Sea level variability in East China Sea and its response to ENSO

Jun-cheng ZUO, Qian-qian HE*, Chang-lin CHEN, Mei-xiang CHEN, Qing XU

1. College of Harbor, Coastal and Offshore Engineering, Hohai University, Nanjing 210098, P. R. China

2. HYDROCHINA Huadong Engineering Corporation, Hangzhou 310014, P. R. China

3. State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, Hangzhou 310012, P. R. China

Sea level variability in East China Sea and its response to ENSO

Jun-cheng ZUO1, Qian-qian HE*2, Chang-lin CHEN3, Mei-xiang CHEN1, Qing XU1

1. College of Harbor, Coastal and Offshore Engineering, Hohai University, Nanjing 210098, P. R. China

2. HYDROCHINA Huadong Engineering Corporation, Hangzhou 310014, P. R. China

3. State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, Hangzhou 310012, P. R. China

Sea level variability in the East China Sea (ECS) was examined based primarily on the analysis of TOPEX/Poseidon altimetry data and tide gauge data as well as numerical simulation with the Princeton ocean model (POM). It is concluded that the inter-annual sea level variation in the ECS is negatively correlated with the ENSO index, and that the impact is more apparent in the southern area than in the northern area. Both data analysis and numerical model results also show that the sea level was lower during the typical El Ni?o period of 1997 to 1998. El Ni?o also causes the decrease of the annual sea level variation range in the ECS. This phenomenon is especially evident in the southern ECS. The impacts of wind stress and ocean circulation on the sea level variation in the ECS are also discussed in this paper. It is found that the wind stress most strongly affecting the sea level was in the directions of 70o and 20o south of east, respectively, over the northern and southern areas of the ECS. The northwest wind is particularly strong when El Ni?o occurs, and sea water is transported southeastward, which lowers the sea level in the southern ECS. The sea level variation in the southern ECS is also significantly affected by the strengthening of the Kuroshio. During the strengthening period of the Kuroshio, the sea level in the ECS usually drops, while the sea level rises when the Kuroshio weakens.

East China Sea; sea level variation; ENSO

1 Introduction

ENSO is considered the strongest ocean and atmosphere inter-annual variability signal, and it can cause global ocean and climate changes. Research shows that sea level variations in China’s coastal sea are closely related to the occurrence of strong ENSO events (Liu et al. 1989; Zuo et al. 1994; Yang et al. 2004). Most early research on this issue has been carried out using only tide gauge data. It showed that during El Ni?o events, the sea level along the coast of China was lower than those in normal years, and the negative anomaly of the water levelexpanded from south to north (Yu 1985; Li et al. 1994), with the sea level dropping a little in the northern area as compared to the southern area (Yu 1985). The sea level of the Yangtze Estuary in El Ni?o years dropped within a range of 9 to 10 cm (Chen et al. 1991). On the southeast coast of China, the annual variation range of the monthly mean sea level became smaller in El Ni?o years (Li 1987). Qiao and Chen (2008) found that the sea level in the East China Sea (ECS) started to decline in 1997, began to rise in early 1998, and attained the maximum in 1999 during the period of 1992 to 2004, and that sea level anomalies (SLAs) in the Yellow Sea and the East Sea showed remarkably stronger responses to ENSO signals than they did in the Bohai Sea based on analysis of satellite altimetry data. Low-frequency components of SLAs in the East Sea are closely related to the southern oscillation index (SOI), and the sea level in the East Sea is dominated by El Ni?o events, but in the northern and southern areas of the East Sea (the 30°N line marks the boundary between the two areas), the responses of sea level variations to SOI are opposite (Liu et al. 2009).

ENSO events affect the sea level variation in the ECS in different ways: climate anomalies (wind stress), ocean waves (Kelvin waves, shelf waves, etc.), the circulation transport, inverse barometer effects, etc., but they do not result in very common results. Zuo et al. (1994) found that El Ni?o plays a role in maintaining the balance of the coastal sea level, and that the changes of the Kuroshio transport also have an impact on the sea level variability in the ECS. Liu et al. (1989) found that during ENSO events, subtropical high over the northern Pacific strengthens the southwest wind, which may be one of the reasons for the decrease of the sea level along the coast. The inter-annual sea level variability in the ECS is closely related to the North Pacific circulation variability (Yamagata et al. 1985; Zhen et al. 1992; Li et al. 1994).

Sea level variation characteristics and its response mechanisms in the ECS during ENSO events need to be further studied. Previous research mainly focused on the analysis of tide gauge and satellite altimetry data, while few numerical simulations were conducted. In this study, the Princeton ocean model (POM) was combined with statistical analysis to study sea level variation characteristics and patterns in the ECS as well as the mechanisms of its response to ENSO events.

2 Data and methods

In this study, satellite altimetry data from maps of sea level anomaly (MSLA) data sets and tide gauge data along the coast of China and over the adjacent shelf seas were used. MSLA was produced by Aviso based on TOPEX/Poseidon, Jason 1, ERS-1, and ERS-2 data. The length of altimetry data was 16 years from 1993 to 2008, and the data consisted of maps produced every 7 d on a 1/3o × 1/3o Mercator grid. Monthly mean sea level data from seven tide gauge stations (Fig. 1) were obtained from the National Marine Data and Information Service of State Oceanic Administration of China. The data duration varied significantly fromstation to station and was usually a record of more than 30 years, starting as early as the 1950s and ending as late as 2008. The monthly mean sea surface temperature (SST) from two stations, the Lianyungang and Xiamen stations, came from the East Sea Information Centre of the State Oceanic Administration of the People’s Republic of China. The multivariate ENSO index (MEI) from the Earth System Research Laboratory of National Oceanic and Atmospheric Administration (NOAA ESRL) of the United States was selected as the ENSO index. SOI from the database of the NOAA Climate Diagnostics Center was also used in this study.

Numerical simulation was performed using POM (Blumberg and Mellor 1987), and the non-uniform Mercator grid (Fig. 2) was adopted in this study. The simulation domain was from 95oE to 60oW and 20oS to 65oN in the Pacific Ocean. For the ECS, with the range from 113oE to 132oE and 22oN to 42oN, which covers the Bohai Sea, the Yellow Sea, and the East Sea, a grid resolution of 0.25o × 0.25o was adopted, so as to obtain precise solution to the ECS.

Fig. 1 Spatial distribution of tide gauge stations (Number in bracket indicates time duration, Unit: year)

Fig. 2 Model grid

The southern and northern boundaries of the simulated domain were located at 20oS and65oN, respectively. The mean wind stress curl at the 20oS line was zero. We considered that the meridional mass transport across the zonal direction was zero (Liu et al. 2001), and the whole simulation domain had closed boundaries. Lateral boundary conditions were no-slip without heat and salt fluxes. The vertical direction was divided into 16 sigma levels.

Time steps for the internal and external modes were 120 s and 6 s, respectively. The model topography data were ETOP05, which were obtained from the National Geophysical Data Center, USA, with a resolution of 1/12o. The depth of the ECS was replaced with data from the Navigation Assurance Ministry of the Chinese Navy Headquarters, and the data were interpolated to the model grid to provide a more accurate topography of the ECS.

The simulation began from a state of rest, with the initial elevation and currents simply set to zero. The climatology data of Levitus et al. (2005) in December were taken as the initial three-dimensional temperature and salt fields, and interpolated to the model grid with the optimal interpolation method. Model surface forces included the wind stress, heat flux, and freshwater flux. Daily mean wind data with a resolution of 1.8o × 1.8o from 1983 to 2008 were obtained from database of the National Centers for Environmental Prediction (NCEP), USA. The surface temperature and salinity were restored values from SST and sea surface salinity (SSS) data of Ishii et al. (2006), and the restoration time scale was 30 d. The POM model run from 1983 to 2008, and monthly mean sea levels from 1993 to 2008 were analyzed in this study.

3 Inter-annual sea level variability in ECS

In order to study the relationship between the sea level variation in the ECS and ENSO events at the inter-annual time scale, the simulated results of the mean sea level were subtracted from the original time series to obtain SLAs, and then annual and semiannual cycles were removed with the least squares analysis. After that, seasonal mean SLA (January through March for winter, April through June for spring, July through September for summer, and October through December for fall) were calculated. Finally, the linear trend was also removed with the least squares regression method. Altimetry data and tide gauge data were processed in the same way as the simulated results.

3.1 Mean SLA in ECS

Fig. 3 shows that the simulated SLA is consistent with the altimetry data, and the correlation coefficient is 0.76. Both the simulated result and altimetry data showed that the sea level was low in the period from 1995 to 1996 and high in 1999. The range of the sea level variation was about 10 cm, and that of the simulated result was a little larger. This result is the same as that of Han and Huang (2008). As Fig. 3 shows, there is a negative correlation between inter-annual sea level variation in the ECS and the ENSO index: the in-phase correlation coefficient is less than –0.1, and the maximum correlation coefficient is –0.4(significantly at the confidence level of 90%), when the variation of SLA lags eight months behind the ENSO index. The sea level was low for most of those years with a more significant positive phase of ENSO, indicating that when strong El Ni?o events occur, the sea level in the ECS is generally lower than those in normal years. Prior to 2002, the ENSO signal was stronger with larger amplitudes, and its negative correlation with the sea level variation was more obvious than that after 2002.

Fig. 3 Comparison of simulated SLA and altimetry SLA in ECS

3.2 Mean SLAs in northern and southern ECS areas

SLAs for the northern and southern ECS areas were computed. The sea area from 30°N to 38°N and 117°E to 131°E was selected to represent the northern ECS area, while the area from 23°N to 30°N and 117°E to 131°E represent the southern ECS area. Fig. 4 shows that the sea level is sensitive to strong ENSO events, and that the inter-annual sea level variations in both areas are negatively correlated with the ENSO index. The impact is more apparent in the southern ECS area than in the northern ECS area.

Fig. 4 Comparison of simulated SLAs and ENSO index in northern and southern ECS areas

By comparing seasonal mean SLAs at the Qinhuangdao, Lüsi, and Xiamen stations varying with the ENSO index (Fig. 5), it can be seen that sea level variations during ENSO events recorded by different tide gauge series are of different types and intensities, showingthe sea level variation responses to ENSO events are remarkably stronger in low latitude areas (Xiamen Station) than in medium-high latitude areas (Qinhuangdao Station). This conclusion is consistent with the simulated result. For the study period, the maximum correlation coefficients of SLA with the ENSO index were –0.32 for the Xiamen Station (SLA with a one-month lag), –0.22 for the Lianyungang Station (SLA with a seven-month lag), and –0.20 for the Qinhuangdao Station (SLA with a ten-month lag). The sea level at the Xiamen Station was the lowest in 1997 during the El Ni?o event, indicating that the sea level of the southern ECS area affected by ENSO events was more evident.

Fig. 5 SLAs at three tide gauge stations and ENSO index

4 Seasonal sea level variations in ECS during El Ni?o events

In this study, the Lianyungang and Xiamen stations were chosen for investigation of seasonal sea level variations in the ECS during El Ni?o events because both the monthly mean SST data and sea level data could be obtained. Sea areas near Lianyungang and Xiamen can represent the northern and southern ECS areas, respectively. We focused on seasonal sea level variations in the ECS when the strong and typical El Ni?o event happened in the period of 1997 to 1998.

4.1 Sea area near Lianyungang

Both the simulated results and tide gauge data indicate that the sea level in the sea area near Lianyungang was lower in the summer of 1996 and higher in the winter of 1997 than those in normal years (Fig. 6). These phenomena were more notable as seen from tide gauge observed results. They also show that the annual variation range of the sea level was smaller in 1997. But SST of the Lianyungang Station features the normal seasonal variation, which can be seen from SST anomaly (SSTA) variation in Fig. 6(b).

Fig. 6 Monthly mean SLAs and SSTA at Lianyungang Station from 1996 to 1999

4.2 Sea area near Xiamen

Both the simulated results and tide gauge observed data show that the peak sea level in the sea area near Xiamen was lower in 1997, and the annual variation range of the sea level at the Xiamen Station was smaller in 1997 than those in normal years (Fig. 7). The lowest sea level often appears when the corresponding ENSO index comes to the maximum positive value. What’s more, the normal SST variations at the Xiamen and Lianyungang stations show that temperature anomalies may not be the major factor causing the sea level variation, but circulation and wind stress may play important roles in the sea level variation during ENSO events.

Fig. 7 Monthly mean SLAs and SSTA at Xiamen Station from 1996 to 1999

In summary, during the typical El Ni?o period of 1997 to 1998, it is apparent that the annual variation range of the sea level was smaller, the peak sea level became lower (Xiamen Station), and the lowest sea level became higher (Lianyungang Station). It can be also inferred that sea level variation responses to El Ni?o signals are remarkably stronger in the southern ECS area than in the northern ECS area.

5 Possible mechanisms of sea level variations in ECS responding to ENSO events

The ECS is a shallow marginal sea of the northwest Pacific Ocean. In addition to local atmospheric forcing and freshwater runoff, the large-scale atmospheric and oceanic variability, such as the Pacific circulation and East Asian monsoon, may impact the seasonal andlonger-term hydrography and circulation variability in this region. In this study, we tried to explain sea level variation mechanisms responding to ENSO events through analysis of the wind stress and ocean current transport.

5.1 Effect of wind stress

The ECS area stretches across the 30°N line in the meridional direction, which is the boundary of the atmospheric circulation divergence zones. Different wind fields on each side may have different impacts on the sea level variability. We picked up the simulated SLA cycle of two to seven years with a low-pass filter. The west-east wind stress anomaly component (Sx) and north-south component (Sy) were calculated from model surface wind force (NECP wind data). LaterSx,Sy, and SOI were treated in the same way as SLA. We tried to calculate the correlation coefficient between wind stress anomalies and SLA and to explore the ways that ENSO events affected the sea level in the ECS through wind stresses. This method is similar to that of Liu et al. (2009).

SOI was obtained from the NOAA Climate Diagnostics Center database. The negative low-frequency component of SOI indicates the occurrence of El Ni?o events, while the positive SOI is due to La Ni?a events. The sea level variations responding to wind stresses from different directions are notably different. In order to find the most optimal response direction, we made a rotation of the coordinate system.

Based on the correlation analysis between SLA and wind stress anomalies (Sx′) in the new coordinate system, it was found that the wind stress most strongly affecting the sea level was in the direction of 70° south of east in the northern ECS area. SLA was negatively correlated withSx′ from 1994 to 2001, and after 2002 they changed in-phase and their relationship was positively correlated (Fig. 8(a)). In the southern ECS area, the wind stress in the direction of 20° south of east had the most significant impact on the sea level. As shown in Fig. 8(b), SLA always maintained a negative correlation withSx′ during the entire study period. During the typical El Ni?o period of 1997 to 1998,Sx′ reached its maximum, indicating that the northwest wind was particularly strong. This caused a large volume of sea water to flow southeastward into the Pacific Ocean, lowering the sea level in the southern ECS area. Thus, the lower sea level occurring in the southern ECS area during El Ni?o events can be explained.

Fig. 8 Low-frequency components of simulated SLAs and NCEPSx′ for northern and southern ECS areas

A conclusion can be drawn that the zonal wind stress affects the sea level significantly, indicating that ENSO impacts the wind field of the ECS by means of atmospheric circulation, hence affecting the sea level. In the northern ECS, the relationship betweenSx′ and SOI is positively correlated, whileSx′ is negatively correlated with SOI and SLA in the southern ECS (Fig. 9). The low-frequency component of SOI came to a minimum in the period of 1997 to 1998 when the strongest El Ni?o event occurred. In the northern ECS, the southeast wind was very strong (Sx′ had a negative value in the direction of 70° south of east), and it caused a large volume of sea water to flow northwestward. The shoreward flowing water was blocked and accumulated when it met the shore land barrier, playing a role in the compensation for the drop of the coastal sea level in the northern ECS. Conversely, the northwest wind (Sx′ had a positive value in the direction of 20° south of east) took abundant sea water southeastward flowing into the Pacific Ocean and lowered the sea level in the southern ECS. The reason that the sea level is lower in El Ni?o years and the phenomenon is more apparent in the southern ECS can be explained. Furthermore, the coastal SLA changes are not only caused by the wind stress, but also caused by the wind stress curl in terms of Ekman pumping (Wang et al. 2002), which needs further study.

Fig. 9 Low-frequency components of NCEPand SOI for northern and southern ECS areas

5.2 Effect of Kuroshio transport

Based on simulated monthly mean flow data, the Kuroshio transport through the PN section was calculated (Fig. 10). The transport volume was about 27.8 Sv on average from 1993 to 2000, which was similar to the observed results (Yuan et al. 2001, 2006; Ichikawa and Chaen 2000). The Kuroshio transport has significant seasonal variation: it is large in spring and summer and small in autumn and winter. Besides the annual cycle, the results also suggest a significant periodical variation of four years in the Kuroshio transport, which is probably associated with ENSO events.

Through analysis of SLA in the southern ECS and the Kuroshio transport through the PN section with a two-year low pass filter, it was found that the Kuroshio transport was negatively related with SLA (Fig. 11), and the maximum correlation coefficient was –0.7 when SLA had a three-month lag behind the Kuroshio transport through the PN section. From 1993 to 2001, when the Kuroshio transport was much larger than the normal value, SLA dropped evidently,and vice versa. On the inter-annual time scale, the Kuroshio transport may affect sea level variations in the ECS.

Fig. 10 Simulated monthly mean Kuroshio transport through PN section from 1993 to 2000

Fig. 11 Low-frequency components of SLA in southern ECS and Kuroshio transport through PN section

6 Conclusions

(1) Results show that the inter-annual sea level variability in the ECS is negatively correlated with the ENSO index. Both data analysis and simulated results showed that the ECS sea level was lower and its annual variation amplitude was much smaller during the typical El Ni?o period of 1997 to 1998. Sea level variation responses to El Ni?o signals are remarkably stronger in the southern ECS than in the northern ECS.

(2) ENSO impacts the wind stress field of the ECS by means of atmospheric circulation, hence affecting the sea level. During the El Ni?o period of 1997 to 1998, the northwest wind was very strong in the southern ECS. This caused a large volume of sea water to flow southeastward into the vast Pacific Ocean and lowered the sea level. Conversely, in the northern ECS, the southeast wind took the sea water northwestward to the shore, which was blocked and accumulated by the shore, compensating the decrease of the sea level during the El Ni?o event. This can explain why the sea level is lower in El Ni?o years and the phenomenon is more apparent in the southern ECS.

(3) The sea level variation in the southern ECS is also much affected by the strength of the Kuroshio. During the strengthening period of the Kuroshio, the sea level in the ECS usually drops, while the sea level rises when the Kuroshio weakens.

Blumberg, A. F., and Mellor, G. L. 1987. A description of a three-dimensional coastal ocean circulation model.Coastal and Estuarine Sciences, 4, 1-16. [doi:10.1029/CO004p0001]

Chen, Z. Y., Huang, Y. H., Zhou, T. H., Tang, E. X., Yu, Y. F., and Tian, H. 1991. A preliminary study on mean sea level of the Changjiang River Estuary.Oceanologia et Limnologia Sinica, 22(4), 315-320. (in Chinese)

Han, G. Q., and Huang, W. G. 2008. Pacific decadal oscillation and sea level variability in the Bohai, Yellow, and East China Seas.Journal of Physical Oceanography, 38(12), 2772-2783. [doi:10.1175/ 2008JPO3885.1]

Ichikawa, H., and Chaen, M. 2000. Seasonal variation of heat and freshwater transports by the Kuroshio in theEast China Sea.Journal of Marine Systems, 24(1-2), 119-129.

Ishii, M., Kimoto, M., Sakamoto, K., and Iwasaki, S. I. 2006. Steric sea level changes estimated from historical ocean subsurface temperature and salinity analysis.Journal of Oceanography, 62(2), 155-170. [doi:10.1007/s10872-006-0041-y]

Levitus, S., Antonov, J. I., and Boyer, T. P. 2005. Warming of the world ocean, 1955-2003.Geophysical Research Letters, 32, L02604. [doi:10.1029/ 2004GL021592]

Li, K. P., Fang, X. Y., Liu, L. H., and Zeng, X. M. 1994. The responds of sea level change to El Ni?o event.Journal of Oceanography of Huanghai and Bohai Seas, 12(2), 10-17. (in Chinese)

Li, L. 1987. Response of sea level along southeast China coast to El Ni?o.Journal of Oceanography in Taiwan Strait, 6(2), 132-138. (in Chinese)

Liu, Q. Y., Yang, H. J., Jia, Y. L., and Gan, Z. J. 2001. The numerical simulation of the seasonal variation of the sea surface height in the South China Sea.Acta Oceanologica Sinica, 23(2), 9-17. (in Chinese)

Liu, X. Y., Liu, Y. G., Guo, L., and Gu, Y. Z. 2009. Change of mean sea level of low-frequency on East China Sea and its relation with ENSO.Journal of Geodesy and Geodynamics, 29(4), 55-63. (in Chinese)

Liu, Z. P., Wang, Y. Z., and Guan, J. X. 1989. The features in the variability of monthly mean sea level and sea surface temperature in the coastal area of China during ENSO events.Marine Sciences, 4, 13-20. (in Chinese)

Qiao, X., and Chen, G. 2008. A preliminary analysis on the China sea level using 11 years TOPEX/Poseidon altimeter data.Marine Science, 32(1), 60-64. (in Chinese)

Wang, D. X., Xie, Q., Du, Y., Wang, W. Q., and Chen, J. 2002. The 1997-1998 warm event in the South China Sea.Chinese Science Bulletin, 47(14), 1221-1227.

Yamagata, T., Shibao, Y., and Umatani, S. 1985. Interannual variability of the Kuroshio extension and its relation to the Southern Oscillation/El Ni?o.Journal of Oceanography, 41(4), 274-281. [doi:10.1007/ BF02109276]

Yang, J., Lu, J. Z., Sha, W. Y., and Chen, X. 2004. The related analysis of the sea surface height abnormally (SSHA) near the China seas.Marine Forecasts, 21(2), 29-36. (in Chinese)

Yu, K. J. 1985. An analysis of mean sea level change along the eastern China coast.Oceanologia et Limnologia Sinica, 16(2), 127-137. (in Chinese)

Yuan, Y. C., Liu, Y. G., and Su, J. L. 2001. Variability of the Kuroshio in the East China Sea during El Ni?o to La Nina: A phenomenon of 1997 and 1998.Chinese Journal of Geophysics, 44(2), 199-210. (in Chinese)

Yuan, Y. C., Yang, C. H., and Wang, Z. G. 2006. Variability of the Kuroshio in the East China Sea and the currents east of Ryukyu Islands, I: Variability of the Kuroshio in the East China Sea and the meso-scale eddies near the Kuroshio in 2000.Acta Oceanologica Sinica, 28(2), 1-13. (in Chinese)

Zhen, W. Z., Yu, J. Y., and Niu, B. 1992. Sea level research in China.Marine Science Bulletin, 11(2), 68-72. (in Chinese)

Zuo, J. C., Yu, Y. F., and Chen, Z. Y. 1994. The analysis of sea level variation factor along China coast.Advance in Earth Sciences, 9(5), 48-53. (in Chinese)

(Edited by Ye SHI)

This work was supported by the National Basic Research Program of China (973 program, Grant No. 2007CB411807), the National Natural Science Foundation of China (Grants No. 40976006 and 40906002), the National Marine Public Welfare Research Project of China (Grant No. 201005019), and the Project of Key Laboratory of Coastal Disasters and Defense of Ministry of Education of China (Grant No. 200802).

*Corresponding author (e-mail:he_qq@ecidi.com)

Received May 30, 2011; accepted Nov. 6, 2011