Seasonal Variability of the Yellow Sea/East China Sea Surface Fluxes and Thermohaline Structure

ADVANCES IN ATMOSPHERIC SCIENCES, VOL. 22, NO. 1, 2005, 1–20

Seasonal Variability of the Yellow Sea/East China Sea Surface Fluxes and Thermohaline Structure

Peter CHU∗1, CHEN Yuchun2 (ŒS), and Akira KUNINAKA1 1Naval Ocean Analysis and Prediction Laboratory, Department of Oceanography Naval Postgraduate School, Monterey, California 2Institute of Cold and Arid Environment and Engineering, Chinese Academy of Sciences, Lanzhou

(Received 27 November 2003; revised 26 April 2004)

ABSTRACT We use the U.S. Navy’s Master Oceanographic Observation Data Set (MOODS) for the Yellow Sea/East China Sea (YES) to investigate the climatological water mass features and the seasonal and non-seasonal variabilities of the thermohaline structure, and use the Comprehensive Ocean-Atmosphere Data Set (COADS) from 1945 to 1989 to investigate the linkage between the fluxes (momentum, heat, and moisture) across the air-ocean interface and the formation of the water mass features. After examining the major current systems and considering the local bathymetry and water mass properties, we divide YES into five regions: East China Sea (ECS) shelf, Yellow Sea (YS) Basin, Cheju bifurcation (CB) zone, Taiwan Warm Current (TWC) region, Kuroshio Current (KC) region. The long term mean surface heat balance corresponds to a heat loss of 30 W m−2 in the ESC and CB regions, a heat loss of 65 W m−2 in the KC and TWC regions, and a heat gain of 15 W m−2 in the YS region. The surface freshwater balance is defined by precipitation minus evaporation. The annual water loss from the surface for the five subareas ranges from 1.8 to 4 cm month−1. The fresh water loss from the surface should be compensated for from the river run-off. The entire water column of the shelf region (ECS, YS, and CB) undergoes an evident seasonal thermal cycle with maximum values of temperature during summer and maximum mixed layer depths during winter. However, only the surface waters of the TWC and KC regions exhibit a seasonal thermal cycle. We also found two different relations between surface salinity and the Yangtze River run-off, namely, out-of-phase in the East China Sea shelf and in-phase in the Yellow Sea. This may confirm an earlier study that the summer fresh water discharge from the Yangtze River forms a relatively shallow, low salinity plume-like structure extending offshore on average towards the northeast.

Key words: Yellow Sea, East China Sea, surface net heat flux, fresh water flux, seasonal variability, thermohaline structure

1. Introduction shelf sea suggests that the water column will be readily affected by seasonally varying atmospheric conditions The combined Yellow Sea and East China Sea such as heating, cooling, and wind stress. Therefore, 2 (called YES) covers roughly 1 250 000 km and is one of the seasonal variation of the water masses is remark- the most developed continental shelf areas in the world ably large. Another feature of the depth distribution is (Yanagi and Takahashi, 1993). While the Yellow Sea the east/west asymmetry. Extensive shoals (< 20 m) (YS) covers a relatively large area, it is quite shallow are located in the western Yellow Sea along the Chi- reaching a maximum depth of about 140 m (Fig. 1). The water depth over most of the area is less than nese coast but are not generally found off the South 50 m. The deepest water is confined to a north-south Korean coastal regions. Also, the 50 m isobath is lo- oriented trench which runs from the northern bound- cated more than 100 km from the Chinese coast, but ary south to the 100 m isobath where it fans out onto only about 50 km from the South Korean coast. This the continental shelf break. The bathymetric gradients asymmetry in bottom depth is important in control- are very small. Such a broad and shallow continental ling the mixed layer depth (Chu et al., 1997b).

*E-mail: [email protected] 2 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

40 mation of the water mass features.

50

38 2. YES Oceanography and subdivision 80 2.1 Water masses

36 YS Korea Earlier studies (e.g., Liu et al., 1992; Su and Weng, 1994; Chen et al., 1995) have shown that a complex 80 water mass structure exists in YES based on temper- 34 ature and salinity features. The water in the shallow Japan Yangtze River parts of YES is affected greatly by the atmospheric (Kyusyu) 32 and geographic conditions, and therefore it does not 50 have the homogeneity as in the open ocean. Su and

Latitude (N) Weng (1994) proposed a concept of modified water 30 China 200 80 mass holding similar physical and chemical character- istics and occupying a certain space. ECS 28 Based on temperature data observed in 1978– 2000 1980, Su and Weng (1994) identified the following wa- 4000 ter masses using a successive clustering method on 26 temperature observations: Kuroshio Surface Water 2000 Okinawa2000 Trough (KSW), East China Sea Mixed Water (ECSMW), East Ryukyu Islands 24 China Sea Deep Water (ECSDW), Yellow Sea and

Taiwan East China Sea Mixed Water (YEMW), Yellow Sea 118 120 122 124 126 128 130 132 Mixed Water (YSMW), Yellow Sea Bottom Cold Wa- Longitude (E) ter (YSCW), Yellow Sea Near-shore Water (YSNW), and Continental Shelf Diluted Water (CSDW). Fig. 1. Topography and isobaths of the East China Sea Since salinity is a more conserved quantity in the and the Yellow Sea. YES than temperature (Su and Weng, 1994), the wa- ter masses are classified into three systems based on their salinity features: (1) the ECS water system (or The East China Sea (ECS) is located south of the the open water system), including KSW, ECSMW, YS to the north of Taiwan (Fig. 1). The ECS is usu- ECSDW, and YEMW (in May only) and originating ally defined as reaching from the northern end of the from the Kuroshio Water Mass, (2) the YS water sys- Taiwan Strait to the southern end of Kyusyu where it tem (or the local water system), including YSMW, adjoins the YS along a line from Kyushu to Shanghai YSCW, YEMW, and (3) the coastal water system, in- (the mouth of the Yangtze River). With the excep- cluding CSDW and YSNW. tion of the Okinawa Trough west of the Ryukyu Is- Since the ECS water system is an open water sys- lands, which reaches 2 700 m depth, the ECS is part of tem, horizontal advection, especially the meandering the continental shelf. The hydrographic character of of the Kuroshio Current, should be an important fac- the water masses in YES depends on the bathymetry, tor in its variability. In contrast, the YS water system the Kuroshio Current, the atmospheric forcing (mon- is a local system. The local atmospheric forcing should soon winds, heat and moisture fluxes), and the degree be responsible for its variability. In fact, the central of mixing of fresh water originating from Asian river region of the YS stands out due to its cold bottom runoff with the intrusion of Kuroshio waters (Tomczak water, which is YSCW. During summer, strong sur- and Godfrey, 1994). face warming generates the sharpest thermocline on the top of the YSCW in the region. During winter, To examine the seasonal variability of the water this water mass provides a significant buffer prevent- mass and circulation, we use the U.S. Navy’s Master ing central YS water temperatures from reaching the Oceanographic Observation Data Set (MOODS) to in- near-freezing temperatures that the near-shore regions vestigate the climatological features and the seasonal experience (Chu et al., 1997a, b). and non-seasonal variability of the thermohaline struc- ture, and use the Comprehensive Ocean-Atmosphere 2.2 Subdivision Data Set (COADS) from 1945 to 1989 to investigate YS and ECS are separated at 32◦N. Water masses the linkage between the fluxes (momentum, heat, and and current systems on the shelf (depth less than 80 m) moisture) across the air-ocean interface and the for- differ from those on the slope (80–200 m) and deeper NO. 1 CHU ET AL. 3

(depth greater than 200 m). Considering the bottom structure of YES. During the winter monsoon season, topography effect and current systems, YES is divided a very cold northwest wind blows over the ECS/YS as into five subregions (Fig. 2): (1) the ECS shelf with the a result of the Siberian High Pressure System. The Jet eastern boundary at the 80 m bathymetry contour line Stream is positioned to the south of the YS and the and the northern boundary at 32◦N, representing the Polar Front to the north of the Philippines. By late coastal current system; (2) the YS shelf with the north- April, the Polar Front has moved northward toward ern boundary at 38◦300N, the southern boundary at Korea with warm, moist air following behind. Numer- 32◦N, the western boundary at 122◦E in the mouth of ous frontally-generated events occur making late April the Gulf of Bohai and the Chinese coast, and the east- and May highly variable in terms of wind speeds and ern boundary at the Korean coast and the 80 m depth cloud amount. During this period storms originating contour line, representing the YS gyre system; (3) the in Mongolia may cause strong, warm westerlies car- Cheju bifurcation (CB) area with the northern and rying yellow desert sand (termed the “Yellow Wind”). western boundaries at the 80 m contour, the eastern By late May and early June, the summer surface atmo- boundary at 127◦150E, and the southern boundary at spheric low pressure system begins to form over Asia. 32◦N, representing the Taiwan Warm Curent (TWC) Initially this low pressure system is centered north branching into the Tsushima current and Yellow Sea of the Yellow Sea producing westerly winds. In late Warm Current; (4) the TWC area bounded by the 80 June, this low begins to migrate to the west setting m and 200 m bathymetric contours, and 32◦N latitude, up the southwest monsoon that dominates the sum- representing the TWC system; and (5) the Kuroshio mer months. The winds remain variable through June area bounded by the 200 m and 4000 m contours, and until strengthening of the Manchurian Low pressure 32◦N latitude, representing the KC system. system occurs. The Jet Stream migrates just south of Korea, and the Polar Front is just south of Kyushu 3. Seasonal variation of the atmospheric forc- and Shikoku. Despite the very active weather systems, ing the mean surface wind speed over the central Yellow Sea in summer is between 3 and 4 m s−1, which is 3.1 General description weaker than in winter. June also marks, historically, a The Asian monsoon strongly affects the thermal jump in precipitation associated with warm, moist air south of the Polar Front (Watts, 1969). Occasionally 40 the Okhotsk High blocks the northerly progression of the Polar Front. By July, however, high pressure (the Bonin High) to the south and the low pressure over 38 Manchuria produce southerly winds carrying warm, moist air over YES. Korea 36 The summer monthly mean surface air tempera- ◦ YS ture (SAT) is usually 1.5–2 C warmer than the mean sea surface temperature (SST) (Van Loon, 1984). The 34 warmer air causes a downward heat flux at the air- Japan CB ocean interface. This heat flux plus the strong down- 32 ward net radiation stabilizes the upper layer of the China water and causes the surface mixed layer to shoal, cre-

Latitude (N) ating a multi-layer structure. Below the thermocline in 30 ECS the YS, there is a cold water mass, commonly referred to as Yellow Sea Bottom Cold Water (YSCW) that 28 TWC remains unchanged and nearly motionless throughout KC the summer (Li and Yuan, 1992). October is the be- ginning of the transition back to the winter condi- 26 tions. The southerly winds weaken, letting the sea surface slope reestablish the winter pattern. The YS ◦ ◦ 24 SST steadily decreases from 12 C in October to 4 C

Taiwan in January (Chu et al., 1997a). 118 120 122 124 126 128 130 132 3.2 Surface climatology Longitude (E) Here, we present a climatological description of the Fig. 2. Subdivisions of ECS/YS. atmospheric conditions over YES and their effect in 4 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

February May 40 40 10(m/s) 10(m/s) 38 38

36 36

34 34 ) ) N N ( (

e 32 e 32 d d u u t t i i t t a a

L 30 L 30

28 28

26 26

24 24 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E ) Longitude (E )

August November 40 40 10(m/s) 10(m/s) 38 38

36 36

34 34 ) ) N N ( (

e 32 e 32 d d u u t t i i t t a a

L 30 L 30

28 28

26 26

24 24 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E ) Longitude (E ) (a)

Fig. 3. (a) Average surface winds for February, May, August, and November from COADS. (b) Average surface air temperature (◦C) for February, May, August, and November from COADS. (c) Average surface relative humidity (%) for February, May, August, and November from COADS. terms of heat exchange and storage, derived from the winds prevail in the southern YES. In summer, weak 1◦ × 1◦ COADS data (1945–1989). The climatological southeast winds prevail. May is the summer monsoon monthly mean values of wind speed, temperature and transition period. The surface winds over the YS relative humidity are shown in Fig. 3a–c for the months are very weak (mean monthly wind speed less of February, May, August, and November, which may than 3 m s−1) both in summer and during the sum- be regarded as representative patterns for each season. mer monsoon transition period. The main characteristic of the COADS wind data The SAT exhibits seasonal fluctuations of about (Fig. 3a) is the seasonal variation of the monsoon 22◦C in the YS and about 10◦C in the ECS, averaging winds. In winter, strong north to northwest winds about 17◦C over the entire domain (Fig. 3b). A nearly prevail in the northern YES and north to northeast latitudinal gradient prevails in all seasons with the NO. 1 CHU ET AL. 5

February May 40 40 2 38 38 14

36 36 4

34 34 ) ) 6 N N 6 ( ( 1

e e 32 8 32 d d u u 8

t t 1 i i t 10 t a a

L 30 L 30 20 12

28 14 28 22

16 26 26 24 18

24 20 24 26 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E ) Longitude (E )

August November 40 40

38 25 38

36 36 12

26 34 34 ) ) 14 N N ( (

27 e 32 e 32 16 d d u u t t i i t 8 t a 2 a 18

L 30 L 30

20 28 28

28 22 26 26

24 24 24 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E ) Longitude (E ) (b)

Fig. 3. (Continued). maximum south-to-north mean SAT difference of 18◦C lead to maximum (minimum) evaporation there. ◦ in February and the minimum SAT difference of 3 C 3.3 Surface fluxes in August. The SAT is uniformly warm (mean SAT around 28◦C) in the ECS during summer (August). Fluxes across the air-ocean interface drive oceanic Surface relative humidity (Fig. 3c) is generally processes. Recent studies (Chu et al., 1998) show that the Haney type sea surface boundary condition higher in the summer season, mainly as a consequence (Haney, 1971) is not valid for regional seas. For the of the southeast summer monsoon bringing moist air YES modeling, it is necessary to use all the fluxes (Chu from the Tropics. The summer field is relatively homo- et al. 2001). geneous (83%) in the ECS, whereas a significant lati- tudinal gradient exists in the other seasons. A relative 3.3.1 Net heat flux humidity minimum (maximum) is present in all sea- Net surface heat flux (Qnet) is computed by sons near Kyushu Island (Chinese coast), which may Qnet = Rs − (Rl + Ql + Qs) (1) 6 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

February May 40 40 H 38 38 82 0

8

36 36 8 7 84 34 34 ) ) 6 N N 7 ( ( 6

8 e e 32 32 d d 7 L u u 0 H t t L i i t t 80 a a

L 30 L 30 7 2 8 2 7 28 4 28 8 5 7 6 83 26 26 0 8 7 8 4 8 24 24 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E ) Longitude (E )

August November 40 40

0 8 38 38 0 7

36 82 36

34 34 ) ) L N N ( ( 6 8

4 8 e e 32 32 d H d u u 7 t t 0 i i t 82 t a a

L 30 3 L 30 8 72 28 28 74

4 26 8 26 76 3 8 7 24 24 8 78 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E ) Longitude (E ) (c)

Fig. 3. (Continued). where Rs is the net downward shortwave radiation, the winter (November to February), and a minimum −2 Rl the net upward longwave radiation, Ql the sensible heat gain of 80 W m in the spring. This long term heat flux, and Qs the latent heat flux. Positive (nega- net surface heat loss will be compensated for by the tive) values of Qnet indicate net heat gain (loss) of the advection of warm tropical waters by KC and TWC. ocean at the surface. The fluxes (Rs,Rl,Ql,Qs) are Besides, there is an increase of net surface heat gain directly obtained from the COADS dataset. The sum- (flux downward) toward the Chinese coast in all sea- mer field is relatively homogeneous (120–160 W m−2) sons, which might be caused by less evaporation there. in the whole YES, whereas a significant horizontal gra- The annual mean heat budget is positive in dient with a minimum center near Kyushu Island ex- the YS region (15 W m−2) and negative elsewhere: ists in the other seasons (Fig. 4). The ocean surface around –30 W m−2 in the ECS and CB regions, and near Kyushu Island belonging to the KC and TWC –65 W m−2 in the KC and TWC regions (Tables 1–5). sub-regions has a maximum heat loss of 280 W m−2 in These values consist with recent calculations (Hirose NO. 1 CHU ET AL. 7

February May 40 40

38 38

36 36 200

−60 −80 34 34

−100

−120 180 32 32 −140 160 −160 −280 140

Latitude (N) 30 (N) Latitude 30 120 −180 100 −200 −260 28 28 −220 80 −240

26 −220 26 −200 100 −180 −160 120 24 −140−120 24 140

−100 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E) Longitude (E)

August November 40 40

38 38

36 36

−160 34 34 −180

−200 32 32

140

160

−220 −280 Latitude (N) Latitude Latitude (N) Latitude 30 30 −260

28 120 28 −240

140 −220 26 120 26 −200 −180 −160 24 24 −140 −120 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E) Longitude (E)

Fig. 4. Average net surface heat flux (W m−2) for February, May, August, and November from COADS. Solid (dashed) lines indicate positive (negative) values. et al., 1999). A low annual mean latent heat flux in the (3) larger annual variation in the shallow water regions YS region causes the positive annual heat budget. Sen- (YS, CB) than in the relatively deep water regions (KC sible heat flux is nearly zero in summer, which implies and TWC). SST is close to SAT. The ratio between sensible and The monthly variation of net surface heat flux (Fig. latent heat fluxes (defined as the Bowen ratio) is listed 5) is nearly sinusoidal and quite similar among the five in the seventh column of these tables. The Bowen ra- sub-areas. The YS region leads the rest of the sub- tio has the following features: (1) a value less than areas by half to one month: Qnet becomes positive in one for any month for all five sub-areas, indicating mid-February in the YS region, in early March in the larger latentheat flux than the sensible heat flux, (2) ECS and CB regions, and in late March in the KC and a seasonal variation with relatively large values (0.33– TWC regions. Qnet peaks in June in the YS region and 0.67) in winter and near zero values in summer, and in July in the rest of the sub-areas. 8 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

Table 1. Monthly surface heat flux components obtained from COADS data for the ECS shelf region.

−2 −2 −2 −2 −2 Month Rs (W m ) Rl (W m ) Qs (W m ) Ql (W m ) Qnet (W m ) Qs/Ql Jan 93.6 78.6 78.3 191.4 –254.8 0.41 Feb 122.0 71.4 64.3 159.3 –173.0 0.40 Mar 165.0 62.4 35.0 116.8 –49.2 0.30 Apr 208.0 51.5 14.0 70.8 71.3 0.20 May 234.8 43.5 5.4 49.5 136.4 0.11 Jun 241.2 34.3 1.5 39.4 166.0 0.04 Jul 269.7 33.1 –2.8 43.7 195.7 -0.06 Aug 257.4 38.4 0.1 72.8 146.1 0.00 Sep 207.5 47.9 8.1 128.6 22.8 0.06 Oct 158.5 62.0 22.4 190.5 –116.6 0.12 Nov 110.5 70.0 43.8 205.5 –208.9 0.21 Dec 89.0 79.3 71.6 212.0 –274.2 0.34 Annual Mean 179.7 56.1 28.5 123.4 –28.2 0.23

Table 2. Monthly surface heat flux components obtained from COADS data for the YS region.

−2 −2 −2 −2 −2 Month Rs (W m ) Rl (W m ) Qs (W m ) Ql (W m ) Qnet (W m ) Qs/Ql Jan 82.1 88.8 76.8 114.2 –197.7 0.67 Feb 117.6 79.1 53.3 78.7 –93.4 0.68 Mar 172.0 68.5 16.0 44.9 42.6 0.36 Apr 222.3 57.3 0.2 23.2 141.6 0.01 May 261.4 21.1 –1.0 13.6 197.7 –0.07 Jun 260.3 39.0 –1.7 14.4 208.6 –0.12 Jul 247.6 32.4 –2.6 15.8 202.0 –0.16 Aug 245.8 39.1 0.7 57.7 148.3 0.01 Sep 194.3 56.5 8.8 114.8 14.2 0.08 Oct 149.7 73.2 23.2 133.6 –80.3 0.17 Nov 99.7 84.6 48.2 137.6 –170.7 0.35 Dec 75.1 90.7 74.7 137.6 –228.0 0.54 Annual Mean 177.3 63.4 24.7 73.8 15.4 0.33

Table 3. Monthly surface heat flux components obtained from COADS data for the CB region.

−2 −2 −2 −2 −2 Month Rs (W m ) Rl (W m ) Qs (W m ) Ql (W m ) Qnet (W m ) Qs/Ql Jan 81.4 88.8 95.7 172.5 –275.6 0.55 Feb 116.6 79.2 70.8 136.1 –169.4 0.52 Mar 166.5 70.1 32.6 94.4 –30.5 0.35 Apr 212.1 57.0 10.0 51.2 93.9 0.20 May 248.0 48.8 2.9 32.4 163.8 0.09 Jun 241.1 36.0 1.6 24.9 178.5 0.06 Jul 247.2 32.0 –1.3 26.6 189.8 –0.05 Aug 245.4 38.8 2.7 68.7 135.1 0.04 Sep 190.7 51.8 12.4 127.9 –1.3 0.10 Oct 150.4 71.2 29.5 180.6 –130.9 0.16 Nov 103.2 83.7 56.0 192.0 –228.5 0.29 Dec 77.8 91.5 90.3 198.5 –302.5 0.45 Annual Mean 173.4 62.4 33.6 108.8 –31.5 0.31 NO. 1 CHU ET AL. 9

Table 4. Monthly surface heat flux components obtained from COADS data for the TWC region.

−2 −2 −2 −2 −2 Month Rs (W m ) Rl (W m ) Qs (W m ) Ql (W m ) Qnet (W m ) Qs/Ql Jan 93.1 81.6 88.1 233.0 –309.7 0.38 Feb 120.5 74.4 74.3 207.1 –235.4 0.36 Mar 162.6 65.2 44.6 159.6 –106.8 0.28 Apr 206.1 53.7 20.3 104.2 27.9 0.19 May 230.2 44.4 9.9 77.5 98.3 0.13 Jun 238.0 34.6 4.2 57.2 142.1 0.07 Jul 269.9 34.5 –0.2 62.6 172.9 0.00 Aug 254.2 39.0 2.6 88.2 124.3 0.03 Sep 207.5 47.3 9.8 136.5 13.9 0.07 Oct 158.5 62.1 25.9 211.8 –141.3 0.12 Nov 110.9 70.4 47.4 235.6 –242.5 0.20 Dec 89.7 57.4 33.6 151.9 –64.5 0.18 Annual Mean 178.4 57.4 33.6 151.9 –64.5 0.18

Table 5. Monthly surface heat flux components obtained from COADS data for the KC region.

−2 −2 −2 −2 −2 Month Rs (W m ) Rl (W m ) Qs (W m ) Ql (W m ) Qnet (W m ) Qs/Ql Jan 99.4 80.5 81.4 248.0 –310.5 0.33 Feb 126.9 73.8 68.9 221.7 –237.6 0.31 Mar 168.5 65.3 43.4 173.9 –114.2 0.25 Apr 211.8 53.7 20.0 118.7 19.5 0.17 May 233.8 44.1 10.5 92.0 87.2 0.11 Jun 243.3 25.1 4.6 67.9 135.7 0.07 Jul 272.9 26.3 1.3 77.3 157.9 0.02 Aug 253.4 39.2 3.3 95.2 115.7 0.03 Sep 213.8 45.6 8.6 130.2 29.4 0.07 Oct 163.3 58.6 24.2 210.0 –129.5 0.12 Nov 116.3 66.7 42.7 238.8 –232.0 0.18 Dec 95.3 78.3 67.6 256.7 –307.2 0.26 Annual Mean 183.2 56.4 31.4 160.9 –65.5 0.20

3.3.2 Fresh water flux with a maximum water mass loss (19 cm month−1 in November) near Kyushu Island. The surface fresh water flux is the difference be- tween precipitation rate (P ) and evaporation rate (E), The climatological monthly surface fresh water flux components for the five sub-areas show similar fea- F = P − E. (2) tures: (1) fresh water gain during the summer mon- Positive values of F indicate a net water mass gain of soon season and fresh water loss during the winter the ocean at the surface. The surface fresh water flux monsoon season, and (2) net annual fresh water loss exhibits distinct winter and summer patterns (Fig. 6). (1.8–4 cm month−1) from the air-ocean interface. Such The summer pattern is characterized by fresh water a long term surface saline process might be compen- gain in the whole area, and a saddle-type distribu- sated for by the river run-off. The seasonal cycle of tion with two low centers, one in the east (less than surface fresh water budget (Fig. 7) shows that E ex- 3 cm month−1 near Kyushu Island) and the other in ceeds P during the winter monsoon period (Septem- the west (less than 1 cm month−1 near the Chinese ber to March) and P exceeds E during the summer coast). Centers of high fresh water flux are found in monsoon period (May to August). The shallow water the north (more than 14 cm month−1 in northern YS) region (YC, ECS, CB) leads the relatively deep region and in the south (8 cm month−1). The winter pattern (KC, TWC) by nearly one month to get fresh water. is characterized by fresh water loss in the whole area F peaks in July in the YS sub-area and in June else- 10 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

300 from a variety of instruments. Due to the shear size ECS (more than six million profiles) and enormous influx YS CB of data to the Naval Oceanographic Office from var- 200 TWC KC ious sources, quality control is a difficult task (Chu et al., 1997c). Our study domain lies inside the area ◦ ◦ ◦ ◦ 24 N to 40 N and 118 E to 132 E (Fig. 8); the dataset 100 within this region consisted of nearly 24 000 profiles af- ter rejecting certain data during quality control. These )

－ 2 primary editing procedures included removal of pro- 0 files with obviously erroneous location, profiles with

( W m large spikes, and profiles displaying features that do －100 n e t not match the characteristics of surrounding profiles. Q In shallow water, this procedure can be partially auto- mated but also involves subjective interpretation be- －200 cause of the under-sampling of MOODS compared to the spatial and temporal variability of the water masses (Chu et al., 1997a, b). －300 The main limitation of the MOODS data is its ir- regular distribution in time and space. Certain periods and areas are over sampled while others lack enough －400 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec observations to provide meaningful insights. Vertical resolution and data quality are also highly variable Fig. 5. Seasonal variations of monthly mean net heat flux −2 depending much on instrument type and sampling ex- (W m ) for the five sub-regions. pertise. A prominent data sparse area is located off where. the eastern coastal region of China (Fig. 8). The pe- riod of 1975 to 1986 is found to have a relatively large 3.3.3 Surface buoyancy flux number of profiles. Yearly temperature (salinity) casts Both Qnet and F are in-phase during the seasonal above 2 000 (400) are 1975, 1976, 1978, 1981, and 1986 variation: positive (negative) values during the sum- (Figs. 9a, b). Within a given year, temporally uneven mer (winter) monsoon season. Thus, the surface buoy- distributions can be seen such as in the 1984 temper- ancy flux, ature casts (Fig. 9c) and salinity casts (Fig. 9d). Spa-

αgQnet tial and temporal irregularities along with the lack of B = + βg(P − E)S0 (3) data in certain regions must be carefully weighed in ρwcpw order to avoid statistically-induced variability (Chu et should have a similar seasonal variation as Q and F . net al, 1997c). Here, g is gravitational acceleration, ρw the character- istic water density, c the water specific heat under pw 5. Seasonal variation of water mass character- constant pressure, α the water thermal expansion co- istics efficient, β the salinity contraction coefficient, and S0 the surface salinity. The surface buoyancy flux, B, 5.1 Relative heat storage for the five sub-regions has a similar quasi-sinusoidal If we are not interested in the actual heat storage seasonal variation with the maximum buoyancy gain values, but rather in its annual cycle, the relative heat (loss) in the summer (winter), which results in a shal- storage (RHS) is useful to arrange our hydrographic low (deep) mixed layer in summer (winter). dataset on a seasonal basis. The definition of RHS is 4. Hydrographic dataset taken from Hecht et al. (1985),  Z 0  D The MOODS is a compilation of observed ocean  cpρw(T − T0)dz , if H < D SRH = H −H (4) data worldwide consisting of (1) temperature-only pro-  R 0 c ρ (T − T )dz , if H D files; (2) both temperature and salinity profiles, (3) −H p w 0 > sound-speed profiles, and (4) surface temperatures where H is the actual water column depth, D the max- from drifting buoys. These measurements are, in gen- imum water depth considered, T the water tempera- eral, irregular in time and space. In this study, we ture, and T0 an arbitrary reference temperature. In ◦ analyze temperature and salinity profiles measured our computations, D = 50 m, and T0 = 5 C. NO. 1 CHU ET AL. 11

February May 40 40

38 38 −3

1 36 36 2 −4

34 34 3

−5

32 32 −6 L 4 2L −7 −16 H Latitude (N) 30 (N) Latitude 30 −8 −15 3 −9 −14 28 28 4 −10 3 −13 −9 −12 26 −11 26 2

1 24 24 0 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E) Longitude (E)

August November 40 40 14 12 10 38 38 −10 8

36 5 6 7 36 −11 4 L1 3 2 34 34 L 32 32 −19 −18

5 Latitude (N) 30 Latitude (N) 30 −17 −16 −15 L −14 28 28 −12−13 3 −11 26 4 26 −10 5 −9 6 −8 24 7 24 −7 8 −6 118 120 122 124 126 128 130 132 118 120 122 124 126 128 130 132 Longitude (E) Longitude (E)

Fig. 6. Average surface fresh water flux (cm month−1) for February, May, August, and November from COADS. Solid (dashed) curves indicate positive (negative) values.

The RHS values for the five sub-areas are shown mer. A significant cooling begins at the end ofsummer in Fig. 10. Two extreme seasons can be clearly identi- with weakening strength with depth. The whole wa- fied, namely, winter from January to April, and sum- ter column becomes well mixed withtemperature near mer from July to November. We can also define the 18◦C in autumn. This is probably due to the entrain- two transition seasons: spring, consisting of May and ment mixing generated by strong surface winds, tidal June, and fall, consisting of December. mixing, as well as strong surface buoyancy loss caused 5.2 Temperature and salinity by the net heat loss of 274 W m−2 (Table 1) and fresh −1 In the ECS shelf region, the entire water column water loss of 13.3 cm month (Fig. 7). From autumn exhibits an evident seasonal thermal cycle (Fig. 11a). to winter, the whole water column is uniformly cooled ◦ A well developed thermocline is present down to 30-m to 13 C. Total seasonal variability at the bottom is depth in spring and extends to the bottom in sum- around 6.5◦C.The surface salinity S exhibits a minim- 12 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

E C S YS ) ) - 1 - 1 20 20 E

10 10 E

P P 0 0

－ 10 － 10 a t e r b l n c ( m o h a t e r b l n c ( m o h P－E (a) P－E (b) W －W 20 － 20 J F M A M J J A S O N D J F M A M J J A S O N D

C B TWC ) ) - 1 - 1 20 20 E E 10 10 P P 0 0

－ 10 － 10 a t e r b l n c ( m o h a t e r b l n c ( m o h (c) － (d) P－E W W P E － 20 － 20 J F M A M J J A S O N D J F M A M J J A S O N D

KC ) - 1 20 E

10 P

0

－ 10

a t e r b l n c ( m o h (e) P－E W － 20 J F M A M J J A S O N D

Fig. 7. Seasonal variations of monthly mean fresh water flux (cm month−1) for the five sub-regions. um value of 32.88 ppt in spring, increases to 33.39 Liu et al. (1992). The surface freshness is in-phase ppt in summer and further increases to its maximum with the surface fresh water flux F (Fig. 7a) in such a value of 34.08 ppt in autumn, and then decreases to way that the maximum (minimum) F and the lowest 33.74 ppt in winter (Fig. 12a). From the T-S diagram (highest) surface salinity occurs in the same season, (Fig. 13a) we can clearly recognize a seasonal layer of spring (autumn). However, the surface dilution is out the East China Sea Surface Water (ECSSW), which of phase with the Yangtze River water discharge. The corresponds to relatively low salinities and high tem- lowest surface salinity (spring) does not occur at the peratures of summer, and an ECSDW layer, which is same time as the maximum Yangtze River discharge. cooled and renewed in winter. From our data, the This may confirm an earlier study by Beardsley et al. ECSDW (open circles in Fig. 13a) has average char- (1985) that the summer fresh water discharge from the ◦ acteristics of T = 12.82 C, S = 33.88 ppt, σt > 25.2 Yangtze River forms a relatively shallow, low salinity kg m−3, which is consistent with an earlier study by plume-like structure extending offshore on average to- NO. 1 CHU ET AL. 13

Temperature (total:24272) Temperature (total:24272) moderately-high salinity water with a weak seasonal (a) variation. The winter density, σt, is greater than 25.6 −3 38oN (a) kg m . These values are consistent with many earlier 38oN studies (Lie, 1985, 1987; Li and Yuan, 1992; Chen et 36oN al., 1994) describing the characteristics of YSCW near 36oN the bottom all year round. The lack of seasonality 34oN might be due to the semi-enclosed basin and the bowl- 34oN type bottom topography (Fig. 1) which limits mixing e

d o with the surrounding waters. The surface waters are u

e 32 N t i d t o freshest in summer (S = 31.42 ppt) but rapidly achieve u

a 32 N t i L t o their saltiest condition in autumn (32.98 ppt). The net

a 30 N

L o surface fresh water flux has the maximum value (4.84 30 N −1 o cm month ) in spring, reduces to a value of –1.96 cm 28 N month−1 (Fig. 7) in summer (July–November) and 28oN o continues to drop to the minimum value of –7.96 26 N cm month−1 (Fig. 7) in autumn. Such a mismatch 26oN between surface salinity and F may suggest that the 24oN dominant factor for determining the near surface salin- 24oN 118oE 121oE 124oE 127oE 130oE ity structure in summer is the river run-off. This co- 118oE 121oE Lo1n2g4iotEude 127oE 130oE incides with Chen et al.’s (1994) results based on hy- Longitude drographic observations during 1986. From the T − S diagram (Fig. 13b) we can clearly recognize a seasonal S alinity (total:5401) variation of the YSCW, which is cooled and renewed in S alinity (total:5401) winter. We may define the YSCW having winter char- ◦ (b) acteristics of T = 6.5 C, S=32.75 ppt, and σt > 25.6 38oN (b) kg m−3. These values are consistent with an earlier 38oN study by Liu et al. 1992. 36oN In the CB region, there is a strong seasonal vari- 36oN ation. A spring-summer thermohaline is present but 34oN weak (Figs. 11c, 12c). From summer to autumn, a o e 34 N significant cooling takes place close to the surface and

d o

u 32 N

t weakens with depth. The whole water column becomes e i t

d o ◦ a well mixed with temperature near 17.0 C in autumn. u 32 N L t

i o t 30 N In winter, the whole water column is uniformly cooled a

L ◦ 30oN to 14.3 C. Total seasonal thermal variability at the 28oN bottom is around 5.0◦C. The surface salinity has a 28oN minimum value of 31.46 ppt in summer, increases to 26oN 33.99 ppt in autumn and further increases to the max- 26oN imum value of 34.41 ppt in winter, and then decreases 24oN to 32.68 ppt in spring (Fig. 12c). Total seasonal haline o variability at the bottom is around 1.69 ppt. Beardsley 24 N 118oE 121oE 124oE 127oE 130oE et al. (1985) indicates that the TWC bifurcates near Longitude 118oE 121oE 124oE 127oE 130oE the CB region into the Yellow Sea Warm Current and Fig. 8. Spatial distributionL ofon MOODSgitude data during 1975– the Tsushima current. There should be some connec- 1986. tion between the CB region and the upper layer (30 m) of the TWC region. However, the similarity between wards the northeast. Thus, the dominant factor in de- CB and upper TWC profiles only occurs in tempera- termining near-surface salinity is the surface fresh wa- ture but not in salinity (Figs. 11c and d). The seasonal ter flux. Besides, a well developed halocline extends to surface salinity variation follows the seasonal cycle of 40-m in spring and summer, being stronger in spring. F : minimum salinity and maximum surface fresh wa- In the YS basin, the spring-summer thermocline is ter flux in summer and the opposite in winter. From formed down to the bottom. The bottom water(Figs. the T − S diagram (Fig. 13c) we can clearly recognize 11b and 12b) is observed as low temperature and the TWC deep water (TWCDW) and YEMW. 14 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

Temperature S alinity 3500 1200 (a) (b) 3000 1000 s e l i 2500 f

o 800 r p

f 2000 o

600 r

e 1500 b

m 400 u

N 1000

500 200

0 0 75 76 77 78 79 80 81 82 83 84 85 86 75 76 77 78 79 80 81 82 83 84 85 86 Year Year

Temperature S alinity 300 100 (c) 1984 (d) 1984 250 80 s e l i f

o 200 r

p 60

f o

150 r e

b 40

m 100 u N 20 50

0 0 J F M A M J J A S O N D J F M A M J J A S O N D Month Month

Fig. 9. Temporally irregular cast distribution: (a) yearly temperature, (b) yearly salinity, (c) monthly temperature in 1984, and (d) monthly salinity in 1984.

In the TWC region, the upper water column (sur- 4.0◦C increase from winter conditions followed by a face to 60 m), usually called the TWC Upper Water 3◦C increase from spring to summer. In summer, the (TWCUW), exhibits an evident seasonal thermohaline surface temperature reaches its maximum of 26.3◦C. cycle (Figs. 11d, 12d). Total seasonal variability of Significant cooling (4.7◦C) occurs in autumn. The sur- temperature (salinity) is around 7◦C (1.02 ppt) at the face salinity has the minimum value of 33.67 ppt in surface and weakens with depth. Below 60 m depth, summer, increases to the maximum value of 34.69 ppt there exists a permanent thermocline with very weak in autumn and slightly decreases to 34.52 ppt in win- seasonal variation. At the bottom (150 m), the water ter, and then decreases to 33.91 ppt in spring. The sur- properties are seasonally invariant; the temperature is face freshness is in-phase with the surface fresh water about 15◦C, and the salinity is around 34.62 ppt. In flux F (Fig. 7d). From the T −S diagram (Fig. 13d) we winter, the surface waters are cool with a temperature can clearly recognize a seasonal layer of TWC, which of 19◦C. Significant warming begins in spring with a is relatively fresh and warm in summer and saline and NO. 1 CHU ET AL. 15

E C S YS 1000 1000

800 800

600 600

400 400

200 200 (a) (b) 0 0 J F M A M J J A S O N D J F M A M J J A S O N D

C B TWC 1000 1000

900 800 )

- 2 800 600 700 400

R H S ( M J m 600

200 500 (c) (d) 0 400 J F M A M J J A S O N D J F M A M J J A S O N D

KC 1000

900

800

700

(e) 600 J F M A M J J A S O N D Fig. 10. Seasonal variations of monthly mean relative heat storage for the five sub- regions. cool in winter. gins in spring. There is a 3.7◦C increase from winter to In the KC region the upper water column (sur- spring and a 2.0◦C increase from spring to summer. In face to 80 m), usually called the Kuroshio Surface Wa- summer,the surface temperature reaches its maximum ter (KSW), exhibits an evident seasonal thermal cy- (27.4◦C).A significant cooling starts in autumn. There cle (Figs. 11e, f; 12e, f). Total seasonal variability is a 3.7◦C decrease from summer to autumn. The sur- of temperature (salinity) is around 5.7◦C (0.47 ppt) face salinity has the minimum value of 34.29 ppt in at the surface and weakens with depth. From 80 to summer, increases to the maximum value of 34.75 ppt 600 m, there exists a permanent thermocline/halocline in winter, and then decreases to 34.56 ppt in spring. with very weak seasonal variation. At 1500 m depth, The surface freshness is in-phase with the surface fresh the temperature is about 3.7◦C and the salinity is water flux F (Fig. 7d). Our data also show the exis- around 34.43 ppt. In winter, the surface has a mini- tence of S-type salinity profiles with a maximum mum temperature (21.7◦C).A significant warming be- salinity in the layer between 150 and 250 m, which 16 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

E C S YS 0 0 －10 － 10

－20 － 20

－30 － 30 D e p t h ( m ) D e p t h ( m ) －40 － 40 (a) (b) －50 － 50 10 15 20 25 30 0 10 20 30

C B TWC 0 0 － 5

－ 10 －50 － 15 － D e p t h ( m ) 20 － 100 D e p t h ( m )

－ 25 (c) (d) － －150 30 10 15 20 25 10 15 20 25 30

KC KC (upper 150m) 0 0 － 200 － － 400 50

－ 600 D e p t h ( m ) D e p t h ( m ) － 100 － 800 (e) (f) －150 －1000 0 10 20 30 o 15 20 25 30 Temperature ( C ) Temperature ( oC )

Fig. 11. Seasonal climatological profiles of temperature for the five sub-regions: (a) the ECS shelf, (b) YS, (c) CB, (d) TWC, (e) KC regions, and (f) upper (0–150 m) KC regions. The symbol “o” denotes winter, “∆” denotes spring, “” denotes summer and “♦” denotes fall. is associated with the Kuroshio Sub-Surface Water the five sub-regions and various ECS/YS water masses (KSSW); and with a minimum salinity at 600 m depth, are well represented (Fig. 14). We can see the connec- which is associated with the Kuroshio Intermediate tions and mixing of water masses among the five sub- Water (KIW). In the sublayer between 150 m and 450 regions. For example, the water masses in the CB re- m, the salinity decreases with depth. From the T − S gion come from TWCDW, YEMW, and YSMW. The diagram (Figs. 13e, f) we can clearly recognize a sea- sonal layer of KC, which is relatively fresh and warm YS and ECS waters are mixed during spring and sum- in summer and saline and cool in winter. Putting five mer. Besides, various ECS/YS water masses are well T − S diagrams (Figs. 13a–e) into one T − S diagram, classified on this T − S diagram. NO. 1 CHU ET AL. 17

E C S YS 0 0

－10 － 10

－20 － 20

－30 － 30 D e p t h ( m ) D e p t h ( m )

－40 － 40 (a) (b) －50 － 50 32.5 33 33.5 34 34.5 31 32 33 34

C B TWC 0 0 －5

－ 10 －50 － 15 －20

D e p t h ( m ) － D e p t h ( m ) 100

－ 25 (c) (d) － － 30 150 31 32 33 34 35 33.5 34 34.5 35

0 KC KC (upper 150m) 0 －200 － 50 －400

－600 D e p t h ( m ) － 100 D e p t h ( m ) －800 (e) (f) －150 －1000 34 34.5 35 34 34.5 35 S alinity (psu) S alinity (psu)

Fig. 12. Same as Figure 11 except for salinity.

7. Conclusions and the formation of the water mass features for the five sub-regions. The work presented here describes seasonal vari- ability of water mass structure of the East China Sea (1) The most important atmospheric forcing is the and Yellow Sea. Based on the current system and strong winter monsoon (north to northeast) and weak bathymetry, we divided the area into five sub-regions: summer monsoon (southeast). The surface air temper- the East China Sea shelf, the Yellow Sea basin, the ature has a nearly latitudinal gradient with a south- ◦ ◦ Cheju bifurcation zone, the Taiwan Warm Current to-north difference of 18 C in February and 3 C in area, and the Kuroshio Current area. We use the U.S. August. The surface relative humidity is generally Navy’s Master Oceanographic Observation Data Set higher in the summer season, mainly as a consequence and the Comprehensive Ocean-Atmosphere Data Set of the southeast summer monsoon bringing moist air to investigate the linkage between the fluxes (momen- from the Tropics. A relative humidity minimum (max- tum, heat, and moisture) across the air-ocean interface imum) is present in all seasons near Kyushu Island 18 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

26 E C S YS 22 20 23 24 22 E ) ) C C C S o o

S 4 ( 22 ( 2 3 W Y

2 S e e M r 20 r 15 W u u t 4 t 5 a 2 a 2 r r e 18 e p 25 p m m 10 e 16 e

T E C T YS S D C 14 W W 26 (a) (b) 12 5 32.5 33 33.5 34 34.5 31 31.5 32 32.5 33 33.5 S alinity (psu) S alinity (psu) C B TWC 1 2 26 22 T 24 2 W 2 C ) ) 23 U C C 24 W o o

22

( ( 23

4 22

e 2 e r 20 r u u t t

a a 20 4

r r 2 Y 5 e 18 E 2 e T p S p 18 W

m M m W C e e 16 D T T 16 25 W TWC DW6 2 26 14 (c) 14 (d) 31 32 33 34 35 33.5 34 34.5 35 S alinity (psu) S alinity (psu) KC KC (upper 150m) 22 28 22 25 KS K K 3 W S ) 2 S ) S W W

C C 26 o o

( 20 24 (

e e r r 23 K u 5 u 24 t 2 t S

a 15 a S r r W e IW e

p K p 26 22 m m

e 10 e T T 24 20 5 (e) 27 (f) 34 34.5 35 34 34.5 35 S alinity (psu) S alinity (psu) Fig. 13. Seasonal climatological T –S diagrams for the five sub-regions: (a) the ECS shelf, (b) YS, (c) CB, (d) TWC, (e) KC regions, and (f) upper (0–150 m) KC regions. The symbol “o” denotes winter, “∆” denotes spring, “” denotes summer and “♦” denotes fall.

(Chinese coast), which may lead to maximum (mini- is nearly zero in summer, which implies SST is close to mum) evaporation there. SAT. The monthly variation of net surface heat flux (2) The air-sea heat budget at the surface is dom- is nearly sinusoidal and quite similar among the five inated by the incoming solar radiation balanced by sub-areas. Qnet peaks in June in the Yellow Sea and latent and longwave heat energy loss. It was found in July in the rest of the sub-areas. that the Yellow Sea has an overall long term heat gain (3) The air-sea fresh water budget exhibits a dis- of 15 W m−2 (caused by low latent heat flux), and tinct winter fresh water loss and summer fresh water the rest of the area has an overall heat loss of gain pattern. The whole area experiences an overall around 30 W m−2 in the East China Sea and Cheju fresh water loss (1.8–4 cm month−1) at the surface. bifurcation zone, and of 65 W m−2 in the Kuroshio and (4) The East China and Yellow Seas do not follow Taiwan Warm Current regions. The sensible heat flux the usual atmospheric seasons. We divided the four NO. 1 CHU ET AL. 19

30 extending offshore on average towards the northeast. K 5 0 S 2 W T Acknowledgments. The authors are grateful to 4 W

C K U W S C. W. Fan at the Naval Postgraduate School for his pro- 25 1 S E W C gramming assistance. This work was funded by the Naval 1 3 S 2 S W Y S Oceanographic Office, the Office of Naval Research NOMP M W T W Program, and the Naval Postgraduate School. 20 2 C 2 Y D W C ) E

o S M 2 W REFERENCES

3 15 2 E C Beardsley, R. C., R. Limeburner, H. Yu, and G. A. Cannon, S D W 1985: Discharge of the Changjiang (Yangtze River)

4 W 2 I T e m p r a t u ( into the East China Sea. Continental Shelf Research, 10 K 4, 57–765.

Y Chen, C. S., R. C. Beardsley, R. Limeburner, and K. S 5 C 2 W Kim, 1994: Comparison of winter and summer hy- 5 drographic observations in the Yellow and East China

8 6 2 2 Seas and adjacent Kuroshio during 1986. Continental 7 2 Shelf Research, 14, 909–929. 0 31. 0 3 1 . 5 3 2 . 0 3 2 . 5 3 3 . 0 3 3 . 5 3 4 . 0 3 4 . 5 3 5 . 0 3 5 .5 Chu, P. C., C. R. Fralick, S. D. Haeger, and M. J. Carron, Salinity (psu) 1997a: A parametric model for Yellow Sea thermal variability. J. Geophys. Res., 102, 10499–10508. Fig. 14. Water mass classification on the climatological Chu, P. C., S. K. Wells, S. D. Haeger, and M. J. Carron, T − S diagram. 1997b: Spatial and temporal scales of the Yellow Sea thermal variability. J. Geophys. Res., 102, 5655–5668. seasons based on the relative heat storage, namely, Chu, P. C., H. C. Tseng, C. P. Chang, and J. M. Chen, winter (January–April), spring (May–June), summer 1997c: South China Sea warm pool detected in spring (July–November), and autumn (December). from the Navy’s Master Oceanographic Observational (5) We have classified the water masses of the East Data Set (MOODS). J. Geophys. Res., 102, 15761– China and Yellow Seas using climatological averages. 15771. The entire water column of the East China Sea shelf Chu, P. C., Y. C. Chen, and S. H. Lu, 1998: On Haney-type surface thermal boundary conditions for ocean circu- water and the Yellow Sea(including the Cheju bifurca- lation models. Journal of Physical Oceanography, 28, tion zone) undergoes an evident seasonal thermal cy- 890–901. cle with maximum values of temperature during sum- Chu, P. C., S. H. Lu, and C. W. Fan, 2001: An air-ocean mer and maximum mixed layer depths during winter. coupled nowcast/forecast system for the East Asian However, only the surface waters of the Taiwan Warm marginal seas. Advances in Mathematical Modeling of Current region (upper 60 m) and the Kuroshio region Atmosphere and Ocean Dynamics, Kluwer Scientific (upper 80 m) exhibit a seasonal thermal cycle. Below Publishing Co., 137–142. these depths, there is no evident seasonal variation. Haney, H. L., 1971: Surface thermal boundary condi- (6) The salt balance of the surface layer is clearly tion for ocean circulation model. Journal of Physical affected by surface fresh water flux and river run- Oceanography., 1, 241–248. Hirose, N., H.-C. Lee, and J.-H. Yoon, 1999: Surface heat off. In the East China Sea shelf region, the surface flux in the East China Sea and Yellow Sea. Journal of freshness (low salinity) is in-phase with the surface Physical Oceanography., 29, 401–417. fresh water flux (P − E) and out-of-phase with the Li, H., and Y. Yuan, 1992: On the formation and Yangtze River water discharge. The lowest surface maintenance mechanisms of the cold water mass of salinity (spring) does not occur at the same time as the the Yellow Sea. Chinese Journal of Oceanology and maximum Yangtze River discharge (July). In the Yel- Limonology, 10(2), 97–106. low Sea, however, the surface freshness is out-of-phase Lie, H.-J., 1985: Wintertime temperature-salinity charac- with the surface fresh water flux and in-phase with teristics in the Southeastern Hwanghae (Yellow Sea). the Yangtze River water discharge. Two different rela- Journal of Oceanographic Society of Japan, 41, 291– 298. tions between surface salinity and Yangtze River run- Lie, H.-J., 1987: Summertime hydrographic features in off, out-of-phase in the East China Sea shelf and in- the Southeastern Hwanghae (Yellow Sea). Progress in phase in the Yellow Sea, may confirm an earlier study Oceanogrophy, 17, 229–242. by Beardsley et al. (1985) showing that the summer Liu, S., X. Shen, Y. Wang, and S. Han, 1992: Prelimi- fresh water discharge from the Yangtze River forms nary analysis of distribution and variation of peren- a relatively shallow, low salinity plume-like structure nial monthly mean water masses in the Bohai Sea, the 20 SEASONAL VARIABILITY OF THE YELLOW SEA/EAST CHINA SEA VOL. 22

Huanghai (Yellow) Sea and the East China Sea. Acta raphy: An Introduction. Daya Publishing House, Ocearologica Sinica, 11, 483–498. Delhi, India, 516pp. Su, Y.-S., and X.-C. Weng, 1994: Water masses in China van Loon, H., 1984: Climates of the Oceans. World Survey Seas. Vol. 1, Oceanology of China Seas, Zhou Di et of Climatology, 15, 453–458. al., Eds., Kluwer Academic Publishers, Boston, 3–16. Watts, I. E. M., 1969: Climates of China and Korea. World Takahashi, S., and T. Yanagi, 1995: A numerical study on Survey of Climatology, 8, 1–117. the formation of circulations in the Yellow Sea during Yanagi, T., and S. Takahashi, 1993: Seasonal variation of summer. La mer, 33, 135–147. circulations in the East China Sea and the Yellow Sea. Tomczack, M., and J. S. Godfrey, 1994: Regional Oceanog- J. Oceanogr., 49, 503–520.