Inconsistency between ship- and Argo float-based pCO2 at the intense upwelling region of the Drake Passage, Southern Ocean

1 Introduction

The Southern Ocean plays an outsized role in the global carbon cycle and therefore regulating the global climate system (Sarmiento & Toggweiler, 1984; Gruber et al., 2019a; Gruber et al., 2019b). The Southern Ocean (south of 30°S) is responsible for a disproportionately large percentage (~40%) of the total oceanic sink of anthropogenic CO₂ (Mikaloff Fletcher et al., 2006; Gruber et al., 2009; DeVries, 2014; Gruber et al., 2019a; Gruber et al., 2019b), as well as the large-scale control of nutrient supplies to the low-latitude oceans and therefore the magnitude of low-latitude biological productivity (Sarmiento et al., 2004). Observations and models have demonstrated large variability in the efficiency of uptake of CO₂ in the Southern Ocean over the past few decades. The strength of the Southern Ocean CO₂ sink was reported to have slackened from the 1980s to the early 2000s due to the increase in Southern Ocean winds which enhances the upwelling and outgassing of natural CO₂ (Le Quéré et al., 2007; Lovenduski et al., 2008). In contrast, recent studies (Landschützer et al., 2015; Munro et al., 2015b) suggest reinvigoration of the Southern Ocean CO₂ uptake since 2002, due to the cooling in the Pacific Ocean, enhanced stratification in the Atlantic and Indian Ocean sectors, and a reduced overturning (Landschützer et al., 2015; DeVries et al., 2017; Gruber et al., 2019b).

The Southern Ocean is uniquely important due in large part to its circulation. Deep waters from both the Atlantic and the Indo-Pacific are upwelled to the surface, and then transformed into intermediate waters or denser bottom waters (Lumpkin & Speer, 2007; Marshall & Speer, 2012; Talley, 2013). The upwelling mostly occurs at the southern portion of the Antarctic Circumpolar Current (ACC) (Orsi et al., 1995; Thorpe et al., 2002; Chapman et al., 2020), where deep waters return to the upper ocean via wind-driven Ekman transport (Marshall & Speer, 2012) and release the remineralized CO₂ that has been accumulated and isolated from the atmosphere over centuries. Such CO₂ release is concentrated to a relatively narrow band between 50°S and 65°S (Chen et al., 2022). Previous studies (e.g., Wu et al., 2019) have demonstrated a strong latitudinal gradient in surface dissolved inorganic carbon (DIC), where DIC is highest in the high-latitude Southern Ocean as a consequence of low temperature and upwelling; and high-DIC subsurface waters outcrop along the upwelling pathway. However, this might not be identical to the partial pressure of CO₂ (pCO₂, which fundamentally matters to air-sea CO₂ exchange) given that pCO₂ is also regulated by the excess total alkalinity (TA), seawater temperature and salinity, resulting in a distinct vertical distribution of pCO₂ (Chen et al., 2022). Therefore, a more comprehensive understanding of the upwelling impacts on surface pCO₂ is required to decipher the mechanism of upwelling-induced deep CO₂ release.

Despite its vital impact on the carbon cycle and climate system, the Southern Ocean remains one of the most poorly sampled regions of the global ocean and characterized by large uncertainty in CO₂ flux estimates (Bakker et al., 2016; Gruber et al., 2019b; Friedlingstein et al., 2022). The Drake Passage is, however, an exception with a lot of high-frequency and high-resolution observational data made since 2002 as part of the Drake Passage Time-series Project. The carbon cycle and carbonate system in Drake Passage is consequently better understood (Munro et al., 2015a; Munro et al., 2015b; Fay et al., 2018) than elsewhere in the Southern Ocean, and suggested to be representative of a broader region in the subpolar Southern Ocean in both seasonality and long-term CO₂ trends.

With the aim of filling data gaps and gaining better understandings of the Southern Ocean carbon cycle, starting since 2014, the Southern Ocean Carbon and Climate Observations and Modeling (SOCCOM) program has deployed the first biogeochemical float array across the Southern Ocean. More than 200 floats, equipped with oxygen, nitrate, pH, and bio-optical sensors (Johnson et al., 2017) have been released (https://soccom.princeton.edu/). The carbonate system data (e.g., pCO₂) were derived from the float-measured pH and algorithm-calculated TA. Based on the first several years of limited SOCCOM data, Williams et al. (2017); Gray et al. (2018) and Bushinsky et al. (2019) showed a surprisingly large CO₂ outgassing signal around the ACC zone, in particular in winter, at a much greater rate (0.36 Pg C year^-1) than previous ship-based estimates (from -0.05 to 0.03 Pg C year^-1; according to Landschützer et al., 2014 and Takahashi et al., 2009). They attributed the increases in the carbon and nutrient content of surface waters in the high-latitude Southern Ocean primarily to the scarcity of pCO₂ observations prior to the float measurements being available as well as interannual variability, e.g., a positive Southern Ocean Annular Mode Index over 2014-2017 resulted in increased wind-driven upwelling (Lovenduski et al., 2007; Lovenduski et al., 2008). However, some recent evidences with independent approaches (e.g., airborne observations, uncrewed surface vehicle observations, reconstructions of winter data, and paired CO₂-O₂ metrics; Long et al., 2021; Mackay & Watson, 2021; Sutton et al., 2021; Wu et al., 2022) showed that the biogeochemical-Argo might have overestimated sea surface pCO₂, thus adding more controversy to the accurate estimates of CO₂ flux. The discrepancy between ship-based and float-based estimates of surface pCO₂ and air-sea exchange therefore motivated us to examine the validity of such strong CO₂ outgassing in the ACC zone of the Southern Ocean. For this purpose, we take advantage of the Drake Passage dataset and its representation of the subpolar Southern Ocean carbon cycle to investigate the ship-based seasonal variability of sea surface pCO₂. In order to gain a mechanistic understanding, we develop a surface carbon balance calculation (accounting for deep entrainment of CO₂ from subsurface water and upwelled water) to estimate the theoretical value of winter surface pCO₂ under the influence of vertical mixing and upwelling.

2 Materials and methods

2.1 Study area: Drake Passage

The strong flow of the zonally unbounded Antarctic Circumpolar Current (ACC) is constricted to as narrow as ~800 km, making Drake Passage a natural laboratory for investigating the entire ACC system over a relatively short distance (Sprintall et al., 2012). We followed the previous studies (Munro et al., 2015a; Munro et al., 2015b) to divide Drake Passage into four regions according to its unique physical oceanography and geographic settings. The four regions (Figure 1A) in Drake Passage are oriented parallel to the Antarctic Polar Front (APF) and the mean flow of the ACC, with two of them (R1 and R2) located north of the APF and the other two (R3 and R4) located south of the APF. The Subantarctic Front (SAF) locates within R1, the APF locates between R2 and R3.

FIGURE 1

Figure 1 The observations from ship-based underway measurements and biogeochemical-Argo float in the Drake Passage. (A) underway pCO₂ measurements indicated by grey lines. The black dashed line indicates the location of SAF, black solid line indicates APF, and black dotted line indicates southern ACC front (front data obtained from Orsi et al., 1995; Thorpe et al., 2002). The cruises with water column sampling are denoted by purple and pink lines, in February 2009 and March 2006/September 2009, respectively. The coordinates of the box corners from the top corner of R1 to the bottom corner of R4 are: 55.06°S, 63.29°W; 57.52°S, 70.97°W; 56.49°S, 61.87°W; 58.95°S, 69.85°W; 57.92°S, 60.39°W; 60.38°S, 68.71°W; 59.34°S, 58.85°W; 61.81°S, 67.54°W; 60.77°S, 57.26°W; and 63.24°S, 66.35°W. (B) Argo floats that intersected the study area. Eight floats were selected.

2.2 Data source

We used several observational datasets containing surface pCO₂ and carbonate system parameters as described below. The austral summer is defined as from January to March, and so on for the other seasons.

2.2.1 The time-series data in Drake Passage

Discrete surface samples for parameters (e.g., salinity, macronutrients, and DIC) as well as high-frequency underway pCO₂ measurements were collected and measured on five to eight transects of Drake Passage per year by the Antarctic Research Supply Vessel Laurence M. Gould. The carbonate system parameters as well as other biogeochemical variables measured onboard (referred to as Drake Passage Time-series, DPT) allow for a comprehensive understanding of the biogeochemistry in Drake Passage. The DPT dataset during 2002 to 2017 (Figure 1A) was downloaded from https://www.ldeo.columbia.edu/res/pi/CO2/. The analytical precision was ±2 µatm for underway pCO₂, and approximately ±1 µmol kg^-1 for DIC (Munro et al., 2015a; Munro et al., 2015b). TA was calculated from DIC, pCO₂, temperature, salinity, phosphate, and silicate data. The estimated accuracy of each TA value was 2 µmol kg^-1 based on the analytical precision of pCO₂ and DIC (Takahashi et al., 2014; Munro et al., 2015a; Munro et al., 2015b). The computed TA values were relatively close to the measured TA values (of the samples with pCO₂, DIC, and TA measurements) by titration (root-mean-square deviation of ±4 µmol kg^-1).

We also took advantage of the existing water column profiles in Drake Passage in March 2006, February and September 2009 (Figure 1A). The data were obtained from the GLODAPv2.2020 dataset (Olsen et al., 2020), with uncertainties of 4 µmol kg^-1 and 6 µmol kg^-1 for DIC and TA, respectively.

2.2.2 SOCCOM float data

The SOCCOM Biogeochemical-Argo float data were downloaded from https://soccom.princeton.edu/, and 8 floats intersecting Drake Passage (from 2014 to 2020) were used in this study (Figure 1B). The float measures water column pH, oxygen, nitrate, fluorescence, and backscattering (Johnson et al., 2017). Carbonate system parameters including pCO₂ and others are first calculated from sensor-measured temperature, salinity, pH, LIAR algorithm-estimated TA, and silicate and phosphate concentrations (Williams et al., 2016; Williams et al., 2017; Carter et al., 2018). All float data were quality controlled as described in Johnson et al. (2017). Specifically, the quality control of pH data is based on the crossover analysis for deep waters between float and shipboard (including available Southern Ocean dataset and SOCCOM deployment cruises) measurements. The empirical algorithm for estimating in-situ pH as a function of temperature, salinity, pressure, and O₂ is determined for shipboard bottle measurements at 1000-2000 m depth, which is then applied to float-measured temperature, salinity, pressure, and O₂. By comparing the two pH values at 1500 m depth, an offset in pH is applied to the entire float profile. The measured pH has a reported uncertainty of 1% (equivalent to 0.005) respectively (Johnson et al., 2017), and the estimated TA and pCO₂ have a reported uncertainty of 5.6 µmol kg^-1 and 2.7% respectively (Williams et al., 2017).

The uncertainty associated with the float-estimated pCO₂ mainly results from three factors: accuracy of the estimated TA, accuracy of the measured pH, and the choice of equilibrium constants used for calculating the carbonate system (Williams et al., 2017). Among them, the accuracy of the measured pH was proven to be the dominant factor (Williams et al., 2017; Takeshita et al., 2018). According to a quality control assessment through December 2016 (Johnson et al., 2017), the pH sensor was recorded to have the lowest percentage (88%) of good data return, while the oxygen sensor had the best percentage (100%) of good data return. Some previous efforts have been made to reveal the mismatch between float-based and ship-based pH values (Álvarez et al., 2020; Wu et al., 2022).

2.3 Data processing

2.3.1 Detrend of pCO₂

In order to prevent the temporal CO₂ trends from generating artificial spatial variability due to anthropogenic CO₂ influence, we detrended the surface pCO₂ and normalized it to a reference year 2005 for a better comparison with global studies (e.g., Takahashi et al., 2014). The rate of trend of 1.67 µatm yr^-1 accounting for increasing atmospheric CO₂ during the 2002-2017 period was determined from the underway pCO₂ measurements used in this study (Figure 2A). The rate varies slightly across the particular region (R1-R4) in the Drake Passage, with deviations less than 0.23 µatm yr^-1 (Figures 2B–E), which equals to 2.76 µatm bias for a 12-year normalization from 2005 to 2017. This is within the uncertainty of pCO₂ measurements and therefore could be negligible. We further applied this rate to the entire Drake Passage when normalizing the ship-based and float-based pCO₂ to 2005.

FIGURE 2

Figure 2 The inter-annual variation in surface pCO₂ in the Drake Passage (A) and specifically in its four regions (B–E) The grey dots are the raw pCO₂ data, the red dots are the monthly mean pCO₂ values. The slope of linear regression is used to detrend surface pCO₂.

2.3.2 Surface carbon balance calculation

Figure 3 shows the seasonal changes in water mass properties in Drake Passage (Evans et al., 2014). In winter, a large amount of Antarctic Winter Water (AAWW) and Antarctic Intermediate Water (AAIW) are formed due primarily to strong surface cooling and consequent deep mixed layers, whilst in summer, AAWW is eroded through surface warming and interior mixing with surface water until the following winter. Compared to the surface water in either summer or winter, AAWW is supposed to exhibit increased salinity and DIC (Table 1). By quantifying the individual contributions of Summer Surface Water (SSW) and AAWW to the surface carbon balance in Winter Surface Water (WSW), we can then reach an estimate of how seasonal water mass transformation changes surface DIC and pCO₂.

FIGURE 3

Figure 3 Schematic showing the changes in the distributions of water masses during (A) summer and (B) winter in depth vs. latitude coordinates. The black circles with dots represent the direction of the zonal winds out of the page. SSW, Summer Surface Water; WSW, Winter Surface Water; AAWW, Antarctic Winter Water; SAMW, Subantarctic Mode Water; AAIW, Antarctic Intermediate Water; UCDW/LCDW, Upper/Lower Circumpolar Deep Water. Figure modified from Evans et al. (2014).

TABLE 1

Table 1 Values of absolute salinity (S_A), conservative temperature (T_cons), TA, and DIC in each endmember water mass.

Due to the fact that most of the float-observed strong CO₂ outgassing occurs in the ACC zone (Williams et al., 2017; Gray et al., 2018), we combined zone R3 and R4 (south of APF and north of southern ACC) in Drake Passage (Figure 1) to investigate the surface carbon balance. During the mixing of SSW and AAWW to form WSW, the fraction of AAWW (f_AAWW) was calculated as the amount necessary to change the surface concentration [X] of a specific tracer from its observed value at the beginning of the entrainment (i.e., summer months) period to the value at the end of the entrainment (i.e., winter months), according to Equations 1 and 2:

$\begin{array}{ll}{\text{f}}_{\text{SSW}}\text{ + } {\text{f}}_{\text{AAWW}}\mathrm{ = 1}& (1)\end{array}$

$\begin{array}{ll}({\text{[X]}}_{\text{SSW}}\times {\text{f}}_{\text{SSW}}) + ({\text{[X]}}_{\text{AAWW}}\times {\text{f}}_{\text{AAWW}}) = {\text{[X]}}_{\text{WSW}}& (2)\end{array}$

where [X]_SSW is the concentration in Summer Surface Water, and [X]_WSW is for Winter Surface Water. In addition, we also considered the mixture of Upper Circumpolar Deep Water (UCDW) and SSW to form WSW which is an extreme scenario that barely happens in the study area (Figure 3A). In this case the endmember AAWW was replaced by UCDW. Two conservative parameters, total alkalinity (TA) and absolute salinity (S_A), were chosen to quantify this process separately. The absolute salinity is defined as the mass fraction of salt in seawater as opposed to practical salinity which is essentially a measure of the conductivity of seawater to describe the salt content of seawater; absolute salinity is an SI unit of concentration with units of g kg^-1.

The AAWW water mass is defined by its characteristic ranges in absolute salinity and conservative temperature (T_cons), with S_A ranging from 34 g kg^-1 to 34.3 g kg^-1, and T_cons ranging from -2°C to 1°C (Evans et al., 2014). The conservative temperature is a new standard for ocean temperature adopted by the oceanographic community as part of the Thermodynamic Equation of Seawater – 2010 (TOES-10), which is based on a hypothetical adiabatic and isohaline change in pressure to the sea surface. Table 1 summarizes the average values of the two tracers as well as the carbonate parameters in each mixing endmember; the values for AAWW and UCDW were taken from depth profile data, and the values for SSW and WSW were taken from the detrended surface underway data. The Gibbs-SeaWater Toolbox was used to calculate absolute salinity and conservative temperature.

Uncertainties for the calculated fractions of different water masses and the subsequent mixed DIC and pCO₂ were estimated from a 1000-iteration Monte Carlo analysis that contained uncertainties of each input parameters as listed in Table 1. The principle of Monte Carlo analysis follows Wu et al. (2019).

3 Results

3.1 Spatiotemporal variability of ship-based surface pCO₂ in Drake Passage

The surface pCO₂ (normalized to reference year 2005) displayed distinct patterns in each season across the four regions in Drake Passage (Figure 4). Spatially, surface pCO₂ exhibit an overall decreasing trend against latitude in spring, autumn, and winter; however, became more dynamic in summer due to the possible interaction with biological activity (Brown et al., 2019). Noteworthy, a hotspot of high pCO₂ was found in R4 particularly in spring and winter, which is due to upwelling around ACC zone. The latitudinal gradient in pCO₂ was sizable in both summer and winter, with spatial variation of up to 50 µatm in winter. Temporally, summer surface waters pCO₂ ranged from 300 to 370 µatm, which were mostly undersaturated with respect to the atmospheric pCO₂ (366 µatm in 2005); however, most of them became supersaturated in winter. The regions south of the APF experienced the greatest seasonal variation in pCO₂ (~80 µatm).

FIGURE 4

Figure 4 The spatial distribution of ship-based and float-based pCO₂ in each season in the surface Drake Passage. (A–D) spring to winter. pCO₂ was normalized to year 2005.

Figures 5, 6 show the seasonal variation of surface pCO₂ and ΔpCO₂ (sea surface pCO₂ minus atmospheric pCO₂) from R1 to R4. Overall, the region acts as a weak sink of CO₂ from the atmosphere (Fig. 6E), with R1 and R2 being a near neutral region (winter CO₂ source counteracted by summer sink), and R3 and R4 being a sink region (persistent CO₂ sink except for a couple of months). The maxima of surface ocean pCO₂ were often reached in austral winter, when upward transport of deep water brings CO₂-rich water to the surface, and the minima were mostly observed in summer, when biological production draws down the surface CO₂ level (Takahashi et al., 2009). The seasonal amplitude of ΔpCO₂ is generally less than 10 µatm throughout Drake Passage (Figures 6A–D), in agreement with the comprehensive study of surface pCO₂ in Drake Passage carried out by Fay et al. (2018) and Munro et al. (2015b).

FIGURE 5

Figure 5 The comparison between ship-based and float-based monthly pCO₂ in each region (A–D) of the Drake Passage.

FIGURE 6

Figure 6 The comparison between ship-based and float-based monthly ΔpCO₂ in each region of the Drake Passage. (A–D) monthly ΔpCO₂ in different regions from R1 to R4. (E) the spatial pattern of ΔpCO₂. ΔpCO₂ means difference between the sea surface and atmospheric pCO₂. The error bars represent the standard deviations of monthly mean values.

3.2 Comparisons between ship- and float-based carbonate parameters

We constrained the 8 floats within R1-R4 regions. For comparison between different sampling years, both the ship-based and float-based pCO₂ were adjusted to the reference year 2005 according to the reported rates of anthropogenic change in the Southern Ocean (see Section 2.3.1). Since the Argo float collected data from surface to deep (2000 m) during a 10-day cycle, it happened that most of the data located in R4, and most of them were collected in winter. Significant disagreement between the ship- and float-based pCO₂ and ΔpCO₂ were found in R4 (Figures 4–6).

In R4, float-based pCO₂ and ΔpCO₂ were significantly higher than ship-based except for a couple of months (November and December). Since the atmospheric pCO₂ is rather stable in the southern hemisphere, the deviations in pCO₂ are therefore equivalent to deviations in ΔpCO₂. Such deviations (float minus ship) seem to be consistent through February to September, ranging from 6 to 24 µatm (averaged 14 µatm), with largest deviations in autumn and winter (May to August). This suggests a possible overestimation of pCO₂ and ΔpCO₂ by float-based observations. In terms of pCO₂, this overestimation means 2-6% uncertainty (6-24 µatm divided by mean pCO₂ of 363 µatm); however, in terms of ΔpCO₂, this would mean 100-400% uncertainty (6-24 µatm divided by mean of absolute ΔpCO₂ value of 6 µatm).

4 Discussion

4.1 The theoretical envelope of upwelling-induced pCO₂

Tracer values from Table 1 were applied to Equations 1-2 to estimate the fractions of AAWW and SSW present in surface water. DIC is conservative regarding the mixing process but affected by biogeochemical processes, therefore we calculated (using Equation 2) the concentrations of ‘mixed’ DIC in WSW just from mixing of AAWW and SSW DIC values; pCO₂ is not conservative regarding the mixing process, the values of ‘mixed’ pCO₂ in WSW were therefore calculated from the carbonate system with the inputs of ‘mixed’ TA and ‘mixed’ DIC, using CO₂SYS (Van Heuven et al., 2011). The choice of dissociation constants was following the global scale studies such as Wu et al. (2019).

Using TA as tracer resulted in calculated fractions of AAWW=13% and SSW=87% (fraction uncertainty of 20%), and subsequent mixed DIC of 2139 ± 13 µmol kg^-1 and pCO_{2 =} 307 ± 27 µatm, both of which are less than the observed levels in the WSW and atmosphere. Using salinity as tracer resulted in calculated fractions of AAWW = 48% (more upwelled waters than TA-derived) and SSW = 52% (fraction uncertainty of 24%), and subsequent mixed DIC of 2153 ± 15 µmol kg^-1 and pCO_{2 =} 337 ± 22 µatm, still less than the observed levels in the WSW and atmosphere.

Moreover, even if we accounted for the mixture between UCDW, AAWW, and SSW to form WSW, the results still yielded lower pCO₂ (‘mixed’ pCO₂ ranged between 310 ± 30 and 330 ± 34 µatm, with UCDW fraction ranging from 6 ± 10% to 14 ± 10% depending on different tracers) than the atmospheric level. Although it seems counterintuitive because UCDW endmember could provide enormous natural carbon to the surface ocean (DeVries et al., 2017), the mixed pCO₂ would be as low as calculated because of the countering effect from high upwelled TA (Wu et al., 2019; Chen et al., 2022).

Despite that the estimated surface pCO₂ from winter entrainment and upwelling exhibited deviations between different choice of tracers, they served as sensitivity test to each other and both suggested much lower levels of pCO₂ than atmospheric level. The physically ‘mixed’ pCO₂ was therefore not capable of inducing strong CO₂ outgassing (consistent with Figure 5); R4 thus remained a CO₂ sink region, which also agrees well with a previous assessment that Drake Passage is overall a persistent CO₂ sink in all regions (Munro et al., 2015b; Fay et al., 2018).

4.2 Is the strong winter CO₂ source plausible?

Our results imply that Drake Passage is a weak oceanic sink for CO₂ (Figure 6) and challenge the finding from SOCCOM float data that high-latitude Southern Ocean releases substantial CO₂ (Williams et al., 2017; Gray et al., 2018; Bushinsky et al., 2019). Both approaches used in this study suggest that it is implausible that winter entrainment/upwelling of deep waters into Drake Passage surface layer could produce a magnitude in ΔpCO₂ of as large as 40 µatm (which was suggested by float estimations; Williams et al., 2017) or a remarkable natural CO₂ outgassing signal (Bushinsky et al., 2019). The upwelling of high pCO₂ subsurface waters is therefore limited in causing large winter outgassing from just beneath the surface mixed layer in Drake Passage of the high-latitude Southern Ocean. Although we did not extend our approach to the wider Southern Ocean due to the lack of ship-based data that prohibited the constructions of surface carbon balance calculation, we presume that the proposed theoretical framework for Drake Passage would also work for the Indo-Pacific sector which are hot spots for upwelling of natural CO₂ from Indo-Pacific Deep Water (Chen et al., 2022; Prend et al., 2022). It is reasonable that some discrepancies between ship- and float-derive pCO₂ could be attributed to episodic or short-term (weekly to monthly) variabilities during the float observation (e.g., storms and mesoscale events) which would cause peaks in pCO₂ (Kwak et al., 2021; Nicholson et al., 2022). However, our study reveals that the inconsistency existed throughout the observation on a longer timescale (yearly; Figure 6D). Furthermore, our study is also in line with some other studies applying novel techniques (Long et al., 2021; Mackay & Watson, 2021; Sutton et al., 2021; Wu et al., 2022) to indicate the possible overestimation of CO₂ outgassing by SOCCOM floats which could be due to erroneous pH measurements or carbonate system-related biases (Álvarez et al., 2020; Wu et al., 2022).

Although it seems contradictory that Fay et al. (2018) and our study have used similar datasets (both used DPT and SOCCOM datasets) but came to different conclusions regarding the consistency between ship- and float-based pCO₂ (Fay et al. claimed that the float-based pCO₂ fall within the range of ship-based values given the uncertainty on the estimates), their study indeed revealed the overall higher float-based pCO₂ than ship-based, particularly in winter, which is consistent with our findings. Fay et al. (2018) showed in their crossover analysis between ship- and float-based pCO₂ in their Figure 9B equally large discrepancies and highly variable difference (i.e., large standard deviation). All of their findings consistently pointed to the fact that bias in float-based pCO₂ is considerable. However, Fay et al. (2018) used a different averaging approach to obtain the final uncertainty value (grey box in their Figure 9B), during which process the overall uncertainty reduced to ±11 μatm, i.e., within the range of observational uncertainties as these authors argue. Noteworthy, ± 11 µatm is by no means an acceptable uncertainty range for the pCO₂ observation, which is five times larger than the typical ship-based pCO₂ uncertainty of ±2 µatm (Pierrot et al., 2009). In all cases, pCO₂ is eventually used to calculate air-sea CO₂ flux based on F_CO2 = k×α×ΔpCO₂, where k is CO₂ gas transfer velocity, α is the CO₂ solubility, and ΔpCO₂ is the difference between sea surface and atmospheric level. Given that atmospheric pCO₂ is stable and can be regarded as a constant at short timescale, the uncertainty of F_CO2 would largely depend on the uncertainty of ΔpCO₂, in their case this would cause ~180% uncertainty in F_CO2 estimates and therefore cannot be ignored.

5 Conclusion

The autonomous biogeochemical-Argo floats have collected tremendous invaluable data to understand the global ocean carbon cycle, and made great efforts to fill the data gaps in the highly under-sampled Southern Ocean. Since the first few carbonate system studies from the SOCCOM project (Williams et al., 2017; Gray et al., 2018; Bushinsky et al., 2019) that revealed a striking CO₂ outgassing signal due to upwelling in the high-latitude Southern Ocean (particularly in winter), there has been debates on the float-based data quality and the magnitude of float-derived Southern Ocean CO₂ flux (e.g., Álvarez et al., 2020; Long et al., 2021; Mackay & Watson, 2021; Sutton et al., 2021; Wu et al., 2022). In this study, we chose the Drake Passage as a window of research to address the above issues, because it has been extensively observed by both ships and floats. We collected all ship-based pCO₂ observations in Drake Passage during 2002-2017 and compared them with 8 series of float-based observations in the same region during 2014-2020. The detrended ship- and float-based pCO₂ showed similar spatiotemporal patterns, however, a discrepancy of 14 µatm (float > ship) was found. We further applied an independent approach based on surface carbon balance to estimate the theoretical value of surface pCO₂ if it was enhanced by upwelling, and suggested that upwelling is insufficient to cause the above discrepancy. These findings are against the previous statements (e.g., Fay et al., 2018) that float-based pCO₂ fall within the range of ship-based values given the uncertainty on the estimates, although we both have found comparable deviations (averaged 14 µatm, with maximum of 24 µatm) to theirs (averaged 11 µatm, with maximum of 25 µatm). This is because the way that Fay et al. (2018) reported their results was based on relative uncertainty, therefore an uncertainty in pCO₂ ± 2.7% (equivalent to ±11 µatm) would sound reasonable, although ±11 µatm is already five times larger than the general ship-based pCO₂ uncertainty. We argue that more cautions have to be made in reporting/estimating the uncertainty, because essentially the ±2.7% pCO₂ uncertainty would mean ~180% uncertainty in CO₂ flux estimates.

Data availability statement

Publicly available datasets were analyzed in this study. This data can be found here: https://www.ldeo.columbia.edu/res/pi/CO2/; https://soccom.princeton.edu/.

Author contributions

Conceptualization, YW. Methodology, YW. Software, YW. Validation, DQ. Writing—original draft preparation, YW. Writing—review and editing, YW and DQ. All authors have read and agreed to the published version of the manuscript.

Funding

This study was funded by the Independent Research Projects of Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) (SML2021SP306), National Key Research and Development Program of China (2019YFE0114800, 2019YFC1509101), National Natural Science Foundation of China (42106222), Natural Science Foundation of Fujian Province, China (2020J05075, 2019J05148), and the Scientific Research Foundation of Ministry of Natural Resources (2019032).

Acknowledgments

We acknowledge the work by the data providers and scientists carrying out data collection, quality-control, and synthesis that led to the production of the datasets used here. We also sincerely thank the two reviewers for their constructive suggestions and comments.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Álvarez M., Fajar N. M., Carter B. R., Guallart E. F., Pérez F. F., Woosley R. J., et al. (2020). Global ocean spectrophotometric pH assessment: consistent inconsistencies. Environ. Sci. Technol. 54 (18), 10977–10988. doi: 10.1021/acs.est.9b06932

CrossRef Full Text | Google Scholar

Bakker D. C. E., Pfeil B., Landa C. S., Metzl N., O'Brien K. M., Olsen A., et al. (2016). A multi-decade record of high-quality fCO2 data in version 3 of the surface ocean CO2 atlas (SOCAT). Earth System Sci. Data 8 (2), 383–413. doi: 10.5194/essd-8-383-2016

CrossRef Full Text | Google Scholar

Brown M. S., Munro D. R., Feehan C. J., Sweeney C., Ducklow H. W., Schofield O. M. (2019). Enhanced oceanic CO2 uptake along the rapidly changing West Antarctic peninsula. Nat. Climate Change 9 (9), 678–683. doi: 10.1038/s41558-019-0552-3

CrossRef Full Text | Google Scholar

Bushinsky S. M., Landschutzer P., Rodenbeck C., Gray A. R., Baker D., Mazloff M. R., et al. (2019). Reassessing southern ocean air-sea CO2 flux estimates with the addition of biogeochemical float observations. Global Biogeochem. Cycles 33 (11), 1370–1388. doi: 10.1029/2019GB006176

CrossRef Full Text | Google Scholar

Carter B. R., Feely R. A., Williams N. L., Dickson A. G., Fong M. B., Takeshita Y. (2018). Updated methods for global locally interpolated estimation of alkalinity, pH, and nitrate. Limnol. Oceanog.: Methods 16 (2), 119–131. doi: 10.1002/lom3.10232

CrossRef Full Text | Google Scholar

Chapman C. C., Lea M.-A., Meyer A., Sallée J.-B., Hindell M. (2020). Defining southern ocean fronts and their influence on biological and physical processes in a changing climate. Nat. Climate Change 10 (3), 209–219. doi: 10.1038/s41558-020-0705-4

CrossRef Full Text | Google Scholar

Chen H., Haumann F. A., Talley L. D., Johnson K. S., Sarmiento J. L. (2022). The deep ocean’s carbon exhaust. Global Biogeochem. Cycles 36, e2021GB007156. doi: 10.1029/2021GB007156

CrossRef Full Text | Google Scholar

DeVries T. (2014). The oceanic anthropogenic CO2 sink: Storage, air-sea fluxes, and transports over the industrial era. Global Biogeochem. Cycles 28 (7), 631–647. doi: 10.1002/2013gb004739

CrossRef Full Text | Google Scholar

DeVries T., Holzer M., Primeau F. (2017). Recent increase in oceanic carbon uptake driven by weaker upper-ocean overturning. Nature 542 (7640), 215–218. doi: 10.1038/nature21068

CrossRef Full Text | Google Scholar

Evans D. G., Zika J. D., Naveira Garabato A. C., Nurser A. J. G. (2014). The imprint of southern ocean overturning on seasonal water mass variability in drake passage. J. Geophys. Research: Oceans 119 (11), 7987–8010. doi: 10.1002/2014JC010097

CrossRef Full Text | Google Scholar

Fay A. R., Lovenduski N. S., McKinley G. A., Munro D. R., Sweeney C., Gray A. R., et al. (2018). Utilizing the drake passage time-series to understand variability and change in subpolar southern ocean pCO₂. Biogeosciences 15 (12), 3841–3855. doi: 10.5194/bg-15-3841-2018

CrossRef Full Text | Google Scholar

Friedlingstein P., Jones M. W., O'Sullivan M., Andrew R. M., Bakker D. C. E., Hauck J., et al. (2022). Global carbon budget 2021. Earth Syst. Sci. Data 14 (4), 1917–2005. doi: 10.5194/essd-14-1917-2022

CrossRef Full Text | Google Scholar

Gray A. R., Johnson K. S., Bushinsky S. M., Riser S. C., Russell J. L., Talley L. D., et al. (2018). Autonomous biogeochemical floats detect significant carbon dioxide outgassing in the high-latitude southern ocean. Geophys. Res. Lett. 45 (17), 9049–9057. doi: 10.1029/2018GL078013

CrossRef Full Text | Google Scholar

Gruber N., Clement D., Carter B. R., Feely R. A., van Heuven S., Hoppema M., et al. (2019a). The oceanic sink for anthropogenic CO2 from 1994 to 2007. Science 363 (6432), 1193–1199. doi: 10.1126/science.aau5153

CrossRef Full Text | Google Scholar

Gruber N., Gloor M., Mikaloff Fletcher S. E., Doney S. C., Dutkiewicz S., Follows M. J., et al. (2009). Oceanic sources, sinks, and transport of atmospheric CO2. Global Biogeochem. Cycles 23 (1), GB1005. doi: 10.1029/2008GB003349

CrossRef Full Text | Google Scholar

Gruber N., Landschützer P., Lovenduski N. S. (2019b). The variable southern ocean carbon sink. Ann. Rev. Mar. Sci. 11 (1), 159–186. doi: 10.1146/annurev-marine-121916-063407

CrossRef Full Text | Google Scholar

Johnson K. S., Plant J. N., Coletti L. J., Jannasch H. W., Sakamoto C. M., Riser S. C., et al. (2017). Biogeochemical sensor performance in the SOCCOM profiling float array. J. Geophys. Research: Oceans 122 (8), 6416–6436. doi: 10.1002/2017JC012838

CrossRef Full Text | Google Scholar

Kwak K., Song H., Marshall J., Seo H., McGillicuddy D. J. Jr. (2021). Suppressed pCO₂ in the Southern Ocean due to the interaction between current and wind. Journal of Geophysical Research: Oceans, 126, e2021JC017884. doi: 10.1029/2021JC017884

CrossRef Full Text | Google Scholar

Landschützer P., Gruber N., Bakker D., Schuster U. (2014). Recent variability of the global ocean carbon sink. Global Biogeochem. Cycles 28 (9), 927–949. doi: 10.1002/2014GB004853

CrossRef Full Text | Google Scholar

Landschützer P., Gruber N., Haumann F. A., Rödenbeck C., Bakker D. C. E., van Heuven S., et al. (2015). The reinvigoration of the southern ocean carbon sink. Science 349 (6253), 1221–1224. doi: 10.1126/science.aab2620

CrossRef Full Text | Google Scholar

Le Quéré C., Rödenbeck C., Buitenhuis E. T., Conway T. J., Langenfelds R., Gomez A., et al. (2007). Saturation of the southern ocean CO2 sink due to recent climate change. Science 316 (5832), 1735–1738. doi: 10.1126/science.1136188

CrossRef Full Text | Google Scholar

Long M. C., Stephens B. B., McKain K., Sweeney C., Keeling R. F., Kort E. A., et al. (2021). Strong southern ocean carbon uptake evident in airborne observations. Science 374 (6572), 1275–1280. doi: 10.1126/science.abi4355

CrossRef Full Text | Google Scholar

Lovenduski N. S., Gruber N., Doney S. C. (2008). Toward a mechanistic understanding of the decadal trends in the southern ocean carbon sink. Global Biogeochem. Cycles 22 (3), GB3016. doi: 10.1029/2007gb003139

CrossRef Full Text | Google Scholar

Lovenduski N. S., Gruber N., Doney S. C., Lima I. D. (2007). Enhanced CO2 outgassing in the southern ocean from a positive phase of the southern annular mode. Global Biogeochem. Cycles 21 (2), GB2026. doi: 10.1029/2006gb002900

CrossRef Full Text | Google Scholar

Lumpkin R., Speer K. (2007). Global ocean meridional overturning. J. Phys. Oceanog. 37 (10), 2550–2562. doi: 10.1175/JPO3130.1

CrossRef Full Text | Google Scholar

Mackay N., Watson A. (2021). Winter air-sea CO2 fluxes constructed from summer observations of the polar southern ocean suggest weak outgassing. J. Geophys. Research: Oceans 126, e2020JC016600. doi: 10.1029/2020JC016600

CrossRef Full Text | Google Scholar

Marshall J., Speer K. (2012). Closure of the meridional overturning circulation through southern ocean upwelling. Nat. Geosci. 5 (3), 171–180. doi: 10.1038/ngeo1391

CrossRef Full Text | Google Scholar

Mikaloff Fletcher S. E., Gruber N., Jacobson A. R., Doney S. C., Dutkiewicz S., Gerber M., et al. (2006). Inverse estimates of anthropogenic CO2 uptake, transport, and storage by the ocean. Global Biogeochem. Cycles 20 (2), GB2002. doi: 10.1029/2005gb002530

CrossRef Full Text | Google Scholar

Munro D. R., Lovenduski N. S., Stephens B. B., Newberger T., Arrigo K. R., Takahashi T., et al. (2015a). Estimates of net community production in the southern ocean determined from time series observations (2002–2011) of nutrients, dissolved inorganic carbon, and surface ocean pCO₂ in drake passage. Deep Sea Res. Part II: Topical Stud. Oceanog. 114, 49–63. doi: 10.1016/j.dsr2.2014.12.014

CrossRef Full Text | Google Scholar

Munro D. R., Lovenduski N. S., Takahashi T., Stephens B. B., Newberger T., Sweeney C. (2015b). Recent evidence for a strengthening CO₂ sink in the southern ocean from carbonate system measurements in the drake passage (2002–2015). Geophys. Res. Lett. 42 (18), 7623–7630. doi: 10.1002/2015GL065194

CrossRef Full Text | Google Scholar

Nicholson S.-A., Whitt D. B., Fer I., du Plessis M. D., Lebéhot A. D., Swart S., Monteiro P. M.S. (2022). Storms drive outgassing of CO2 in the subpolar Southern Ocean. Nature Communications, 13(1), 158. doi: 10.1038/s41467-021-27780-w

CrossRef Full Text | Google Scholar

Olsen A., Lange N., Key R. M., Tanhua T., Bittig H. C., Kozyr A., et al. (2020). An updated version of the global interior ocean biogeochemical data product, GLODAPv2.2020. Earth System Sci. Data 12 (4), 3653–3678. doi: 10.5194/essd-12-3653-2020

CrossRef Full Text | Google Scholar

Orsi A. H., Whitworth T., Nowlin W. D. (1995). On the meridional extent and fronts of the Antarctic circumpolar current. Deep Sea Res. Part I: Oceanographic Res. Papers 42 (5), 641–673. doi: 10.1016/0967-0637(95)00021-W

CrossRef Full Text | Google Scholar

Pierrot D., Neill C., Sullivan K., Castle R., Wanninkhof R., Lüger H., et al. (2009). Recommendations for autonomous underway pCO₂ measuring systems and data-reduction routines. Deep Sea Res. Part II: Topical Stud. Oceanog. 56 (8-10), 512–522. doi: 10.1016/j.dsr2.2008.12.005

CrossRef Full Text | Google Scholar

Prend C. J., Gray A. R., Talley L. D., Gille S. T., Haumann F. A., Johnson K. S., et al. (2022). Indo-pacific sector dominates southern ocean carbon outgassing. Global Biogeochem. Cycles 36, e2021GB007226. doi: 10.1029/2021GB007226

CrossRef Full Text | Google Scholar

Sarmiento J. L., Gruber N., Brzezinski M. A., Dunne J. P. (2004). High-latitude controls of thermocline nutrients and low latitude biological productivity. Nature 427 (6969), 56–60. doi: 10.1038/nature02127

CrossRef Full Text | Google Scholar

Sarmiento J. L., Toggweiler J. R. (1984). A new model for the role of the oceans in determining atmospheric pCO₂. Nature 308 (5960), 621–624. doi: 10.1038/308621a0

CrossRef Full Text | Google Scholar

Sprintall J., Chereskin T. K., Sweeney C. (2012). High-resolution underway upper ocean and surface atmospheric observations in drake passage: Synergistic measurements for climate science. Oceanography 25 (3), 70–81. doi: 10.5670/oceanog.2012.77

CrossRef Full Text | Google Scholar

Sutton A. J., Williams N. L., Tilbrook B. (2021). Constraining southern ocean CO2 flux uncertainty using uncrewed surface vehicle observations. Geophys. Res. Lett. 48 (3), e2020GL091748. doi: 10.1029/2020GL091748

CrossRef Full Text | Google Scholar

Takahashi T., Sutherland S. C., Chipman D. W., Goddard J. G., Ho C., Newberger T., et al. (2014). Climatological distributions of pH, pCO₂, total CO₂, alkalinity, and CaCO3 saturation in the global surface ocean, and temporal changes at selected locations. Mar. Chem. 164 (0), 95–125. doi: 10.1016/j.marchem.2014.06.004

CrossRef Full Text | Google Scholar

Takahashi T., Sutherland S. C., Wanninkhof R., Sweeney C., Feely R. A., Chipman D. W., et al. (2009). Climatological mean and decadal change in surface ocean pCO₂, and net sea–air CO2 flux over the global oceans. Deep Sea Res. II 56 (8), 554–577. doi: 10.1016/j.dsr2.2008.12.009

CrossRef Full Text | Google Scholar

Takeshita Y., Johnson K. S., Martz T. R., Plant J. N., Sarmiento J. L. (2018). Assessment of autonomous pH measurements for determining surface seawater partial pressure of CO2. J. Geophys. Research: Oceans 123 (6), 4003–4013. doi: 10.1029/2017JC013387

CrossRef Full Text | Google Scholar

Talley L. D. (2013). Closure of the global overturning circulation through the Indian, pacific, and southern oceans: Schematics and transports. Oceanography 26 (1), 80–97. doi: 10.5670/oceanog.2013.07

CrossRef Full Text | Google Scholar

Thorpe S. E., Heywood K. J., Brandon M. A., Stevens D. P. (2002). Variability of the southern Antarctic circumpolar current front north of south Georgia. J. Mar. Syst. 37 (1), 87–105. doi: 10.1016/S0924-7963(02)00197-5

CrossRef Full Text | Google Scholar

Van Heuven S., Pierrot D., Rae J. W. B., Lewis E., Wallace D. W. R. (2011). MATLAB program developed for CO₂ system calculations. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tennessee. doi: 10.3334/CDIAC/otg.CO2SYS_MATLAB_v1.1

CrossRef Full Text | Google Scholar

Williams N. L., Juranek L. W., Feely R. A., Johnson K. S., Sarmiento J. L., Talley L. D., et al. (2017). Calculating surface ocean pCO₂ from biogeochemical argo floats equipped with pH: An uncertainty analysis. Global Biogeochem. Cycles 31 (3), 591–604. doi: 10.1002/2016GB005541

CrossRef Full Text | Google Scholar

Williams N. L., Juranek L. W., Johnson K. S., Feely R. A., Riser S. C., Talley L. D., et al. (2016). Empirical algorithms to estimate water column pH in the southern ocean. Geophys. Res. Lett. 43 (7), 3415–3422. doi: 10.1002/2016GL068539

CrossRef Full Text | Google Scholar

Wu Y., Bakker D. C. E., Achterberg E. P., Silva A. N., Pickup D. D., Li X., et al. (2022). Integrated analysis of carbon dioxide and oxygen concentrations as a quality control of ocean float data. Commun. Earth Environ. 3 (1), 92. doi: 10.1038/s43247-022-00421-w

CrossRef Full Text | Google Scholar

Wu Y., Hain M. P., Humphreys M. P., Hartman S., Tyrrell T. (2019). What drives the latitudinal gradient in open-ocean surface dissolved inorganic carbon concentration? Biogeosciences 16 (13), 2661–2681. doi: 10.5194/bg-16-2661-2019

CrossRef Full Text | Google Scholar