Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1225-6692(Print)
ISSN : 2287-4518(Online)
Journal of the Korean earth science society Vol.40 No.5 pp.487-501
DOI : https://doi.org/10.5467/JKESS.2019.40.5.487

Steric Sea Level Variability in the East Asian Seas estimated from Ocean Reanalysis Intercomparison Project Data

You-Soon Chang*, Min-Ji Kang
Department of Earth Science Education, Kongju National University, Kongju 32588, Korea
Corresponding author: yschang@kongju.ac.kr Tel: +82-41-850-8292
July 15, 2019 August 19, 2019 October 24, 2019

Abstract


In this study, steric height variability in the East Asian Seas (EAS) has been analyzed by using ocean reanalysis intercomparison project (ORA-IP) data. Results show that there are significant correlations between ocean reanalysis and satellite data except the phase of annual cycle and interannual signals of the Yellow Sea. Reanalysis ensemble derived from 15-different assimilation systems depicts higher correlation (0.706) than objective analysis ensemble (0.296) in the EAS. This correlation coefficient is also much higher than that of the global ocean (0.441). For the longterm variability of the thermosteric sea level during 1993-2010, a significant warming trend is found in the East/Japan Sea, while cooling trend is shown around the Kuroshio extension area. For the halosteric sea level, a dominant freshening trend is found in the EAS. However, below 300 m depth around this area, the signal-to-noise ratio of the linear trend is generally less than one, which is related to the low density of observation data.



초록


    Introduction

    According to the assessment report version 5 (AR5) of the Intergovernmental Panel on Climate Change (IPCC), one of the major harmful factors to the human society is the sea level rise associated with the global warming. Annual mean sea level rise on a global scale is estimated as 1.7±2 mm/year for 1901- 2010 and 3.2±4 mm/year for 1993-2010 (IPCC, 2014). However, large uncertainties remain for the regional scale sea level budget (Chang et al., 2010; Chang, 2012).

    Sea level rise in the East Asian Seas (hereafter EAS) is very vulnerable to climate change by monsoons, El Nino, and other major climate signals (Pacific Decadal Oscillation (PDO), Arctic Oscillation (AO), and so on) (Lee et al., 2012;Chang and Shin, 2014). Therefore, it is very important to continuously investigate sea level rise associated with oceanic variability. While sea level change may have different causes, regional sea level rise is generally dominated by the steric component (Fukumori and Wang, 2013). Steric height is calculated by the density change associated with water temperature and salinity change. If we consider temperature (salinity) contribution only, it is called as thermosteric (halosteric) height (Tabata et al., 1986; Levitus et al., 2005).

    Objective analyzed (OA) and reanalysis (REA) data have been used to investigate the global or regional mean steric height change because oceanic observing system is very rare in the global ocean for the long time. OA data are interpolated products with in situ observations usually blended with climatology, or a persistence background fields (Ishii et al., 2006; Chang, 2012; Levitus et al., 2012). They are relatively simple and computationally less expensive, but they do not guarantee any dynamic process of the ocean and strongly depend on the correlation function during the interpolation process. OA data can also be unrealistically close to the climatology fields especially in the data sparse regions. An alternative method is the oceanic reanalysis based on the data assimilation model. However, REA data also have chronic limitation due to inaccurate numerical models associated with sensitive initial/boundary conditions, imperfect assimilation methods, and the nonlinearity of nature. Therefore, continuous efforts need to be necessary for the intercomparison of various OA and REA data.

    Recently, many attempts have been devoted to the assessment of OA and REA data, which have now reached some degree of maturity (Lee et al., 2009). With the same purpose, Ocean Reanalysis Intercomparison Project (ORA-IP) was successfully coordinated (Chang, 2015; Balmaseda et al., 2015). Important 8 different oceanic variables (steric height, heat content, sea surface height, surface heat flux, mixed layer depth, salinity change, 20 °C isothermal depth, sea ice content) have already been validated. However, very limited numbers of studies for the evaluation of OA or REA data around the EAS have been published (Chang, 2012; Seo et al., 2015; Han et al., 2016; Sim et al., 2018; Lee et al., 2018).

    Storoto et al. (2017) published a comprehensive evaluation of the steric sea level variability from ORA-IP data in the global ocean. They initially generated ensemble means of 16 REA and 4 OA, and compared them with a satellite-derived (altimetry minus gravimetry) steric height for a short period (2003-2010). Extended periods (1993-2010) were also analyzed for the interannual trends of global steric sea level. In this study, we intentionally follow their configuration by selecting same analysis method and periods expect for the study area, EAS. We expect that this consistent configuration with previous study leads to an effective comparison between global and regional characteristics of ORA-IP products.

    This study is organized as follows. Section 2 describes the data and methods used in this study, and results are followed in section 3. A comparison against reference steric sea level estimated from satellite data for a short period (2003-2010) is present. Another intercomparison is extended to the 1993-2010 period for analyzing long-term steric change. Finally, the main summary and conclusion are given in the last section 4.

    Data and Methods

    Ocean reanalysis intercomparison project

    ORA-IP was suggested and agreed at the Climate and Ocean Variability, Predictability, and Change (CLIVAR)/Global Synthesis and Observations Panel (GSOP) meeting held at Woods Hole Oceanographic Institution (WHOI) in 2012, and European Centre for Medium-Range Weather Forecasts (ECMWF) in 2013. There are two different working groups of the ORAIP. One is the data producing group and the other is the data processing group. Total 26 different products have participated this project. Six of them are objective analysis data, and twenty of them are assimilation data.

    Table 1 is the summarized information about product name, model information (base model, resolution, and assimilation method), and product institute for each ORA-IP data used in this study. It includes not only assimilation model data, but also OA data. Fifteen of them are REA including four coupled data assimilation systems (CFSR by NOAA/NCEP, ECDA by NOAA/ GFDL, GEOS5 by NASA/GMAO, and MOVEC by MRI/JMA).

    Validation data

    For the validation, this study used two satellite datasets representing total and mass component of sea level. The total sea level roughly equals to the summation of the sea level change by the density and mass. Each sea level component can be observed by different observation platforms (Cazenave and Llovel., 2010; Chang et al., 2010; Church et al., 2011; Chang, 2014). If we have two satellite platforms providing total and mass component of sea level, we can simply estimate satellite-based steric component by subtracting mass component from total component of sea level. This satellite-based steric data can be used to evaluate steric height based on temperature and salinity data of ORA-IP products. Therefore, this study used two satellite data (Archiving, Validation and Interpretation of Satellite Oceanographic data (AVISO) is for the total sea level, and Gravity Recovery and Climate Experiment (GRACE) Tellus is for the mass sea level).

    For the total sea level, satellite altimeter gridded data from the Jason-1 measurements distributed by AVISO have been used, that is the monthly mean delayed time version on 1/4° by 1/4° grid point. We converted them to 1° by 1° in the EAS from 2003 to 2010 for the comparison with ORA-IP data. For the mass component of sea level, we used monthly mean RL04 version on 1° by 1° gridded GRACE data provided by Center for Space Research, University of Texas (CSR) and GeoForschungsZentrum, the German Research Centre for Geosciences (GFZ). GRACE is a twin satellite system and it is used to make gravitational field measurement by estimating the changes in the distance between the twin satellites. However, it is known that mass variation in the ocean measured by the GRACE satellite is vulnerable to leakage of land hydrologic signals and their magnitudes are larger than the real ocean signals (Chambers, 2006; Chang et al., 2010;Leuliette and Willis, 2011). The GRACE Tellus used in this study made an effort to remove the hydrologic signals, but possible land contamination of oceanic signals may remain especially on the reginal scale. In this study, we do not address steric sea level uncertainties from the gravimetric data errors, as it is beyond the scope of this work. Two different datasets have been averaged in the EAS for 2003-2010. By subtracting gridded GRACE data from AVISO data (hereafter ALT-GRV), we can obtain satellite-based steric component for the validation of ORA-IP.

    Methods

    Steric sea level component (ηs) approximately consists of thermosteric (ηt) and halosteric (ηh) sea level components:

    η s η t  + η h = H 0 ρ ( T , S ) ρ 0 d z H 0 ρ ( T , S ) ρ 0 d z

    where the ρ is the ocean density that is the function of temperature (T) and salinity (S). Over-bar denotes the time-averaged climatology value, ρ0 is the reference density, and H is the ocean depth, respectively. Thermosteric is the derived sea level based on observed (or assimilated) T and climatological S, and halosteric is the derived sea level based on observed (or assimilated) S and climatological T. This study calculates both thermosteric and halosteric trends from surface to 300, 700, 1500, 3000 m, and bottom depth, respectively.

    In this study, we selected total 19 products of ORAIP data covering same periods from 1993-2010. This study defines the EAS from 24° N to 52 °N, and 115 °E to 143 °E, which is consistent to the regional World Ocean Atlas (WOA) area provided by National Oceanographic Data Center (NODC). We gathered gridded temperature and salinity profiles from global ORA-IP and converted them to a regional version for the EAS by regridding process based on 1° by 1° WOA grid. Finally, we estimated the monthly mean steric, thermosteric, and halosteric sea level in the EAS for 1993-2010. We also calculated ensemble mean of OA (REA) that is denoted as OAENS (REAENS).

    The strategy used in this work consist of two separate phases of steric sea level intercomparison, which span different periods: a first validation period (2003-2010) and extended comparison period (1993- 2010), which is followed by Storto et al. (2017). For the first validation period, we estimated temporal and spatial correlation coefficients with satellite data and checked the annual and interannual phase and amplitude. For 1993-2010 period, long-term trends of all product and their ensemble means (OAENS, and REAENS) are presented together with signal-to-noise ratio (SNR) indicating the uncertainty of regression result. SNR is defined as the ratio between ensemble mean and the ensemble spread (standard deviation) for the linear trend.

    Results

    Validation period (2003-2010)

    In this section, steric sea level anomalies obtained from ORA-IP averaged over the EAS are compared to satellite derived sea level for the 2003-2010. Seasonal cycle of each sea level component as a function of latitude is represented in Fig. 1. Top panel is the total sea level by AVISO satellite. Middle is the mass component by GRACE and the bottom indicates the steric component by ALT-GRV. Total sea level shows a strong seasonal change with maximum in September and minimum in March. Steric sea level is very similar to total sea level, but there is a little modification of phase and amplitude by the mass component. Mass sea level shows the maximum in September due to freshwater discharge in this area, but the amplitude is relatively small compared to that of total sea level. Tayler diagram shown in Fig. 2 represents mismatch between satellite and ORA-IP products. Black dot with black line denotes satellite-based steric height (ALT-GRV). The maximum peak appears around October as shown in Fig. 1(c), while most ORA-IP products have maximum peaks around September which is consistent with the maximum period of total sea level as shown in Fig. 1(a). It is most likely that the modification effect by the mass component was not fully considered in the most reanalysis data. It is worth to investigate which one is correct representing the real annual cycle of the EAS in the following studies because it has been known that there are many uncertainties for the mass component especially around coastal areas (Chambers, 2006; Chang et al., 2010;Leuliette and Willis, 2011).

    Time series of steric sea level around the EAS for the period of 2003 to 2010 are depicted in Fig. 3. All data have been smoothed by 6-months moving average after eliminating strong annual cycle in this area, which contained dominant interannual signals. Contrary to expectation that this area would be affected by global warming, most products except for GLOSEA5 and K7OC show no significant increase trends during the period of 2003-2010, which reflects the importance of regional climatology study. We will present detailed intercomparison results about linear trends of each product for the longer extended period (1993-2010) in the following chapter. Around 2008-2009, a noticeable decline signals are found in the most products. Unfortunately, this period is not directly related to the major climate signals such as strong ENSO, PDO, and AO. Further investigations would require to determine possible causes explaining decreasing trend of the most reanalysis products in this area during this period (2003-2010). In addition, we cannot find specific correspondence (small spreads) between similar products grouped by the same base model or assimilation method, e. g., OA group (ARMOR, CORA, EN3, IK09), coupled data assimilation group (CFSR, ECDA, GEOS5, MOVEC), and so on. MOM group (PEDAS, GODA, K7OC) shows the largest spread pattern, and it is related to model bias, resolution, and the difference of initialization period among each model (Palmer et el., 2017). The temporal correlation of each product with the satellite is also shown on the regional and global scale and detailed analysis is provided in Fig. 4.

    Figure 4 shows the same result as Fig. 3, but different display for the correlation coefficient of each product both in regional and global analysis results. Blue bar denotes correlation coefficient averaged only for the EAS and sky-blue bar is the global mean result denoted as GLO in Fig. 4 and Figs. 6-8. The interesting fact is that the correlation coefficient of REAENS is estimated to be about 0.706, which is higher than that of OAENS (0.296). The correlation coefficient of REAENS in the EAS is also much higher than that of the global mean (0.441). This comparison clearly shows the importance of the REAENS especially in this area. Unfortunately, there is no significant difference among different assimilation methods. For example, no improvement is found in the coupled data assimilation products such as CFSR, ECDA, GEOS5, and MOVEC.

    Point-by-point spatial correlation map between ensemble means (REAENS and OAENS) and the satellite observation has been calculated in Fig. 5. Difference between the correlation coefficient of REAENS and OAENS are also compared. Left panels of Fig. 5 show full signal including strong annual cycle and right panels indicate the interannual time scale only after eliminating the seasonal cycle. Most regions show significant high correlations with satellite except for the interannual variation of the Yellow Sea, which is related to the low density of observation data. Due to the intense level of fishing and trawling in the shallow Yellow Sea, there are very little temperature and salinity observed profiles including Argo for the data assimilation and OA in the Yellow Sea (Lie and Cho, 2016). Most areas of the difference map have positive values indicating that steric height from the REA products (mean correlation coefficient of full signal: 0.834, interannual signal: 0.667) are better than OA products (mean correlation coefficient of full signal: 0.747, interannual signal: 0.469) especially around northern East/Japan sea and Kuroshio areas of the EAS. This result suggests that reanalysis ensemble could be an important index parameter in the analysis of interannual variability of the steric height in the EAS.

    Extended intercomparison period (1993-2010)

    Long-term linear trends of total steric sea level (Fig. 6), thermosteric component (Fig. 7), and the halosteric height (Fig. 8) have been compared for the period of 1993 to 2010, respectively. Blue bars denote the regional mean trend in the EAS and sky-blue is the global mean trend published by Storto et al. (2018). Most data show increasing sea level trends for 1993- 2010. There is a significant contribution of thermosteric in the global ocean, while there is a significant contribution of halosteric in the EAS. This result reflects the importance of regional ocean climatology study indicating that significant warming is not uniform for the global ocean. However, the spread of the ensemble means (both OA and REA) of the EAS is much larger than that of the global ocean.

    Another objective of the intercomparison is the quantification of the spatial pattern on the thermal and haline sea level contributions to the total steric sea level in the EAS. Therefore, this study estimates the spatial linear trend from the ensemble mean of all products for the steric, thermosteric, and halosteric sea level, respectively (Fig. 9). There is positive (warming) thermosteric linear trend in the East/Japan sea, while the negative (cooling) trend in the Kuroshio extension area (Fig. 9(b)). The total steric level and thermosteric sea level look similar especially in the central East/ Japan sea, i.e., the sea level trend around this area is generally controlled by the temperature change, while other areas reveal that total steric can be modified by the haline effect. It is well known that thermosteric change is strongly correlated with the heat content changes in the upper ocean (Palmer et el., 2017). Na et al. (2012) analyzed heat content variability by using different datasets from this study, World Ocean Database (WOD) and Simple Ocean Data Assimilation (SODA) data. They showed a steady increasing trend in the upper ocean (0-300 m depth) of East/Japan sea that is consistent with our results. Possible physical relationship between heat content variability and surface circulation pattern including Tsushima warm current has been suggested in their study. Yoon et al. (2016) also quantified the spatial distribution of heat content variability in the upper ocean (0-500 m) of East/Japan sea. However, research areas of most studies are limited in the East/Japan sea and analyzed periods are difference each other. For the Yellow and south coast of Korea including Korea Strait, as previously mentioned, there have been very little studies because observations are insufficient and water depth is not deep enough to calculate subsurface heat content. Therefore, it is hard to directly compare the previous results with this result covering total areas in the EAS for 1993-2010.

    Positive halosteric linear trend is depicted in the most area of the EAS, which represents general freshening event during the period of 1993-2010 (Fig. 9(c)). However, SNR is less than one indicating that ensemble spread of each product is greater than their mean signal. Therefore, we cannot determine that this trend is a statistically significant. Storto et al. (2017) already reported that halosteric trends are non-significant (SNR<1) almost everywhere. They investigated the percentage of the global ocean area with significant trend as a function of the starting and ending year for the trend computation. For the halosteric component, only a small portion (below 7% of the global ocean) show significant halosteric trends except for the last 5- year trend (2006-2010). In order to explain the major reason about low SNR, they suggested a close relationship between SNR and the number of observations, questioning the reliability of the halosteric trend estimates especially before the full deployment of the Argo floats. This study provides a consistent result that only central East/Japan sea and several Kuroshio extension areas where high density Argo salinity profiles can be obtained show significant linear trends among reanalysis products.

    In order to check the contributions of temperature and salinity variability of each vertical levels to total steric height change, we estimate the linear trends of steric, thermosteric, and halosteric sea level in different depth levels as shown in Figs. 10-12. In the upper oceans (0-300 m depth), most oceans show significant increasing linear trends (SNR>1) except for the halosteric in the East/Japan Sea (Fig. 12(a)). Around 300-700 m, we find overall decreasing steric trend less than 2 mm/year in the Kuroshio meandering area except for the southern coast of Japan and most of East/Japan Sea. Especially for the southern coast of Japan around 32 °N, 136 °E, robust warming trend more than 2 mm/year is found from most reanalysis data (SNR>1) as shown in Figs. 10-11(b). In addition, the steric sea level trend is similar that of thermosteric trend, which shows a predominant role of the thermosteric sea level, however we cannot determine this signal is significant because of low SNR below 700 m depth (Figs. 10-11(c~e)). As for the halosteric trend, upper 700 m and below 3000m, increase (freshening) trend is shown (Fig. 12(a, b, d)). For the middle depth layer between 700 to 3000 m, increase (freshening) trend is depicted in the East/Japan sea, and decrease (salting) trend is in the Kuroshio area, but they are all insignificant. Distribution of the halosteric sea level around 1500 m has not been shown because of relative very sparse salinity data distribution, which is closely associated with the low SNR in this area.

    Summary and Perspective

    Ocean reanalysis intercomparison project was successfully initiated and important different oceanic variables including steric sea level have been validated. However, most analyzed results have been limited to the global scale assessments. In this study, steric sea level variability around the EAS has been analyzed by using total 19 different ORA-IP data. This study found significant correlations between steric sea level from ORA-IP and satellite data (ALTGRV) except for the phase of annual cycle and interannual signals of the Yellow Sea. Correlation coefficient of REAENS is calculated as 0.706, which is much higher than that of either OAENS (0.296) and global ocean (0.441) that was published from the previous study (Storto et al., 2017). This new estimation implies that reanalysis ensemble could be an important index parameter in the analysis of interannual variability of the steric height in the EAS.

    However, SNR is less than one for the most linear trends in the EAS except for thermosteric sea level upper 300 m depth. General warming (cooling) trends are found in the East/Japan Sea (Kuroshio extension area) for 1993-2010, but these tendencies are different with depth levels. Significant freshening trends are found in the EAS for 1993-2010, and it is also different with depth levels. We expect that detailed physical mechanisms about the steric sea level variability based on the results of this study will be examined in the following studies.

    Acknowledgments

    We acknowledge Dr. Andrea Storto of CMRE (Center for Maritime Research and Experiment) for providing the ORA-IP data. We also thank three anonymous reviewers for their helpful comments to improve this manuscript. This research was a part of the Chungcheong Sea Grant Program funded by Korean ministry of Oceans and Fisheries and National Research Foundation of Korea (2016R1D1A1B0393 1519, 2019R1A2C1008490).

    Figure

    JKESS-40-5-487_F1.gif

    Seasonal cycle (2003-2010 mean) as a function of latitude for the (a) total sea level observed by satellite altimetry, (b) mass component derived by GRACE satellite, and (c) steric sea level denoted as ALT-GRV. ALT-GRV is derived by subtracting the mass component of sea level from the total sea level. Units are cm.

    JKESS-40-5-487_F2.gif

    Amplitude and phase of annual mean of steric height around the East Asian Seas. Radius represents amplitude (mm) and the angle (month) with respect to the x-axis indicates the phase (corresponding to the maximum reached in the annual cycle).

    JKESS-40-5-487_F3.gif

    Time series of steric sea level with the interannual time scale only around East Asian Seas for the period of 2003-2010. The correlation coefficients with the satellite derived dataset are also shown. The numbers in parenthesis are the correlation coefficients on a global scale.

    JKESS-40-5-487_F4.gif

    Comparison of correlation coefficients between steric sea level as shown in Fig. 3 and satellite (ALT-GRV) in the East Asian Seas (EAS, dark blue) and global ocean (GLO, skyblue).

    JKESS-40-5-487_F5.gif

    Left panel shows the correlation map of the (top) REAENS and (middle) OAENS with the satellite (ALT-GRV), and (bottom) their difference between REAENS and OAENS. Right panel shows the same information except for interannual time scale only after seasonal signal removed.

    JKESS-40-5-487_F6.gif

    Comparison of linear trend for steric sea level between the East Asian Seas (EAS, dark blue) and global (GLO, skyblue). Black error bars correspond to the spread (standard deviation) of the linear trends from the OAENS and REANES in the EAS and global oceans, respectively. Units are mm/year.

    JKESS-40-5-487_F7.gif

    The same as Fig. 6. except for thermosteric sea level.

    JKESS-40-5-487_F8.gif

    The same as Fig. 6. except for halosteric sea level.

    JKESS-40-5-487_F9.gif

    Map of linear trend ensemble mean (including all the products) for (a) steric, (b) thermosteric and (c) halosteric sea level from the surface to bottom during 1993-2010. Units are mm/year and contour lines denote signal-to-noise ratio.

    JKESS-40-5-487_F10.gif

    Map of linear trend ensemble mean (including all the products) for steric sea level in the (a) 0-300 m, (b) 300-700 m, (c) 700-1500 m, (d) 1500-3000 m, (e) 3000 m to bottom level during 1993-2010. Units are mm/year and contour lines denote signal-to-noise ratio.

    JKESS-40-5-487_F11.gif

    The same as Fig. 10 except for the thermosteric sea level.

    JKESS-40-5-487_F12.gif

    The same as Fig. 11 except for the halosteric sea level. Distribution around 1500m has not been shown because of relative sparse salinity data.

    Table

    Data information used in this study (The symbol * indicates that the computational grid has a resolution refinement in the Equatorial region. See Appendix for the full names of acronyms.)

    Acronyms and their full names shown in Table 1

    Reference

    1. Balmaseda, M.A. , Hernandez, F. , Storto, A. , Palmer, M.D. , Alves, O. , Shi, L. , Smith, G.C. , Toyoda, T. , Valdivieso, M. , Barnier, B. , Behringer, D. , Boyer, T. , Chang, Y-S. Chepurin, G.A. , Ferry, N. , Forget, G. , Fujii, Y. , Good, S. , Guinehut, S. , Haines, K. , Ishikawa, Y. , Keeley, S. , Köhl, A. , Lee, T. , Martin, M.J. , Masina, S. , Masuda, S. , Meyssignac, B. , Mogensen, K. , Parent, L. , Peterson, K.A. , Tang, Y.M. , Yin, Y. , Vernieres, G. , Wang, X. , Waters, J. , Wedd, R. , Wang, O. , Xue, Y. , Chevallier, M. , Lemieux J-F. Dupont, F. , Kuragano, T. , Kamachi, M. , Awaji, T. , Caltabiano, A. , Wilmer-Becker, K. , and Gaillard, F. ,2015, The Ocean Reanalyses Intercomparison Project (ORA-IP). Journal of Operational Oceanography, 8, 80-97.
    2. Cazenave, A. and Llovel, W. ,2010, Contemporary sea level rise. Annual Review of Marine Science, 2, 145-173.
    3. Chambers, D.P. ,2006, Evaluation of new GRACE timevariable gravity data over the ocean. Geophysical Research Letter, 33, L17603.
    4. Chang, Y.-S. , Rosati, A. , and Vecchi, G. ,2010, Basin patterns of global sea level changes for 2004-2007. Journal of Marine Systems, 80, 115-124.
    5. Chang, Y.-S. ,2012, A review of the quality control of global ocean temperature and salinity data, Journal of Korean Earth Science Society, 33(6), 554-566. (in Korean)
    6. Chang, Y.-S. ,2014, Basin pattern of the upper ocean warming for 1993-2009. Ocean Dynamics, 64, 823-832.
    7. Chang, Y.-S. , and Shin. H.-R. ,2014, Vertical gradient correction for the oceanographic Atlas of the East Asian Seas, Journal of Geophysical Research, 119, 5546-5554,
    8. Chang, Y.-S. ,2015, Intercomparison of the global ocean reanalysis data (review). The Sea, Journal of the Korean Society of Oceanography, 20, 102-118. (in Korean)
    9. Church, J. A. , Gregory, J. M. , White, N. J. , Platten, S. M. , and Mitrovica, J. X. ,2011, Understanding and projecting sea level change. Journal of Operational Oceanography, 24, 130-143.
    10. Fukumori, I. , and Wang, O. ,2013, Origins of heat and freshwater anomalies underlying regional decadal sea level trends, Geophysical Research Letter, 40, 563-567,
    11. Han, S. , Hirose, N. , Usui, N. , and Miyazawa, Y. ,2016, Multi-model ensemble estimation of volume transport through the straits of the East/Japan Sea, Ocean Dynamics, 66(1), 59-76.
    12. Ishii, M. , Kimoto, M. , Sakamoto, K. , and Iwasaki, S.-I. ,2006, Steric sea level changes estimated from historical ocean subsurface temperature and salinity analyses. Journal of Oceanography, 62, 155-170.
    13. IPCC,2014, Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 p.
    14. Lee, T. , Awaji, T. , Balmaseda, M.A. , Grenier, E. , and Stammer, D. ,2009, Ocean state estimation for climate research, Oceanography, 22, 160167.
    15. Lee, J.-H. , Matsuna, T. , Wimbush, M. , Yuan, D. , and Chang, K.-I. ,2012, The PAMS (Pacific-Asian Marginal Seas) Study, Progressive Oceanography, 105, 1-3.
    16. Lee, J.-H. , Lim, B.-J. , Kim, D.-Y. , Park, S.-H. Park, Chang, Y.-S. ,2018, The validation of MOHID regional ocean circulation model around the East Asian Seas in 2016. Journal of Korean Earth Science Society, 39(5), 436-457.
    17. Leuliette, E. , and Willis, J. ,2011, Balancing the sea level budget. Journal of Operational Oceanography, 24, 122- 129.
    18. Levitus, S. , Antonov, J. I. , Boyer, T. P. , Garcia, H. E. , and Locarnini, R. A. ,2005, Linear trends of zonally averaged thermosteric, halosteric, and total steric sea level for individual ocean basins and the world ocean, (1955-1959)-(1994-1998). Geophysical Research Letters, 32.
    19. Levitus, S. , Antonov, J.I. , Boyer, T.P. , Baranova, O.K. , Garcia, H.E. , Locarnini, R.A. , Mishonov, A.V. , Reagan, J.R. , Seidov, D. , Yarosh, E.S. , and Zweng, M.M. ,2012, World ocean heat content and thermosteric sea level change (0-2000 m), 1955-2010. Geophysical Research Letter, 39, L10603,
    20. Lie, H.J. , and Cho, C.-H. ,2016, Seasonal circulation patterns of the Yellow and East China seas derived from satellite-tracked drifter trajectories and hydrographic observations. Progress Oceanography, 146, 121-141.
    21. Na, H. , Kim, K.Y. , Chang, K.I. , Park, J.J. , Kim, K. , and Minobe, S. ,2012, Decadal variability of the upper ocean heat content in the East/Japan Sea and its possible relationship to northwestern Pacific variability, Journal of Geophysical Research: Oceans, 117, C02017,
    22. Palmer, M.D, Roberts, C.D., Balmaseda, M., Chang, Y.-S., Chepurin, G., Ferry, N., Fujii, Y., Good, S., Guinehut,S., and Haines, K., et al.,2017, Ocean heat content variability and change in an ensemble of ocean reanalyses. Climate Dynamics, 49, 909-930.
    23. Seo, G.-H. , Choi, B.-J. , Cho, Y.-K. , Kim, Y.H. , and Kim S. ,2015, Evaluation of a regional ocean reanalysis system for the East Asian Marginal Seas based on the ensemble Kalman filter, Ocean Science Journal, 50(1), 29-48.
    24. Sim, J.-E. , Shin, H.-R. , and Hirose, N. ,2018, Comparative analysis of surface heat flux in the East Asian marginal seas and its acquired combination data. Journal of Korean Earth Science Society, 39(1), 1-22.
    25. Storto, A. , Masina, S. , Balmaseda, M. , Guinehut, S. , Xue, Y. , Szekely, T. , Fukumori, I. , Forget, G. , Chang, Y.S. , Good, S.A. , Köhl, A. , Vernieres, G. , Ferry, N. , Peterson, K.A. , Behringer, D. , Ishii, M. , Masuda, S. , Fujii, Y. , Toyoda, T. , Yin, Y. , Valdivieso, M. , Barnier, B. , Boyer, T. , Lee, T. , Gourrion, J. , Wang, O. , Heimback, P. , Rosati, A. , Kovach, R. , Hernandez, F. , Martin, M.J. , Kamachi, M. , Kuragano, T. , Mogensen, K. , Alves, O. , Haines, K. , and Wang, X. ,2017, Steric sea level variability (1993-2010) in an ensemble of ocean reanalyses and objective analyses. Climate Dynamics, 49, 709-729,
    26. Tabata, S. , Thomas, B. , and Ramsden, D. ,1986, Annual and interannual variability of steric sea level along line P in the northeast Pacific Ocean. Journal of Physical Oceanography, 16, 1378-1398.
    27. Yoon, S.T. , Chang, K.I. , Na, H. , and Minobe, S. ,2016, An eastwest contrast of upper ocean heat content variation south of the subpolar front in the East/Japan Sea, Journal of Geophysical Research: Oceans, 121, 6418-6443,