Seasonal Variations of Surface fCO2 and Sea-Air CO2 Fluxes in the Ulleung Basin of the East/Japan Sea

Temperature, salinity, chlorophyll a, and surface CO2 fugacity (fCO2) were extensively investigated in the Ulleung Basin of the East/Japan Sea during four seasonal cruises. In spring, surface fCO2 showed large variations ranging from 260 to 356 μatm, which were considerably lower than the atmospheric CO2 levels. Surface fCO2 was highest (316 to 409 μatm) in summer. The central part of the study area was undersaturated with respect to atmospheric CO2, while the coastal and easternmost regions were oversaturated. In autumn, the entire study area was fairly undersaturated with respect to atmospheric CO2. In winter, surface fCO2 ranged from 303 to 371 μatm, similar to that in autumn, despite the much lower sea surface temperature. The seasonal variation in surface fCO2 could not be explained solely by seasonal changes in sea surface temperature and salinity. The vertical mixing, lateral transport, and sea-air CO2 exchange considerably influenced the seasonal variation in surface fCO2. The Ulleung Basin of the East/Japan Sea was a sink of atmospheric CO2 in spring, autumn, and winter, but a weak source of CO2 to the atmosphere in summer. The annual integrated sea-air CO2 flux in the Ulleung Basin of the East/ Japan Sea was -2.47 ± 1.26 mol m-2 yr-1, quite similar to a previous estimate (-2.2 mol m-2 yr-1) in the south East/Japan Sea. This indicates that the Ulleung Basin of the East/Japan Sea acts as a strong sink of atmospheric CO2.


INTRODUCTION
Marginal seas play important roles in the global carbon cycle, with those having high biological activities serving as annual net sinks (Borges et al. 2005;Omar et al. 2007;Chen and Borges 2009).However, while marginal seas at temperate and high latitudes act as net sinks, those at tropical and subtropical latitudes may act as net sources (Borges et al. 2005;Chen and Borges 2009).Previous studies on the sea-air CO 2 fluxes of marginal seas have reported large variability ranging from 0.1 ~ 0.45 Pg C yr -1 , which is attributable to the complex and heterogeneous ecosystems and hydrodynamics of these seas (Liu et al. 2000;Thomas et al. 2004;Borges et al. 2005;Chen and Borges 2009).For the same reason, estimates of sea-air CO 2 flux for marginal seas still contain much uncertainty.Therefore, spatially and temporally high-resolution CO 2 measurements in marginal seas are essential for improving estimates of global sea-air CO 2 fluxes.
The East/Japan Sea (hereafter, East Sea) is a semi-enclosed marginal sea surrounded by Korea, Japan, and Russia.It consists of three major basins: the Japan, Yamato, and Ulleung basins.The average depth of the East Sea is 1740 m, and it connects to the western North Pacific through four shallow straits with depths less than 140 m.Because of the shallow depths of these straits, subsurface waters below the thermocline (located at about 100 ~ 200 m) cannot be directly exchanged between the East Sea and North Pacific.Due to high biological productivity and accumulation of anthropogenic CO 2 (Yamada et al. 2005;Park et al. 2006;Yoo and Park 2009), the East Sea could be an important marginal sea in which to study oceanic carbon cycles.However, relatively few studies have investigated sea-air CO 2 fluxes in the southwestern part of the East Sea, the Ulleung Basin (Oh 1998;Kang 1999;Choi et al. 2011).Oh (1998) and Kang (1999) estimated daily and monthly averaged sea-air CO 2 flux in the East Sea using a multifactor mathematical model tuned by observational data in summer and winter.Choi et al. (2011) reported the surface fCO 2 distribution and sea-air CO 2 flux in the Ulleung Basin of the East Sea in the summer of 2005.Until now, however, seasonal observations of seaair CO 2 fluxes have not been performed in the East Sea.
We investigated the spatial distribution and seasonal variations of surface fCO 2 in the Ulleung Basin of the East Sea based on data obtained in April 2006, August 2007, and February and October 2008.Here we evaluate the major physical and biological factors controlling the distribution of surface fCO 2 in the study area, and estimate the sea-air CO 2 fluxes for four seasonal surveys.

Analytical Methods
The data were obtained during four seasonal cruises on board the R/V Eardo in spring (15 ~ 16 April 2006), summer (5 ~ 9 August 2007), autumn (9 ~ 14 October 2008), andwinter (20 ~ 23 February 2008).The study area was the Ulleung Basin of the East Sea (35 ~ 37°N, 129 ~ 132°E, Fig. 1).Continuous measurements were made of fCO 2 , temperature, and salinity in the surface seawater, which was pumped aboard from a 5-m depth during the surveys.Surface fCO 2 and atmospheric CO 2 were determined every minute and every hour, respectively, using an underway CO 2 measurement system consisting of a flowing pCO 2 system and a showerhead equilibrator.The fCO 2 measurements were described in detail by Shim et al. (2007).The system was calibrated every 12 hours with working standard gases (~250, 380, and 450 ppm CO 2 in the air, Korea Industrial Gases, Ltd., Shihung City, Korea), which were subsequently calibrated with the National Oceanic and Atmospheric Administration (NOAA) and World Meteorological Organization (WMO).Consistency between pre-and post-cruise analyses for the working standard gases was 1 μatm.Repeated analyses also indicated that the fCO 2 measurement had a precision of ±1 μatm.Vertical profiles of temperature, salinity, and density were measured with a SeaBird conductivity-temperature-depth profiler (CTD; SBE 9/11 plus, SeaBird Inc., Bellevue, WA, USA).Seawater samples for chlorophyll a analyses were collected using a Rosette sampler with 10-L Niskin bottles.The seawater for the chlorophyll a analysis was filtered through GF/F filter paper (47 mm, Whatman), and the filters were then immediately frozen using liquid nitrogen.The chlorophyll a concentra-tion in the extracted filtrate mixed with 90% acetone for 24 h was determined using a Turner-designed fluorometer (10-006R, Turner BioSystems, Sunnyvale, CA, USA).

Calculation of the Sea-Air CO 2 Flux
The sea-air CO 2 fluxes across the sea-air interface can be calculated based on the equation where k is the gas transfer velocity (cm h -1 ), s is the solubility of CO 2 gas in seawater (mol kg -1 atm -1 ; Weiss 1974), and ΔfCO 2 is the sea-air difference of CO 2 fugacity.A positive flux indicates that the sea is acting as a CO 2 source; a negative flux means that the sea is acting as a CO 2 sink.We adapted the formula for k and the wind speed relationships from Wanninkhof (1992) to allow comparison of our results with those of most other studies.Wind speed is the key force driving gas exchange at the sea-air interface.Here, we used the spatially averaged QuikSCAT wind speed data for the cruise dates, obtained from the Physical Oceanography Distributed Active Archive Center of the Jet Propulsion Laboratory, US National Aeronautics and Space Administration (PO.DAAC, JPL, NASA: http://podaac.jpl.nasa.gov).

Surface Currents in the Ulleung Basin of the East Sea
The Ulleung Basin is located in the southwestern part of the East Sea.Although the Korea/Tsushima Strait is the only entrance to the Ulleung Basin, several surface currents occur in the basin.The Tsushima Warm Current branches from the Kuroshio Current into the Ulleung Basin and then splits into two or three branches (Chang et al. 2002).Especially in summer, the Tsushima Warm Current carries both warm, low salinity water originating from the shelf of the East China Sea and high salinity water of Kuroshio origin (Chang et al. 2004).The northward branch along the east coast of Korea has been called the East Korean Warm Current (Uda 1934).
Figure 2 shows the surface temperature and salinity overlaid with currents for the four survey periods from the hybrid coordinate ocean model (HYCOM, http://www.hycom.org/dataserver/glb-analysis).In spring (April 2006), the Tsushima Warm Current, which had salinity higher than 34.4,passed along the Japanese coast, and there was a distinct anticyclonic eddy with 10°C core temperature and 34.2 salinity in the center of the Ulleung Basin (Figs. 2a and e).In summer (August 2007), the Tsushima Warm Current transported warm (> 25°C) and less saline (< 33.2) water into the Ulleung Basin through both channels in the Korea/ Tsushima Strait (Figs. 2b and f).In autumn (October 2008), the Tsushima Warm Current became weaker than in summer, and an anticyclonic eddy also became weaker (Figs.2c  and g).In winter (February 2008), the surface current through the Korea/Tsushima Strait was the weakest, but an anticyclonic eddy was well developed compared to other seasons (Figs.2d and h).The surface temperature in the Ulleung Basin was warmer in winter than in spring.

Spatial and Seasonal Variations of Surface fCO 2
Surface fCO 2 is determined as a function of physical factors (temperature and salinity), chemical factors (dissolved inorganic carbon, DIC and total alkalinity, TA), and others (vertical mixing, lateral mixing, biological uptake, sea-air CO 2 exchange, etc.).fCO 2 is expressed approximately as where NfCO 2ave (T ave , S ave ) represents the mean temperatureand salinity-normalized fCO 2 , ΔfCO 2 (T obs -T ave ) the thermodynamic effect of SST change on fCO 2 , ΔfCO 2 (S obs -S ave ) the thermodynamic effect of salinity change on fCO 2 , and ΔNfCO 2 (T ave , S ave ) the change in the normalized fCO 2 due to the changes in non-thermodynamic effect of changes in DIC, TA, vertical mixing, lateral mixing, biological uptake, and sea-air CO 2 exchange, etc.The value of ΔNfCO 2 (T ave , For the normalization of SST and SSS, we used the fCO 2 -SST and fCO 2 -SSS relationships of Takahashi et al. (1993): (a @ using a seasonal mean of SST and SSS.The thermodynamic effect ratio for seasonal change (TER seasonal ) is the same as above using annual mean of SST and SSS instead of seasonal mean.The surface distributions of temperature, salinity, fCO 2 , and chlorophyll a for the four cruises are shown in Fig. 3.In spring (April 2006), surface measurements were conducted in a narrow area covering from 129.5 to 131.5°E along 37°N.Due to the narrow coverage, the sea surface temperature (SST) and salinity (SSS) were confined to within a narrow range from 9.4 to 11.7°C and from 34.2 to 34.5, respectively (Figs. 3a and b).However, surface fCO 2 showed large variations, ranging from 260 to 356 μatm, which were considerably lower than atmospheric CO 2 (376.6 μatm).The ΔfCO 2 (T obs -T ave ), ΔfCO 2 (S obs -S ave ), ΔNfCO 2 (T ave , S ave ), TER spatial , and TER seasonal for the four cruises are shown in Fig. 4. ΔfCO 2 (T obs -T ave ) and ΔfCO 2 (S obs -S ave ) were in small ranges, from -17 to 13 μatm and from -2 to 1 μatm,  respectively, whereas ΔNfCO 2 (T ave , S ave ) was a large range, from -57 to 59 μatm (Figs.4a, b and c).Thus, TER spatial was as low as 0.22 ± 0.20, implying that SST and SSS were not the major factors in controlling the spatial distribution of surface fCO 2 in spring (Fig. 4d).Among non-thermodynamic factors controlling surface fCO 2 , vertical and lateral mixing, and sea-air CO 2 exchange will be discussed in section 3.3.Lower surface fCO 2 (less than 300 μatm) was observed in the area where chlorophyll a concentrations in the surface waters were relatively high (Figs.3c and d). Figure 5 shows the relationship between ΔNfCO 2 (T ave , S ave ) and surface chlorophyll a for the four surveys.The changes of normalized fCO 2 by non-thermodynamic factors showed a strong negative correlation with chlorophyll a only in spring (r 2 = 0.75).It suggested that spatial distribution of surface fCO 2 is largely influenced by biological activities in spring.
In summer (August 2007), SST showed a wide range from 21.4 to 26.2°C, and SSS varied from 32.8 to 33.8 (Figs.3a and b).Surface fCO 2 had a wide range from 316 to 409 μatm, the highest among the four seasons (Fig. 3c).Lower surface fCO 2 was observed in the central part of the study area where SSS was also relatively low (< 33.0).Despite that a variation of ΔfCO 2 (T obs -T ave ) became larger than that in spring, ranging from -36 to 38 μatm, spatial distribution of surface fCO 2 was much more similar to ΔfCO 2 (S obs -S ave ) (Figs. 4a and b).Spatial distribution of ΔNfCO 2 (T ave , S ave ) corresponded with that of ΔfCO 2 (S obs -S ave ) (Figs. 4b  and c).Moreover, ΔNfCO 2 (T ave , S ave ) showed a larger range from -48 to 62 μatm than both ΔfCO 2 (T obs -T ave ), from -36 to 38 μatm and ΔfCO 2 (S obs -S ave ), from -5 to 5 μatm.Thus, TER spatial was 0.45 ± 0.18, representing SST and SSS were not the primary factors to determine the spatial distribution of surface fCO 2 in summer (Fig. 4d).The Tsushima Warm Current became less saline in summer because it included low salinity shelf water from the East China Sea.Furthermore, the Changjiang Diluted Water could reach the East Sea through the Korea Strait in summer (Chen et al. 2003).Wang and Chen (1996) reported that the normalized alkalinity NTA TA S 35 # = ^h and salinity had a linear relationship with NTA shooting up (up to 2480 μmol kg -1 ; normal range of surface NTA in the East China Sea and East Sea was 2320 ~ 2360 μmol kg -1 ) at lower salinity by riverine alkalinity input in the shelf area of the East China Sea.High NTA and low DIC of surface water might be the major non-thermodynamic factors to diminish the surface fCO 2 in summer.The central part of the study area was undersaturated with respect to atmospheric CO 2 (371.0 μatm), whereas oversaturation was observed in coastal regions and in the far east of the study area.Oversaturation was found only in summer.Choi et al. (2011) measured surface fCO 2 in the Ulleung Basin of the East Sea on July 2005 and reported that the western and eastern parts of the Ulleung Basin were oversaturated with respect to atmospheric CO 2 , while the central part was undersaturated.They suggested that the undersaturation resulted from low SSS and high biological activity.In this study, relatively high surface chlorophyll a concentrations were also observed in the central region (Fig. 3d).A rough anti-correlation showed between ΔNfCO 2 (T ave , S ave ) and surface chlorophyll a (Fig. 5), indicating that biological activity slightly influenced the spatial variability of surface fCO 2 in summer.
In autumn (October 2008), SST ranged from 20.4 to 23.8°C, somewhat lower than in summer, and SSS varied from 32.3 to 33.6, rather similar to that in summer (Figs.3a  and b).Less saline waters were observed in the coastal regions; these waters were probably associated with the East Korea Warm Current that branches from the Tsushima Warm Current (Chang et al. 2004).Surface fCO 2 ranged from 298 to 355 μatm, a somewhat lower variation than in summer (Fig. 3c).In autumn, the study area was fairly undersaturated with respect to atmospheric CO 2 (376.6 μatm).Lower surface fCO 2 was observed in the central part of the study area, which was characterized by moderate SST, SSS, and surface chlorophyll a. ΔNfCO 2 (T ave , S ave ) showed a smallest range from -20 to 40 μatm among 4 seasonal observations (Fig. 4c).TER spatial was 0.58 ± 0.16, which was the highest value (Fig. 4d).It meant that SST and SSS were the major factors to control the surface fCO 2 in autumn.Distributions of ΔfCO 2 (T obs -T ave ) and ΔfCO 2 (S obs -S ave ) were almost opposite in phase, thus the variations of surface fCO 2 by SST and SSS changes were canceled out (Figs. 4a and b).Due to both a compensation of surface fCO 2 variation by SST and SSS changes, and small amount of non-thermodynamic effect on surface fCO 2 , surface fCO 2 in the central part of the study area was low.Surface fCO 2 was relatively high in the coastal areas, where SSS was relatively low and surface chlorophyll a concentrations were high (Figs.3b, c and d).In summer, low SSS and high chlorophyll a led to a decrease in surface fCO 2 , but these two factors were less important for controlling surface fCO 2 in autumn.ΔNfCO 2 (T ave , S ave ) was high in the coastal area where SSS was low (Figs.3b  and 4d).In the coastal areas where the water̓s depth was shallow, vertical mixing actively occurred in autumn when the surface stratification was weakened by the decrease in SST.In this study, the mixed layer depth increased from 10 m in summer to 25 m in autumn.It was deeper in the coastal areas than offshore (Fig. 6).Thus, the high surface fCO 2 in coastal areas was ascribed to vertical mixing, which brought CO 2 -rich subsurface waters to the surface.Shim et al. (2007) suggested that the high surface fCO 2 in the northern East China Sea in autumn was the result of vertical mixing with deep waters rich in CO 2 .
In winter (February 2008), the SST ranged from 10.1 to 14.4°C, about 10°C lower than in autumn, and SSS varied from 34.0 to 34.4, somewhat higher than in autumn (Figs.3a and b).Surface fCO 2 in winter ranged from 303 to 371 μatm, quite similar to that in autumn, despite the much lower SST (Fig. 3c).ΔfCO 2 (S obs -S ave ) varied in a small range (Fig. 4b).ΔfCO 2 (T obs -T ave ) and ΔNfCO 2 (T ave , S ave ) showed reversed distributions (Figs.4a and c).Due to the anti-correlation between ΔfCO 2 (T obs -T ave ) and ΔNfCO 2 (T ave , S ave ), surface fCO 2 range was similar with that in au-tumn in spite of the much lower SST.TER spatial in winter was 0.45 ± 0.18 (Fig. 4d), showing SST and SSS were not the primary factors to control the spatial distribution of surface fCO 2 like in summer.In winter, the surface mixed layer was deeper than 100 m in the study area (Fig. 6), implying active vertical mixing within the upper 100 m.Active mixing might have led to an increase in surface fCO 2 , which could have offset the decrease due to lower SST.Higher surface fCO 2 was observed in the northwestern part of the study region, where the surface mixed layer was deeper than in other areas and surface chlorophyll a concentrations were lowest.Highest ΔNfCO 2 (T ave , S ave ) in the northwestern part implied strong vertical mixing and/or weak biological activity.Lower surface fCO 2 was found at the southern and eastern parts of the study area, where surface chlorophyll a concentrations were relatively high.ΔNfCO 2 (T ave , S ave ) in the southern and eastern parts were negative, representing relatively high biological activity.ΔNfCO 2 (T ave , S ave ) was vaguely correlated with surface chlorophyll a (Fig. 5).In winter, surface fCO 2 was influenced, to some degree, by biological activity.

Factors Influencing the Seasonal Variability of Surface fCO 2
During the four seasonal surveys, the spatial mean SST showed a large variation of 13.2°C, while the spatial mean SSS showed a variation of 1.32 (Table 1).To validate the major influences of SST and SSS on surface fCO 2 , we plotted the TER seasonal for the four seasonal surveys (Fig. 4e).TER seasonal were 0.50 ± 0.09, 0.79 ± 0.08, 0.68 ± 0.05, and  0.40 ± 0.02 in the spring, summer, autumn, and winter, respectively.SST and SSS played major roles to control surface fCO 2 in autumn and especially in summer.In the northern reaches of the South China Sea, however, the seasonal variations of surface fCO 2 were mainly influenced by the seasonal variations of SST (Zhai et al. 2005).In the northern East China Sea, where the Kuroshio Current passed through, the seasonal variations in surface fCO 2 were affected by the seasonal changes in SST (Shim et al. 2007).
The measured surface fCO 2 values in spring and winter were not significantly different from those in summer and autumn, despite the large differences in SST (Table 1).Considering only the temperature effect of 4.23% °C-1 (Takahashi et al. 1993), the surface fCO 2 would show a difference of about 200 μatm between winter and summer.However, the surface fCO 2 varied seasonally by 62 μatm (Table 1).The small seasonal variability of surface fCO 2 resulted from several processes, such as vertical and lateral mixing, biological activity, and sea-air CO 2 exchange (Ishii et al. 2001;Ríos et al. 2005;Shim et al. 2006).We plotted the relationships among ΔNfCO 2 (T ave , S ave ), SST, and SSS obtained during the four seasonal surveys (Fig. 7).A good inverse relationship between ΔNfCO 2 (T ave , S ave ) and SST in winter represented non-thermodynamic increment of surface fCO 2 by vertical mixing.
To elucidate the effects of vertical mixing on the surface fCO 2 , the degree of stratification in the water column was calculated using the potential energy anomaly (PEA; Simpson et al. 1977;Shim et al. 2007).The low PEA indicated that the water column was unstable and well mixed.PEA was an order of magnitude higher in summer and autumn than in winter and spring (Table 1), indicating that the water column was unstable and well mixed in winter and spring.Low TER seasonal in winter and spring were consistent with low PEA.In winter and spring, therefore, the active vertical mixing brought CO 2 -rich subsurface waters to the surface, and thus caused the high surface fCO 2 .Shim et al. (2007) explained the high surface fCO 2 observed in the East China Sea during spring and autumn by vertical mixing with CO 2 -rich water masses.Consequently, the small seasonal variability of surface fCO 2 was ascribed to the high surface fCO 2 due to active vertical mixing in winter and spring.
The Tsushima Warm Current entered the East Sea through the Korea/Tsushima Strait, transporting warm and salty water into the study area (Chang et al. 2004).The Tsushima Warm Current branched from the Kuroshio Current and passed through the East China Sea before entering the study area.Kim et al. (2012) measured the surface fCO 2 in the northern East China Sea during four seasons; the spatial mean fCO 2 values were 311 ± 31 μatm in spring, 309 ± 53 μatm in summer, 376 ± 37 μatm in autumn, and 335 ± 17 μatm in winter.To identify the effects of lateral advection on the surface fCO 2 , the spatial mean surface fCO 2 values measured in the study area were compared with those in the East China Sea during four seasons (Table 1).In winter and spring, the surface fCO 2 values measured in the study area were rather similar to those in the East China Sea.However, they were somewhat higher in summer and lower in autumn than those in the East China Sea.In summer, the Tsushima Warm Current Water with lower surface fCO 2 was transported into the study area.A relationship between ΔNfCO 2 (T ave , S ave ) and SSS in summer represented decline of non-thermodynamic changes on surface fCO 2 by intrusion of low-salinity and low-fCO 2 waters into the study area.Therefore, the small seasonal variability of surface fCO 2 resulted from the lateral transport of water masses with lower surface fCO 2 in summer.
In Fig. 5, the relationship was random, indicating that the seasonal variation of the surface fCO 2 was not affected by biological activities.The primary production estimated in the study area also showed little seasonal variation (Noh, personal communication).Therefore, the small seasonal variability of surface fCO 2 was not related to the biological activity.
Surface fCO 2 is also influenced by the sea-air CO 2 exchange.The changes in surface fCO 2 due to sea-air CO 2 exchange were quantitatively estimated from the mean seasonal mixed layer depths (70, 25, 50, and 92 m in spring, summer, autumn, and winter, respectively), surface mean DIC (2100 μmole kg -1 ), and the Revelle factor (10).Values of 19.3, -1.6, 10.3, and 20.2 μatm were found for spring, summer, autumn, and winter, respectively.The increase in surface fCO 2 due to sea-air CO 2 exchange was largest in winter, the season with the highest CO 2 influx (Table 2).In summer, however, the surface fCO 2 decreased slightly by CO 2 outflux.Therefore, the small seasonal variability of surface fCO 2 was influenced, to some extent, by the sea-air CO 2 exchange.

Sea-Air CO 2 Flux
Table 2 shows the averaged sea-air differences of CO 2 fugacity (ΔfCO 2 ), wind speeds, and calculated sea-air CO 2 fluxes for the four seasons.The CO 2 fluxes had large seasonal variation.The Ulleung Basin of the East Sea was a sink of atmospheric CO 2 in spring, autumn, and winter, but a small source of CO 2 to the atmosphere in summer.
In spring, the CO 2 influx (negative sign) was calculated to be 10.4 ± 5.43 mmol m -2 day -1 .The lowest ΔfCO 2 was observed in spring among the four seasons, probably due to the spring phytoplankton bloom.Thus, the large CO 2 influx in spring resulted from high biological activities.The calculated CO 2 influx was somewhat higher than that (5.9 mmol m -2 day -1 ) calculated in the southern part of the East Sea during April (Oh 1998).
In summer, the Ulleung Basin of the East Sea acted as a source of CO 2 to the atmosphere, with CO 2 flux of 0.26 ± 2.29 mmol m -2 day -1 , which was almost the same as that (0.33 ± 2.48 mmol m -2 day -1 ) estimated in July 2005 (Choi et al. 2011).The CO 2 outflux was also rather similar to that (1.7 mmol m -2 day -1 ) calculated for the southern part of the East Sea in August (Oh 1998).
In autumn, the CO 2 influx was calculated to be 3.83 ± 0.70 mmol m -2 day -1 .The autumn ΔfCO 2 was similar to that in spring, but the autumn CO 2 influx was less than half that in spring due to the low wind speed (Table 2).The CO 2 influx was quite similar to that (2.9 mmol m -2 day -1 ) calculated in the southern East Sea in October (Oh 1998).
The largest CO 2 influx (13.3 ± 3.62 mmol m -2 day -1 ) was estimated in winter (Table 2).This high CO 2 influx was mainly attributable to high wind speeds in winter, because Table 2. Sea-air differences of CO 2 fugacity (ΔfCO 2 ), wind speeds, and sea-air CO 2 flux in the Ulleung Basin of the East Sea during the four seasonal observations.a Mean ΔfCO 2 along the cruise tracks expressed as the mean ± standard deviation (S.D.).b Mean wind speed of the study area (35 ~ 37.5°N, 129 ~ 132°E) from QuikSCAT satellite data during each observation period, expressed as the mean ± S.D. c Mean sea-air CO 2 fluxes based on the transfer coefficient of Wanninkhof (1992), expressed as the mean ± S.D. Positive values represent CO 2 emission from the sea to the atmosphere, while negative values represent CO 2 absorption from the atmosphere to the sea.  the winter ΔfCO 2 was higher than that in spring or autumn (Table 2).The winter CO 2 influx was somewhat lower than that (17.4 mmol m -2 day -1 ) calculated for the southern part of the East Sea in February (Oh 1998).
The annual integrated sea-air CO 2 flux in the Ulleung Basin of the East Sea was -2.47 ± 1.26 mol m -2 yr -1 (Table 2), quite similar to the previous estimate (-2.2 mol m -2 yr -1 ) for the southern East Sea (Oh 1998).The annual CO 2 uptake rate in this study area was considerably larger than the estimate for worldwide continental shelves (-1.1 mol m -2 yr -1 ; Chen and Borges 2009) and the global mean (-0.51 mol m -2 yr -1 ; Takahashi et al. 2009).Kim et al. (2012) reported that the annual sea-air CO 2 flux in the northern East China Sea was -2.2 ± 2.1 mol m -2 yr -1 , which is comparable to our result.Therefore, the Ulleung Basin of the East Sea, like the East China Sea, acts as a strong sink of atmospheric CO 2 .

CONCLUSIONS
Observations from four seasonal cruises showed that the Ulleung Basin of the East Sea acts as a strong sink for atmospheric CO 2 .The sea-air CO 2 flux displayed large seasonal variation, with CO 2 emitted to the atmosphere in summer and absorbed from the atmosphere in other seasons.This finding is consistent with modeling results showing the East Sea emitting CO 2 into the atmosphere from June to September and absorbing CO 2 from October through May (Oh 1998;Kang 1999).In the Ulleung Basin of the East Sea, the seasonal variation of surface fCO 2 could not be explained solely by seasonal changes in SST and SSS.Considering only the temperature effect of 4.23% °C-1 , the surface fCO 2 would show a difference of about 200 μatm between winter and summer, but the surface fCO 2 varied only by 62 μatm.The small seasonal variability of surface fCO 2 was attributed to the high surface fCO 2 due to the active vertical mixing in winter, the lateral transport of water masses with the lower surface fCO 2 in summer, and the sea-air CO 2 exchange.The Ulleung Basin of the East Sea adsorbed atmospheric CO 2 at an annual rate of 2.47 ± 1.26 mol m -2 yr -1 , which was comparable with the previous model result (-2.2 mol m -2 yr -1 ) for the southern East Sea (Oh 1998).The annually integrated CO 2 flux for worldwide continental shelves was -1.1 mol m -2 yr -1 (Chen and Borges 2009), which was less than half the CO 2 influx estimated for the Ulleung Basin of the East Sea.Therefore, the Ulleung Basin of the East Sea acts as a strong sink for atmospheric CO 2 compared to other continental shelves.

Fig. 1 .
Fig. 1.Study area and locations of sampling stations in the East/Japan Sea.TWC indicates the Tsushima Warm Current, EKWC East Korea Warm Current, and NKCC North Korea Cold Current.
Fig. 4. Surface distribution of ΔfCO 2 (T obs -T ave ) (a), ΔfCO 2 (S obs -S ave ) (b), ΔNfCO 2 (T ave , S ave ) (c), TER spatial (d), and TER seasonal (e) in April 2006, August 2007, October 2008, and February 2008.TER spatial (thermodynamic effect ratio for spatial distribution) of each season is a ratio of sheer fCO 2 change by thermodynamic effect of SST and SSS variation to total fCO 2 change by thermodynamic and non-thermodynamic effect CO f Tobs 2 D 6 CO CO CO N CO , T f S S f T T f S S f T S ave o bs ave obs ave o bs ave a ve ave 2 2 2 2D D D D -+ --+ -+ ^^^ĥ h h h h h 6 @@ using a seasonal mean of SST and SSS.TER seasonal (thermodynamic effect ratio for seasonal change) is the same as above using annual mean of SST and SSS.The figures in (d) and (e) represent the mean ± standard deviation (S.D.).

Fig. 5 .
Fig. 5. Relationship between ΔNfCO 2 (T ave , S ave ) and surface chlorophyll a for the four seasonal surveys.
94. Thermodynamic effect ratio for spatial distribution (TER spatial ) of each season is a ratio of sheer fCO 2 change by a thermodynamic effect of SST and SSS variations to total fCO 2 change by thermodynamic and non-thermodynamic

Table 1 .
Seasonal surface water properties of the study area.