Interannual variation of the southern limit in the Yellow Sea Bottom Cold Water and its causes

    Interannual variation of the southern limit in the Yellow Sea Bottom Cold Water and its causes Hee-Won Yang, Yang-Ki Cho, Gwang-Ho Seo, Sung Hyup You, JangWon Seo PII: DOI: Reference:

S0924-7963(14)00127-4 doi: 10.1016/j.jmarsys.2014.05.007 MARSYS 2547

To appear in:

Journal of Marine Systems

Received date: Revised date: Accepted date:

21 November 2013 6 May 2014 13 May 2014

Please cite this article as: Yang, Hee-Won, Cho, Yang-Ki, Seo, Gwang-Ho, You, Sung Hyup, Seo, Jang-Won, Interannual variation of the southern limit in the Yellow Sea Bottom Cold Water and its causes, Journal of Marine Systems (2014), doi: 10.1016/j.jmarsys.2014.05.007

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Interannual variation of the southern limit in the Yellow

SC R

IP

T

Sea Bottom Cold Water and its causes

Hee-Won Yang1, 2, Yang-Ki Cho1, Gwang-Ho Seo1, Sung Hyup You3, Jang-

MA

NU

Won Seo3

School of Earth and Environmental Sciences/Research Institute of

TE

D

Oceanography, Seoul National University, Seoul 151-742, Korea1

Seoul, Korea2

CE P

Korea research institute of climate change countermeasure strategies,

Marine Meteorology Division, Observation and Infrastructure Bureau,

AC

Korea Meteorological Agency, Seoul 156-720, Korea3

Corresponding Author: School of Earth and Environmental Sciences / Research Institute of Oceanography, Seoul National University, Seoul 151-742, Republic of Korea. Tel: +82 2 880 0364 / fax: +82 2 880 6749. E-mail address: [email protected] (Gwang-Ho Seo). 1

ACCEPTED MANUSCRIPT Abstract The Yellow Sea Bottom Cold Water (YSBCW) occupies a wide region below the Yellow

T

Sea (YS) thermocline in summer. The southern limit of the YSBCW shows pronounced

IP

interannual variability. A regional ocean model with realistic forcing was used to identify the

SC R

structure of the YSBCW and to investigate the causes of its interannual variability from 1981 to 2010. Sea surface temperature (SST) in winter is strongly correlated with the southern

NU

limit of the YSBCW in summer. The correlation coefficient between the August southern limit and the February SST is -0.884. This result suggests that cold SST is associated with the

MA

increased southern limit in the following summer. Linear regression suggests that the southern limit increases by about 55 km when the SST in February decreases by 1 °C. The

D

southern limits are more extended to the south in August than in June in some years despite

TE

surface heating. The difference in southern limits between June and August is positively

CE P

correlated with the summer southerly wind stress with a correlation coefficient of 0.529. The contribution of SST in winter on the southern limit of the YSBCW in summer is larger than the wind stress in summer. The SST in winter is mainly determined by the air temperature

AC

and wind speed in winter. The other factor affecting winter SST is the previous year’s bottom water temperature. The winter SST is significantly correlated with the bottom water temperature in previous year. The southern limit of the YSBCW in the observed data in the limited area has relatively weak correlation with the winter SST and summer southerly wind stress possibly due to observation error and uncertainty of the reanalysis wind.

Keywords: Yellow Sea, bottom cold water, interannual variation, wind stress, surface temperature, air temperature, models, correlation 2

ACCEPTED MANUSCRIPT 1. Introduction The Yellow Sea (YS) is a shallow and semi-enclosed marginal sea located between

T

Korea and China (Figure 1). Adjoining rivers bring sediments rich in nutrients to the YS

IP

(Park et al., 2011). High primary productivity, abundant marine life, and dominating

SC R

monsoon conditions demonstrate the importance of the YS for the marine ecosystem and geography (Teng et al., 2005). Shallow depths of less than 100 m and pronounced seasonal

NU

changes in the YS cause large variation of the water temperature.

Strong northerly wind drives deep convection in winter, resulting in vertically well-

MA

mixed waters from the surface to the bottom during winter (Figure 2). This vertically homogeneous temperature structure can be observed until spring. On the contrary, increased

D

solar radiation drives strong stratification in summer. Because this strong stratification

TE

prevents vertical mixing between the sea surface and the bottom in summer, cold water that

CE P

forms in winter maintains its temperature beneath the stratified thermocline. This cold water occupies the central region of the YS at depths deeper than 50 m (T < 10 °C) (Youn et al., 1991; Zhang et al., 2008) and is known as the Yellow Sea Bottom Cold Water (YSBCW)

2011).

AC

(Guan, 1963; Hur et al., 1999; Kim et al., 1999; Hu et al., 2004; Zhang et al., 2008; Park et al.,

Previous researches have explored several aspects of this unique water mass, which includes studies of interannual and long-term changes in sea water temperature in the YS (Hu et al., 2004; Zhang et al., 2008; Wei et al., 2010). Winter air temperature may affect variability of the YSBCW temperature (Guan, 1963; Park et al., 2011). Park et al. (2011) reported that changes in atmosphere forcing are strongly correlated with variability of the YSBCW temperature. From the historical studies, the YSBCW temperature was affected by 3

ACCEPTED MANUSCRIPT air temperature. This theory has been widely accepted. The bottom temperature in the YS is higher in winter than in summer due to the

T

intrusion of the Yellow Sea Warm Current (YSWC) in winter and southward flow of the

IP

YSBCW in summer (Xu et al., 2003).

SC R

The northward flow of the YSWC is driven by strong northerly wind in winter (Moon et al., 2009). The YSWC transports warm and saline water into the YS (Ma et al., 2006). The

NU

winter temperature may be affected by the intrusion of the YSWC (Wei et al., 2010). Southward migration of the YSBCW in summer has been reported recently. Zhang et al.

MA

(2008) used short-term observed data to show that the cold bottom water moves from the north to the central area of the YS during the summer, and Jacobs et al. (2000) used a

D

numerical model to reveal that the southeasterly summer monsoon generates a southward

TE

flow of the bottom layer. The seasonal migration of the YSBCW has a large impact on

CE P

primary products and marine habitats due to dramatic changes in water temperature of the southern YS (Wang et al., 2003). Despite many previous studies, long-term variation of the southern limit of the YSBCW

AC

is still unclear. And its driving mechanisms are poorly understood. Therefore, the purpose of this study is to investigate the interannual variations of the southern limit of the YSBCW and its causes by using long-term numerical model results based on realistic data.

2. Data and numerical model The Regional Ocean Modeling System (ROMS), which is a three-dimensional ocean circulation model, was used for this study. The model domain is from 18.5°N to 48.5°N and from 117.5°E to 154.5°E and includes the YS, the East China Sea, the Japan/East Sea, and the 4

ACCEPTED MANUSCRIPT Northwestern Pacific (Figure 1). The horizontal grid has a resolution of 0.1° with 20 vertical levels. The open boundary data of the model were provided from a regional northwest Pacific

T

(NWP) model (Cho et al., 2009) nested within a global model known as Estimation of the

IP

Circulation and Climate of the Ocean (ECCO; www.ecco-group.org). The initial values for

SC R

temperature, salinity, velocity, and sea surface height were obtained from the NWP model (Cho et al., 2009). The model was run from January 1981 to December 2010. The monthly

NU

mean values from the European Center for Medium-range Weather Forecasting (ECMWF) reanalysis data were used as the surface forcing data. Bulk-flux formulae (Fairall et al., 1996)

MA

were adapted. Tidal forcing with ten major tidal components was applied (Egbert and Erofeeva, 2002). Vertical mixing was calculated by using the Mellor–Yamada turbulence

D

closure scheme (Mellor and Yamada, 1974). The horizontal viscosity coefficient was 300

TE

m2/s. Further details on the model have been reported by Cho et al. (2009, 2013).

CE P

Temperatures observed by the National Fisheries Research and Development Institute (NFRDI) during the last 30 years (1981–2010) were analyzed. Temperature data has been routinely observed bimonthly at the standard depths in the sea near the Korean Peninsula.

AC

Most offshore stations of each observed line corresponding to the deep trough in the YS were selected to study the variation of the southern limit of the YSBCW in August. These stations are shown as filled red circles in Figure 1. In the Figure 1, the number at each station represents the line observed by NFRDI. We analyzed the temperature selected from model grid corresponding to the observation. From 34°N to 37°N and from 125°E to 127°E, spatially averaged SSTs were used to examine the correlation between the southern limit of the YSBCW and the winter SST. To relate the YSBCW interannual variations with changes in atmospheric conditions, we 5

ACCEPTED MANUSCRIPT used monthly mean air temperature and wind stress from the ECMWF reanalysis data obtained during the last 30 years, from 1981 to 2010. For winter air temperature, we averaged

T

the December, January, and February air temperatures over a domain bounded by 33°N, 42°N,

IP

117°E, and 127°E. Winter wind speed and summer wind stress were calculated by using the

SC R

V (north–south) component wind speed spatially averaged from 30°N to 42°N and from 117°E to 127°E in December, January, and February and in June, July, and August,

NU

respectively .

To identify the previous year’s oceanic conditions that affect YSBCW variability, the

MA

bottom temperature, i.e., the spatial mean temperature at a depth of 75 m in October was

3. Result

CE P

3.1 Model validation

TE

D

analyzed.

Numerical models can indicate the entire distribution of the YSBCW and effectively correlate its dynamical interactions with the causes (Riedlinger et al., 2000). Monthly mean

AC

model temperature can effectively define the distribution of the YSBCW in summer, represented in Figure 1 by the shaded area, whereas the observed temperatures are obtained from a limited space, represented in the figure by filled circles. In order to overcome the limitations associated with the sparse spatial and temporal samplings in the observation data, numerical model results were analyzed. Meridional sections of observed temperature along the outermost observation stations averaged over 30 years showed noticeable seasonal variation (Figure 2, left). In particular, strong stratification was observed in August, and well-mixed structures appeared from winter 6

ACCEPTED MANUSCRIPT to early spring. Simulated temperature sections (Figure 2, right) along the same stations as those in the observation showed the strongest stratification in August and well-mixed

T

temperatures from the surface to the bottom in winter and spring. In other to explore the

IP

interannual variability of the location of the YSBCW, the southern limit of the YSBCW is

SC R

defined as the distance of 10 °C isothermal line from the observation line 307 by National Fisheries Research and Development Institute (NFRDI) along the sea floor.

NU

SSTs obtained from the National Oceanic and Atmospheric Administration/Advanced Very High Resolution Radiometer (NOAA/AVHRR) satellite and from model results in each

MA

season averaged over 29 years from 1982 to 2010 are compared in Figure 3. Satellite data in 1981 was not available. Both SSTs clearly indicated the intrusion of the warm Yellow Sea

D

Warm Current in February. Although the seasonal temperature amplitude of the model along

CE P

were effectively captured.

TE

the coast was larger than that of the satellite, the overall changes in seasonal temperature

3.2 Correlation with SST of the previous winter

AC

The model results showed large interannual variation of the YSBCW southern limit during the last 30 years (Figure 4). The YSBCW southern limit retreats to the north or extends to the south from year to year. Maximum and minimum distances were detected at 354 km in 2008 and at 196 km in 1999, respectively. Because the YSBCW is formed during the previous winter, the bottom temperature in summer is closely related to the SST of the previous winter (Kang et al., 1987). To explore the relationship between the southern limit variability and the oceanic condition of the previous winter, the correlation between February SST and the southern limit 7

ACCEPTED MANUSCRIPT in August was calculated. Figure 4 shows the southern limit of the model results and SSTs in February from 1981 to 2010, respectively. The interannual variations of two factors show

T

good correlation. The correlation coefficient between the August southern limit and the

IP

February SST was -0.884, (Significant at the 95% confidence level, r > 0.340). This negative

SC R

correlation indicates that higher SST in February cause shrinkage of the YSBCW. Linear regression suggests that the southern limit is reduced by about 55 km when the SST in

NU

February rises by 1 °C.

The area of the YSBCW in August was calculated by using model results to understand

MA

its relationship with SST in winter. The area showed large interannual variation from about 3 × 104 km2 to 12 × 104 km2. The area of the YSBCW in August had a strong correlation (0.895)

D

with the SSTs in winter (Figure 5), which suggests that cold SST produced the large volume

TE

of the YSBCW. Such increased volume should result in YSBCW expansion further to the

CE P

south of its southern limit in August.

3.3 Correlation with southerly wind stress in summer

AC

As first suggested by Park (1986), the seasonal evolution of the southern limit after its formation is an important factor controlling the variability of the location of the YSBCW (Pang et al., 2003; Zhang et al., 2008).. In order to reveal the interannual variations of the cold-water displacement in summer, the southern limit in August was compared with that in June (Figure 6a). During a 30-year period, the August southern limit was located further south than the June southern limit for seven years, which indicates that the southern limit in August moves to the south after June. The sea surface in August is generally warmer than that in June, and 8

ACCEPTED MANUSCRIPT the ocean becomes more stratified. Therefore, it is difficult to explain abnormal expansion of the southern limit in August by using the heat transport between the sea surface and the

T

bottom layer. Previous studies suggest that summer wind plays a role in the further expansion

IP

of the YSBCW (Zhang et al., 2008; Jacob et al., 2000). Model results by Jacobs et al. (2000)

SC R

suggest that southerly wind in summer generates southward flow at the bottom. The sea level increases in the northern areas of the YS are attributed to the southeasterly wind stress. The

NU

higher surface waters drive a southward flow due to a pressure gradient at the bottom layer.

MA

In order to explore the relationship between the displacement of the southern limit and summer wind stress for the last 30 years, we calculated the differences between the southern

D

limits in August and June and estimated the correlation with southerly wind stress in summer

TE

(Figure 6b). The simulated difference between the June and August southern limit positively correlated with the summer southerly wind stress with a correlation coefficient of 0.529. This

CE P

result proves that the anomalously stronger southerly wind stress in summer tends to push the

AC

southern limit further south in August.

3.4 Reconstructed southern limit based on winter SST and summer wind In the previous sections, we showed that the SST in February and summer wind stress both affect the extent of the southern limit of the YSBCW. In order to quantify the contribution of the two factors on the change in the southern limit, we reconstructed the southern limit through multiple regression analysis by using the normalized SST (nSST) and wind stress (nWind). ---------------(1) This relation suggests that the SST in February has a greater impact than the summer 9

ACCEPTED MANUSCRIPT wind. The model southern limit and reconstructed southern limit are shown in Figure 7 by the solid and dashed lines, respectively. The correlation coefficient is 0.857, and the southern

4. Discussion 4.1. Factors affecting SST in February

SC R

IP

T

limit variability was effectively captured by these two factors.

NU

The correlation analysis between the southern limit in August and SST in February reveals the importance of winter seawater temperature in setting the southern limit of the

MA

YSBCW in the following summer. The SST may be changed significantly by the heat flux between the atmosphere and the sea (Hirose et al., 1999).

D

The air–sea heat exchange significantly affected the water masses on the shelf region of

TE

the YS. The air–sea interaction is described by the various physical relationship of the surface

CE P

atmosphere and the ocean. The surface heat flux response to SST is the most important factor in air–sea interaction. The net heat flux through the sea surface

can be expressed as

,

AC

where Qs is short-wave radiation flux, Qb is long-wave radiation flux, Qh is sensible heat flux, and Ql is latent heat flux. We analyzed the surface heat flux using the bulk method, which uses wind, humidity, pressure, air temperature, and SST as variables (Subrahmanyam et al., 2007). Air temperature and wind speed in winter, which are crucial to heat flux, are analyzed to understand their effects on the SST in interannual variation (Chu et al., 2005). Figure 8a shows the correlation between winter air temperature and February SST obtained from model. Simulated SSTs are positively correlated with air temperatures in winter. This result suggests that the YSBCW shrinks as a result of high air temperature and 10

ACCEPTED MANUSCRIPT SSTs, as shown in Figure 4. Air temperature may be associated with anomalous SSTs in winter. Through its influence on winter SST, winter air temperature is an important factor

T

controlling the location of the YSBCW.

IP

Previous studies identified two different roles of winter wind stress. Kim et al. (1999)

SC R

revealed that strong northerly winds introduce a large vertical contrast between air temperature and SST, which produces a large amount of heat loss. Conversely, the model

NU

results reported by Hsueh and Yuan (1997) showed that the increase in sea level induced by the northerly strong monsoon generates inflow of warm and salty water from the south,

MA

which may result in a decrease of the YSBCW in the following summer. Mask and O’Brien (1998) showed that the warm and salty water intrudes northward by northerly wind in winter.

D

To explore the role of winter winds, we plotted a scatter diagram between SSTs in February

TE

and winter wind speeds (Figure 8b). The correlation coefficient was 0.554 significant at the

CE P

95% confidence level; r > 0.340. This result suggests that the stronger wind speed may decrease SSTs through latent heat loss in February. As a result, wind speed and air temperature in winter play crucial roles in interannual variation in the southern limit of the

AC

YSBCW. Cold air appears usually with strong wind in winter; thus, wind speed is closely related to air temperature in winter (not shown). High air temperatures and weak wind in February are correlated with short southern limits (Figure 9a). However, the southern limit in 2007 did not follow this trend; it was peculiarly extended past the fitted line, although the air temperature was high and the wind speed was low in winter (Figure 9b). In addition, the summer southerly wind stress in 2007 at about 1.92 dyne/cm2 was weaker than the 30-year mean value of about 2.08 dyne/cm2. However, the SST in February 2007 of about 8.9 °C was lower than the 30-year mean SST of 11

ACCEPTED MANUSCRIPT about 9.1 °C. The long southern limit in 2007 was due to the low SST in winter. Therefore, factors other than winter air temperature and wind speed may affect SSTs in February.

T

We analyzed the previous October’s bottom temperature because the YSBCW remains

IP

beneath the thermocline until October (Figure 2). The area of the YSBCW in October (Figure

SC R

10c) was uniquely larger in 2006 than that averaged over the 30 years (Figure 10a). The southern limit in August 2007 (Figure 10d) was extended 38 km more than that

NU

averaged over the 30 years (Figure 10b) despite high air temperatures and weak wind speed during the previous winter. The bottom water in October 2006 was about 2 °C cooler than the

MA

30-year mean temperature. This result suggests that the heat storage of the bottom water in the previous year affects SSTs in winter and, in turn, the southern limit of the YSBCW.

D

In order to explore the effects of the previous year’s bottom water temperature on winter

TE

SSTs, the correlation between SSTs in February and bottom water temperature during the

CE P

previous October was calculated (Figure 11). The correlation was significant at the 95% confidence level by 0.633. This result reveals that the temperature in October affects SSTs in February. The temperature of the bottom water remaining from the previous year may be a

AC

contributing factor in determining the SSTs during vertical mixing in winter, as shown in Figure 2. Thus, the previous year’s bottom temperature is an important factor affecting the southern limit of the YSBCW.

4.2 Analysis of observed data and their limitations The southern limit of the observed data was calculated using temperature recorded at the stations along the lines selected in the model analysis. The maximum and minimum distances were 422 km in 1984 and 146 km in 1991, respectively (Figure 12). Although the details of 12

ACCEPTED MANUSCRIPT the variability differed quantitatively, the observed southern limit agreed reasonably well with the simulated variability.

T

Figure 13a shows the correlation between SSTs in February and the southern limit of the

IP

YSBCW in August determined on the basis of observation. The correlation coefficient was -

SC R

0.516, which is lower than the simulation-based estimate. This result is consistent with the model result, and the former result also indicates that increased SSTs are associated with the

NU

decreased southern limit in the following summer. We used linear regression of the observation data to determine that an increase in air temperature of 1 °C tends to reduce the

MA

southern limit by about 47 km, which is 8 km shorter than that determined by the model result.

D

The difference in the observed southern limit between June and August shows a

TE

significantly reduced correlation with the summer southerly wind stress (r = 0.121; Figure

CE P

13b). Observation error of water temperature and sparse observations in time and space may have resulted in this low correlation in interannual variation. Such low correlation may also be attributed to the bias of air temperature and wind of the reanalysis model result.

AC

Nevertheless, the qualitative agreement suggests that the extension of the August southern limit may be closely related to the summer winds.

5. Conclusion The southern limit of the YSBCW in summer shows pronounced interannual variability over the last 30 years. The model and observation results show that the maximum difference of the southern limit is more than 150 km. A numerical model experiment that used realistic forcing revealed that SSTs in February are strongly correlated with the southern limit of the 13

ACCEPTED MANUSCRIPT YSBCW in summer. The correlation coefficient between the southern limit in August and the SST in February was -0.884. Linear regression suggests that the southern limit extended

T

about 55 km when the SST in February decreased 1 °C. A large volume of the YSBCW by

IP

attributed to the cold SST caused the extension further south of its southern limit in August.

SC R

The southern limit of the YSBCW extended further south in August than in that June in several years, and this could be attributed to the southerly wind in summer. The southerly

NU

wind stress in summer had significant correlation (r = -0.529) with the difference of the southern limit of the YSBCW between June and August. This correlation suggests that the

MA

southern limit tends to extend in summer when the southerly winds are strong, which is consistent with the results of previous studies such that southerly wind in summer generates

TE

surface at the north.

D

southward flow at the bottom by the pressure gradient to the south due to the increase of sea

CE P

The southern limit of the YSBCW was successfully reconstructed by using the SSTs in February and southerly wind stress in summer recorded from 1981 to 2010. The correlation coefficient of the reconstructed southern limit with the simulated was 0.808, which implies

AC

that winter SST and summer wind are crucial factors in determining the locations of the southern limit each year. The contribution of SST in winter to the southern limit of the YSBCW in summer is larger than the wind stress in summer. Air temperature and wind speed are major factors in determining SST in winter. It is believed that the temperature of the YSBCW in the previous year is also a contributing factor. A significant correlation of 0.633 between the SSTs in February and the bottom water temperatures in the previous October suggests that the bottom temperature in the previous year contributes to determination of the SSTs in winter when the vertical mixing occurs. 14

ACCEPTED MANUSCRIPT Although we detected the same trend of the southern limit of the observed data according to the winter SST and summer southerly wind stress, the correlation coefficients of

T

the observed southern limit were lower than those derived from the model result. Observation

IP

error of the water temperature, sparse sampling, and uncertainty of the reanalysis wind may

SC R

have caused such weak correlation. Additional dynamical explanations on the variations of

MA

NU

the YSBCW in volume and location are essential for in future research.

D

Acknowledgment

TE

This research was supported by the Korea Meteorological Administration Research and

AC

CE P

Development Program under the grant CATER 2012-2080.

15

ACCEPTED MANUSCRIPT References

T

Cho, Y.-K., Seo, G.-H., Choi, B.-J., Kim, S., Kim, Y.-G., Youn, Y.-H., Dever, E.P.,

IP

2009. Connectivity among straits of the northwest Pacific marginal seas. J. Geophys.

SC R

Res. 114, C06018.

Cho, Y.-K., Seo, G. H., Kim, C. S., Choi, B.-J., Shaha, D. C., 2013. Role of wind

NU

stress in causing maximum transport through the Korea Strait in autumn. J. Mar. Syst. 115-116, 33-39.

MA

Chu, P., Yuchun, C. H. E. N., Kuninaka, A., 2005. Seasonal variability of the Yellow Sea/East China Sea surface fluxes and thermohaline structure. Adv. Atmos. Sci.

D

22(1), 1-20.

TE

Egbert, G. D., Erofeeva, S. Y., 2002. Efficient inverse modeling of barotropic ocean

CE P

tides. J. Atmos. Oceanic Technol. 19(2), 183–204. Fairall, C. W., Bradley, E. F., Rogers, D. P., Edson, J. B., Young, G. S., 1996. Bulk parameterization of air-sea fluxes for tropical ocean-global atmosphere Coupled-Ocean

AC

Atmosphere Response Experiment. J. Geophys. Res. 101, 3747–3764. Hirose, N., Lee, H. C., Yoon, J. H., 1999. Surface heat flux in the East China Sea

and the Yellow Sea. J. Phys. Oceanogr. 29(3), 401-417. Hsueh, Y., Yuan, D., 1997. A numerical study of currents, heat advection, and sealevel fluctuations in the Yellow Sea in winter 1986. J. Phys. Oceanogr. 27(11), 23132326. Hu, D. and Wang, Q., 2004. Interannual variability of the southern Yellow Sea cold water mass. Chin. J. Oceanol. Limnol. 22(3), 231-236. 16

ACCEPTED MANUSCRIPT Hur, H.B., Jacobs, G.A., Teague, W.J., 1999. Monthly variations of water masses in the Yellow and East China Seas. J. Oceanogr. 55(2), 171–18.

T

Jacobs, G. A., Hur, H. B., Riedlinger, S. K., 2000. Yellow and East China Seas

IP

response to winds and currents. J. Geophys. Res. 105 (C9), 21947-21.

SC R

Kang, Y.Q., Kim, H.-K., 1987. Relationships between the winter-time surface water temperature and the summer-time bottom water temperature in the West Sea of

NU

Korea. J. Oceanol. Soc. Korea. 22 (4), 228–235.

Kim, Y. S., Kimura, R., 1999. Estimation of the Heat Flux Exchange between the

MA

Air-Sea Interface over the Neighbouring Seas of Korean Peninsula. J. Korean Meteor. Soc. 35(4), 501-510.

D

Ma, J., Qiao, F., Xia, C.S., Kim, C.S., 2006. Effects of the Yellow Sea Warm

CE P

111, C11C04.

TE

Current on the winter temperature distribution in a numerical model. J. Geophys. Res.

Mask, A.C. and O’Brien, J.J., 1998. Wind-driven effects on the Yellow Sea Warm Current. J. Geophys. Res. 103(C13), 30, 713-30129.

AC

Mellor, G., Yamada, T., 1974. A hierarchy of turbulence closure models for planetary boundary layers. J. Atmos. Sci. 31, 791–1806. Park, S., Chu, P. C., Lee, J. H., 2011. Interannual-to-interdecadal variability of the Yellow Sea Cold Water Mass in 1967–2008: characteristics and seasonal forcings. J. Mar. Syst. 87(3), 177-193. Park, Y.H., 1986. Water characteristics and movements of the Yellow Sea Warm Current in summer. Prog. Oceanogr. 17(3), 243–254. Pang, I. C., Hong, C. S., Chang, K. i., Lee, J. C., Kim, J. T., 2003. Monthly 17

ACCEPTED MANUSCRIPT Variarion of Water Mass Distribution and Current in the Cheju Strait. J. Korean Soc. Oceanogr. 38(3), 87-100.

T

Riedlinger, S. K., Jacobs, G. A., 2000. Study of the dynamics of wind‐driven

IP

transports into the Yellow Sea during winter. J. Geophys. Res. 105(C12), 28695-28708.

SC R

Subrahamanyam, D. B., Ramachandran, R., Rani, S. I., Kumar, B. P., 2007. Air-sea interaction processes over the East Asian marginal seas surrounding the Korean

NU

peninsula. Ann. Geophys. 25(7), 1477-1486.

Teng, S.K., Yu, H., Tang, Y., Tong, L., Choi, C.I., Kang, D., Liu, H., Chun, Y.,

MA

Juliano, R.O., Rautalahti-Miettinen E., Daler, D., 2005. Global international waters assessment: Yellow Sea. GIWA Regional Assessment, 34 University of Kalmar,

D

Sweden.

TE

Wang, R., Zuo, T., Wang, K. E., 2003. The Yellow Sea cold bottom water—an

169-183.

CE P

oversummering site for Calanussinicus (Copepoda, Crustacea). J. Plankton Res. 25(2),

Wei, H., Shi, J., Lu, Y., Peng, Y., 2010. Interannual and long-term hydrographic

AC

changes in the Yellow Sea during 1977-1998. Deep Sea Res. II, 57, 1025-1034. Xu, D., Yuan, Y., Liu, Y., 2003. The baroclinic circulation structure of Yellow Sea Cold Water Mass. Sci. China Ser. D, 46(2), 117-126. Youn, Y.-H., Park, Y.-H., Bong, J.-H., 1991. Enlightenment of the Characteristics of the Yellow Sea Bottom Cold Water and Its Southward Extension. J. Korean Earth Sci. Soc. 12(1), 25-37. Zhang, S.W., Wang, Q.Y., Lu, Y., Cui, H., Yuan, Y. L., 2008. Observation of the seasonal evolution of the Yellow Sea Cold Water Mass in 1996–1998. Cont. Shelf Res. 18

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

28(3), 442–457.

19

ACCEPTED MANUSCRIPT Figure captions

T

Figure 1 Bathymetric map of the Yellow Sea (YS) and schematic diagrams of the Yellow Sea

IP

Bottom Cold Water (YSBCW; gray) determined on the basis of model results. Numbers

SC R

represent observation lines routinely observed by the National Fisheries Research and Development Institute. The insert box shows model area.

NU

Figure 2 Mean meridional section of temperature along the outermost observation stations of Figure 1. (left) Observation and (right) model results from 1981 to 2010.

MA

Figure 3 Comparison of sea surface temperature between (up) satellite observation and (down) model results in each season averaged from 1982 to 2010.

TE

D

Figure 4 Comparison of sea surface temperature in February with the southern limit of the Yellow Sea Bottom Cold Water (YSBCW) in August.

CE P

Figure 5 Correlation between sea surface temperature in February and the area of the Yellow Sea Bottom Cold Water (YSBCW) in August.

AC

Figure 6 (a) Southern limit of the Yellow Sea Bottom Cold Water (YSBCW) in June and August and (b) correlation between southerly wind stress in summer and the difference of the southern limit in June and August. Figure 7 Reconstructed southern limit of the Yellow Sea Bottom Cold Water (YSBCW) determined on the basis of sea surface temperature in February and southerly wind stress in summer. Figure 8 Correlation of sea surface temperature in February with (a) winter air temperature and (b) winter wind speed. 20

ACCEPTED MANUSCRIPT Figure 9 Relationships of the southern limit of the Yellow Sea Bottom Cold Water (YSBCW)

T

in August with (a) air temperature and (b) wind speed in winter.

IP

Figure 10 Horizontal temperature distributions in bottom layer of 75 m or deeper in (a)

SC R

October; (b) August, averaged over 30 years; (c) October 2006; and (d) August 2007. The black points on the figure represent the outermost observation station listed in Figure 1.

and sea surface temperature in February.

NU

Figure 11 Correlation between the bottom water temperature in October of the previous year

MA

Figure 12 Comparison of southern limits of the Yellow Sea Bottom Cold Water (YSBCW) between numerical model result and observation in August from 1981 to 2010.

TE

D

Figure 13 (a) Relationships of the observed southern limit of the Yellow Sea Bottom Cold Water (YSBCW) in August and sea surface temperature in February. (b) Correlation between

AC

June.

CE P

southerly wind stress in summer and the difference in observed southern limits in August and

21

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

TE

Figure 1 Bathymetric map of the Yellow Sea (YS) and schematic diagrams of the Yellow Sea

CE P

Bottom Cold Water (YSBCW; gray) determined on the basis of model results. Numbers represent observation lines routinely observed by the National Fisheries Research and

AC

Development Institute. The insert box shows model area.

22

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

Figure 2 Mean meridional section of temperature along the outermost observation stations of Figure 1. (left) Observation and (right) model results from 1981 to 2010.

23

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 3 Comparison of sea surface temperature between (up) satellite observation and (down) model results in each season averaged from 1982 to 2010. 24

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 4 Comparison of sea surface temperature in February with the southern limit of the

AC

CE P

TE

D

MA

Yellow Sea Bottom Cold Water (YSBCW) in August (a) and correlation (b).

25

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 5 Correlation between sea surface temperature in February and the area of the Yellow

AC

CE P

TE

D

Sea Bottom Cold Water (YSBCW) in August

26

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

MA

Figure 6 (a) Southern limit of the Yellow Sea Bottom Cold Water (YSBCW) in June and August and (b) correlation between southerly wind stress in summer and the difference of the

AC

CE P

TE

D

southern limit in June and August.

27

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 7 Reconstructed southern limit of the Yellow Sea Bottom Cold Water (YSBCW)

TE

D

determined on the basis of sea surface temperature in February and southerly wind stress in

AC

CE P

summer.

28

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

MA

Figure 8 Correlation of sea surface temperature in February with (a) winter air temperature

AC

CE P

TE

D

and (b) winter wind speed.

29

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 9 Relationships of the southern limit of the Yellow Sea Bottom Cold Water (YSBCW)

AC

CE P

TE

D

in August with (a) air temperature and (b) wind speed in winter.

30

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 10 Horizontal temperature distributions in bottom layer of 75 m or deeper in (a)

AC

October; (b) August, averaged over 30 years; (c) October 2006; and (d) August 2007. The black points on the figure represent the outermost observation station listed in Figure 1.

31

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

TE

Figure 11 Correlation between the bottom water temperature in October of the previous year

AC

CE P

and sea surface temperature in February

32

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 12 Comparison of southern limits of the Yellow Sea Bottom Cold Water (YSBCW)

AC

CE P

TE

D

between numerical model result and observation in August from 1981 to 2010.

33

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 13 (a) Relationships of the observed southern limit of the Yellow Sea Bottom Cold

D

Water (YSBCW) in August and sea surface temperature in February. (b) Correlation between

TE

southerly wind stress in summer and the difference in observed southern limits in August and

AC

CE P

June.

34

ACCEPTED MANUSCRIPT Highlights  The southern limit of the YSBCW shows a pronounced inter-annual variability.

T

 The variation is closely related to the SST in winter and the wind stress.

AC

CE P

TE

D

MA

NU

SC R

IP

 The SST in winter is mainly determined by the air temperature and wind speed.

35