Spatial variability in annual sea level variations around the Korean peninsula

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 35, L03603, doi: /2007gl032527, 2008 Spatial variability in annual sea level variations around the Korean peninsula Sok Kuh Kang, 1 Josef Y. Cherniawsky, 2 Michael G. G. Foreman, 2 Jae-Kwi So, 1 and Sang Ryong Lee 3 Received 1 November 2007; revised 4 December 2007; accepted 28 December 2007; published 2 February [1] Analyses of annual sea level variations (Sa) around the Korean peninsula reveal systematic spatial patterns in the phase lags and amplitudes. Sa amplitudes in this region vary between 10 and 20 cm, with atmospheric pressure accounting for only 7.7 to 9.4 cm. These patterns are dominated by annual changes in thermosteric sea levels in the Yellow Sea (YS) and East/Japan Sea (EJS), with only minor contributions from seasonal wind-driven set up. Citation: Kang, S. K., J. Y. Cherniawsky, M. G. G. Foreman, J.-K. So, and S. R. Lee (2008), Spatial variability in annual sea level variations around the Korean peninsula, Geophys. Res. Lett., 35, L03603, doi: /2007gl Introduction [2] An analysis of sea level data at coastal tidal stations around the Korean peninsula (KP) (Figure 1) reveals that amplitudes of the annual component Sa vary from 20 cm at the YS stations to 10 cm at the EJS stations. Variations in Sa in the YS and EJS were previously examined in a simple analytical manner by Kang and Lee [1985]. The global distribution of Sa was examined using coastal tidal data and found to be comprised of large-scale features superimposed upon a complexity of regional spatial variability [Tsimplis and Woodworth, 1994]. In relation to climate variability, the temporal variability of Sa has also been investigated in the North Sea and the Baltic [Plag and Tsimplis, 1999]. However, there have been no previous detailed studies with regard to Sa around the Korean peninsula. [3] The equilibrium tidal amplitude of Sa is much less than the observed amplitudes of 12 to 20 cm, implying that the major driving force is not of astronomical tidal origin. Thus, this spatial variability in Sa must be related to seasonal changes in temperature and salinity, and river runoff, as well as other atmospheric forcing, such as sea level pressure (SLP) and wind-stress. [4] In fact, each of the seas surrounding the KP has their own characteristic bottom topographic and oceanic features. The YS is shallow and tidal current is dominant, while the EJS is quite deep and has several oceanic currents (Figure 1). Specifically, the Tsushima Warm Current (TWC) enters EJS through Korea/Tsusima Strait (KTS) [Naganuma, 1977] and then, branches into the East Korea Warm Current 1 Korea Ocean Research and Development Institute, Seoul, Korea. 2 Institute of Ocean Sciences, Fisheries and Oceans Canada, Sidney, British Columbia, Canada. 3 Department of Marine Science, Pusan National University, Pusan, Korea. Copyright 2008 by the American Geophysical Union /08/2007GL (EKWC) and a nearshore Tsusima branch (NB) along the coast of Japan (Figure 1). [5] In this study, we examine the driving mechanisms behind the regionally-dependent, and rather systematic features of the Sa in the YS, northeastern ECS, and the EJS. In particular, spatial variations in the thermosteric sea level (TSL) are investigated and compared with atmospherically induced sea level (ASL) change and wind induced sea level (WSL). WSL was examined using a depth averaged 2D-numerical model with explicit calculation of Sa by wind set up. 2. Data Sources and Processing [6] Tidal data were collected at 23 permanent Korean, 5 Chinese and 5 Japanese tidal stations, while meteorological data were collected at representative stations in each sea (Figure 1). Each year of tidal data (over 20 years of data were available at each station), were subjected to harmonic analysis [Foreman, 1977], yielding 61 average harmonic constants, including Sa. In addition, sea levels computed through the inverted barometric relation from anomalies of SLP were also subject to harmonic analysis, yielding averaged harmonic constants for the ASL. [7] TSL is the variation of mean sea level due to changes in water-column temperature. Over 40 years of temperature data at hydrographic stations in the YS and EJS are available from the Korean Ocean Data Center (KODC) ( and the Japan Ocean Data Center (JODC) ( Tidal data along the Korean and Japanese coasts are available from same sources while those along the Chinese coast are available from the Hawaii sea level center ( data set. TSLs were computed following the method of Tabata et al. [1986], and as shown by Kang et al. [2005]. These were computed at 147 KODC stations and 34 JODC stations around KP (Figure 1). KODC data were sampled every 2 months, or longer, while JODC data had more irregular sampling intervals. The TSLs were integrated down to depths less than 100 m in the shallow YS, down to 100 or 125 m in the northeastern ECS, and mostly down to 300 m in the EJS. Halosteric sea levels were not computed since salinity data are generally not of sufficiently good quality, as described by Kang et al. [2005]. [8] Wind stress data from NOAA NCEP2 ( were used to force a finite-difference numerical ocean model for estimating the contributions to Sa due to wind set up. Ten years ( ) of monthly NCEP2 wind L of7

2 Figure 1. Locations of tidal (large circles) and meteorological (triangles) stations around Korean peninsula and of hydrographic stations of KODC (squares) and JODC (small circles). Depth contour of 2000 m is shown in the EJS. Schematic summer surface currents [after Naganuma, 1977] for the EJS are marked for North Korea Cold Current (NKCC), nearshore branch (NB) and Korean branch (EKWC) of Tsusima Warm Current. data with spatial resolution were interpolated to the model grid. [9] Almost 10 years (Sep. 23, 1992 to Aug. 11, 2002) of 10-day sampled sea level data from the TOPEX/Poseidon (T/P) altimeter were also analyzed to provide the Sa harmonic values around the KP. The harmonic analysis method and the altimeter data were described by Cherniawsky et al. [2001] and these data were previously corrected for the inverse barometer effect. 3. Results 3.1. Spatial Pattern of the Sa Component in the Seas Around the Korean Peninsula [10] Amplitudes and phase lags for the Sa at tidal stations around KP, from harmonic analyses over periods as short as 7 years at PT to 45 years at MH (Figure 1), are presented in Figure 2a. The amplitudes range from 10 to 20 cm, while the phase lags are between 210 and 230. These phases are with respect to January 1, so a value of 210 means a maximum value at August 1. The amplitudes generally decrease from the stations such as IC in the YS toward SC in the EJS through the stations JJ to US in the KTS. Phase lags increase from the YS stations to the stations in the EJS. The Sa equilibrium tidal amplitude at 40 N is about 0.4 mm, much less than the observed amplitude. So the driving forces of the Sa component are not of astronomical tidal origin. The Sa tidal amplitudes (Figure 2b) along the Chinese coast show larger values moving northward, while amplitudes at Japanese coast are relatively uniform. The leading additional candidates are investigated separately in the following sections Atmospherically Induced Sea Level Contribution to Sa Amplitude [11] Meteorological data at three stations around the KP (Figure 1) were collected as representative. This is justifiable as SLP values do not change much over the peninsula. In order to examine the contribution to Sa from SLP, pressure anomalies were calculated by subtracting the observed SLP values from mean pressures. Using the inverse barometric relationship of 1 mb being equivalent to 1 cm, these anomalies were converted into ASL (Figure 3a), and subjected to harmonic analysis. [12] The mean Sa amplitude of the ASL is 9.4 cm at IC over the 10 years, while the amplitudes at YU and TH are 8.5 cm and 7.7 cm, respectively. As the total Sa amplitudes at Korean tidal stations range between 12 and 20 cm, other contributions are required to make up the difference. [13] The phase lags of the Sa components in the tidal data range from 210 in the YS stations to 225 in the EJS stations and to nearly 220 at the KTS stations. On the other hand, the ASL phase lags at IC, YU, and TH are 184.0, 186.2, and 180.6, respectively, implying that minimum 2of7

3 Figure 2. Spatial variations of the Sa harmonic constants (a) at tidal stations around the Korean peninsula and (b) at Chinese and Japanese stations (see Figure 1). Phases are with respect to January 1, with 184 denoting July 1. Vertical bars show standard deviations. 3of7

4 Figure 3. (a) Atmospheric sea levels (ASL) computed by the inverse barometric relation of SLP anomalies (observed minus mean) at three stations IC, YU, and TH. (b) Mean Sa amplitudes and phase lags (white contours) of wind-driven sea level computed by a model forced with a 10 years (1990 to 1999) of wind data. 4of7

5 Figure 4. Contours of Sa (a) amplitude (in cm) and (b) phase lags (degrees), from harmonic analysis of TSLs. SLP under the East Asia Monsoon takes place in late June or early July. Such differences between the total Sa and ASL phase lags indicate again that the remaining contributions to Sa delay the ASL signals. [14] Wind-driven effects were estimated separately using a 10-year run of a 2D barotropic numerical model, forced with 1990 to 1999 monthly wind stress from NCEP2 reanalysis. The modeled 10-year mean Sa amplitudes and phase lags of the WSL are shown in Figure 3b. WSL Sa amplitudes in the central and eastern YS range between 1.0 and 2.5 cm, while in the EJS they are between 1.0 and 1.5 cm. Sa amplitude along the southern Chinese coast are relatively large, up to 3 6 cm. Phase value of WSL in the central and EJS are about 180, similar to those of ASL. The Sa amplitudes in our region of interest are relatively small, compared with those of ASLs, with larger values at southern Chinese stations Contribution to Sa From TSL and Comparison to T/P Analysis Results [15] The TSLs were computed at the hydrographic stations (Figure 1) around the KP, as in the work of Kang et al. [2005]. The annual cycle of TSL at the most western stations in the YS is rather regular like a sinusoidal curve, with the highest TSL values occurring during the summer and lowest values in the winter. However, the TSLs at the southwestern stations of the EJS are less regular, implying that the mechanism for TSL changes in the EJS differs from that in the YS. [16] The Sa co-amplitude and co-phase contours from tidal analyses of TSL are shown in Figures 4a and 4b, while those from IB-corrected T/P altimeter are in Figures 5a and 5b. The TSL amplitudes in the YS (Figure 4a) are spatially homogeneous, about 6 cm, except for smaller values in the shallow coastal regions (depths of 10 to 20 m). The larger amplitudes correspond to phases of about 240 (Figures 4b and 5b), indicating a late August maximum heat content in the YS region. This spatial homogeneity in the phase lags (240 ) and amplitudes (6 cm) in the central YS indicates a nearly homogeneous thermosteric signal due to seasonal solar heating. [17] Compared to the YS, phase lags of TSLs in the EJS and northeastern ECS are larger, with values up to 270 or so (Figure 4b). Amplitudes in the southwestern boundary of the EJS and in the KTS are larger, by up to more than 10 cm in the region of the Tsusima Current (TC), reflecting the amplitude structure of a geostrophic current during summer season, with higher amplitudes along the Japanese coast of the KTS. Since the peak TSL values in the EJS occur during late September (270 ), this annual change in heat content is not due to solar heating. As Sa amplitudes along the southeastern EJS (Figure 4a) show a pattern like that of the TC shown in Figure 1, the corresponding phase lags of TSLs are expected to reflect the phase lags of the TC. Therefore, the heating mechanism in the southwestern EJS may be due to the advection of warm water by the branches of TC from the Northwest Pacific Ocean. Furthermore, the phase lags off the southwestern Korean coast of the EJS show a more delayed peak of TSL than the phase lags in other EJS regions. This may be related to the effects of colder water masses, such as NKCC, penetrating into the southwestern EJS from the north (Figure 1). [18] The spatial distribution of the TSL component of Sa (Figures 4a and 4b) compares well with results obtained through the T/P data analysis (Figures 5a and 5b). Amplitudes are about 6 cm in the central YS and about 4 cm in the southwestern Korean stations of the EJS, while phase lag patterns are similar in the central YS and southwestern EJS. The Sa amplitude errors from T/P analyses, computed as in the work of Cherniawsky et al. [2001], are typically of the order of 1 cm or less over most of this area, except for the northern YS (Pohai Sea) and the Kuroshio extension, where they increase to 3 4 cm due to very strong sea level variability and an irregular seasonal cycle. [19] The smaller Sa amplitudes from TSL analysis along the southeastern Korean coast of the EJS match those from T/P analysis results well, though there is some discrepancy in the southeastern Yellow Sea (around Jeju), probably due to halosteric effects from the fresh water discharge of the Yangtze River, which are localized near the Chinese coast. 5of7

6 Figure 5. Sa (a) amplitudes and (b) phases from analysis of TOPEX/Poseidon sea-level data corrected for the inversebarometer effect. 6of7

7 larger Sa phase lag (225 ) at EJS tidal stations than the phase lag (210 ) at YS stations (Figure 2) appears to be due to a delayed heat transport by the TC from the Pacific Ocean and an additional contribution of the NKWC which flows south in the summer into the southwestern EJS. Figure 6. Complex superposition of ASL, WSL, and TSL of Sa at the YS station (IC) (black) and at the EJS station (MH) (red). Their combined (SUM) values agree to within 2 cm and 10 with the observations (OBS) in these areas. The contributions to Sa due to this river runoff appear to be minor over most of the adjacent seas, considering the amplitudes of 7 to 11 cm and their spatial pattern in the ECS (Figure 5a) Combination of TSL, ASL, and WSL Contributions to Sa [20] Adding the ASL, TSL, and WSL contributions at IC yields the Sa value of 16.5exp(204 i) cm (in complex form), which is relatively close to the observed 18.1exp(211 i) cm, as is displayed in Figure 6. Similarly, at MH in the EJS, the addition of ASL, WSL and TSL effects yields 9.9exp(210 i), which approaches the observed 11.3exp(222 i). Figure 6 thus shows that the annual variations observed in two areas in the YS and EJS are explained reasonably well by combining the corresponding Sa due to TSL, WSL and ASL, with a minor contribution from WSL. Therefore, the annual cycle of observed sea levels mainly consists of an atmospheric contribution, which peaks in late June or early July in the YS and EJS, combined with a two-month delayed thermosteric forcing in the YS, which is due to solar heating, or a 3-month delayed heating due to advection by ocean currents in the EJS and KTS. [21] The systematic amplitude decrease at Korean tidal stations in the EJS is mainly due to smaller TSL amplitudes or inhomogeneous TSLs, with phase lags larger than the 240 values in the YS (Figures 4b and 5b). The ASL and wind set up effects contribute little in the EJS. The relatively 4. Conclusions [22] An investigation of annual sea levels (Sa) in seas surrounding the KP reveals systematic patterns in phase lag and amplitudes varying from 20 to 10 cm that can be largely explained by atmospheric pressure and thermal forcing mechanisms. The systematic Sa variability is driven by the TSL features in the YS and EJS. Wind set up and river runoff contribute little around the Korean coast. The general agreement between TSL and satellite altimeter observations of Sa also supports this view. [23] Acknowledgments. We thank the two anonymous reviewers for valuable suggestions. We are grateful to Brian Beckley and Richard Ray for T/P altimeter data. We also thank M. K. Kim and E. J. Kim for drafting some of the figures. This work has been supported by KORDI program (PG46000, PE97604). References Cherniawsky, J. Y., M. G. G. Foreman, W. R. Crawford, and R. F. Henry (2001), Ocean tides from TOPEX/Poseidon sea level data, J. Atmos. Oceanic Technol., 18, Foreman, M. G. G. (1977), Manual for tidal analysis and prediction, Pac. Mar. Sci. Rep , 66 pp., Inst. of Ocean Sci., Sidney, B. C., Canada. Kang, Y. Q., and B. D. Lee (1985), On the annual variation of mean sea level along the coast of Korea, J. Oceanol. Soc. Korea, 20, Kang, S. K., J. Y. Cherniawsky, M. G. G. Foreman, H. S. Min, C.-H. Kim, and H.-W. Kang (2005), Patterns of recent sea level rise in the East/Japan Sea from satellite altimetry and in situ data, J. Geophys. Res., 110, C07002, doi: /2004jc Naganuma, K. (1977), The oceanographic fluctuations in the Japan Sea (in Japanese with English abstract), Mar. Sci., 9, Plag, H.-P., and M. N. Tsimplis (1999), Temporal variability of the seasonal sea-level cycle in the North Sea and Baltic Sea in relation to climate variability, Global Planet. Change, 20, Tabata, S., B. Thomas, and D. Ramsden (1986), Annual and interannual variability of steric sea level along line P in the northeast Pacific Ocean, J. Phys. Oceanogr., 16, Tsimplis, M. N., and P. L. Woodworth (1994), The global distribution of the seasonal sea level cycle calculated from coastal tide gauge data, J. Geophys. Res., 99(C8), 16,031 16,039. J. Y. Cherniawsky and M. G. G. Foreman, Institute of Ocean Sciences, Fisheries and Oceans Canada, Sidney, BC, Canada V8L 4B2. S. K. Kang and J.-K. So, Korea Ocean Research and Development Institute, Ansan, P.O. Box 29, Seoul , Korea. S. R. Lee, Department of Marine Science, Pusan National University, Pusan , Korea. 7of7