dimension 2; and 137Cs dominated variations in Dimension 3. (c) The coordinates of individual results in Dimension 1 were consistent with the primary peak of 137Cs defined along the stratified water column. (d) The peak coordinate of Dimension 2 was consistent with the secondary 137Cs peak along the σθ gradient.

Table 2 Principal components.

The highest coordinate of Dimension 2 corresponds to the secondary peak of 137Cs in the surface waters (Fig. 7d), implying that Dimension 2 represents a factor that led to the secondary 137Cs peak. The plot of Dimension 2 coordinates against σθ (Fig. 7d) can be divided into a high σθ (> 22 kg m−3) arm and a low σθ (< 22 kg m−3) arm. As both temperature and salinity were positively correlated with Dimension 2 (Fig. 7a, Table 2), we hypothesize that the high σθ arm was caused by upwelling which transports high salinity water toward the surface and reduces the thickness of the lens of warm surface water in this study area. Upwelling near the coast of Taiwan and in the Taiwan Strait is known to transport subsurface waters (as deep as 80 m to 100 m) to the near-surface layer. The vertical velocity of upwelling off northeastern Taiwan (Fig. 1)42,43,44,45 has been estimated to be 15 m day−1 on the shelf and over 40 m day−1 at the shelf edge29,46. The σθ of water with salinity = 34.3 psu, temperature = 18.3 °C, and σθ = 25.0 kg m−3 at a depth of 200 m, can decrease to less than 22 kg m−3 if the water temperature increases to 28 °C at 1 m during the upwelling process (Fig. 2a). This annual upwelling to the northeastern Taiwan may lead to the higher binned 137Cs between 25 and 26° N than the others to the east of 121°E in the near-surface waters (Fig. 4a). Another driving force of annual upwelling is internal tide, which can induce upwelling in the waters off northeastern Taiwan47 and off southern Taiwan48,49,50. Moreover, this transition σθ of 22 kg m−3 in Fig. 7d was also consistent with the transition range of σθ where NTA positively deviated from 2309 ± 5 μmol kg−1 in the near-surface water (Fig. 3d). This deviation of NTA in low σθ waters indicates freshwater inputs from terrestrial runoff. To sum up, we suggest that mixing between riverine freshwater and seawater is responsible for the lower σθ arm while upwelling drives the higher σθ arm in Fig. 7d.

Dimension 3 accounted for no more of the variance than 137Cs itself (Table 2, Fig. 7b), suggesting that it was controlled by the chemical characteristics of 137Cs. MacKenzie et al.51 have argued that high freshwater discharge from land can remobilize 137Cs from surface sediments (< 10 cm). Future work should be conducted towards a dynamic representation of radionuclide transfer among freshwater, seawater, sediment, and the biological compartments11.

The seasonal variation of the 137Cs activity concentration in the Taiwan Strait (Fig. 6a) reflects the seasonal intrusions of waters from Kuroshio intrusion, ECS, and SCS. Wu et al.19 corrected their 137Cs activity concentrations to the same date as this study, leading to a mean of 0.71 ± 0.27 Bq m−3 in the surface ECS and an average of 0.92 ± 0.28 Bq m−3 in the surface SCS. These results are consistent with our observation that southward-flowing cold waters with low 137Cs values intruded into the northern half of the shallow Taiwan Strait during the winter while the reverse took place during the summer (Fig. 6a)22,24,25. In addition to the Kuroshio offshoot transporting warm waters with higher 137Cs values in the summer, the Pearl river plume can also intrude into the southern Taiwan Strait during the summer52. It follows that the slightly lower 137Cs activity concentration during July indicates that the effect of the intrusion of Kuroshio is more than offset by the amount of warm water with lower 137Cs activity originating from the SCS.

The maximum in 137Cs activity concentration was observed in a specific range of σθ in the subsurface and deep waters (Fig. 3e,f), implying a lateral transport along the 125–400 m depth horizon in addition to local atmospheric fallout. Local and modern atmospheric 137Cs fallout can only affect the average 137Cs values in the surface waters (Table 1). 137Cs activity concentrations in the subsurface water of the study area displayed characteristics (σθ = 25.2 and 26.1 kg m−3) which were similar to those of STMW and CMW (σθ = 25.3 to 26.3 kg m−3) in NPO4 (Fig. 3). Some surface NTA values and also the average NTA value in subsurface waters (2309 ± 5 μmol kg−1) of this study were consistent with NTA values previously reported in the surface western NPO (2301 ± 9 to 2299 ± 5.4 μmol kg−1)53,54. The maximum 137Cs activity concentration at a depth of 300 to 400 m at 165° E before the FDNPP event should be 1.65 Bq m−3 (corrected to January 1st, 2020)3. After the FDNPP event in June/July 2012, 137Cs activity concentrations at depths of between 0 and 600 m in the western NPO (25 to 45° N, 165° E) were between 2.1 and higher than 8.4 Bq m−3 (corrected to January 1st, 2020)4. As the half-life of 134Cs is shorter than that of 137Cs, 134Cs/137Cs is assumed to be 1.000 at 165° E; it became 0.095 after 7.5 year (July 2012 to December 2020). Since 137Cs activity concentration is already low (< 2.1 Bq m−3) in this study area, 134Cs is expected to be < 0.2 Bq m−3 which is lower than the detection limit (0.5 Bq m−3) in this study.

The increase between the estimated 137Cs activity concentration before FDNPP (1.65 Bq m−3) and the maximum measured 137Cs value after FDNPP (2.1 Bq m−3) in this study (both corrected to January 1st, 2020) implies an additional 137Cs activity concentration of 2.1–1.65 = 0.45 Bq m−3 during this time period. This increase is the result of complex physical transportation from NPO to the study area. Kamidaira et al.55 reported that approximately 43% of FDNPP-derived 137Cs could be delivered to below the mixed layer through eddy processes. In addition, 137Cs in STMW and CMW appears to be transported clockwise toward the western boundary of the NPO3,56,57,58,59. This subsurface or deep-water layer corresponding to σθ = 26.7 kg m−3 can further rise westward and reach the shelf break surrounding Taiwan60,61,62,63,64,65,66. East of Taiwan, the Kuroshio is a swift and powerful current that can reach as deep as 400–600 m67. While the primary 137Cs maximum centered around σθ of 25.3 to 26.1 kg m−3 is suspected to be from lateral transportation from the NPO, there is still a research gap between pelagic studies in the NPO and shelf break data in this study area. For example, impacts of interleaving68 and meso-scale eddies69 on the cross-Kuroshio transport mechanism of 137Cs are still unclear. Direct evidence to constrain the origin and evolution of the 137Cs maxima along 165° E to this study area is needed in the future. We suggest integrating multiple chemical tracers to study the complex circulation across the pelagic NPO to its western shelf boundary in the future.