Long-Term Characterization of Sea Conditions in the East China Sea Using Significant Wave Height and Wind Speed

ZHENG Kaiwen, OSINOWO Adekunle Ayodotun, SUN Jian, *, and HU Wei



Long-Term Characterization of Sea Conditions in the East China Sea Using Significant Wave Height and Wind Speed

ZHENG Kaiwen1), OSINOWO Adekunle Ayodotun1), SUN Jian1), *, and HU Wei2)

1),,266100,2),266061,

In this study, the statistical characterization of sea conditions in the East China Sea (ECS) is investigated by analyzing a significant wave height and wind speed data at a 6-hour interval for 30 years (1980–2009), which was simulated and computed using the WAVEWATCH III (WW3) model. The monthly variations of these parameters showed that the significant wave height and wind speed have minimum values of 0.73m and 5.15ms−1and 1.73m and 8.24ms−1in the month of May and December, respectively. The annual, seasonal, and monthly mean sea state characterizations showed that the slight sea generally prevailed in the ECS and had nearly the highest occurrence in all seasons and months. Additionally, the moderate sea prevailed in the winter months of December and January, while the smooth (wavelets) sea prevailed in May. Furthermore, the spatial variation of sea states showed that the calm and smooth sea had the largest occurrences in the northern ECS. The slight sea occurred mostly (above 30%) in parts of the ECS and the surrounding locations, while higher occurrences of the rough and very rough seas were distributed in waters between the southwest ECS and the northeast South China Sea (SCS). The occurrences of the phenomenal sea conditions are insignificant and are distributed in the northwest Pacific and its upper region, which includes the Southern Kyushu-Palau Ridge and Ryukyu Trench.

significant wave height; wind speed; sea state; occurrence

1 Introduction

The sea state refers to the overall condition of the surface of a large area of an open ocean or sea resulting from the combined effects of wind-generated waves, swells, and surface currents. It is described in terms of the roughness of the waters based on significant wave height.

Considering that the sea state is of primary importance to mariners, as high waves can cause many tragic losses, large waves have been identified as the highest risk factor in recreational boating deaths by drowning and immersion hypothermia. Coastal hazards associated with ocean surface waves and storm surges can also cause terrible disasters; therefore, public awareness of coastal hazards has been increasing. For example, the Boxing Day tsunami that devastated South Asia in 2004.

Long term characterization analysis is an important method to assess the properties of a certain sea region. It is relevant for the planning, design, and construction of coastal protection constructions, harbors, and navigational routes (Schneggenburger, 2000; Sundar and Ananth, 1988), and it contributes to fishery activities, navigation,and marine management.

In recent years, various techniques and methods have been used by the researchers to study the long-term characterization of sea conditions in different regions. Jahns and Wheeler (1973) have given a description of the long and complicated path from historical meteorological data to long-term wave statistics, and determination of the distribution of significant wave height during a short time interval from the spectrum of the sea during that interval is a major stage in the hind casting method. Thompson (1980) analyzed wave records from nine locations along the United States Atlantic, Pacific, Gulf, and Great Lakes coasts and observed that the occurrence of different range significant wave height varies from different locations. Lucas(2014) discussed the generation and validity of 21 years (1958–1978) daily maximum significant wave height and associated peak period distributions off Azores in the North Atlantic Ocean, which was extracted from the 44-year HIPOCAS database.

Ocean wave height observation data are only available for the last couple of decades at a limited number of buoy observations and the satellite data span only the last couple of decades (Young, 2011, 2012); consequently, most of the study of historical significant wave height trends are based on data sets of ocean waves for the past couple of decades (Wang and Swail, 2001, 2002; Wang, 2009; Dodet, 2010; Semedo, 2011). The length of the wave reanalysis data sets is limited to the record of the atmospheric data observations. Ocean wave models can provide more reliable simulation results of wave fields with the recent improvements of model parameterizations (Ardhuin and Roland, 2012), development of obstruction grids (Chawla and Tolman, 2008) and faster computers. Additionally, many marine scientists have studied the long-term characterization of sea conditions using a third-generation wave model.

The aim of this paper is to utilize the Douglas Sea states scale for various intervals of the significant wave height and 10-m wind speed to categorize the sea surface conditions in the ECS from a 6-hour, 30-year (January 1, 1980–December 31, 2009) significant wave height data obtained from numerical simulations by the WAVEWATCH III (WW3) model and wind speed data sets calculated by the model for the same period. Furthermore, the sea surface characteristics in different locations of the ECS are also examined.

The paper is arranged as follows. Section 2 provides information on the model used, the wind field data, and a thorough statistical validation of both significant wave height and computed wind speed from the model with the same parameters obtained from six Chinese buoy stations around the ECS. Section 3 presents the methodologies used in the sea surface characteristics and a detailed discussion of results obtained. This study is concluded in Section 4.

2 Model Configurations

In this study, the recent WAVEWATCH III version 4.18 is used, and new parameterizations related to the source terms are embedded in this official version. Parameterizations for the spectral dissipation of wind-generated waves, wave-current interactions, white capping, depth-limited wave breaking, refraction, and shoaling have significant improvement (Tolman, 2009). A couple of experiments were conducted to test different combinations of source terms. Details of various experiments published in different papers are as follows: The water depth field was processed by the Gridgen 3.0 packet according to the National Geophysical Data Center ETOPO 1 data, for which the resolution is 0.1˚×0.1˚. Source terms for energy spectra in the model are set as default(Arun, 2013). The model integrates the spectrum to a cut-off frequency and a parametric tail is applied above this frequency. The other option settings are 36 directions and 24 discrete wave-numbers (0.0412–0.4060Hz, 2.4–24.7s). Forced bythe ERA-Interim reanalysis wind field data sets, the model output includes a 10-m wind speed and significant wave height at each grid point with a period spanning from January 1, 1980 to December 31, 2009 at a 6-hour interval.

2.1 Data and Methodology

Long-term ECMWF, ERA-Interim wind field data sets within the space scope of 15˚–47˚N and 105˚–135˚E at a spatial resolution of 0.125˚ by 0.125˚ was used to force the WW3 model. The 6-hour, zonal (u10), and meridional (v10) wind field data spanned from 00.00 on January 1, 1980 to 18.00 on December 31, 2009, and the model outputs including the 10-m wind speed were obtained at a spatial resolution of 0.1˚ by 0.1˚. The wind speed is computed as the square root of sum of the squares of theandwind components.

The simulation region was set at 15˚–47˚N, 105˚–135˚E for the removal of boundary effects and precise simulation of wave parameters. Significant wave height with 10-m wind speed data sets were extracted within the study area 15˚–35˚N, 115˚–135˚E, which contains the ECS and regions surrounding it, as shown in Fig.1. The calculating time was from 1980-01-01 00.00 to 2009-12-31 18.00 with a calculation time step of 360, 180, 180, and 60s and a 6-hour output data(Panteleev, 2015).

Fig.1 Geography and bathymetry of the ECS.

2.2 Data Validation

Observational data of significant wave height and wind speed from January 1, 2015 to December 31, 2015, were obtained from six Chinese buoy stations. The time series showing the synchronous comparison between observations and simulations for significant wave height and wind speed are shown in Fig.2a, and the corresponding scatter plots are presented in Fig.2b. For two parameters, there is a good agreement between the buoy and model data on the curve trends.

The accuracy of the model was evaluated through a conventional statistical analysis that calculates the following:

wheremeans the correlation coefficient,means the root mean square error,xrepresents the buoy data,yrepresents the model data,andrepresent the mean values of buoy and model data, andis the total number of observations.

As seen in Fig.2b, the model computed (ECMWF) wind speed and the buoy values are highly correlative, with thelarger than 0.8 for the six cases and significant at the 99% level. The calculated negative biaseswhich ranged between −0.483 and −0.105ms−1and low value showed that the model data are slightly larger than the buoy data. The errors of model data, which ranged between 1.406 and 1.928ms−1, remain low when analyzed by RMSE. Also for the six cases, thebetween the significant wave height for the buoys and model is larger than 0.85 and significant at the 99% level, which the MBE ranged between −0.124 and 0.086m. The negative biases of −0.013, −0.077, −0.124, and −0.025m for cases 2, 3, 5, and 6 showed that the model data are slightly larger than buoy data. The RMSE values are low and ranged between 0.177–0.309m.

In general, the simulation results are consistent with observations, which indicate that the WW3 can reproduce the SWH and wind speed well and be a dependable model to simulate surface waves in the ECS.

Fig.2a Time series between simulations and observations for significant wave height and wind speed. On the-axis is the number of data points.

Fig.2b Scatter plots between simulations and observations for significant wave height and wind speeds.

3 Analysis Methods

The WW3 model has been set to output several parameters at 6-hour intervals for the 30-year hindcast. Out of the main output parameters, the significant wave height and wind speed were used to characterize the sea conditions in the ECS. For the different categories of sea states and for the different ranges of significant wave height () shown in Table 1, the information on the annual, seasonal, and monthly mean sea states in the ECS is presented as the cumulative frequencies() as a percentage as follows:

whererepresents the frequency ofthat satisfies each of the defined range offor various sea states andrepresents the total number ofvalues. Also, for different categories of sea states and for the different range of wind speed () shown in Table 1, the information on the annual, seasonal, and monthly mean sea states in the ECS is presented as the cumulative frequencies() as a percentage as follows:

whererepresents the frequency ofthat satisfies each of the defined range offor the various sea states andrepresents the total number ofvalues.

The seasonal mean sea state characterization was determined by combining: the December–February monthly mean sea states for winter, March–May for spring, June–August for summer, and September–November for au-tumn. The annual and monthly mean sea states are presented in Tables 2 and 3 for the significant wave height () and Tables 4 and 5 for the wind speeds (). Also, for a spatial distribution, the percentage cumulative frequencies of wave height () and wind speeds () for the different categories of sea states were computed for each grid point of the ECS domain and surrounding regions.

4 Results and Discussion

4.1 Sea States Characterization Using the Wave Height (hs)

The overall monthly mean significant wave height andwind speed in the ECS are shown in Fig.3. For both parameters, minimum values of 0.73m and 5.15ms−1occurred in May, respectively, while maximum values of1.73m and 8.24ms−1occurred in December, respectively.

Table 1 Douglas Sea state scale for significant wave height and wind speed

Table 2 Average seasonal hs values for the ECS (f is computed using Eq. (4))

Table 3 Average monthly hs values for the ECS (f is computed using Eq. (4))

Table 4 Average seasonal u values for the ECS (f is computed using Eq. (5))

Table 5 Average monthly u values for the ECS (f is computed using Eq. (5))

Fig.3 Monthly variation of significant wave height and wind speed in ECS.

As a basis for the assessment of sea conditions, the seasonal and monthly average percentage cumulative fre-quencies offor the different sea states have been plotted for the ECS, as shown in Figs.4(a)–(c).

The annual cumulative percent frequency distribution offor different categories of sea states shown in Fig.4a indicate that the slight sea state has the highest percentage cumulative frequency of 36.39% in the ECS. The calm (rippled) and rough sea states showed much lower occurrences of 4.6% and 8.23%, respectively, in comparison to the smooth and moderate sea states, which are 19.95% and 29.7%, respectively. The occurrences of the very rough and high sea conditions are very insignificant (<5%) and are hardly the sea state of the ECS in the calm (glassy) or very high conditions.

For the seasons shown in Fig.4b, the moderate sea state occurred for the highest percentage of time (43.52%) during winter. The slight sea state has the highest occurrence during spring, summer, and autumn with the frequencies of 43.34%, 38.88%, and 34.75%, respectively. Higher occurrence of the rough (15.66%) sea state during winter in comparison with other seasons showed that the sea appears to be rougher during winter, which is because of the prevalence of strong northeast winds and the influence of cold fronts during the winter monsoon. During spring, the least values of 3.25% and 0.12% for the rough and very rough sea states, respectively, suggest the sea is most calm at spring. The occurrences of the rest sea states from very rough to phenomenal are very insignificant.

Fig.4 Cumulative percent frequency distribution of hs for thedifferent sea states in the ECS.

Monthly characterization in Fig.4c shows that the rough sea has higher occurrences of 15.24%, 19.54%, and 15.05% from November to January as compared to other months. The very rough sea state in November and December also has higher frequencies of 2% and 2.57%, respectively, in comparison to the other months. The slight sea condition has the highest occurrence in March through October, with respective values of 41.34%, 45.44%, 43.32%, 43.23%, 37.89%, 35.64%, 38.77%, and37.1%. Furthermore, the moderate sea condition has highest frequencies in November through February, with respective values of 41.04%, 42.56%, 46.1%, and 41.75%. The smooth (wavelets) sea state has the highest occurrence of 34.99% in May, while the calm (rippled) sea state also showed higher frequencies of 6.49%, 7.31%, and 6.52% in May, July, and August, respectively, in comparison with other months.

4.2 Spatial Variation of Sea States Using Significant Wave Height (hs)

The overall regional distributions of the percentage fre- quencies of significant wave height () representatives of the different sea states are shown in Fig.5.

The figures show that the percentage frequency of calm (rippled) sea is below 5% in the ECS and most of the surrounding locations. Slightly higher occurrences (10%–20%) are distributed in the Yellow Sea, Pusan, Korean Strait, Shikoku Island, and around the northeastern part of SCS. Relatively large regions of the smooth (wavelets) sea with percentage frequencies less than 15% in the Taiwan and Ryukyu Island and their surrounding regions. In most parts of the ECS, the percentage frequencies of smooth (wavelets) sea are less than 22%. Higher occurrences, up to 50%, are found in the Yellow Sea and in some sporadic waters in the northeastern ECS. Areas that had percentages above 35% were found in every part of the ECS except some parts around the Yellow Sea and Shikoku Island, in which the frequencies were less than 20%. Peak occurrences are particularly seen on the onshore borders of the ECS and waters close to Taiwan Island and Taiwan Strait, and the occurrences are between 25% and 30% in the northeastern part of SCS, while an occurrence of up to 33% extends from the Philippine Basin to the Northwest Pacific. In most parts of the ECS and surrounding locations, the moderate sea occurred at frequencies between 32% and 39%. The frequency is largest, up to 39%, around Diaoyu Island, Ryukyu Island, and waters close to the Taiwan Strait in the southern ECS. Areas with low frequencies are distributed in the northeast SCS, Jeju Island, and in some few waters in the northwest Pacific. Areas with the percentages of less than 8% are found close to the Yellow Sea and to some islands in the northeastern part of ECS. The rough sea had the largest occurrence (14%–16%) in the Bashi Channel and in regions in the northeastern part of SCS. Relatively large regions with frequencies between 10%–14% in the Philippine Basin, Northwest Pacific, and in some waters around the southern ECS. The areas with the least percentages of less than 4% are found in the northern ECSand around Luzon and Taiwan Island. Peak occurrences of the very rough sea between 2.5 and 3% are also distributed in the Bashi Channel and in some regions in the northeastern part of SCS. Areas with fewer frequencies, approximately 2%, are found in the Philippine Basin. Most regions in the northern ECS and few locations in the southern ECS such as Taipei, Taiwan Strait, and Luzon Island showed insignificant occurrences of less than 0.5%. The occurrences of the high and very high sea are generally insignificant, of less than 0.5% in the ECS and surrounding regions. In addition, slight areas with a high sea state (0.3%–0.35%) are distributed in the central and southern ECS, while areas of the very high sea of approximately 0.05% occurred around the southern Kyushu- Palau ridge. Hardly is there any calm (glassy) or phenomenal sea in the ECS and surrounding regions.

4.3 Sea States Characterization Using the Wind Speed (u)

Fig.6a shows the annual cumulative percent frequency distribution offor different categories of sea states. The figure shows that the slight sea state category has the highest percentage cumulative frequencies of 30.73% in the ECS. Lower frequencies occurred in the moderate sea (22.09%) and high sea conditions (5.24%). The smooth (wavelets) sea also showed a lower frequency of 19.1% and the occurrences of the very high and phenomenal sea conditions are insignificant when the occurrences are below 1%. A seasonal characterization (Fig.6b) shows thatthe slight sea has the largest occurrence during spring, summer, and autumn with frequencies of 36.12%, 37.18%, and 28.44%, respectively. The moderate sea only has the highest frequency (23.05%) during winter, while higher occurrence from moderate (23.05%) to high (9.81%) sea states during winter than in other seasons suggests that the sea is generally rougher at winter. However, the fewest occurrences ranging between 19.72%–5.02% from the moderate to very rough seas, respectively, during the summer combined with the largest frequency of the smooth (wavelets) sea of 26.61% suggests that the sea is most calm in this season. The occurrences of the very high and phenomenal seas are negligible in all seasons when the occurrences are below 2%. Monthly characterization (Fig.6c) shows that the rough sea has higher occurrences in November through February with respective frequencies of 17.3%, 17.62%, 18.48% and 16.6%. Its occurrence is least (6.53%) in May, and the occurrences of the smooth (wavelets) sea are higher in April through September with respective frequencies of 23.18%, 28.5%, 25.18%, 26.56%, 28.06% and 25.25%. Consequently, larger frequencies exist for the rough to high sea conditions from November to February. Through these months, the rough sea occurred for 17.3%, 17.62%, 18.48%, and 16.6%, respectively, very rough sea occurred for 16.87%, 19.59%, 18.91% and 15.57%, respectively, while the high sea occurred for 9.63%, 13.12%, 8.86% and 7.23%. The occurrences of the very high and phenomenal seas are insignificant in all months when the occurrences are below 2%. In all, the slight sea prevailed in February throughOctober with respective occurrences of 24.22%, 30.84%, 36.99%, 40.57%, 40.66%, 36.78%, 34.21%, 35.22% and 28.63%. From November through January, the moderate sea state has the highest frequencies of 23.61%, 21.67% and 23.75%, respectively, and the smooth (wavelets) sea occurred the most (28.5%) in May.

Fig.6 Cumulative percent frequency distribution of u for the different sea states in the ECS.

4.4 Spatial Variation of Sea States Using Wind Speed (u)

The overall regional variation of the percentage frequency of the wind speed () representative of the different sea states is displayed in Fig.7.

The figure shows that very low occurrences of the smooth (wavelets) sea of less than 25% distribute over every part of the ECS and its surrounding locations except around Shikoku, Taiwan, and Luzon Islands where there are higher occurrences of greater than 30%. Frequencies up to 80% occur in a small region around Shikoku Island in the northeastern ECS. The slight sea distributes over a large part of the ECS and in the surrounding locations at frequencies above 30%. Peak occurrences between 40% and 45% are seen around the Yellow Sea, Taiwan, and Luzon Island. Frequencies less than 28% extend from the upper SCS, through Taipei to the central ECS, and relatively large region of the moderate sea with a frequency of approximately 25% locates in the ECS and regions above the Northwest Pacific with few locations in the southern ECS. Occurrences between 20% and 23% dominate most part of the central and northern ECS with the northwest Pacific, while areas with low frequencies between 18% and 19% are distributed in most waters between the southwestern ECS and northeastern SCS. Peak frequencies of the rough sea (14%–16.5%) occurred in waters below the Taiwan Strait extending to Bashi Channel and in regions covering Taipei and Diaoyu Island. Rough sea occurrences between 10% and 14% dominate most parts of the ECS and surrounding locations, while frequencies less than 7% are seen in sporadic waters around the Yellow Sea, some islands in the northeastern ECS, Taiwan, and Luzon Islands.

Fig.7 Spatial distribution of the percentage cumulative frequencies of u for the different sea states in the ECS. Values on the side bars are in percentage.

The very rough sea occurred at peak frequencies (15%–18%) in waters adjacent to Bashi Channel and near Taipei and the Taiwan Strait. Lower occurrences between 10% and 12% occupy a large part of ECS and the locations, while occurrences below 8% are found in some waters around some islands in the northeastern ECS, the Yellow Sea, Taiwan, and Luzon Island. The largest occurrences of the high (16%) and very high seas (2%) are then seen in the Taiwan Strait and Bashi channel, respectively. Relatively large regions of both sea states with frequencies of approximately 6% for the high sea and 0.8% for the very high sea are in most waters of the ECS and surrounding locations. In contrast, the least frequencies for both seas are distributed in the Yellow Sea and in some islands such as Taiwan and Luzon Islands in the southern ECS and Kyushu and Shikoku Islands in the northern ECS. Phenomenal sea conditions and with negligible occurrences (0.04%–0.06%) are distributed over a relatively large part of the Northwest Pacific and its upper region, including the Southern Kyushu-Palau Ridge and Ryukyu Trench.

5 Conclusions

In this paper, significant wave height and wind speed data obtained for 30-year (1980–2009) from WAVEWAT- CH III (WW3) the model has been utilized to investigate the sea conditions in the ECS. The monthly variations of these parameters have also been investigated. Results showed that significant wave height and wind speed have minimum values in May and peak values in December. The ECS wave fields are under the control of monsoons, so the variety ofanddisplays strong seasonal features. Furthermore, statistical analysis of the daily significant wave height, wind speed, and subsequent characterization of the annual, seasonal, and monthly mean sea state based on these was also performed. Results showed that the slight sea generally prevailed in the ECS and had nearly the highest occurrence in all seasons and months. The moderate sea condition prevailed in winter months, while the smooth (wavelets) sea prevailed in May.

Furthermore, spatial variation of sea states showed that calm and smooth sea conditions have largest occurrences around the Yellow Sea and islands in the northeastern ECS. The slight sea condition showed the largest occurrence (above 30%) over most parts of the ECS and its surrounding locations. Higher occurrences of the rough and very rough seas are distributed in waters between the southwest ECS and the northeast SCS, and the occurrences of the phenomenal sea conditions are insignificant and are distributed in the Northwest Pacific and its upper region, including the Southern Kyushu-Palau Ridge and Ryukyu Trench. We also found that the distribution of significant wave height has a strong correlation with local wind speed, and we understand that surface waves in ECS are composed of wind sea and swell. Wind seas are locally generated by the surface wind, and it seems that stronger wind energies correspond to a higher wind sea wave energy. Thus, we will study in future studies the separation of swell and wind sea energy from the mixed sea state and focus on understanding the relationship between wind energy and significant wave height in ECS.

Acknowledgements

This study is supported by the National Key Research and Development Program of China (No. 2016YFC1401405) and the National Natural Science Foundation of China (No. 41376010). The buoy data is provided by the North China Sea Branch of State Oceanic Administration.

Ardhuin, F., and Roland, A., 2012. Coastal wave reflection, directional spread, and seismoacoustic noise sources., 117: C00J20, DOI:10.1029/2011JC007832.

Arun, C., Deanna, M. S., and Hendrik, L. T., 2013. Validation of a thirty year wave hindcast using the climate forecast system reanalysis winds., 70: 189-206, DOI:10.1016/j.ocemod.2012.07.005.

Chawla, A., and Tolman, H. L., 2008. Obstruction grids for spectral wave models.,22: 12-25.

Dodet, G., Bertin, X., and Taborda, R., 2010. Wave climate variability in the North-East Atlantic over the last 6 decades., 31: 120-131, DOI:10.1016/j.ocemod.2009.10.010.

Jahns, H.O., and Wheeler, J. D., 1973. Long-term wave probabilities based on hindcasting of severe storms., 25: 473-486, DOI: 10.2118/3934-PA.

Lucas, C., Muraleedharan, G., and Guedes, S.C., 2014. Outliers identification in a wave hindcast dataset used for regional frequency analysis., 140: 1317-1327, DOI: 10.1201/b17494-178.

Panteleev, G., Yaremchuk, M., and Rogers, E. W., 2015. Adjoint-free variational data assimilation into a regional wave model., 32: 1386-1399, DOI:10.1175/JTECH-D-14-00174.1.

Schneggenburger, C., Gunther, H., and Rosenthal, W., 2000. Spectral wave modelling with non-linear dissipation: Validation and applications in a coastal tidal environment., 41: 201-235, DOI:10.1016/S0378-3839(00)00033-8.

Semedo, A., Suselj, K., Rutgersson, A., and Sterl, A., 2011. A global view on the wind sea and swell climate and variability from ERA-40., 24: 1461-1479, DOI: 10.1175/2010JCLI3718.1.

Sundar, V., and Ananth, P.N., 1988. Wind climate for Madras Harbor, India., 31: 323-333.

Thompson, E. F., 1980. Energy Spectra in Shallow U.S. Coastal Waters.Technical Paper. U.S. Army, Corps of Engineers, Coastal Engineering Research Centre, 80-2.

Tolman, H. L., 2009.. Marine Modelling and Analysis Branch, Technical Report. 276, NOAA, 220pp.

Wang, X. L., and Swail, V. R., 2001. Changes of extreme wave heights in Northern Hemisphere Oceans and related atmospheric circulation regimes., 14(10): 2204-2221.

Wang, X. L., and Swail, V. R., 2002. Trends of Atlantic wave extremes as simulated in a 40-year wave hindcast using kinematically reanalyzed wind fields., 15(9): 1020-1035.

Wang, X. L., Swail, V. R., Zwiers, F. W., Zhang, X.B., and Feng, Y., 2009. Detection of external influence on trends of atmospheric storminess and northern oceans wave heights.s, 32(2-3): 189-203, DOI: 10.1007/s00382-008-0442-2.

Young, I. R., Vinoth, J., Zieger, S., and Babanin, A. V., 2012. Investigation of trends in extreme value wave height and wind speed., 117: C00J06, DOI: 10.1029/2011JC007753.

Young, I. R., Zieger, S., and Babanin, A. V., 2011. Global trends in wind speed and wave height., 332(6028): 451-455, DOI: 10.1126/science.1197219.

(Edited by Xie Jun)

(Received March 17, 2017; revised May 27, 2017; accepted June 2, 2017)

© Ocean University of China, Science Press and Springer-Verlag GmbH Germany 2018

. Tel: 0086-532-66786228E-mail: sunjian77@ouc.edu.cn