The oceans of the world are massive carbon sinks, and play an important role in absorbing and storing greenhouse gases. Oceanic models have been the main tools used to examine the oceanic circulations and carbon cycles. Also, passive tracers have usually been utilized to test the performances of oceanic circulation models.

Chlorofluorocarbons (CFCs, mainly CFC-11 and CFC-12) are man-made carbon emissions in the atmo-sphere, which were caused by the increasing human activities during the early 20th century. CFCs entered into the oceans' surfaces through gas exchanges, and have short air-to-sea exchange times, high boiling points, low water solubility, good stability, and low noise. They have been used as passive and conservative tracers of the circulation and mixing processes on a ten-year scale in the oceans (Bullister and Weiss, 1983; Weiss et al., 1985; Wallace and Lazier, 1988; Warner and Weiss, 1992; Doney and Bullister, 1992; Smethie, 1993; Roether et al., 1993; Rhein, 1994; Warner et al., 1996; Smythe-Wright and Boswell, 1998; Andrie et al., 1999). Therefore, CFCs are able to participate in air-sea exchanges, oceanic cycles, and mixing processes as tracers on an inter-decadal time scale. They are also one of the most important tracers for studying oceanic circulation models.

Although the concentrations of CFCs in the atmosphere have been increased exponentially since the 1930s, only 30% have remained in the atmosphere, with the rest having been absorbed by the oceans and transported downwards. Many previous studies have used passive tracer CFCs to assess oceanic circulation models (England, 1995; Robitaille and Weaver, 1995; England and Hirst, 1997; Heinze et al., 1998; Dutay, 1998; Haine and Gray, 1999). Bullister et al. (1983) summarized and reviewed the process of using CFCs as tracers for oceanic circulations, and noted that CFCs are a valuable tool for assessing oceanic circulation models. England et al. (1999) used low-resolution ocean models to simulate CFCs, and the results indicated that the classical sea-air exchange formula could potentially greatly enhance the simulation depths of the CFCs. Dutay et al. (2002) compared the abilities of 13 models from Germany, Australia, Japan, France, United States, Norway, Switzerland, Britain, and Belgium, which were used to participate in CFC-11 simulation experiments of the Second Ocean Carbon Model Inter-Comparison Plan (OCMIP-2). It was found that the simulation results of the 13 models displayed large differences, especially in the Southern Ocean. Seferian et al. (2013) assessed the abilities of three ocean biogeochemical models. The results confirmed that the ventilation capabilities of the oceanic models could be obtained directly by examining the CFC-11. The global oceanic models L30T63 and LICOM, which were developed by the Institute of Atmospheric Physics of the Chinese Academy of Sciences, were used by Li et al. (2006) and Zhao et al. (2012), respectively, to study the distribution and uptake of CFC-11, as well as the possible determinants and sensitivities to the transfer velocity. Fang et al. (2014) investigated the influences of different wind fields on the uptake and transport of CFC-11 in the global oceans using a LICOM model.

In the 1980s and 1990s, the World Ocean Circulation Experiment (WOCE) observed and collected a large amount of CFC-11 concentration data. However, for global oceans, the data were very limited. Global oceanic circulation models have been used to fill in the gaps in order to compensate for the lack of observational samples. Furthermore, CFC-11 has been widely used as an inert passive tracer to inspect and assess oceanic models. The Modular Ocean Model (MOM) of the US Geophysical Fluid Dynamics Laboratory (GFDL) is one of the most widely used oceanic circulation models in the world. On the basis of Modular Ocean Model Version 4 (MOM4), a 40-level ocean model, referred to as MOM4 L40, was developed by the National Climate Center of the China Meteorological Administration (Li et al., 2015). The MOM4 L40 has been utilized to simulate the distributions of carbons and nutrients in the oceans under natural conditions (Li et al., 2015). In this study, the distribution of the CFC-11 in MOM4 L40 was simulated using CFC-11 as a tracer. The basic performance and ventilation capacities of MOM4 L40 were tested and evaluated through comparisons of the simulation results and the observed data. This study also provided reference information for future improvements and development of the model.

In the following sections, the model's framework, as well as the CFC-11 simulation experiment, are first introduced. Then, the model's results are compared with the observations of the oceanic surfaces, and vertical profiles are used to investigate the model's ventilation capacities and pathways. Finally, the study's conclusions and discussions are detailed.

The MOM4 L40 was developed by the National Climate Center of the China Meteorological Adminis-tration. It was based on the fourth version of the modular ocean model (MOMO4) constructed by the US Geophysical Fluid Dynamics Laboratory (GFDL) (Li et al., 2015). The original GFDL MOM4 model had only two versions. One version contained 28 vertical layers, and the other contained fifty vertical layers. Then, in accordance with its own computer ability, the National Climate Center of the China Meteorological Adminis-tration designed and developed the global oceanic circulation model MOM4 L40 with 40 vertical layers. In this model, a three-pole grid was used, containing one South Pole and two North Poles on the North American and Eurasian continents, respectively. The model had a horizontal resolution of (1/3)° to 1° latitude ×1° longitude, of which the resolution along the longitude circle between 30°S and 30°N was (1/3)°, and that from 30°N to the North Pole, and from 30°S to the South Pole was 1°. The upper 200 m of the oceans were divided into 20 layers, and the underneath section was divided unevenly into twenty layers, with a maximum depth of 5334 m. The vertical grid of the model was made up of (1/10)° terrain data. The important straits, such as the Taiwan Strait, were specially treated. Since the vertical and horizontal resolutions of the MOM4 L40 differed from the original GFDL MOM4 model, the MOM4 L40 also selected the best parameterization schemes of the physi-cal processes through the experimental process. These included a third-order advection scheme based on the tracer, isometric tracer mixing and diffusion scheme, Laplace horizontal friction scheme, KPP vertical mixing scheme, complete convection scheme, and short-wave penetration scheme based on the spatial distribution of chlorophyll concentration. An overflow scheme for processing the submarine boundaries and steep slopes on the seafloor, as well as the topography changes with vertical resolution and grid distribution, were all adopted in the model. This resulted in improved simulations in the regions with very gentle or steep slopes. In addition, the MOM4 L40 model was comprised of a dynamic-thermodynamic sea-ice model, of which the horizontal resolution was the same as the MOM4 L40. The sea-ice model had three vertical layers, which included a layer of snow, and two layers of ice. Therefore, the concentrations, thicknesses, temperatures, and salinity of the sea-ice could be calculated on the basis of the ice model. A detailed description of the MOM4 L40 model was presented by Li et al. (2015).

The CFC-11 entered the oceans through air-sea interactions, and was transported by physical processes, such as the horizontal and vertical movements of the oceans. In this study, by referring to the parameterization scheme for the CFC-11 air-sea flux provided by the Ocean Carbon Cycle Model Inter-Comparison Project-Phase 2 (OCMIP2), the CFC-11 exchange flux on oceans' surfaces could be expressed as the following formula:

(1)

where K_w is the gas transfer (piston) velocity (m·s^-1); C_sat is the concentration of CFC-11 in the air with saturated water vapor (mol·m^-3); C_surf represents the model's simulated concentration of CFC-11 on the oceans' surfaces (mol·mm^-3); and the C_sat is calculated as follows:

(2)

where αlpha is the solubility of the CFC-11 in the air when the water vapor becomes saturated, which was computed by using the modeled SST and SSS according to the solubility equation presented by Warner and Weiss (1985); P_CFC is the atmospheric partial of the CFC-11 in dry air at 1 atm; P is the total atmospheric pressure of the local sea surface; the monthly mean climate values from Esbensen and Kushnir (1981) were used; and P₀ equals 1 atm.

This study referred to the results of Wanninkhof (1992), and the air-sea gas transmission velocity was computed as follows:

(3)

where F_ice is the fraction of the sea surface covered by ice, which ranged from 0.0 to 1.0, and was given as monthly climatology averages by Walsh (1978) and Zwally et al. (1983); S_c is the Schmidt number which was computed using the modeled SST; X_conv and a are the constants, with values of 1/(3.6×10⁵) and 0.337, respectively; u² is the square of the monthly averaged sea surface wind speed; and represents the variance in the monthly instantaneous wind velocity. The wind speed information was obtained from the first and second Special Sensor Microwave Imager (SSMI) satellite data.

The simulation period of the atmospheric CFC-11 in this study was from 1931 to 1999. The atmo-spheric CFC-11 concentrations were provided by the United States Carbon Dioxide Information Analysis Cen-ter (CDIAC) (http://cdiac.ornl.gov/ftp/oceans/CFCATMHist/CFCATMHist2014/). The concentrations of CFC-11 in the northern hemisphere were found to be higher than those in the southern hemisphere (Fig. 1). This was due to the fact that the world's major industrial countries were mainly concentrated in the northern hemisphere. The CFCs in the southern hemisphere lagged behind the northern hemisphere for approximately two years. However, the rate of increase was the same as in the southern hemisphere (Walker et al., 2000), since the atmospheres in the southern and northern hemispheres were completely mixed through the equator in approximately two years. The atmospheric CFC-11 increased nonlinearly from the 1930s onwards, and reached maximums of 264 ppt (parts per trillion) in the northern hemisphere, and 260 ppt in the southern hemisphere (Walker et al., 2000). The atmospheric circulation, along with the larger amounts of emissions in the northern hemisphere caused the gradient between the two hemispheres to be basically maintained at 4 ppt.

Fig. 1

Fig. 1 Time series of atmospheric CFC-11 concentrations (unit: parts per trillion)

In this study's experiment, the MOM4 L40 model was first integrated for 1000 years, from the stationary state to the quasi-equilibrium state, under climatolog-ical force fields with seasonal variations, including ra-diation, air temperatures, pressure, specific humidity, wind, and so on. Then, by taking the 1000th model year as 1931, the model was forced by the atmospheric CFC-11, and integrated from 1931 to 1999, using the CFC-11 as a passive tracer. In 1931, the concentrations of CFC-11 in the oceans and atmosphere were zero.

In this study, the simulation results were compared with the observational data provided by the Global Ocean Data Analysis Project (GLODAP). The GLODAP is a comprehensive marine research project supported by the US National Oceanic and Atmospheric Administration, as well as the Department of Energy, and the National Science Foundation. It provides three-dimensional climatological analysis data, such as CFC-11, CFC-12, salinity, and so on. The data were obtained through data assimilation, based on the 10-year aeronautical data of the World Ocean Circulation Experiment (WOCE), which were averaged in the 1990s (Key et al., 2004). In this study, the average values of the model simulations for the corresponding time periods (1990 to 1999) were selected and compared with the observed data of the CFC-11 provided by the GLODAP. As a passive tracer, its absorption and storage were determined to be mainly influenced by the oceans' physical fields. Therefore, not only the CFC-11, but also the oceans' temperatures and stream functions, were analyzed.

Figure 2 details the annual mean sea surface CFC-11 concentrations of the simulations and observations, along with their differences. As can be seen in the figure, the distribution of the CFC-11 concentrations which were simulated by the model was consistent with observational results. The higher values of the sea surface concentrations were mainly distributed in the Southern Ocean, the Northwest Atlantic Ocean, the Northwest Pacific Ocean. The concentrations of CFC-11 in the middle and high latitudes were found to be higher than those in the low latitudes. Since the sea surface solubility of CFC-11 was mainly a function of temperature (Warner and Weiss, 1985), the distribution of sea surface CFC-11 concentrations was strongly influenced by the sea surface temperatures (SST), and opposite to the SST distribution (figure omitted). In the tropical regions, the temperatures were higher, and the concentrations of CFC-11 were found to be lower. In contrast, the temperatures were lower in the high latitudes, while the concentrations of CFC-11 were higher. The meridional temperature gradient in the Kuroshio area, which is located near the western Pacific, was larger. Therefore, the gradient of CFC-11 in the western Pacific Ocean was found to be larger than that in the eastern Pacific. Meanwhile, the distribution of the sea surface CFC-11 was also determined to be related to the oceans' physical fields. For example, the deep water of the low CFC-11 concentrations was transported upwards by the upwelling flow, which led to lower CFC-11 concentrations on the sea surface in the eastern Pacific Ocean. The CFC-11 concentrations were high in both the cold SST, and the strong vertical mixing zones (for example, the Southern Ocean, the Atlantic Ocean, and the northwest Pacific Ocean). For instance, the concentrations of CFC-11 in the mid-high latitude regions of the Southern Ocean were found to be the highest, and were one to two times lower in the subtropical areas.

Fig. 2

Fig. 2 Global sea surface CFC-11 concentrations (unit: pmol/kg) (a) Observations; (b) Simulations; (c) Differences between the simulations and observations.

The differences between the simulations and observational data (Fig. 2c) also showed that the simulated CFC-11values were larger than the observed values in the majority of the global oceanic areas. The simulated values were found to be 2 to 4 pmol/kg higher than the observed values in the Okhotsk and Bering Strait, and 1 to 2 pmol/kg in the Labrador Sea and the North Atlantic Ocean, as well as the South Atlantic and Pacific Oceans in the vicinity of 50°S to 70°S, which may be related to deviations in the SST simulations. The spatial correlation coefficients between the observed and simulated values in the oceans of the northern and southern hemispheres were determined to be 0.93 and 0.98, respectively. These results indicated that the simulations in the southern oceanic regions were in better agreement with observational data than those of the northern hemispheric regions.

Figure 3 shows the variations of the zonal integral inventory of the sea surface CFC-11 with latitudes. It can be seen from the figure that the concentrations of CFC-11 in the southern hemisphere were larger than that of the northern hemisphere, and the maximum observed value of approximately 68 mol·m^-2 was located near 55°S. The maximum simulated value of approximately 78 mol·m^-2 was located near 60°S, and was mainly due to the subduction of the southern hemispheric mode water (Seferian et al., 2013). These simulations were generally consistent with the observations be-tween 60°S and 70°S and 20°S to 40°S. Meanwhile, the simulations were found to be slightly larger than obser-vations in the other latitudes. The largest deviation in the southern hemisphere was located in the vicinity of 55°S to 60°S, and the largest deviation in the northern hemisphere was located in the vicinity of 40°N to 65°N, where the simulation values were larger than the ob-servation values by approximately 10 mol·m^-2. There were no observations at 65°N to 90°N, where the sim-ulation values ranged from 2 to 25 mol·m^-2. In the oceanic model, the absorption levels of CFC-11 in the southern hemisphere were higher than that of the northern hemisphere. The CFC-11 zonal integral inventories within the two hemispheres were asymmetric, which was mainly due to the land-sea distributions. When compared with the simulation results of the three models evaluated by Seferiand et al. (2013), the MOM4 L40 simulation results in the Southern Ocean were found to be in higher agreement with the observational data.

Fig. 3

Fig. 3 Zonal integral inventory of the global sea surface CFC-11 (unit: mol·m-2) Solid and dashed curves denote the observations and simulations, respectively.

The total CFC-11 column inventory is one of the most important means to test the absorption of CFC-11 in the oceans. As detailed in Fig. 4, the spatial distribution of the simulated CFC-11 inventories (Fig. 4b) was essentially consistent with the observations (Fig. 4a). In other words, the contents in the mid-high latitudes were larger than those in the low latitudes, and the content in the North Atlantic was higher than that in the North Pacific. The minimum values were located in the Arabian Sea, Bay of Bengal, eastern Indian Ocean near the equator, equatorial Pacific Ocean, and eastern Atlantic Ocean. Meanwhile, the maximum values were located between 50°N and 60°N in the northern Atlantic Ocean. However, the simulation values were found to be generally higher than observations. It was clear that there were slender and denser contours in the observations at the eastern and western boundaries of the North Atlantic Ocean. The simulation results also showed the same characteristics. However, the meridional gradient of the simulated CFC-11 was smaller, and the absorption of the CFC-11 was found to be underestimated in the model.

Fig. 4

Fig. 4 Global oceanic CFC-11 inventory (unit: 103 pmolm/kg) (a) Observations; (b) Simulations; (c) Differences between the simulations and observations.

From the viewpoint of both the observational and simulated data, it could be seen that the high CFC-11 concentrations were mainly concentrated in the Labrador Sea of the northwestern section of the Atlantic Ocean, as well as the Southern Ocean. The mixed ocean layer of the Labrador Sea extends downwards to between 800 m and 1300 m (Canuto et al., 2004), and is an important source of deep water in the northern Atlantic Ocean. At the same time, due to the presence of a strong convective mixing process in the Northwest Atlantic Ocean, the seawater with low temperatures and high concentrations can be readily transported northwards and downwards to nearly 2000 m. In this manner, the main storage area of CFC-11 was formed. In the Labrador Sea, Danish Channel, and the Icelandic-Scottish overflow area, a subduction zone of water masses from the Atlantic Ocean exists. In the deep seas of these areas, the CFC-11 has been transported southwards along the eastern and western borders of the Atlantic Ocean. It was observed that another region of high CFC-11 concentration was located in the southern oceanic regions, with a maximum of 3500 pmolm/kg. Also, the CFC-11 column inventories in the Southern Ocean (40°S to 60°S) were determined to be relatively larger than those in the northern hemisphere oceans. This was mainly due to the large amounts of CFC-11 being stored in the deep mixing layer, and the sub-polar mode water in the Southern Ocean, as well as the strong subsidence flow near the Antarctic continent where the CFC-11 was unsaturated. The simulations effectively reproduced the observational distribution characteristics. In addition, the simulation results showed that the column inventories of the CFC-11 in the Arctic Ocean were approximately between 500 and 3000 pmolm/kg. The column inventories of the CFC-11 along the southeastern coast of China, and the seawaters near the Philippines, were found to be between 500 and 1000 pmolm/kg. There were no comparable observations for these areas.

In the North Atlantic Ocean, the simulated CFC-11 column inventory was significantly smaller than the observations (Fig. 4c). It is known that the majority of models tend to underestimate the CFC-11 inventories in the North Atlantic and Southern Ocean (Ishida et al., 2007). Similarly, the MOM4 L40 model also showed an underestimation in the North Atlantic Ocean, which indicated that there was stronger meridional transport, or even excessive transport, in the North Atlantic Ocean. However, it was found that in the Southern Ocean, the MOM4 L40 overestimated the CFC-11 inventories, especially near the Vidder and Nanshai Seas. These results were possibly due to the fact that the MOM4 L40 model is a thermo-dynamic sea-ice model. The simulation results of the CFC-11 displayed a certain dependence on the model itself, as well as the forced data. Therefore, improvements in the model's physical field could potentially lead to the improvements in the temperature and salt simulation results. The basic performance of the model determined the simulation results of the basic physical fields, such as temperature and salinity, and simulation deviations in temperature and salinity will lead to simulation deviations of the CFC-11 concentrations. In this research study, the CFC-11 simulation showed large errors in the Atlantic Ocean, which were consistent with the conclusion put forward by Li et al. (2015) that the MOM4 L40 simulated sea temperature displayed large deviations in the Atlantic Ocean.

In order to conduct an improved examination of the CFC-11 transportation in the upper oceans, the global zonal profile averaged CFC-11 column inventories in the oceans above 1200 m were analyzed. Then, according to the conditions of the observational data, the CFC-11 column averaged inventories along the North Atlantic Ocean at 30°W, North Pacific Ocean at 179°E, South Pacific Ocean at 170°W, South Indian Ocean at 20°E, and South Atlantic Ocean at 0°E were investigated (Fig. 5).

Fig. 5

Fig. 5 Zonal mean CFC-11 column inventories in the global oceans (a) (unit: 1010 pmol·m2/kg); North Atlantic Ocean (b), North Pacific Ocean (c), South Pacific Ocean (d), South Indian Ocean (e), South Atlantic Ocean (f) (unit: 103pmol·m/kg) Solid and dashed curves denote the observational and simulated data, respectively.

It can be seen from Fig. 5a that, in the region between 30°S and 30°N, the zonal average of the simulated CFC-11 column inventory was close to the observational results. The simulation was more significant in the range between 70°S and 30°N, while smaller in the range between 30°N and 60°N. At approximately 60°N, the observed CFC-11 concentration was 4400 pmol·m/kg, whereas the simulated value was only 2400 pmol·m/kg. To the north of 60°N, the global average value of the simulated GFG-11 column inventory was approximately between 1400 to 2400 pmol·m/kg. The simulated CFC-11 column inventory was found to be slightly larger than the observed data in the southern oceanic regions. In the region between 60°S and 80°S, the simulations were larger than observations by 200 to 1200 pmol·m/kg.

In the North Atlantic, North Pacific, South Pacific, South Indian, and South Atlantic Oceans (Figs. 5b, 5c, 5d, 5e, 5f), the distributions of the simulated zonal mean CFC-11 column inventory was found to be in good agreement with the observations. Along the profile at 30°W in the Atlantic Ocean (Fig. 5b), the simulated values were smaller, which indicated that the model underestimated the uptake of CFC-11 in the North Atlantic Ocean. The lowest content levels of CFC-11 were observed in the vicinity of 10°N, and along the profile at 179°E in the North Pacific Ocean (Fig. 5c). At 10°N, an equatorial wind-free zone and strong upward current exist. The upwelling has brought the seawater with low CFC-11 concentrations to the surface, which has resulted in the situation that the CFC-11 column inventory in the north being higher than that in the south. Along the profile at 170°W in the South Pacific Ocean, both the observed and simulated CFC-11 column inventories were larger in the high latitudes, and smaller near the equator, with a minimum of 600 pmol·m/kg observed (Fig. 5d), which was consistent with the results presented by Dixon et al. (1996).

Along the profile at 29°E in the South Indian Ocean (Fig. 5e), an area of high CFC-11 was located near 40°S to 50°S, and the contours of the CFC-11 column inventory were reduced towards the high latitudes and equator, respectively. Along the 0°E section of the South Atlantic Ocean (Fig. 5f), the simulation results were mostly close to the observed values.

In regard to the zonal mean CFC-11 column inventory in the global oceans, the MOM4 L40 model es-sentially captured the ventilation process of the CFC-11. Two important CFC-11 storage areas which were simulated by the model were located in the southern oceanic regions at 45°S, and in the North Atlantic Ocean near 60°N. When compared with the simulation results of Danabasoglu et al. (2009), Vinu Valsala et al. (2008), and Li et al. (2007), the distribution of CFC-11 column inventory which was simulated in this study was found to be similar. For example, the MOM4 L40 model also underestimated the uptake of CFC-11 in the North Atlantic Ocean at approximately 60°N. However, the deviation was smaller, which indicated that this study's model had certain advantages in the North Atlantic Ocean.

Furthermore, the vertical distribution of the CFC-11 concentration was analyzed in this study, at the 30°W section of the northern Atlantic Ocean, 179°E section of the northern Pacific Ocean, 170°W section of the southern Pacific Ocean, 29°E section of the southern Indian Ocean, and the 0°E section of the southern Atlantic Ocean.

As previously mentioned, the northwestern Atlantic Ocean contains the main storage area for the CFC-11. Therefore, the vertical distribution of the CFC-11 concentration along the 30°W profile of the North Atlantic Ocean was further analyzed in this study (Fig. 6). The major water mass in the North Atlantic is made up of the North Atlantic Deep Water (NADW), including the Lower North Atlantic and Upper North Atlantic Deep Water (LANDW). The seawater below 1000 m is mainly determined by the renewal of the NADW. The MOM4 L40 model simulations (Fig. 6a) basically exhibited the above properties, which were also in good agreement with the observations (Fig. 6b). The CFC-11 was transported downwards from the high latitudes, with a water tongue extending from the high latitudes to the low latitudes. This is mainly due to the fact that a deep-core area is transported downwards and southwards. The area was formed due to the seawater sinking from the mixed layer into the deep layer in the North Atlantic Ocean.

Fig. 6

Fig. 6 CFC-11 concentration in the northern Atlantic Ocean along the 30°W profile (unit: pmol/kg) (a) Observations; (b) Simulation; (c) Differences between the simulation and observations.

However, there were also biases observed in the simulation (Fig. 6c). For example, a deep-core area of 2 pmol/kg was observed at 2000 m. However, this was simulated at approximately 1200 m. Similarly, the maximum penetration depth of the 0.1 pmol/kg contour was approximately 3000 m, while the simulated depth was approximately 2200 m. The simulated penetration depth was not deep enough, which suggested that when MOM4 L40 model was used to simulate the CFC-11, there were excessive southward transports of CFC-11, and the CFC-11 in the high latitudes was transported to the low latitudes. However, the depth of the downward transport was insufficient, which resulted in the simulated CFC-11 column inventory in the high-latitude regions of the Atlantic being lower than the observed values. Also, the simulations of Vinu Valsala et al. (2008) encountered a similar problem.

Figure 7 illustrates a vertical cross-section of the CFC-11 concentration along the North Pacific Ocean at 179°E. In view of the overall pattern of CFC-11 concentration distribution, the simulation results and observations were found to be consistent. The maximum concentration of the simulated CFC-11 appeared in the subsurface layer. A water tongue of the 3 pmol/kg contour line extended southwards, from the sub-polar region to 20°N, and the depth of penetration was approximately 400 m. To the north of 40°N, both the simulated and observed maximum values were 4 pmol/kg. In contrast to the temperature gradients, the high-latitude regions displayed high concentrations of CFC-11. With the ocean circulation transporting southwards, as well as the northern equatorial current and Kuroshio transport, the sea water with low CFC-11 concentrations was located to the north of the equator. The seawater carrying CFC-11 subsided from the high latitudes to the Kuroshio area, where the seawater was found to have high temperatures and low CFC-11, and reached the maximum depth between 25°N and 35°N. The maximum penetration depth was approximately 1000 m, and the contours lifted slowly southward. The depth of the penetration near 10°N was minimized, and the penetration depth to the south was found to be deepened. This is mainly due to the fact that the density of sea water near 10°N was the lowest. When compared with the observations, there were deviations observed in the southward extension of the simulated high-concentration water-tongue, while the maximum penetration depth was found to be more consistent. It can be seen from the differences between simulation and observations (Fig. 7c) that, to the south of 30°N, when compared with the observations, the simulated CFC-11 concentration was larger by 0.5 pmol/kg, and the simulation was slightly smaller in the ocean beneath 1000 m. In the range between 30°N and 50°N, the simulation above 400 m was somewhat too large. However, below 400 m it was larger by 1.5 pmol/kg. To the north of 50°N, the simulation above 200 m was also too large. However, it was determined to be 0.5 pmol/kg less below 200 m. The reasons for these biased results were probably due to a thermocline simulation bias.

Fig. 7

Fig. 7 CFC-11 concentration in the North Pacific Ocean along the 179°E profile (unit: pmol/kg) (a) Observations; (b) Simulation; (c) Differences between the simulation and observations.

Figure 8 illustrates the vertical distribution of the CFC-11 which was averaged along the profile of 170°W in the South Pacific Ocean. As seen from the figure, the simulations were in overall agreement with the observations extending from the high latitudes to the equator, and showed a "V"-shaped pattern. There was a high CFC-11concentration in the region to the south of 50°S, with the central value of approximately 3.5 pmol/kg. The CFC-11 had been transported downwards to the north from the high latitudes. The maximum penetration depths of the observed and simulated contours of 0.05 pmol/kg were both approximately 4000 m. The concentration of CFC-11 was found to decrease drastically in the area near the Antarctic continental shelf, which was reflected well in the simulations near 60°S. The circulation of the South Pacific Ocean is related to the Antarctic Circumpolar Current (ACC) in the southwest Atlantic Ocean. It is known that the ACC mainly moves in an eastwardly direction, and greatly hampers the exchange of material and energy between the north and south in the upper and middle layers of the Southern Ocean. Meanwhile, the narrower the width of the subtropical front (38°S), and the subarctic front in the West Wind Drift, the shallower the penetration depth of the CFC-11. An observed water tongue of 2 pmol/kg extended northwards to 20°S. Meanwhile, a simulated water tongue of 2 pmol/kg extended northwards to 5°S. As can be seen in Fig. 8c, to the north of 40°S, the simulation was larger than observational values above 1000 m. However, it was less than observations below 1000 m. Also, the simulation was less than observations to the south of 60°S. The simulation in the surface layer was found to be larger than the observational values by approximately 0.5 pmol/kg between 50°S and 60°S. However, from the subsurface to 1000 m, the simulation was smaller than the observational values by approximately 0.5 pmol/kg.

Fig. 8

Fig. 8 CFC-11 concentration in the South Pacific Ocean along the 170°W profile (unit: pmol/kg) (a) Observations; (b) Simulation; (c) Differences between the simulation and observations.

As illustrated in Fig. 9, along the profile at 29°E in the Indian Ocean, the CFC-11 simulation was similar to the observations. The sea surface CFC-11 of high concentration was transported northwards and downwards from the sub-polar at approximately 65°S, and the counters of the CFC-11 concentrations were relatively smooth. The penetration depths of the 0.5 pmol/kg contours were shallow, and approximately 1200 m in both the simulation and observations. However, the core region of maximum CFC-11 in the simulation extended to approximately 42°S, while it extended only to the vicinity of 46°S in the observations. The simulated values began to fall sharply to the south of 65° due to the deep penetration of the CFC-11 near the continental shelf, which may have been related to the vertical movements of the oceans. When compared with the results of Li et al. (2006), the MOM4 L40 was found to effectively reproduce the motion of the ocean in this section, and the simulation results had certain advantages. As can be seen from Fig. 9c, to the south of 60°S, the simulated value was substantially greater than the observed value. To the north of 60°S, the error was tilted downwards and northwards. In the range between 52°S and 60°S, the simulated values above 1000 m were mainly larger than the observational values. Also, to the north of 44°S, the simulated values above 1000 m were larger than the observational values.

Fig. 9

Fig. 9 CFC-11 concentration in the South Indian Ocean along the 29°E profile (unit: pmol/kg) (a) Observations; (b) Simulation; (c) Differences between the simulation and observations.

As can be seen from the vertical distribution of the CFC-11 along the 0°E section of the South Atlantic Ocean (Fig. 10), the observed and simulated CFC-11 maxima of approximately 4 pmol/kg were located to the south of 44°S. The concentration of CFC-11 to the south of 54°S was concentrated in the vicinity of 200 m, and extended northwards and southwards to the south of 54°S, with a maximum penetration depth of 1200 m located near 40°S. The simulated concentrations of CFC-11 above 1000 m in the South Atlantic Ocean were approximately 0.4 pmol/kg higher than the observed values (Fig. 12c). The vertical distribution of the CFC-11 simulated in this study was found to be similar to those of Dutay et al. (2002), and Vinu Valsala (2008), which were very similar to the observed values. The results indicated that the MOM4 L40 model displayed a good simulation performance. It is also found that the vertical distribution of the CFC-11 along the profile of 0°E in the South Atlantic Ocean was similar to the distribution along the profile of 29°E in the South Indian Ocean. This was mainly due to the fact that the southern oceans are influenced by the Antarctic flow, and the differences in the physical characteristics of each ocean are minimal (Li et al., 2006).

Fig. 10

Fig. 10 CFC-11 concentration in the South Atlantic Ocean along the 0°E profile (unit: pmol/kg) (a) Observations; (b) Simulation; (c) Differences between the simulation and observations.

As previously stated, the main CFC-11 storage areas have been determined to be located in the Northwest Atlantic Ocean, the subtropical North Pacific Ocean, and the Southern Ocean. Figs. 11a and 11b detail the simulated and observed CFC-11 penetration depths, which were defined as the vertically integrated CFC-11 concentrations of the upper 500 m divided by the surface concentrations (Dutay et al., 2002). These represented the maximum penetration depth of CFC-11 from surface to the reference depth of the vertical integration. As can be seen in Fig. 11, overall, the simulated CFC-11 penetration depth agreed well with the observations. The penetration depths in the mid-high latitude regions of the North Pacific Ocean, the North Atlantic Ocean, and the Southern Ocean were found to be larger. Meanwhile, they are smaller in the equatorial Indian Ocean, equatorial Pacific Ocean, equatorial Atlantic Ocean, the North Pacific Ocean between 50°N and 60°N, the Southern Ocean between 50°S and 70°S, Arabian Sea, and the Bay of Bengal. The maximum penetration depths were located in the mid-latitude region of the southeastern Indian Ocean, and the western section of the northern Atlantic Ocean. The penetration depths in the mid-latitude North Atlantic and the North Pacific Oceans were greater than those in the South Atlantic and South Pacific Oceans, respectively. It could be seen from the differences between simulation and observations (Fig. 11c), that the simulated values were larger than the observed values in most of the marine areas, with large deviations observed in the subtropics. For instance, in the North Pacific Ocean between 0° and 20°N, the South Pacific Ocean between 20°S and 40°S, and the West Indian Ocean between 0° and 20°S, the simulated values were approximately 100 m to 200 m larger than the observed values. This may have been due to the excessive simulation of the vertical mixing in these regions (Valsa et al., 2008).

Fig. 11

Fig. 11 Penetration depth of the surface CFC-11 concentrations with respect to the 500 m vertical integral concentration (unit: m) (a) Observations; (b) Simulation; (c) Differences between simulation and observations.

This study further analyzed the CFC-11 penetration depth in the upper oceans above 1200 m, along the global oceanic meridional circle, the 30°W section in the North Atlantic Ocean, the 179°E section in the North Pacific Ocean, the 170°W section in the South Pacific Ocean, the 29°E section in the South Indian Ocean, and the 0°E section in the South Atlantic Ocean, as illustrated in Fig. 12.

Fig. 12

Fig. 12 Mean CFC-11 penetration depths of the global oceans and along oceanic sections (unit: m) (a) Global; (b) North Atlantic Ocean; (c) North Pacific Ocean; (d) South Pacific Ocean; (e) South Indian Ocean; (f) South Atlantic Ocean. The solid and dashed curves denote the observations and simulation, respectively.

In general, in regard to the penetration depths, the simulations in the mid-high latitude region of the Southern Ocean were better than those in the mid-high latitude regions of the North Ocean. Also, the simu-lations in the South Indian Ocean and the South Atlantic Ocean were better than those in the South Pacific Ocean, which was consistent with the simulation results of Sen Gupta A et al. (2004). When combined with the CFC-11 column inventory previously discussed, the results indicated that the larger the CFC-11 column inventory, the deeper the penetration depth. Also, the smaller the column inventory, the shallower the penetra-tion depth. For example, the minimum value appeared in the area of 10°N in the North Pacific Ocean, which was caused by the boundary between the North Equatorial Current and the Counter Current (Sen Gupta A et al., 2004). The simulation effectively reproduced the process in which the CFC-11 penetration depth became shallow. However, it did not simulate the depth shallow enough. This may be related to the weak intermediate water in the North Pacific Ocean which was simulated in the model.

The oceanic meridional overturning circulation is the main channel for the meridional exchange of tem-perature and salt in the oceans. In terms of zonal av-erages, the oceanic meridional overturning circulation is a closed circulation composed of the meridional and vertical motions in the oceans, including the upper and deep circulations. This directly affects the transport of tracers within the ocean. Fig. 13 shows the MOM4 L40 model's simulation of the annual mean meridional over-turning circulation in the global oceans. The positive value denotes the clockwise flow, and the negative value denotes the counter-clockwise flow. As can be seen in Fig. 13, there were two distinct reversal currents near the equator in the upper oceans, which were generated by the Ekman transport. The maximum downward ex-tending depth of the Deacon Cell, which was approxi-mately located between 35°S and 70°S, was 3500 m in the Southern Hemisphere, while the simulated central intensity exceeded 32 Sv. The Antarctic Bottom Water (AABW) was transported northwards below the Deacon Cell. The central intensity of the simulated AABW was approximately 16 Sv. There were positive values observed between 30°N and 60°N in the northern hemisphere, which mainly rose from the deep water of the North Atlantic Ocean. There was a strong meridional flow observed from the north to the south, at the depth of 1000 m to 2000 m in the northern hemisphere, which mainly reflected the transport of the North Atlantic Deep Water (NADW). The simulation of the global averaged stream function was found to be consistent with the fact that CFC-11 was transported southwards in the northern hemisphere, and northwards in the southern hemisphere, with a significant uplift in the vicinity of 30°N to 60°N. As can be seen in Fig. 13, the center of the North Atlantic meridional flow which was simulated by using the MOM4 L40 was located between 40°N and 50°N, and at a depth of approximately 1000 m, with a central strength of 20 Sv. Roemmich and Wunsch (1985) estimated that the southward transport of deep water at 24°N in the North Atlantic Ocean was 20 Sv. Other research studies (Döös and Webb, 1994; Ganachaud et al., 2000; Holfort and Siedler, 2001; Talley et al., 2003; Zhu et al., 2014; Hu et al., 2015) also estimated the maximum center strength of the NADW as between 17 Sv and 22 Sv. Therefore, the oceanic meridional circulation was found to be reasonably simulated by the MOM4 L40 model in the current study.

Fig. 13

Fig. 13 Meridional overturning circulation for the global oceans (unit: Sv)

In this research study, a high resolution global ocean circulation model (MOM4 L40), which was coupled with a CFC-11 tracer mode, was used to simulate and analyze the distribution of global oceanic CFC-11, and the ventilation capacity of the model was further evaluated.

The results showed that the sea surface CFC-11concentration was significantly affected by sea surface tem-perature. The distribution of the sea surface CFC-11 concentration was found to be similar to the distribution of the sea surface temperature. However, the gradient of the CFC-11 concentration was the opposite of the temperature. The simulated sea surface CFC-11 concentrations of the MOM4 L40 model was high in the South Ocean, Northwest Atlantic Ocean, and Northwest Pacific Ocean, which were in agreement with observational results. The MOM4 L40 model simulated the CFC-11 concentrations in the South China and Philippine Seas, as well as the Arctic Ocean, where no CFC-11 observation data was available for comparison with the simulations. However, it also indicates that in the areas where observational data were scarce, the simulations had the ability to make up for the deficiencies. Since the global oceans' synchronous survey of CFC-11 is difficult, the observational data in the range of 60°N to 90°N were lacking in those areas. The MOM4 L40 model was successful in simulating the CFC-11 concentrations in this region due to the effects of sea ice being taken into account in the model. The results indicated that the MOM4 L40 model had certain advantages in the high-latitude regions.

The zonally integrated inventory of the annual mean concentrations of the global sea surface CFC-11 in both the simulation and observations indicated that the CFC-11 absorption was higher in the southern hemisphere than in the northern hemisphere, and also that the asymmetry of the zonally integrated CFC-11 inventory between the two hemispheres was mainly caused by the distribution of the land and sea. When compared with the three models evaluated by Seferian et al. (2013), the simulation of the MOM4 L40 model was determined to be much closer to observations in the southern oceanic regions.

The distribution of the global CFC-11 column inventory simulated by the MOM4 L40 model was similar to observations. However, the simulated values were found to be smaller than the observed values. Also, the model significantly underestimated the uptake of CFC-11 in the North Atlantic Ocean. The further analyses of the column inventory, vertical distribution, and penetration depth along the profile of 30°W in the North Atlantic Ocean, the profile of 179°E in the North Pacific Ocean, the profile of 170°W in the South Pacific Ocean, the profile of 29°E in the South Indian Ocean, and the profile of 0°E in the South Atlantic Ocean revealed that the simulations were very close to the observations. However, it was also found that the model over-simulated the vertical movements in the majority of the sea areas, which resulted in the penetration depth being deeper than the observational data. When compared with the results of previous modeling studies (Danabasoglu et al., 2009; Vinu Valsala et al., 2008; Li et al., 2007), the MOM4 L40 model also underestimated the uptake of CFC-11 in the North Atlantic Ocean. However, the deviation was less significant. This demonstrated that the MOM4 L40 model used in this study had certain advantages.

In general, the MOM L40 model reasonably reproduced the distribution characteristics of the passive tracer CFC-11 in the global oceans, and showed a better ventilating capacity. The penetration depth of the model was found to be close to the observational value. When compared with other models, the MOM4 L40 displayed its own advantages. However, the simulated values of the CFC-11 column inventory were higher than the observed values. The model underestimated the uptake of CFC-11 in the North Atlantic Ocean, which was probably related to the strong convection mixing of the simulation in that area. The MOM4 L40 simulation showed an excessively southward transport of the CFC-11. That is to say, the CFC-11 was transported from the high latitudes to the low latitudes. However, the depth of the downward transport was not sufficiently simulated, which resulted in the CFC-11 column inventory being less than the observations in the high-latitude regions of the Atlantic Ocean. Therefore, it could be seen in this study that the deviations between the simulation and the observations of the oceanic model's physical field (such as the circulation, temperature, and so on) may have caused biases in the CFC-11 simulation (Li et al., 2006). On the other hand, the atmospheric forcing data (for example, the wind fields) may also have affected the distribution and uptake of the CFC-11 in the global oceans (Fang et al., 2014). These biases will gradually be improved in the future related research endeavors.

Thanks to two reviewers for their valuable suggestions. This work was jointly supported by the National Program on Key Basic Research Project of China (2016YFA0602204, 2016YFA0602201, 2013CB430202, 2012CB955203).