• For Contributors +
• Journal Search +
Journal Search Engine
ISSN : 1225-6692(Print)
ISSN : 2287-4518(Online)
Journal of the Korean earth science society Vol.39 No.1 pp.1-22
DOI : https://doi.org/10.5467/JKESS.2018.39.1.1

# Comparative Analysis of Surface Heat Fluxes in the East Asian Marginal Seas and Its Acquired Combination Data

Jung-Eun Sim1, Hong-Ryeol Shin2*, Naoki Hirose3
1GeoSystem Research Corporation, Gunpo 15807, Korea
2Department of Atmospheric Science, Kongju National University, Kongju 32588, Korea
3Research Institute for Applied Mechanics, Kyushu University, Kasuga 816-0811, Japan
Corresponding author: hrshin@kongju.ac.kr+82-41-850-8529+82 41 856 8527
20171124 20180131 20180205

## Abstract

Eight different data sets are examined in order to gain insight into the surface heat flux traits of the East Asian marginal seas. In the case of solar radiation of the East Sea (Japan Sea), Coordinated Ocean-ice Reference Experiments ver. 2 (CORE2) and the Objectively Analyzed Air-Sea Fluxes (OAFlux) are similar to the observed data at meteorological stations. A combination is sought by averaging these as well as the Climate Forecast System Reanalysis (CFSR) and the National Centers for Environmental Prediction (NCEP)-1 data to acquire more accurate surface heat flux for the East Asian marginal seas. According to the Combination Data, the annual averages of net heat flux of the East Sea, Yellow Sea, and East China Sea are −61.84, −22.42, and −97.54Wm−2 , respectively. The Kuroshio area to the south of Japan and the southern East Sea were found to have the largest upward annual mean net heat flux during winter, at −460- −300 and at −370- −300Wm−2 , respectively. The long-term fluctuation (1984-2004) of the net heat flux shows a trend of increasing transport of heat from the ocean into the atmosphere throughout the study area.

## Introduction

Heat exchange between the atmosphere and the ocean through the sea surface has significant influences on numerous meteorological phenomena. Furthermore, surface heat flux plays an essential role in understanding the modeling of water circulation and energy conservation. Uncovering the temporal and spatial distribution characteristics of heat flux is considered a key factor in predicting synoptic weather phenomena as well as climate change and various physical phenomena occurring in the ocean (Hong et al., 2005).

Thus far, studies estimating the heat flux of the sea surface in the East Asian marginal seas, namely, the East Sea (Japan Sea), Yellow Sea, East China Sea, and the Kuroshio, have been completed by numerous researchers. Yet, they have demonstrated significant differences in the heat flux calculated for each area, depending on the method of heat flux calculation, data, and coefficients used for the calculation. Methods of calculating heat flux include the eddy correlation method that analyzes turbulent flows directly, and bulk aerodynamic method that utilizes average weather variables. Furthermore, there is a method that derives weather variables and heat flux on the sea surface from a model based on atmospheric boundary layer dynamics (Kang et al., 1994; Park et al., 1995; Ahn et al., 1997). The data used in each study varied over time; prior to the 1980s, most studies utilized data collected from coastal areas, vessels, and ocean buoys, but by the 1990s the use of satellite data increased as well as the use of other data.

Even though heat fluxes are calculated using observations, the calculated heat fluxes are quite different because of different calculation methods. For example, Ahn et al. (1997) used the Comprehensive Ocean-Atmospheric Data Set (COADS) data in order to calculate the heat flux in the sea surrounding the Korean Peninsula according to the calculation methods of both Bunker (1976) and Kondo (neutral and diabatic methods) (1975). The net heat flux in the East Sea was −90, −22, and −61Wm −2 , respectively. Different bulk aerodynamic methods with different constants were used for the latent and sensible heat flux calculations. As such, a broad range of annual average net heat flux appears due to various calculation methods and data: −110- −25Wm −2 in the East Sea, −15-10Wm−2 in the Yellow Sea, and −106 -3Wm−2 in the East China Sea.

Despite the large differences in the net heat flux derived from the various studies, similarities can be found in the East Asian marginal seas. That is, heat usually travels from the atmosphere into the ocean during summer, while a significant amount of heat is released from the ocean into the air in the form of latent heat flux and sensible heat flux during winter (Yun et al., 1998; Kang et al., 2001). The greatest heat loss appears during January, and heat exceeding 200- 400Wm−2 is released from the ocean into the atmosphere in the East Asian marginal seas, especially along the Kuroshio axis and in the central part of the Yellow Sea (Kang et al., 1994; Na et al., 1999).

Previous studies mostly calculated and assessed the surface heat flux based on observation data (Yun et al., 1998; Kang et al., 2001). In recent years, performance enhancements in computing and numerical modeling have resulted in high-quality reanalysis data that is considered reliable. In comparison to observation data, reanalysis data offers numerous advantages, such as lengthier data time frame, data continuity, and higher data resolution and consistency. Hence, the use of reanalysis data for analysis of surface heat flux is becoming increasingly popular.

In this study, eight data sets from various organizations were adapted in order to compare and analyze the characteristics of the surface heat flux in the East Asian marginal seas (Fig. 1). The data are classified into radiant flux (shortwave, longwave) and turbulent flux (latent heat, sensible heat), and were compared according to appropriate criteria. This study provides improved data on the surface heat flux of the East Asia marginal seas. Using the improved data, the characteristics of seasonal and annual mean surface heat flux and the long-term variation trend were analyzed for the East Asian marginal seas, and horizontal heat transport was analyzed in the East Sea. A list of acronyms and their definitions is provided in the appendix.

## Data and Methodology

Equation (1), a basic and widely used formula, is used for net heat flux calculation through the sea surface (Hirose et al., 1996; Ahn et al., 1997)

$Q n e t = Q s − ( Q b + Q e + Q h )$
(1)

where Qnet, Qs, Qb, Qe and Qh indicates net heat flux, shortwave radiation, longwave radiation, latent heat flux, and sensible heat flux, respectively. The heat flux calculation of each item uses various numerical models, empirical equations, bulk formulae and input data. Various numerical models were used to calculate heat flux for five data sets: 40-yr ECMWF Re-Analysis (ERA-40), the Modern-Era Retrospective analysis for Research and Applications (MERRA), the National Centers for Environmental Prediction (NCEP)-1, NCEP-2, and the Climate Forecast System Reanalysis (CFSR) (Table 1 and Appendix). Three data sets, the Objectively Analyzed air-sea Fluxes (OAFlux), Coordinated Ocean-ice Reference (CORE) 2, and the National Oceanography Centre Southampton (NOCS), calculated the heat flux with bulk formulas using various observational data including satellite and reanalysis data.

ERA-40 heat flux data used the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric model with a resolution of T159 (approximately 125 km) and 60 vertical layers (Uppala et al., 2005; Brunke et al., 2011). Data assimilation of the ERA-40 model was performed using a 3DVar system, and this model includes improvements to the parameterizations of deep convection and radiation and a new representation of sea ice. The observations used in ERA-40 were collected from many sources and these data cover conventional sources (surface land, ship observations, aircraft etc.) and satellite measurements.

The National Aeronautics and Space Administration (NASA) Goddard Earth Observing System Model, Version 5 (GEOS-5) atmospheric model is used for MERRA reanalysis data (Brunke et al., 2011; Rienecker et al., 2011). To be specific, this model includes shortwave and longwave radiation schemes and turbulent flux parameterizations (Roberts et al., 2012). It has a resolution of 1/2°latitude × 2/3°longitude with 72 vertical layers. Assimilation of the model is completed using GSI and a new 3DVar analysis. MERRA utilizes input data from various conventional observations (surface land, ship and buoy observations, wind profilers, aircraft, etc.) and many satellite observations.

NCEP-1 of the NCEP National Center for Atmospheric Research (NCAR), utilized NCEP’s operational global forecasting model of the mid-1990s with a horizontal resolution of T62 (approximately 210 km) and 28 vertical layers (Kalnay et al., 1996; Brunke et al., 2011). Assimilation of the model was completed by spectral statistical interpolation (SSI), an older 3DVar technique, using global radiosonde, COADS surface marine, aircraft, surface land synoptic, and satellite sounder data and SSM/I surface wind speed, satellite cloud drift winds, etc. NCEP-2 (NCEP?DOE) is an updated version of NCEP-1 improved to correct errors discovered in the processing of NCEP-1. NCEP-2 uses a T62 spectral resolution model with 28 vertical layers (Brunke et al., 2011). Some improvements were made to the model physics; for example, updates such as boundary layer turbulence and radiation. Data similar to NCEP-1 were incorporated into NCEP-2, except for sea ice, seasurface temperature (SST), and new ozone climatology.

CFSR uses the CFS, a fully coupled atmosphereocean- sea ice-land model with a spectral resolution of T382 (approximately 38 km) and 64 vertical layers (Saha et al., 2010; Brunke et al., 2011). There are three programs in the CFS that have their own data flow: the atmospheric model (GFS), the ocean model (GFDL MOM version 4), and the coupler. Data similar to NCEP-1 and NCEP-2 reanalysis data were also incorporated into the CFSR with the addition of SSM/I, ERS, QuikSCAT, WindSat ocean surface winds, etc.

OAFlux is produced by the Woods Hole Oceanographic Institution (WHOI). The OAFlux project was intended to develop an enhanced global analysis of air-sea latent heat, sensible heat, net shortwave, and net longwave radiation fluxes over the past 50 years through an appropriate combination of satellite retrievals, ship reports from COADS, and surface meteorology from Numerical Weather Prediction (NWP) analysis/ reanalysis outputs (Yu and Weller, 2007; Yu et al., 2008). In OAFlux, the three reanalysis (NCEP-NCAR, NCEP-DOE, and ERA-40) data are blended with satellite data to determine a best-guess estimate using objective analysis. The latent and sensible heat fluxes are calculated using the COARE 3.0 algorithm (Brunke et al., 2011; Roberts et al., 2012).

CORE2 uses bulk formulae with various data to calculate air-sea fluxes (Griffies et al., 2009; Large and Yeager 2009). The input data of CORE2 incorporate numerous and various observed and reanalysis data, either directly or indirectly: NCEP reanalysis for the atmospheric state, the International Satellite Cloud Climatology Project (ISCCP)-FD for the radiation data, QuikSCAT for winds, Hadley-OI for SST and NOC, etc.

NOCS utilizes observations of bulk variables from dedicated buoys, research vessels, and voluntary observing ships (COADS data) to produce space and time-averaged fields of air-sea fluxes (Berry and Kent, 2009). Following this, the observation fields are used with the bulk flux parameterization to calculate daily averaged estimates of the net shortwave (Payne, 1972; Reed, 1977) and longwave radiation (Clark et al., 1974) and the latent and sensible heat fluxes (Smith, 1980, 1988).

In total, eight ocean heat flux data sets, all of which are monthly mean data, were utilized in this study (Table 1). The time span of the data differs slightly. With the exception of OAFlux, the rest of the data covers a time span of more than thirty years. The data from Hirose et al. (1996, 1999) is presented as a comparison, and the observational values of COADS, the Japan Oceanographic Data Center (JODC), the National Oceanographic Data Center Home Page (NODC), and the Far Eastern Regional Hydrometeorological Research Institute (FERHRI) were plugged into the bulk air dynamics method so as to calculate the heat flux over the East, Yellow, and the East China seas.

In the case of shortwave radiation, positive values mean an influx of heat into the ocean, and, in the case of other fluxes (longwave radiation, latent heat flux, and sensible heat flux), positive values indicate an outflow of heat from the ocean. Therefore, there is heat gain in the ocean when net heat flux is of a positive value while when it is a negative value there is heat loss from the ocean. The research area is 24- 50°N , 118-142°E (Fig. 1). The area of East Sea used in this study is 1.012×10 6 km 2 , for the Yellow Sea it is 0.462×10 6 km 2 and for the East China Sea it is 0.809×10 6 km 2 . Areas of each sea were used for calculation of mean heat flux in each sea.

## Characteristics of the Heat Flux of the Reanalysis Data

### Annual average

Shortwave radiation is highly susceptible to the influence of the number of clouds, and the distribution of its isolines generally parallels latitudes in the open ocean. In the southeastern part of the research area, isolines of shortwave radiation are distributed relatively in an east-west direction (Fig. 2). However, northsouth isolines of shortwave radiation are predominant in the East Asian marginal seas. These spatial distribution patterns can generally be found in all of the eight data sources. ERA-40, MERRA, NCEP-1, NCEP-2, CFSR, and NOCS share similar values (140- 180Wm−2 ) in the East Sea and the Yellow Sea. On the other hand, OAFlux and CORE2 show 120-150 Wm−2 , smaller in comparison. Nevertheless, the values of the aforementioned two data sources resemble the shortwave radiation value (128Wm−2 ) of Hirose et al. (1996), which was calculated using observational COADS data. In the East China Sea, shortwave radiation from ERA-40, MERRA, NCEP-1, NCEP-2, CFSR, and NOCS ranged from 160 to 190Wm−2 , while those of OAFlux and CORE2 ranged from 140 to 160Wm−2 . The influence of the number of clouds may be responsible for small values in the entrance area of the Yangtze River and in the area between theYangtze River and Taiwan.

Longwave radiation is known to be determined by sea surface temperature (SST), atmospheric watervapor pressure, and the extent of clouds (Cho, 1968; Yun et al., 1998). In the East Sea, Yellow Sea, and northwestern Pacific Ocean to the south of Japan, MERRA, NCEP-2, and CFSR demonstrate relatively larger values of greater than 80Wm−2 , while NOCS and OAFlux demonstrate smaller values ranging from 40 to 60Wm−2 (Fig. 3). As with shortwave radiation, the extent of clouds may also be responsible for smaller longwave radiation in the Taiwan coastal area of the East China Sea.

Latent and sensible heat fluxes are caused by an interaction between the ocean and the atmosphere, whose values are mainly determined in terms of thermal differences and wind speed. They are important factors governing the thermal equilibrium of the atmosphere (Hong et al., 2005). In all of the data sources, the Kuroshio and the southern East Sea affected by the Tsushima Warm Current appear to exhibit significant latent heat flux (Fig. 4). In the Kuroshio area (31-35°N, 131-142°E) to the south of Japan, NCEP-2 indicates the highest latent heat flux with an average of 205.9Wm−2 , and NCEP-1, CFSR, OAFlux, CORE2, and NOCS indicate a similar distribution with an average of 179.5-191.3Wm−2 . In contrast, ERA-40 and MERRA have a relatively smaller value than the others with an average of 142.3-164.4Wm−2 . Roberts et al. (2012) reported that the latent heat flux (142.3Wm−2 ) of MERRA in the Kuroshio area is 40Wm−2 less than the average values of three data sources (NOCS, GSSTF2b, and OAFlux) that are based on observation. Further, Hirose et al. (1999) reported that the average latent heat flux calculated using observational data is approximately 185Wm−2 in the Kuroshio to the south of Japan. Therefore, NCEP-1, CFSR, OAFlux, CORE2, and NOCS seem to show consistent values with previous studies (Hirose et al., 1999; Roberts et al., 2012), whose average latent heat flux indicates 179.5- 191.3Wm−2 in the Kuroshio area. ERA-40, NCEP-2, CFSR, OAFlux, and NOCS appropriately simulate the effects of the sub-polar front and the Tsushima Warm Current in the southern East Sea. In the Yellow Sea, NCEP-1, NCEP-2, CFSR, OAFlux, and CORE2 demonstrate a distribution indicating an inflow of warm currents from the East China Sea to the Yellow Sea.

All of the eight data sources pertaining to sensible heat flux (Fig. 5) indicate large values both in the Kuroshio and in the East Sea. NCEP-1 and NCEP-2 indicate 40-80Wm−2 ; CFSR, OAFlux and CORE2 indicate 25-70Wm−2 ; and ERA-40, MERRA and NOCS indicate 25-45Wm−2 . Roberts et al. (2012) reported that sensible heat flux of MERRA in the Kuroshio was approximately 15Wm−2 less than the average values of three data sources based on observation (NOCS, GSSTF2b and OAFlux). In the Korea Strait (Tsushima Strait) and in the southeastern East Sea, ERA-40, NCEP-1, CFSR and OAFlux have a large sensible heat flux indicative of the influences of the Tsushima Warm Current. In contrast, both OAFlux and CORE2 share a region with low sensible heat flux (25-40Wm−2 ) in the central East Sea.

The net heat flux calculated using Eq. (1) mostly represents negative values, indicating a transfer of heat from the ocean into the atmosphere (Fig. 6). Yet, NOCS has positive net heat transfer values in the northern East Sea and in the Yellow Sea, indicating a transfer of heat from the atmosphere into the ocean. The Kuroshio area, and the southern East Sea affected by the Tsushima Warm Current, have the highest net heat flux of 100-190 and 60-120Wm−2 , respectively. Among the data sources, CORE2 has the highest annual average value of 79.01Wm−2 (Table 2). In contrast, ERA-40, MERRA, and NOCS have the lowest net heat flux (14.61-32.89Wm−2 ) as well as indicate the lowest latent heat flux. These match the description of Li et al. (2011) that states the net heat flux is mostly influenced by latent heat flux. CFSR and OAFlux are closest to the net heat flux (−53.2 Wm−2 ) of Hirose et al. (1996). NCEP-1, NCEP-2, CFSR, OAFlux, and CORE2 appear to indicate an inflow of the Tsushima Warm Current from the southern East Sea.

In all data sources, net heat fluxes related to the East Sea have negative values, suggesting a transfer of heat from the ocean into the atmosphere (Table 2). NOCS has the smallest value of −3.71Wm−2 , whereas CORE2 has the largest value of −72.38Wm−2 . Net heat flux of four data sources (CFSR, NCEP-1, NCEP-2, and OAFlux) is distributed in a range of −55.60 to −45.56Wm−2 . Despite the values being small, both positive and negative values appeared in the Yellow Sea. ERA-40, MERRA, and NOCS have positive values ranging from 10.0 to 27.47Wm−2 , while CFSR, NCEP-1, NCEP-2, OAFlux, and CORE2 have negative values from −27.45 to −6.5Wm−2 . Hirose et al. (1999) stated that the positive net heat flux of the Yellow Sea, meaning a transfer of heat from the atmosphere into the ocean, is overestimated. In the East China Sea, all of the data sources show negative values with the greatest differences among the data sources. MERRA indicates the smallest value (−38.91Wm−2 ), while NCEP-2 indicates the largest value (−123.60Wm−2 ).

The surface heat flux of the East Asian marginal seas varies significantly depending on each data set due to the following factors: (1) the eight heat flux data sets use different physical models, variables, parameterizations, calculation methods, reanalysis data, and observational data; (2) the input data sources may contain errors; (3) sampling problems can arise when using the data; and (4) although model reanalysis data are constrained to satisfy both model physics and observations, errors occur from both deficiencies in surface layer variables and turbulent flux parameterizations (Josey, 2001; Wang and McPhaden, 2001; Renfrew et al., 2002; Sun et al., 2003; Cronin et al., 2006; Robertson et al., 2011; Roberts et al., 2012).

### Comparison of shortwave radiation between reanalysis and observed data

Long-term direct surface heat flux measurements from the ocean are few, if any. Fortunately, shortwave radiation has been observed for a long time along the coast of the East Sea. In order to verify the shortwave radiation of the reanalysis data used in this research, they were compared to solar radiation observed in the coastal area near the East Sea from 1984 to 2006 (Fig. 7). Observed solar radiation data includes 11 data sets, two of which are from the Korea Meteorological Administration (KMA) and the remaining from the Japan Meteorological Agency (JMA). The closest areas to the observed sites are used for reanalysis data of different resolutions. Albedo used in this study is 0.06 in all calculations (Payne, 1972; Hirose et al., 1996; Knauss, 1997). In general, the monthly variation trend is more or less similar among all the data with the maximum value found during May and the minimum value found during December (Fig. 8). CORE2 and OAFlux are closely matched with observed solar radiation in terms of both seasonal and interannual variability, as well as monthly and annual averages. The remaining reanalysis data show larger values than the observational data. The data from Hirose et al. (1996) also have similar values but are slightly smaller.

### Criteria for selecting heat flux data

We attempted to find more suitable data for East Asian marginal seas by comparing the eight data according to criteria for radiant flux (shortwave and longwave radiation) and turbulent flux (latent and sensible heat flux) (Table 3). The identical conditions previously mentioned, which stated that net heat flux has negative values in the Yellow Sea, are applied together to the criteria for radiant and turbulent flux. The data that satisfy the conditions in the Yellow Sea are NCEP-1, NCEP-2, OAFlux, CFSR, and CORE2 (Table 3a). As verified previously, the data with negligible differences from the observed solar radiation in the East Sea are those of OAFlux and CORE2.

Because there are no in situ observations of latent heat and sensible heat from the ocean, we have no other choice but to make indirect comparisons using previous research results (Hirose et al., 1996, 1999; Robert et al., 2012) and various data to determine whether the calculated latent and sensible heat flux are appropriate. Three more conditions other than the net heat flux in the Yellow Sea are added for the assessment of turbulent flux (Table 3b). Considering the latent heat flux of Hirose et al. (1999) and Li et al. (2011) in the Kuroshio area to the south of Japan, latent heat flux is taken into account to ensure it satisfies 179.5-191.3Wm−2 condition in the Kuroshio (31-35°N, 131-142°E). In addition, other conditions are analyzed to determine if a sub-polar front appears in the latent heat flux of the central East Sea, and if sensible heat flux values in the southern East Sea indicate higher values than those in the northern East Sea due to the influence of the Tsushima Warm Current. As a result, CFSR and OAFlux seemed to be the only sources that satisfied all of the four conditions previously mentioned; NCEP-1 met three of the four conditions.

Considering net heat flux, OAFlux satisfied all five criteria (Table 3). However, OAFlux has an imbalance of 30Wm−2 over the period 1984-2004 with the oceans gaining heat (Yu and Weller, 2007). Therefore, in this study, two or three data sets close to each criterion were selected. Then we averaged OAFlux and CORE2 for the radiant flux and OAFlux, CFSR, and NCEP-1 for the turbulent flux to produce a new data set (called Combination Data) and analyzed seasonal and long-term changes.

## Characteristics of the Combination Data

### Annual averages

Items concerning radiant and turbulent flux selected from the data sources (OAFlux, CORE2, CFSR, and NCEP-1) are combined to derive a new data set: Combination Data. In deriving the data, data of different resolutions are re-latticized (interpolation) using a one-degree interval. The Combination Data covers 1984 through 2004, given that OAFlux, the set with the shortest time span, only covers as many years.

In the shortwave radiation of the Combination Data, the average value (136.91Wm−2 ) in the East Sea resembles the observation value (133.63Wm−2 ) calculated as previously described (Fig. 9a). Longwave radiation has a larger value near land and reaches its maximum in areas to the east and west of Korea and to the south of Japan (Fig. 9b). Furthermore, latent heat flux suitably shows a sub-polar front in the East Sea, and its value in the Kuroshio area satisfies the values determined in the criteria (179.5-191.3Wm−2 ) (Fig. 9c). In addition, the effects of the Tsushima Warm Current are well displayed in the latent and sensible heat flux (Fig. 9c and 9d). The net heat flux indicates negative values in all sea areas, including the Yellow Sea (Fig. 9e). The effects of latent and sensible heat flux are also well represented, clearly demonstrating a significant release of heat in the Kuroshio area and a sub-polar front in the East Sea. The annual average net heat flux values in the East Sea, Yellow Sea, and East China Sea were −61.84, −22.42, and −97.54 Wm−2 , respectively (Table 4).

The average surface heat flux of the Combination Data pertaining to the East Sea, Yellow Sea, and East China Sea are calculated for comparison to previous studies (Table 4). In the East Sea, the values in the Combination Data most closely matched those of Kato and Asai (1983) for shortwave radiation (Table 4a), Park et al. (1995) and Ahn et al. (1997) for longwave radiation, Park et al. (1995) and Hirose et al. (1996) for latent heat flux, Kato and Asai (1983) and Ahn et al. (1997) for sensible heat flux, and Hirose et al. (1996) for net heat flux. The net heat flux in the East Sea indicates negative values in all of the aforementioned studies.

In the Yellow Sea (Table 4b), the Combination Data demonstrate similar values to those of Hirose et al. (1999) for shortwave radiation, Hong et al. (2005) for latent heat flux, and Ahn et al. (1997) and Hong et al. (2005) for sensible heat flux. For the net heat flux, only Ahn et al. (1997) has a negative value (-15Wm−2 ) similar to that found in the Combination Data. Hirose et al. (1996) and Hong et al. (2005) have positive values.

In the East China Sea (Table 4c), shortwave radiation, latent heat flux, sensible heat flux, and net heat flux of the Combination Data most closely matched those of Hirose et al. (1996). The net heat flux in the three studies, excluding Hong et al. (2005), had negative values, indicating a release of heat from the ocean into the atmosphere. The net heat flux in Hong et al. (2005) seems positive, as its latent heat flux is much smaller than that of other studies. Such differences may be a result of different data sources and the margin of error for the coefficients of the bulk method used in calculating turbulent flux in each study.

### Seasonal variations

The shortwave radiation of the Combination Data is at its greatest value in the East China Sea and the western Pacific to the south of Japan during summer and at its least value in the East Sea during winter (Fig. 10a). In contrast, unlike other types of heat flux, longwave radiation demonstrates less seasonal variation (Fig. 10b). Its value is at a minimum (20-50Wm−2 ) during summer and a maximum (60-95Wm−2 ) during winter. In particular, the Kuroshio area to the south of Japan, the western East Sea, and the Korea Strait demonstrate a significant value (80-95Wm−2 ). The type of distribution found during winter closely resembles the annual average distribution (Fig. 9b).

In the case of latent heat flux (Fig. 11a), the maximum mean value appears during winter: 151.56Wm−2 in the East Sea, 124.84Wm−2 in the Yellow Sea, and 235.84Wm−2 in the East China Sea. The influence of the latent heat flux on the net heat flux is greater in the East China Sea than that in the East Sea or Yellow Sea (Hong et al., 2005). A significant amount of heat is released into the atmosphere from the ocean mostly in the Kuroshio: more than 100Wm−2 of heat is released annually, while the daily maximum heat loss can reach approximately 900Wm−2 during a cold air outbreak in the East China Sea (Kim and Kwon, 2003).

The maximum latent heat flux in the research area is approximately 332.74Wm−2 for the Kuroshio area during winter; on the other hand, the minimum value appears during summer. Fig. 11a shows that the Tsushima Warm Current flows into the East Sea through the Korea Strait during winter and summer while the Yellow Warm Current flows into the Yellow Sea during winter. The latent heat flux is greater during fall than it is during spring (not shown here). The variability in the sensible heat flux is insignificant from the spring through the fall (not shown here).Especially, the variability is near zero during summer (Fig. 11b): −3.91Wm−2 in the East Sea, −4.41Wm−2 in the Yellow Sea, and −1.61Wm−2 in the East China Sea. In contrast, the sensible heat flux significantly increases in all areas during winter (Fig. 11b): 132.49 Wm−2 in the East Sea, 81.24Wm−2 in the Yellow Sea, and 93.27Wm−2 in the East China Sea. Among these, the sensible heat fluxes of coastal areas in the vicinity of Russia indicate significantly high values. This may be attributed to the significant differences between atmospheric temperature and sea surface temperature as well as influences from wind.

The net heat flux appears to signal a transport of heat from the atmosphere into the ocean during spring in most of the sea areas with the exception of the Kuroshio area (Fig. 12). During summer, an influx of heat from the atmosphere into the ocean appears in all sea areas, including the Kuroshio area, with small spatial variability (80-160Wm−2 ). In general, the ocean is heated at a similar intensity: 135.47Wm−2 in the East Sea, 134.34Wm−2 in the Yellow Sea, and 104.46Wm−2 in the East China Sea. In contrast, heat is released from the ocean into the atmosphere in all sea areas again during fall, greatly influenced by an increase in the latent heat flux (not shown here). During winter, a significant amount of net heat flux is released from the ocean into the atmosphere due to the significant temperature differences between the ocean and the atmosphere. The most significant value appears in the Kuroshio area (300-460Wm−2 ) and in the southern East Sea (300-370Wm−2 ). The average value of each ocean is −292.72Wm−2 in the East Sea, −191.21Wm−2 in the Yellow Sea, and −312.98Wm−2 in the East China Sea, respectively.

### Long-term variation trend

In order to comprehend the long-term variation trend in heat flux, the Combination Data from 1984 through 2004 was analyzed (Fig. 13). The radiant flux (shortwave and longwave radiation) does not have as great a range of variability as the turbulent flux (latent and sensible heat flux). Shortwave radiation demonstrated a slightly decreasing trend in the East Sea and the Yellow Sea, while an increasing trend was found near the coastal areas of Taiwan. For longwave radiation, a decreasing trend can be seen in the coastal areas of the continents, while an increasing trend is shown in the open ocean. The latent heat flux shows an increasing trend in all areas except for the area to the west of Korea. Especially, a significant increasing trend can be found in the central area near the subpolar front of the East Sea and in the Kuroshio area to the south of Japan. The sensible heat flux demonstrates an increasing trend in the Kuroshio area and in the eastern East Sea, yet it nearly remains unchanged in the Yellow Sea and in the East China Sea. In general, an increasing trend in negative net heat flux, where heat is released from the ocean into the atmosphere, appears in nearly all of areas (Fig. 13e). In particular, a significant range of increases with a maximum −2Wm−2 year−1 was found in the eastern East Sea and in the Kuroshio area to the south of Japan.

### Horizontal heat transport in the East Sea

Changes in thermal capacity (DH) in the ocean are determined by the vertical heat flux transport (QN) via the sea surface and the horizontal heat transport (CN) via the ocean current as follows:

$D H = C N + Q N$
(2)

We assumed that the temporary change in heat capacity (DH) is 0 (zero) in the East Sea, and the net heat flux of the Combination Data was also used as vertical heat transport (QN). In the upper layer of the East Sea, the Tsushima Warm Current flows into the East Sea through the Korea Strait (Tsushima Strait), and is released into the Pacific Ocean through the Tsugaru Strait and the Sea of Okhotsk through the Soya Strait. The horizontal heat transport (CN) of the Tsushima Warm Current in the Korea, Tsugaru, and Soya straits is calculated using a formula as follows (Hirose et al., 1996):

$C N = ρ w C w T i V i ( i = 1 , 2 , 3 )$
(3)

where ρw represents the oceanic density and Cw the specific heat, approximately 1025.6 kg m−3 or 3993.1 J kg−1 C−1 , respectively. Ti= 1, 2, 3 and Vi= 1, 2, 3 are the average temperature and the volume transport in each ocean. The subscripts 1, 2, and 3 represent the Korea, Tsugaru, and Soya straits, respectively. Also, climatological monthly mean temperature of the NODC, provided by NOAA, is adopted as sea temperature in each strait. The sea temperature is depth averaged from the surface to 100 m. The temperature in the Korea, Tsugaru, and Soya straits is 17.16±1.36, 12.79±1.39, and 6.88±0.74°C, respectively.

In this study, in addition to using the same method of Hirose et al. (1996), the net heat flux value (61.84Wm−2 ) of the Combination Data of the East Sea has been used as the vertical heat transport from the ocean into the atmosphere. The difference in thenet horizontal heat transport (NHHT) due to the ocean current was calculated as follows:

$N H H T = K H T − T H T − S H T$
(4)

where KHT represents the Korea Strait heat transport, THT the Tsugaru Strait heat transport, and SHT the Soya Strait heat transport.

The horizontal heat transport was directly calculated in the same manner for all three straits. In the Korea Strait, only the horizontal heat transport to the East Sea was calculated. Conversely, in the Tsugaru and Soya straits, only the horizontal heat transport from the East Sea was calculated. We calculated the range of volume transport in each strait that is suitable for the net heat flux from the sea surface in the East Sea towards the atmosphere. Previous studies have used this exact method (Miyazaki, 1952; Leonov, 1961; Kato and Asai, 1983; Kim, 1992). Among the calculation results of several cases, we found that the inflow through the Korea Strait is 2.4-2.5 Sv (1 Sv = 10 6 m 3 s−1 ), and the outflow through the Tsugaru Strait and the Soya Strait is 1.4-1.6 Sv and 0.8-1.0 Sv, respectively. These are suitable for the volume transport of the Tsushima Warm Current in the East Sea. In these cases, the differences in the horizontal heat transport (NHHT), 62.31-67.16Wm−2 , are similar to the net heat flux of the Combination Data in the East Sea, 61.84Wm−2 . The volume transport of the Tsushima Warm Current in the Korea Strait, 2.4-2.5 Sv, is slightly smaller than the 2.64 Sv of Takikawa et al. (2005) but similar to the 2.42 Sv of Han et al. (2016), and the 1.4-1.6 Sv in the Tsugaru Strait is similar to the 1.5 Sv of Nishida et al. (2003).

## Summary and Conclusion

Eight data sets were examined in order to identify the characteristics of the surface heat flux of the seas surrounding East Asia (the East Sea, Yellow Sea, East China Sea, and the Kuroshio to the south of Japan). All of the data sources have shortwave radiation greater than the observed solar radiation in the coastal East Sea, except for CORE2 and OAFlux. The Kuroshio area has the greatest latent heat flux values of all the data sources. The latent heat flux distribution represented in NCEP-1, CFSR, OAFlux, and CORE2 seems to be the most appropriate in the Kuroshio area. The sub-polar front of the central East Sea is clearly demonstrated in ERA-40, NCEP-2, CFSR, OAFlux, and NOCS. In the Korea Strait and in the southeastern East Sea, ERA-40, NCEP-1, CFSR, and OAFlux show substantial sensible heat flux, indicating the influence of the Tsushima Warm Current. The net heat flux of the East Sea and of the East China Sea demonstrate negative values for all data sources, indicating a transfer of heat from the ocean into the atmosphere. However, the net heat flux of the Yellow Sea has both positive and negative values; NCEP-1, NCEP-2, CFSR, OAFlux, and CORE2 have negative values, whereas ERA-40, MERRA, and NOCS have positive values.

OAFlux and CORE2 have been selected as the most suitable data sources for the radiant flux in the East Asian marginal seas, while CFSR, OAFlux, and NCEP-1 were found to be the most suitable for turbulent flux among the eight surface heat flux data sources. Each data was averaged in deriving a new data set; the Combination Data. The seasonal variation characteristic of the heat flux in the Combination Data is that heat is transferred into the ocean from the atmosphere in most of the sea areas, excluding the Kuroshio area, during spring, and in most sea areas, including the Kuroshio, during summer. In contrast, heat is transferred into the atmosphere from the ocean during fall and winter. In particular, a significant net heat flux (−300- −460Wm−2 ) appears in the southern East Sea and in the Kuroshio area during winter. According to Kim et al. (2016), in the central part of the Kuroshio axis (25-30°N, 125-130°E), both heat loss in the upper mixed layer by surface heat flux and vertical heat advection mainly cause the decrease in heat storage (heat release) during autumn and winter.

A characteristic of the long-term (1984 through 2004) variation trend of the Combination Data is that its turbulent flux has a larger range of variation than that of radiant flux. Latent heat flux and sensible heat flux increased in all areas, especially the central East Sea and the Kuroshio to the south of Japan that have the greatest range in variation. The net heat flux demonstrated a trend of increasing release of heat from the ocean into the atmosphere in all areas. Most notably, the eastern East Sea and the Kuroshio have the greatest variation range of net heat flux at a maximum of −2Wm−2 year−1 .

Differences in the horizontal heat transport due to ocean currents along the straits in the East Sea were calculated, using the surface net heat flux (61.84Wm−2 ) of the Combination Data as the vertical heat transport from the ocean into the atmosphere. The differences (62.31-67.16Wm−2 ) in the horizontal heat transport indicate the volume transport of the Tsushima Warm Current as follows: the inflow through the Korea Strait is responsible for 2.4-2.5 Sv, while the outflow through the Tsugaru Strait and the Soya Strait is responsible for 1.4-1.6 Sv and 0.8-1.0 Sv, respectively.

## Figure

Map of the study area.

Spatial distributions of the annual mean shortwave radiation over the East Asian marginal seas. Contour interval is 5Wm -2.

Spatial distributions of the annual mean longwave radiation over the East Asian marginal seas. Contour interval is 5Wm -2.

Spatial distributions of the annual mean latent heat flux over the East Asian marginal seas. Contour interval is 10Wm -2.

Spatial distributions of the annual mean sensible heat flux over the East Asian marginal seas. Contour interval is 5Wm -2.

Spatial distributions of the annual mean net heat flux over the East Asian marginal seas. Contour interval is 10Wm -2.

Observational and reanalysis data points in the East Sea. Square points indicate the observation and circle points indicate each of the reanalysis data.

(a) Seasonal variation of the shortwave radiation. (b) Interannual variation of annual mean values of the shortwave radiation. In both figures, the black line indicates the observational value.

Spatial distributions of the annual mean heat fluxes using the Combination Data: (a) shortwave radiation, (b) longwave radiation, (c) latent heat flux, (d) sensible heat flux, and (e) net heat flux. Unit isWm -2. Contour interval is 5Wm -2 for (a), (b), and (d) and 10Wm -2 for (c). For (e), the contour interval is 10Wm -2 from -100 to 100Wm -2 and 20Wm -2 from -200 to - 101Wm -2.

Spatial distribution of the long-term seasonal mean radiant flux: (a) shortwave and (b) longwave radiation. Contour interval is 5Wm -2.

Spatial distribution of the long-term seasonal mean turbulent flux: (a) latent and (b) sensible heat flux. Contour interval is 10Wm -2 from 0 to 200Wm -2 and 20Wm -2 from 201 to 360Wm -2 for (a). Contour interval is 10Wm -2 for (b).

Spatial distribution of the long-term seasonal mean net heat flux. Contour interval is 20Wm -2.

Spatial distributions of the annual mean heat flux trend from 1984 to 2004 using the Combination Data. Contour interval is 0.1Wm -2 year-1 for (a), (b), (c), and (d) and 0.5Wm -2 year-1 for (e).

## Table

Sources of the Data used in the study

The annual mean values of the net heat flux in each area

Satisfaction of various conditions in the radiant, turbulent and net heat flux

Comparisons of annual mean heat fluxes in the East Asian Marginal Seas

## Reference

1. AhnB. RyuJ.H. YoonY.H. (1997) Comparative Analysis and Estimates of Heat Fluxes over the Ocean Around Korean Peninsula. , J. Atmos. Sci., Vol.33 ; pp.725-736
2. BerryD.I. KentE.C. (2009) A new air-sea interaction gridded dataset from ICOADS with uncertainty estimates. , Bull. Am. Meteorol. Soc., Vol.90 ; pp.645-656
3. BrunkeM.A. (2011) An Assessment of the Uncertainties in Ocean Surface Turbulent Fluxes in 11 Reanalysis, Satellite-Derived, and Combined Global. , J. Clim., Vol.24 (21) ; pp.5469-5493
4. BunkerA.F. (1976) Computations of surface energy flux and annual air-sea interaction cycles of the North Atlantic Ocean. , Mon. Weather Rev., Vol.104 ; pp.1122-1140
5. ClarkN.E. EberL. LaursR.M. RennerJ.A. SaurJ.F.T. (1974) Heat exchange between ocean and atmosphere in the eastern North Pacific for 1961-1971. NOAA Technical Report NMFS SSRF-682., US Department of Commerce Washington,
6. ChoH.K. (1968) Radiation balance over Korea. , Journal of the Korean Meteorological Society, Vol.4 ; pp.8-12
7. CroninM.F. FairallC.W. McPhadenM.J. (2006) An assessment of buoy-derived and numerical weather prediction surface heat fluxes in the tropical Pacific. , J. Geophys. Res., Vol.111 ; pp.C06038
8. GriffiesS.M. BiastochA. BöningC. BryanF. DanabasogluG. ChassignetE.P. EnglandM.H. GerdesR. HaakH. HallbergR.W. HazelegerW. JungclausJ. LargeW.G. MadecG. PiraniA. SamuelsB.L. ScheinertM. GuptaA.S. SeverijnsC.A. SimmonsH.L. TreguierA.M. WintonM. YeagerS. YinJ. (2009) Coordinated Ocean-ice Reference Experiments (COREs). , Ocean Model., Vol.26 ; pp.1-46
9. HanY.S. HiroseN. UsuiN. MiyazawaY. (2016) Multi-model ensemble estimation of volume transport through the straits of the East Sea. , Ocean Dyn., Vol.66 (1) ; pp.59-76
10. HiroseN. KimC.H. YoonJ.H. (1996) Heat budget in the Japan Sea. , J. Oceanogr., Vol.52 ; pp.553-574
11. HiroseN. LeeH.C. YoonJ.H. (1999) Surface Heat Flux in the East China Sea and the Yellow Sea. , J. Phys. Oceanogr., Vol.29 ; pp.401-417
12. HongG.M. KwonB.H. KimY.S. (2005) Heat Fluxes in the Marine Atmospheric Surface Layer around the Korean Peninsula based on satellite Data. , Journal of Fisheries and Marine Sciences Education, Vol.17 (2) ; pp.210-217
13. JoseyS.A. (2001) A comparison of ECMWF, NCEPNCAR, and SOC surface heat fluxes with moored buoy measurements in the subduction region of the northeast Atlantic. , J. Clim., Vol.14 ; pp.1780-1789
14. KalnayE. KanamitsuM. KistlerR. CollinsW. DeavenD. GandinL. IredellM. SahaS. WhiteG. WoollenJ. ZhuY. LeetmaaA. ReynoldsR. ChelliahM. EbisuzakiW. HigginsW. JanowiakJ. MoK.C. RopelewskiC. WangJ. JenneR. JosephD. (1996) The NCEP/NCAR 40-Year Reanalysis Project. , Bull. Am. Meteorol. Soc., Vol.77 ; pp.437-471
15. KangI.S. KimM.K. ShimT.B. (1994) Seasonal Variation of Surface heat budget and Wind Stress Over the Seas Around the Korean Peninsula. , Journal of the Korean Society of Oceanography, Vol.29 ; pp.325-337
16. KangY.J. HwangS.O. KimT.H. NamJ.C. (2001) Estimation of Air-Sea Heat Exchange Using BUOY Data at the Yellow Sea, Korea. , Journal of the Korean Earth Science Society, Vol.22 (1) ; pp.40-46
17. KatoK. AsaiT. (1983) Seasonal Variations of Heat Budgets in Both the Atmosphere and the Sea in the Japan Sea Area. , J. Meteorol. Soc. Jpn., Vol.61 ; pp.222-237
18. KimH.J. KimC.H. ShinH.R. (2016) Analysis of Sea Surface Temperature Simulation in the Northwestern Pacific and the East Asian Marginal Seas using HadGEM2-AO. , Ocean Polar Res., Vol.38 (2) ; pp.89-102
19. KimY.S. (1992) Estimate of heat transport across the sea surface near Japan with bulk methods. Ph.D. Thesis., Tokyo University,
20. KimY.S. KwonB.H. (2003) Variations of heat fluxes over the East China Sea and the southern part of the east sea based on the buoy data. , Journal of the Korean Meteorological Society, Vol.39 ; pp.337-345
21. KnaussJ.A. (1997) Introduction to physical oceanography., Prentice Hall,
22. KondoJ. (1975) Air-sea bulk transfer coefficients in diabatic conditions. , Boundary-Layer Meteorol., Vol.9 ; pp.91-112
23. LargeW.G. YeagerS.G. (2009) The global climatology of an interannually varying air-sea flux data set. , Clim. Dyn., Vol.33 ; pp.341-364
24. LeonovA.K. (1961) Basic Features of the Japan Sea Geology and Hydrology., P.P. Shirshov Institute of Oceanography Moscow,
25. LiG. RenB.H. ZhengJ.Q. YangC.Y. (2011) Net air-sea surface heat flux during 1984-2004 over the North Pacific and North Atlantic oceans (10oN-50oN): Annual mean climatology and trend. , Theor. Appl. Climatol., Vol.104 ; pp.387-401
26. MiyazakiM. (1952) The heat budget of the Japan Sea. , Bull Hokkaido Regional Fisheries Research Laboratory, Vol.4 ; pp.1-54
27. NaJ.Y. SeoJ.W. LieH.J. (1999) Annual and Seasonal Variations of the Sea Surface Heat Fluxes in the East Asian Marginal Seas. , J. Oceanogr., Vol.55 ; pp.257-270
28. NishidaY. KanomataI. TanakaI. SatoS. TakahashiS. MatsubaH. (2003) Seasonal and interannual variations of the volume transport through the Tsugaru Strait. , Umi no Kenkyu (Oceanography in Japan), Vol.12 (5) ; pp.487-499
29. ParkW.S. OhI.S. ShimT.B. (1995) Temporal and spatial distributions of heat fluxes in the East Sea (Sea of Japan). , Journal of the Korean Society of Oceanography, Vol.30 ; pp.91-115
30. PayneR.E. (1972) Albedo of the sea surface. , J. Atmos. Sci., Vol.29 ; pp.959-970
31. ReedR.K. (1977) On estimating insolation over the ocean. , J. Phys. Oceanogr., Vol.7 ; pp.482-485
32. RenfrewI.A. MooreG.W.K. GuestP.S. BumkeK. (2002) A comparison of surface layer and surface turbulent flux observations over the Labrador Sea with ECMWF analyses and NCEP reanalyses. , J. Phys. Oceanogr., Vol.32 ; pp.383-400
33. RieneckerM.M. SuarezM.J. GelaroR. TodlingR. BacmeisterJ. LiuE. BosilovichM.G. SchubertS.D. TakacsL. KimG.K. BloomS. ChenJ. CollinsD. ConatyA. SilvaA.D. GuW. JoinerJ. KosterR.D. LucchesiR. MolodA. OwensT. PawsonS. PegionP. RedderC.R. ReichleR. RobertsonF.R. RuddickA.G. SienkiewiczM. WoollenJ. (2011) MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Applications. , J. Clim., Vol.24 (14) ; pp.3624-3648
34. RobertsJ.B. RobertsonF.R. ClaysonC.A. BosilovichM.G. (2012) Characterization of Turbulent Latent and Sensible Heat Flux Exchange between the atmosphere and ocean in MERRA. , J. Clim., Vol.25 ; pp.821-838
35. RobertsonF.R. BosilovichM.G. ChenJ. MillerT.L. (2011) The effect of satellite observing system changes on MERRA water and energy fluxes. , J. Clim., Vol.24 ; pp.5197-5217
36. SahaS. MoorthiS. PanH.L. WuX. WangJ. NadigaS. TrippP. KistlerR. WoollenJ. BehringerD. LiuH. StokesD. GrumbineR. GaynoG. WangJ. HouY.T. ChuangH.Y. JuangH.M. SelaJ. IredellM. TreadonR. KleistD. DelstP.V. KeyserD. DerberJ. EkM. MengJ. WeiH. YangR. LordS. DoolH. KumarA. WangW. LongC. ChelliahM. XueY. HuangB. SchemmJ.K. EbisuzakiW. LinR. XieP. XhenM. ZhouS. HigginsW. ZouC.Z. LiuQ. ChenY. HanY. CucurullL. ReynoldsR. RutledgeG. GoldbergM. (2010) The NCEP Climate Forecast System reanalysis. , Bull. Am. Meteorol. Soc., Vol.91 ; pp.1015-1057
37. SmithS.D. (1980) Wind stress and heat flux over the ocean in gale force winds. , J. Phys. Oceanogr., Vol.10 (5) ; pp.709-726
38. SmithS.D. (1988) Coefficients for sea surface wind stress, heat flux, and wind profiles as a function of wind speed and temperature. , J. Geophys. Res., Vol.93 ; pp.15467-15472
39. SunB. YuL. WellerR.A. (2003) Comparisons of surface meteorology and turbulent heat fluxes over the Atlantic: NWP model analyses versus moored buoy observations. , J. Clim., Vol.16 ; pp.679-695
40. TakikawaT. YoonJ.H. ChoK.D. (2005) The Tsushima Warm Current through Tsushima Straits estimated from Ferryboat ADCP data. , J. Oceanogr., Vol.35 ; pp.1154-1168
41. UppalaS.M. A...llbergK. SimmonsP.W. AndraeA.J. Da CostaU. BechtoldV. FiorinoM. GibsonJ.K. HaselerJ. HernandezA. KellyG.A. LiX. OnogiK. SaarinenS. SokkaN. AllanR.P. AnderssonE. ArpeK. BalmasedaM.A. BeljaarsA.C.M. Van de BergL. BidlotJ. BormannN. CairesS. ChevallierF. DethofA. DragosavacM. FisherM. FuentesM. HagemannS. HA3lmE. HoskinsB.J. IsaksenL. JanssenP.A.E.M. JenneR. McNallyA.P. MahfoufJ-F. MorcretteJ-J. RaynerN.A. SaundersR.W. SimonP. SterlA. TrenberthK.E. UntchA. VasiljevicD. VitebroP. WoolenJ. (2005) The ERA-40 Re-Analysis. , Q. J. R. Meteorol. Soc., Vol.131 ; pp.2961-3012
42. WangW. McPhadenM.J. (2001) What is the mean seasonal cycle of surface heat flux in the equatorial Pacific? , J. Geophys. Res., Vol.106 (C1) ; pp.837-857
43. YunY.H. KimT.H. MoonJ.I. (1998) On the Seasonal Variations of the Heat Budget in the Yellow Sea. , Journal of Korean Earth Science Society, Vol.19 ; pp.1-8
44. YuL. WellerR.A. (2007) Objectively analyzed air-sea heat fluxes for the global ice-free oceans (1981-2005). , Bull. Am. Meteorol. Soc., Vol.88 ; pp.527-539
45. YuL. JinX. WellerR.A. (2008) Multi-decade Global Flux Datasets from the Objectively Analyzed Air-sea fluxes (OAFlux) Project: latent and sensible heat fluxes, ocean evaporation, and related surface meteorological variables. , OAFlux Project Technical Report (OA-2008-01),