SEARCH

SEARCH BY CITATION

Keywords:

  • carbon dioxide;
  • oceanic sink;
  • North Atlantic

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

[1] Between 1983 and 2005, continuous oceanic CO2 observations at two time series sites in the North Atlantic Ocean near Bermuda indicate that surface seawater dissolved inorganic carbon (DIC) and pCO2 increased annually at rates similar to that expected from oceanic equilibration with increasing CO2 in the atmosphere. In addition, seawater pH, CO32− ion concentrations, and CaCO3 saturation states have also decreased over time. There was considerable seasonal asymmetry in the oceanic CO2 sink or source rates, with wintertime air-to-sea CO2 influx greater than the summertime sea-to-air CO2 efflux. On an annual basis, the region was an oceanic sink for CO2, with a mean net annual air-sea CO2 flux rate of −815 ± 251 and −1295 ± 294 mmol CO2 m−2 yr−1, respectively, estimated using different synoptic and data assimilation model wind speed data sets. Peak-to-peak variability of ∼850–1950 mmol CO2 m−2 yr−1 represented an interannual variability of ∼0.2–0.3 Pg C yr−1 in the oceanic CO2 sink scaled to the subtropical gyre of North Atlantic Ocean. The long-term trend over the 1983–2005 period was a slight increase in the oceanic CO2 sink, associated primarily with a gradual increase in wind speed over the same period. Interannual variability of summertime (June–September) and fall (October–December) air-sea CO2 flux rates were correlated to the North Atlantic Oscillation (NAO) and strongly influenced by wind events such as hurricanes. Wintertime (January–May) air-sea CO2 flux rates were poorly correlated with the NAO and Arctic Oscillation (AO), although gas exchange rates were ∼11–40% higher during concurrent El Niño periods compared to La Niña periods.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

[2] Continuous time series of atmospheric CO2 over the last 50 years show upward trends of CO2 due to the release of anthropogenic CO2 and its accumulation in the atmosphere. The ocean uptake of atmospheric CO2 through air-sea CO2 gas exchange each year is estimated at ∼1.9–2.2 Pg C (Pg = 1015 g C [Takahashi et al., 2002; Le Quére et al., 2003]), while tracers indicate that ∼118 ± 19 Pg C of anthropogenic CO2 has accumulated in the global ocean since pre-industrial times [Sabine et al., 2004a]. Although the rate of surface ocean CO2 change due to the accumulation of anthropogenic CO2 in the atmosphere can be theoretically calculated (i.e., from the surface ocean buffer factor assuming that near-surface waters have residence times long enough to equilibrate entirely with the anthropogenic perturbation in atmospheric CO2), direct determination of long-term oceanic trends is limited to a few oceanic observational studies.

[3] There are several continuous ocean time series observations of seawater CO2 of sufficient length to evaluate long-term interannual trends of seawater CO2 and air-sea CO2 fluxes. These ocean time series, include, for example: (1) BATS (Bermuda Atlantic Time series Study), located near Bermuda (31°43′N, 64°10′W) in the NW Atlantic Ocean [e.g., Bates et al., 1996a, 2002]; (2) Hydrostation S (32°10′N, 64°30′W) located near Bermuda in the NW Atlantic Ocean [e.g., Keeling, 1993]; (3) ESTOC (European Station for Time series in the Ocean Canary Islands (ESTOC), located near Gran Canaria in the NE Atlantic Ocean [e.g., Gonzalez-Dávila et al., 2003; Santana-Casiano et al., 2007]; (4) station ALOHA (A Long-term Oligotrophic Habitat Assessment), located near Hawaii (22°45′N, 158°W) in the North Pacific Ocean [e.g., Karl et al., 2001; Dore et al., 2003; Keeling et al., 2004; Brix et al., 2004]; (5) Ocean Weather Station P/line P in the North-east Pacific [Wong et al., 1999], and; (6) Kyodo North Pacific Ocean Time series (KNOT; 44°N, 155°E) in the western North Pacific subpolar region [Wakita et al., 2005]. There are other studies in which intermittent repeat occupations of regions or sections allow estimation of interannual changes in oceanic CO2 and air-sea CO2 fluxes [e.g., Midorikawa et al., 2003; Takahashi et al., 2003; Olsen et al., 2003; Sabine et al., 2004b; Lefèvre et al., 2004; Yoshikawa-Inoue and Ishii, 2005; Friis et al., 2005].

[4] The long-term trends of oceanic CO2 at the BATS site in the North Atlantic Ocean near Bermuda have been examined previously [Bates et al., 1996a; Bates, 2001; Bates et al., 2002; Gruber et al., 2002]. The Hydrostation S and BATS CO2 sampling time series began in 1983 and 1988, respectively, and early studies suggested that surface seawater layer dissolved inorganic carbon (DIC) and partial pressure of CO2 (pCO2) have increased at a rate higher than expected with equilibrium with the anthropogenic perturbation of atmospheric CO2 [Bates et al., 1996a; Bates, 2001]. However, as the length of the BATS record has grown, surface seawater DIC and pCO2 appear to increase at a rate closer to equilibrium with increasing atmospheric CO2 [Bates et al., 2002; Gruber et al., 2002]. Previous studies on air-sea CO2 gas exchange have indicated that the subtropical gyre of the North Atlantic is a net oceanic sink for atmospheric CO2, but with highly variable rates (∼0.3–1.9 mol CO2 m−2 yr−1 [Bates et al., 1996a, 1998a; Gruber et al., 2002]). Furthermore, the air-sea CO2 exchange appears highly influenced by strong wind events associated with tropical cyclones (i.e., hurricanes) or winter storms [Bates et al., 1998b; Bates and Merlivat, 2001; Bates, 2002].

[5] Here interannual trends of oceanic CO2 variability are examined from a composite data set of BATS and Hydrostation S spanning a period of 22 years from 1983 to 2005. Surface DIC and pCO2 observations are evaluated to determine whether the subtropical gyre of the North Atlantic increases at an equilibrium rate with atmospheric CO2. Seawater trends of total alkalinity (TA), pH and CO32− ion concentration are examined for geochemical evidence of marine calcification [e.g., Bates et al., 1996b], and long-term changes in carbonate chemistry in response to the anthropogenic perturbation of atmospheric CO2. These long-term trends are compared with previous studies in the North Atlantic Ocean [Bates et al., 1996a, Bates, 2001; Bates et al., 2002; Gruber et al., 2002] and with oceanic CO2 time series records from other oceanic basins (e.g., ALOHA, KNOT, ESTOC).

[6] Rates of air-sea CO2 flux for the North Atlantic Ocean near Bermuda are determined each year for the 1984 to 2005 period using synoptic meteorological observations from the Bermuda Weather Service (BWS), Comprehensive Ocean-Atmosphere Data Set (COADS), and data assimilation models such as NCEP/NCAR reanalysis 2 (NNR) and European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-40 reanalysis. The impact of high wind events (such as hurricanes) and changing wind speeds on air-sea CO2 fluxes are examined over the 1984 to 2005 period.

[7] Here we examine whether physical changes in the North Atlantic that manifest as low-frequency modes of variability correlate with interannual changes in air-sea CO2 flux. The low-frequency modes of climate variability relevant to this study include the North Atlantic Oscillation (NAO) [Rogers, 1990; Hurrell, 1995; Jones et al., 1997; Osborn et al., 1999; Hurrell and Van Loon, 1997; Hurrell et al., 2001, 2002], Arctic Oscillation (AO) [Thompson and Wallace, 1998; Wallace, 2000; Marshall et al., 2001], and El Niño−Southern Oscillation (ENSO) [Zhang et al., 1996; Bojariu, 1997; Penland and Matrosova, 1998]. Low-frequency modes such as NAO, AO, and ENSO cannot themselves exert influence on air-sea CO2 fluxes, but, rather on physical phenomena such as sea surface temperature and wind speed that can in turn influence seawater pCO2 and air-sea CO2 flux. Previous studies on interannual changes in the subtropical gyre of the North Atlantic have shown that changes in the physical properties of the region [e.g., Jenkins, 1982; Dickson et al., 1996; Klein and Hogg, 1996; Joyce and Robbins, 1996; Rodwell et al., 1999; Alfutis and Cornillon, 2001; Curry and McCartney, 2001; McKinley et al., 2004] and biogeochemical dynamics of CO2, inorganic nutrients, and phytoplankton community structure are correlated to NAO variability [e.g., Bates, 2001; Oschlies, 2001; Gruber et al., 2002; Bates and Hansell, 2004; Lomas and Bates, 2004].

2. Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

2.1. Data Sources and Analyses

[8] A composite time series of oceanic CO2 data from Hydrostation S and BATS was used in this study. Both time series sites are located only ∼50 km apart in the subtropical gyre of the North Atlantic near Bermuda. Recent analysis of physical properties from Hydrostation S and BATS indicates that both sites have very similar seasonal and annual patterns of physical variability [Phillips and Joyce, 2006]. With this consideration in mind, it is assumed that both time series CO2 data sets are representative of this region of the subtropical gyre; combining both data sets thereby extends the seawater CO2 observational record back to 1983.

[9] For the period from September 1983 to September 1988, surface DIC and total alkalinity data were collected at the Hydrostation S (or Panularis) site; hydrographic sampling has been conducted at Hydrostation S continuously at biweekly intervals since 1954. DIC and TA were determined analytically using manometric and potentiometric techniques, respectively, at Scripps Institution of Oceanography [Keeling, 1993]. For the period, September 1988 to December 2005, DIC and TA samples were collected at the BATS site [Bates et al., 1996a; Bates, 2001] and analyzed at the Bermuda Institute of Ocean Sciences (BIOS). A comprehensive monthly hydrographic and biogeochemical sampling program commenced at the BATS site in September 1988. DIC was determined by gas extraction and coulometry [Bates et al., 1996a, 2002] using a Single-Operator MultiMetabolic Analyzer (SOMMA) system [Johnson et al., 1987, 1993; Dickson and Goyet, 1994].

[10] Although different analytical techniques were used for DIC analyses at Hydrostation S and BATS, both methodologies (i.e., manometry and coulometry) have very high precision (∼0.03%; 0.5 μmoles kg−1). In addition, the accuracy of BATS DIC data was evaluated using routine analysis of certified reference materials (CRMs), traceable standards for both DIC and total alkalinity [Dickson et al., 2003]. Since the CRM values were certified by manometry at Scripps Institution of Oceanography, the long-term accuracy of DIC for both Hydrostation S and BATS data was estimated at ∼0.04% (∼0.8 μmoles kg−1). In addition, Gruber et al. [2002] found a small mean difference (<1 μmoles kg−1) between BATS and Hydrostation S surface DIC samples for the period 1988−1998.

[11] Total alkalinity samples from BATS were routinely analyzed using potentiometry at BIOS along with CRMs since 1992 [Bates et al., 1996b], with a potential accuracy of ∼0.1% (∼2 μmoles kg−1). Analysis of total alkalinity samples from Hydrostation S using similar potentiometric techniques predated the introduction of CRM's, and evaluation of long-term inaccuracy is not possible. However, the long-term trend of salinity-normalized total alkalinity (i.e., TA corrected to a constant salinity of 36.6 to account for local evaporation and precipitation changes), over the 1983 to 2005 time period, was close to 0 (Table 1) providing evidence that inaccuracy of total alkalinity over time was small (<2−3 μmoles kg−1).

Table 1. Long-Term Trends (1983–2005) of Surface Seawater Temperature, Salinity, DIC, nDIC, TA, nTA, Seawater pCO2 (pCO2sea), Atmospheric pCO2 (pCO2atm), pH, CO32− Ion Concentration, and Saturation States of Aragonite (Ωarag) and Calcite (Ωcalcite) in the North Atlantic Ocean Near Bermudaa
ParameterPeriod22 Year ChangeSlope and Standard Errornr2p-Value
  • a

    From September 1983 to September 1988, surface DIC, TA, temperature and salinity data were collected at Hydrostation S (32° 10′N, 64° 30′W) [Keeling, 1993]. From September 1988 to December 2003, water-column DIC, TA, temperature and salinity data were collected at the BATS site (31° 50'N, 64° 10'W) [Bates, 2001; Bates et al., 2002]. For the period September 1988-December 2000, TA was computed from salinity data using the salinity–total alkalinity relationships observed at the BATS site [Bates et al., 1996]; nDIC and nTA were computed from DIC and TA data corrected to a constant salinity of 36.6 to account for local evaporation and precipitation changes. Seawater pCO2, pH, CO32− ion concentration and saturation states for calcite and aragonite were calculated from DIC, TA, temperature and salinity data. In addition, long-term trends were determined for atmospheric pCO2 data from two atmospheric monitoring sites (Terceira Island, Açores, September 1983 to September 1988; Bermuda, August 1988 to December 2003). Regression statistics (slope, error, r2 and p-value) were determined using a least squares fitting routine using singular value decomposition method. Surface layer is 0–10 m.

Temp.09/1983–12/2005+0.37 °C+0.017 ± 0.0302740.000.56
Salinity09/1983–12/2005+0.19+0.0084 ± 0.00152740.11<0.01
DIC09/1983–12/2005+30.6 μmoles kg−1+1.37 ± 0.162740.21<0.01
nDIC09/1983–12/2005+19.1 μmoles kg+0.86 ± 0.112740.18<0.01
TA09/1983–12/2005+13.4 μmoles kg−1+0.61 ± 0.112430.11<0.01
nTA09/1983–12/2005+1.4 μmoles kg−1+0.06 ± 0.042430.010.15
pCO2sea09/1983–12/2005+37.3 μatm+1.67 ± 0.282740.11<0.01
pCO2atm09/1983–12/2005+39.2 μatm+1.78 ± 0.0211020.91<0.01
pH09/1983–12/2005−0.037−0.0017 ± 0.00032740.10<0.01
CO32−09/1983–12/2005−10.4 μmoles kg−1−0.47 ± 0.092740.10<0.01
Ωarag09/1983–12/2005−0.16−0.007 ± 0.0022740.06<0.01
Ωcalcite09/1983–12/2005−0.25−0.011 ± 0.0022740.09<0.01

[12] Seawater pCO2, pH and CO32− ion concentrations were calculated from DIC, TA, temperature and salinity data using the computations of Lewis and Wallace [1998] the CO2 solubility equations of Weiss [1974], and dissociation constants for carbonic acid, borate [Dickson, 1990], and phosphate [Dickson and Goyet, 1994]. The carbonic acid dissociation constants of Mehrbach et al. [1973] (as refit by Dickson and Millero [1987]) were used to determine seawater pCO2 and other carbonate parameters. Other dissociation constants were also used [Goyet and Poisson, 1989; Roy et al., 1993; F. J. Millero et al., Dissociation constants of carbonic acid in seawater as a function of salinity and temperature, submitted to Marine Chemistry, 2006] to examine the potential inaccuracy of these calculations. For example, the difference in computed seawater pCO2 values using different carbonic acid dissociation constant was less than 12 μatm. The pH was computed on the total scale at in situ temperature. The saturation states for calcium carbonate minerals, aragonite and calcite, were also computed using the equations detailed by Lewis and Wallace [1998].

[13] For the period September 1988 to December 1990, TA was computed from salinity data using the salinity-total alkalinity relationships observed at the BATS site [Bates et al., 1996a, 1996b]. Water-column hydrographic data was used from both Hydrostation S and BATS. The BATS data set is available at http://bats.bios.edu, and Hydrostation S data prior to 1988 is archived at the Carbon Dioxide Information and Analysis Center (CDIAC; http:///www.cdiac.org).

[14] Salinity normalized DIC (i.e., nDIC) and TA (i.e., nTA) data were used to account for local evaporation and precipitation changes. A salinity of 36.6 was chosen as the constant to correct DIC and TA data as this represents the mean salinity observed at the BATS site. It is assumed that the preformed total alkalinity of subtropical gyre waters did not change significantly over time. Since there is a small range of surface salinity values (∼0.6) observed at BATS/Hydrostation S from 1983 to 2005, it is assumed that normalizing DIC and TA data to a constant salinity does not introduce potential biases [Friis et al., 2003].

2.2. Air-Sea CO2 Gas Exchange Calculations

[15] The net air-sea CO2 flux (F) was determined by the following formula:

  • equation image

where k is the transfer velocity, s is the solubility of CO2 and, ΔpCO2 is the partial pressure difference between atmosphere and ocean. The ΔpCO2, or air-sea CO2 disequilibrium, sets the direction of CO2 gas exchange while k determines the rate of air-sea CO2 transfer. The gas transfer velocity, k, is dependent on a variety of physical variables such as wind speed, bubbles, boundary layer stability and drag coefficients [e.g., Liss and Merlivat, 1986].

[16] Here a gas transfer velocity–wind speed relationship for short-term wind conditions based on a quadratic (U2) dependency between wind speed and k [i.e., Wanninkhof, 1992] was used to determine air-sea CO2 fluxes,

  • equation image

where U10 is wind speed corrected to 10 metres, and Sc is the Schmidt number for CO2. The Schmidt number was calculated using the equations of Wanninkhof [1992] and s (solubility of CO2 per unit volume of seawater) was calculated from the observed temperature and salinity using the equations of Weiss [1974]. Net air-sea CO2 fluxes were also estimated using a cubic (U3) relationship between wind speed and k [Wanninkhof and McGillis, 1999],

  • equation image

Both equations define steady/short-term gas transfer–wind speed relationships that can be used with high-frequency synoptic meteorological observations.

[17] The ΔpCO2 data sets used here to estimate air-sea CO2 fluxes were computed from seawater pCO2 data at BATS/Hydrostation S sites, and atmospheric pCO2 data from Bermuda and Terçeira Island, Açores. From August 1988 to December 2005, weekly flask samples for atmospheric xCO2 were collected at two sites on the island of Bermuda (data from http://www.cmdl.noaa.gov; National Oceanographic and Atmospheric Administration, NOAA, Climate and Meteorological Diagnostics Laboratory, CMDL). Atmospheric pCO2 data were calculated from xCO2 data by computing the contribution of pH2O to xCO2 using seawater temperature and salinity data. Prior to 1988, atmospheric pCO2 data collected from Terçeira Island, Açores, was used to compute ΔpCO2 (NOAA CMDL; http://www.cmdl.noaa.gov).

2.3. Comparison of Synoptic and Data Assimilation Model Wind Speed Observations

[18] Synoptic meteorological observations (including wind speed) was collected each hour for the period 1983 to 2005 from the Bermuda Airport, Bermuda (64.38°W, 32.37°N) by the Bermuda Weather Service (BWS). The BWS data (data courtesy of the National Climatic Data Center, NCDC, and BWS) was the primary meteorological data set used here to compute air-sea CO2 fluxes. The anemometers at the Bermuda Airport have been located at the same site and height (∼11 m) for the entire period from 1 January 1983 to 31 December 2005.

[19] For comparative purposes, 6-hourly averaged surface wind speed for the period 1983−2003 from NCEP/NCAR reanalysis 2 (NNR; http://www.cdc.noaa.gov/cdc/data.ncep.html) data assimilation model products were also used in this study. The spatial resolution of the NNR data assimilation model is 2.5°, with data from the 2.5° box centered on the island of Bermuda used in this analysis. In addition, a subset of the ECMWF ERA-40 reanalysis meteorological data (http://data.ecmwf.int/data/) was used. Similar to NNR data, 6-hourly averaged surface wind speed data centered on the island of Bermuda was used.

[20] Differences between synoptic and data assimilation model meteorological data have been reported previously. For example, Doney [1996] reported that the synoptic BWS wind speed data was systematically ∼0.38−0.53 m s−1 lower than the Comprehensive Ocean-Atmosphere Data Set (COADS) surface buoy and ship reports, and 6 hourly assimilated ECMWF weather products. More, recently Johnson [2002] compared the hourly BWS wind speed observations at the Bermuda Airport, hourly offshore observations in the vicinity of the BATS site from the Bermuda Testbed Mooring (BTM; ∼32°20′N, 64°30′W), and 6-hourly NCEP products. Although a good agreement was found in general, wind speed observations at the Bermuda Airport were significantly underestimated compared to BTM and NCEP during times when wind direction occurred between the northeast and northwest sector. This was attributed to local orographic effects at the Bermuda Airport site, where low hills (<30 m) are located to the northeast and north. To account for this effect, Johnson [2002] corrected the BWS wind speed data using the following equation:

  • equation image

where U10 is the uncorrected wind speed, and θ is the wind direction. In this study, the BWS wind speed data was also corrected using equation (4).

[21] In this paper, analysis of the relationships between air-sea CO2 fluxes and low-frequency modes of climate variability were evaluated using the corrected BWS wind speed. Both corrected and uncorrected BWS wind speed data sets exhibited similar patterns and relationships. Annual net air-sea CO2 fluxes were ∼215 ± 94 mmol CO2 m−2 yr−1 lower using uncorrected BWS wind speed data.

2.4. Analysis of Low-Frequency Modes of Climate Variability

[22] The relationship between of low-frequency modes climate variability and air-sea CO2 fluxes was also examined. Monthly Southern Oscillation Indices (SOI) of sea level pressure (SLP) differences in the tropical Pacific Ocean were obtained from the Climate Prediction Center (http://www.cpc.ncep.noaa.gov and ftp://ftp.cpc.ncep.noaa.gov/wd52dg/data/indices/soi). ENSO exerts its influence on the tropical Caribbean Sea and western Atlantic Ocean with a typical lag or persistence time of 6–9 months [e.g., Zhang et al., 1996; Bojariu, 1997; Penland and Matrosova, 1998; Bates, 2001]; here SOI values with a lag of 6 months are primarily used.

[23] Monthly and seasonal (i.e., January to May, JFMAM; June to September, JJAS; October to December, OND) means for the principal modes in the North Atlantic sector, such as the AO and NAO, as well as other relevant modes, were obtained from http://www.cpc.ncep.noaa.gov (ftp://ftp.cpc.ncep.noaa.gov/wd52dg/data/indices/tele_index.nh). For example, the monthly NAO index was based on SLP differences between Lisbon, Portugal and Stykkisholmur, Iceland http://www.cpc.ncep.noaa.gov). Relationships between air-sea CO2 fluxes and low-frequency modes were examined through regression analyses and empirical orthogonal function analysis (EOF) using monthly and seasonal means.

3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

3.1. Long-Term Trends of Seawater CO2 at the BATS and Hydrostation S Sites

3.1.1. Long-Term Trends of Hydrographic and Seawater CO2 Properties

[24] Long-term observations at the combined BATS/Hydrostation site show distinct trends in hydrographic and seawater CO2 properties. Over the 1983 to 2005 period, linear regression analysis indicates that surface temperature and salinity increased by ∼0.37°C and ∼0.19, respectively (Table 1), while DIC and total alkalinity increased by ∼30.6 μmoles kg−1 and ∼13.4 μmoles kg−1, respectively. Annually, DIC has increased at a rate of +1.37 ± 0.16 μmoles kg−1 yr−1, while total alkalinity has increased at a rate of +0.61 ± 0.11 μmoles kg−1 yr−1 (Table 1 and Figure 1). This long-term increase of DIC and TA is partially due to an increase in surface salinity over the last 22 years. Accounting for salinity changes, surface nDIC has increased at a rate of +0.86 ± 0.11 μmoles kg−1 yr−1, while nTA did not change significantly (Table 1). Over the 1983 to 2005 period, seawater and atmospheric pCO2 increased at similar rates of +1.67 ± 0.28 μatm yr−1 and +1.78 ± 0.02 μatm yr−1, respectively (Table 1 and Figure 1). The total change in seawater and atmospheric pCO2 in the vicinity of Bermuda from 1983 to 2005 was ∼37–39 μatm; a change of ∼10% during the last 22 years. With the exception of surface temperature and nTA, all long-term trends were statistically significant (p-values <0.01; Table 1).

image

Figure 1. Long-term trends of oceanic CO2 properties observed at the BATS (Bermuda Atlantic Time series Study; 31°43′N, 64°30′W) and Hydrostation S (32°50′N, 64°10′W) sites located near Bermuda in the NW Atlantic Ocean. Slopes and statistics of regressions are listed in Table 1. (a) Sea surface temperature °C (red line) and salinity (grey line). (b) Surface dissolved inorganic carbon (DIC, μmoles kg−1, blue line) and salinity normalized DIC (nDIC, μmoles kg−1, grey line) changes; nDIC data is normalized to a constant salinity of 36.6 (the average salinity observed at the BATS site). (c) Seawater pCO2 (μatm, blue line) and atmospheric pCO2 changes (ppm, grey line). (d) Surface saturation state of aragonite (Ωarag, green line) and surface total alkalinity (TA, μmoles kg−1, grey line). (e) Surface seawater pH (orange line) and CO32− ion concentrations (μmoles kg−1, grey line) changes.

Download figure to PowerPoint

3.1.2. Long-Term Trends of Seasonally Detrended Hydrographic and Seawater CO2 Properties

[25] In evaluating long-term trends, consideration should also given to the annual cycle of oceanic CO2, which is influenced by a variety of processes including seasonal temperature, salinity, and density changes, vertical mixing/diffusion and biological production. The amplitude of the annual cycle of CO2 is large compared to the amplitude of lower frequency decadal signals. DIC, nDIC, TA, and pCO2 data can therefore be seasonally detrended by fitting the data to a function that includes a constant, a linear trend and harmonic terms with periods of 12, 6, and 4 months. Surface temperature and salinity increased by 0.75°C and 0.19 over the 22-year period (Table 2 and Figure 2). Long-term trends for DIC, nDIC, TA, and seawater pCO2 were +1.27 ± 0.08, +0.80 ± 0.06, +0.58 ± 0.09 μmoles kg−1 yr−1 and +1.80 ± 0.013 μatm yr−1, respectively, similar to the trends computed for observed CO2 data (Table 2). All long-term trends were statistically significant (p-values <0.01; Table 2), except for nTA and temperature.

image

Figure 2. Long-term trends of seasonally detrended oceanic CO2 properties observed at the BATS (Bermuda Atlantic Time series Study; 31°43′N, 64°30′W) and Hydrostation S (32°50′N, 64°10′W) sites located near Bermuda in the NW Atlantic Ocean. Slopes and statistics of regressions are listed in Table 2. (a) Sea surface temperature °C (red) and salinity (grey). (b) Surface dissolved inorganic carbon (DIC, μmoles kg−1, blue) and salinity normalized DIC (nDIC; μmoles kg−1, grey) changes; nDIC data is normalized to a constant salinity of 36.6 (the average salinity observed at the BATS site). (c) Seawater pCO2 (μatm, blue) and atmospheric pCO2 changes (ppm, grey). (d) Surface saturation state of aragonite (Ωaραγ, green) and total alkalinity (TA, μmoles kg−1, grey) changes. (e) Surface seawater pH (orange) and CO32- ion concentrations (μmoles kg−1, grey) changes.

Download figure to PowerPoint

Table 2. Long-Term Trends (1983–2005) of Seasonally Detrended Surface Seawater Temperature, Salinity, DIC, nDIC, TA, nTA, Seawater pCO2 (pCO2sea), Atmospheric pCO2 (pCO2atm), pH, CO32− Ion Concentration, and Saturation States of Aragonite (Ωarag) and Calcite (Ωcalcite) in the North Atlantic Ocean Near Bermudaa
ParameterPeriod22 Year ChangeSlope and Standard Errornr2p-Value
  • a

    Further details on data are given in Table 1. Regression statistics (slope, error, r2, and p-value) were determined using a least squares fitting routine using singular value decomposition method (see section 2). Surface layer is 0–10 m.

Temp.09/1983–12/2005+0.75°C+0.034 ± 0.0082740.060.56
Salinity09/1983–12/2005+0.19+0.0079 ± 0.0012740.13<0.01
DIC09/1983–12/2005+28.4 μmoles kg−1+1.27 ± 0.082740.48<0.01
nDIC09/1983–12/2005+17.8 μmoles kg−1+0.80 ± 0.062740.40<0.01
TA09/1983–12/2005+13.4 μmoles kg−1+0.58 ± 0.092430.13<0.01
nTA09/1983–12/2005+1.4 μmoles kg−1+0.06 ± 0.042430.000.19
pCO2sea09/1983–12/2005+40.1 μatm+1.80 ± 0.132740.43<0.01
pCO2atm09/1983–12/2005+39.6 μatm+1.80 ± 0.0211020.98<0.01
pH09/1983–12/2005−0.037−0.0017 ± 0.00012740.19<0.01
CO32−09/1983–12/2005−9.4 μmoles kg−1−0.52 ± 0.022740.18<0.01
Ωarag09/1983–12/2005−0.14−0.006 ± 0.0012740.15<0.01
Ωcalcite09/1983–12/2005−0.22−0.010 ± 0.0012740.19<0.01

[26] The equilibrium rate of DIC increase due to anthropogenic CO2 has previously been calculated as +0.9 μmoles kg−1 yr−1 for the North Atlantic Ocean subtropical gyre [Bates et al., 2002; Gruber and Sarmiento, 2002]. Within the 95% confidence levels, the observed rate of surface ocean nDIC increase (+0.80 μmoles kg−1 yr−1) at Hydrostation S/BATS was similar to that expected from oceanic equilibration with increasing CO2 in the atmosphere. The upward trend in seawater pCO2 (+1.80 μatm yr−1) was identical to the rate of atmospheric CO2 increase (+1.80 μatm yr−1) over the last twenty years.

[27] Long-term observations at the BATS site also indicate that the nDIC of surface and deeper water layers have increased at divergent rates over time since water-column sampling began in 1988. In the upper water column (0−250 m), DIC and nDIC both increased at slightly higher rates of +0.72 ± 0.08 and +0.69 ± 0.10 μmoles kg−1 yr−1 (Table 3) compared to the surface layer [Bates, 2001]. The mean rate of change in the 0- to 250-m layer was computed from vertically integrated profiles of DIC and nDIC. The total change in nDIC from 1983 to 2005 was ∼19 μmoles kg−1; a change of ∼0.8%. In deeper subtropical mode waters (STMW), the mean rate of change of nDIC, over the 1988−2001 time period, was significantly higher than surface waters, increasing at a rate of +2.2 ± 0.26 μmoles kg−1 yr−1 [Bates et al., 2002] (Table 3). Here STMW is defined as those waters with temperatures ranging from 17.8° to 18.4°C, by a salinity of ∼36.5 ± 0.05, and by a minimum in the vertical gradient of isopycnic potential vorticity [Klein and Hogg, 1996; Jenkins, 1998; Hanawa and Talley, 2001; Alfutis and Cornillon, 2001]. The rate of increase of nDIC in STMW from 1988 to 2003 is similar at +2.1 ± 0.28 μmoles kg−1 yr−1 to earlier estimates [Bates et al., 2002]. The STMW of the North Atlantic Ocean is formed each winter by cooling and convective mixing at the northern edges of the subtropical gyre south of the Gulf Stream, filling the shallow depths of the subtropical gyre (∼250–400 m deep) [Klein and Hogg, 1996; Hazeleger and Drijfhout, 1998]. The cause of the divergence between surface-ocean and deeper STMW oceanic CO2 trends is not certain, but is presently thought to relate to atmospheric/climatic variability of the North Atlantic subtropical gyre [Bates et al., 2002; Gruber et al., 2002].

Table 3. Long-Term Oceanic CO2 Changes Observed at Four Time-Series Sitesa
SitePeriodYearsLayernDIC, μmoles kg−1 yr−1pCO2,bμatm yr−1Reference
  • a

    The four sites include A Long-term Oligotrophic Habitat Assessment (ALOHA), located near Hawaii in the North Pacific Ocean; Bermuda Atlantic Time-Series Study (BATS), located near Bermuda in the NW Atlantic Ocean; Hydrostation S (Hydro S) and BATS combined, located near Bermuda in the NW Atlantic Ocean; and European Station for Time-series in the Ocean Canary Islands (ESTOC), located near Gran Canaria in the NE Atlantic Ocean.

  • b

    Here na means not applicable, and ND means no data.

  • c

    Surface and upper ocean samples were used in this analysis. Keeling et al. [2004] analyzed samples at 5 m and 25 m until 1999, and at 5 m thereafter. Dore et al. [2003] analyzed all samples from 0 to 30 m.

  • d

    DIC data was normalized to salinity of 35, close to the average salinity observed at the ALOHA site. DIC data was also seasonally detrended by Keeling et al. [2004] but not by Dore et al. [2003].

  • e

    Here pCO2 was not seasonally detrended.

  • f

    DIC data was normalized to salinity of 36.6, close to the average salinity observed at the BATS site. DIC data was not seasonally detrended [Bates, 2001].

  • g

    DIC data was normalized to constant salinity of 36.6, close to the average salinity observed at the BATS site. DIC data was seasonally detrended [Bates, 2001].

  • h

    DIC changes in subtropical mode water (STMW) which occurs at a depth of ∼250–400 m deep at the BATS site. STMW is characterized by a temperature of ∼18 ± 0.2°C, salinity of 36.5 ± 0.03, and a potential vorticity minima [Bates et al., 2002].

  • i

    DIC data was normalized to constant salinity of 35 [Gonzalez-Dávila et al., 2003].

North Pacific Ocean
ALOHA (HOT) (22°45′N, 158°W)10/88–12/01130–30 mc1.19d ± 0.142.5e ± 0.3Dore et al. [2003]
 10/88–12/02140–10 mc1.22c ± 0.082.5e ± 0.1Keeling et al. [2004]
KNOT (44°N, 155°E)1992–200190–10 mc1.3 to 2.3naWakita et al. [2005]
 
North Atlantic Ocean
BATS (32°N, 64°W)10/88 to 12/98100–10 mc1.60f ± 5.61.4e ± 5.1Bates [2001]
 10/88 to 09/01130–10 mc1.25g ± 0.14ndBates et al. [2002]
 10/88 to 09/0113STMWh2.22 ± 0.25ndBates et al. [2002]
 10/88 to 12/03150–250c0.72 ± 0.13ndthis study
Hydro S/BATS (32°N, 64°W)05/83 to 09/01185, 25 mc0.69 ± 0.051.5 ± 0.1Keeling et al. [2004]
 05/83 to 12/05220–10 mc0.80 ± 0.081.80 ± 0.13this study
ESTOC (29°10′N, 15°30′W)10/95 to 12/0050–10 mc0.4i ± 1.60.71 ± 5.1Gonzalez-Dávila et al. [2003]
 10/95 to 06/04100–10 mc0.99 ± 0.21.55 ± 0.43Santana-Casiano et al. [2007]
3.1.3. Long-Term Trends of Alkalinity

[28] In the total alkalinity record at BATS/Hydrostation S, nTA remained generally constant (mean of 2389 μmoles kg−1) within a typically small range of values (∼2387–2393 μmoles kg−1), reflecting the conservative nature of total alkalinity with respect to salinity in the subtropical gyre of the North Atlantic [Bates et al., 1996a; Millero et al., 1998]. Exceptions to this general rule were observed in October 1985, February 1992, and February 2001, when nTA values were lower by >20 μmoles kg−1 than the mean nTA value. The nonconservative drawdown of total alkalinity and low nTA in February 1992 was attributed to transient marine calcification by coccolithophores [Bates et al., 1996b] rather than salinity or source water changes. The two other low nTA periods (observed in October 1985 and February 2001) also probably reflect drawdown of total alkalinity by marine calcifying phytoplankton (e.g., coccolithophores), although, anomalously high biomasses of calcifying foraminifera or pteropods cannot be ruled out [Bates et al., 1996b]. There were also several other occasions during the 1983 to 2005 period when nTA values were lower by ∼10–20 μmoles kg−1 than the mean nTA concentration; these events also likely reflect the geochemical impact of marine calcification.

3.1.4. Long-Term Trends of CO32−, CaCO3 Saturation States, and pH

[29] During the 1983 to 2005 period, observed and seasonally detrended CO32− ion concentration decreased by 0.47 ± 0.09 μmoles kg−1 yr−1 and 0.52 ± 0.02 μmoles kg−1 yr−1 respectively (Tables 1 and 2 and Figures 1 and 2). Furthermore, the saturation states for aragonite (Ωarag) and calcite (Ωcalcite) also decreased over the period of observation. At the same time, haptophyte biomass data (estimated from 19′ hexanoyloxyfucoxanthin pigment data which can be used as a proxy for coccolithophore variability) did not change significantly over time at the BATS site (1992–2002) [Lomas and Bates, 2004]. To date, there is insufficient evidence to suggest that marine calcification has been affected by the recent gradual decrease in CO32− ion concentrations and carbonate mineral saturation states in the North Atlantic Ocean.

[30] The ocean near Bermuda has also become more acidic, with a decrease in seawater pH of 0.0012 ± 0.0006 pH units yr−1 (Table 1 and Figure 1). This represents a decline of ∼0.025 pH units (∼8.125 to ∼8.100) over the last 20 years, about one third of the total 0.06 pH unit increase in ocean acidity observed since the pre-industrial times [e.g., Caldeira and Wickett, 2003; Sabine et al., 2004a]. The long-term trends of pH, CO32− ion concentration and saturation states of calcium carbonate minerals were statistically highly significant (p-values <0.01; Table 1).

3.2. Comparison of Long-Term Trends of Seawater CO2

[31] In the early stages of observations at the BATS site (1988–1998), nDIC and seawater pCO2 increased at a rate of +1.6 μmoles kg−1 yr−1 (Table 3 and Figure 1), higher than equilibrium with anthropogenic CO2 accumulation in the atmosphere [Bates, 2001; Bates et al., 2002]. A significant component of the interannual variability of nDIC and seawater pCO2 was attributed to physical forcing that correlated with the NAO [Bates, 2001; Gruber et al., 2002]. It seems though that over longer decadal timescales that the surface of the North Atlantic subtropical gyre increased in equilibrium with the anthropogenic CO2 transient in the atmosphere.

[32] Long-term trends of oceanic CO2 in other oceanic regions are quite variable, with some areas exhibiting rates of seawater CO2 increase higher or lower than equilibrium with the anthropogenic CO2 transient in the atmosphere. This variability reflects the shortness of most time series observations and the difficulties associated with determining trends where the amplitude of the annual CO2 cycle is much greater than the amplitude of lower frequency decadal signals.

[33] In the northeast Atlantic Ocean, observations at the ESTOC site near Gran Canaria (29°N, 15°W), also show upward trends of surface ocean salinity normalized dissolved inorganic carbon (nDIC) and seawater pCO2. nDIC and seawater pCO2 increased at a rate of +0.99 ± 0.2 μmoles kg−1 yr−1, and +1.55 ± 0.43 μatm yr−1, respectively (Table 3) [Santana-Casiano et al., 2007] for the 1995–2004 period. The observed rate of change of surface ocean DIC, for example, was slightly lower than the expected oceanic equilibration with anthropogenic CO2 in the atmosphere. The causes for the lower than anticipated oceanic CO2 increase are not certain, but probably relate to the relatively short period of observation (i.e., 1995–2004 data reported), and subdecadal variability of the region close to upwelling off the African coast [Santana-Casiano et al., 2007].

[34] In the North Atlantic subpolar gyre, upward trends of DIC and seawater pCO2 in the last decade were higher than expected from oceanic equilibration with anthropogenic CO2 (1982 to 1993/1998 [Olsen et al., 2003], 1988 to 1998 [Lefèvre et al., 2004]). This has been attributed to a decrease in oceanic CO2 uptake due to a gradual decrease in biological productivity [Lefèvre et al., 2004].

[35] In the North Pacific Ocean, observations at the ALOHA site near Hawaii (22°45′N, 158°W), show upward trends of salinity normalized dissolved inorganic carbon (nDIC) and seawater pCO2 higher than at BATS/Hydrostation S. Surface ocean nDIC and seawater pCO2 increased at a rate of +1.2 ± 0.1 μmoles kg−1 yr−1, and +2.5 ± 0.3 μatm yr−1, respectively (Table 3) [Dore et al., 2003; Keeling et al., 2004] for the 1988–2002 period. Similarly, at time series site KNOT (44°N, 155°E) in the western North Pacific subpolar region, the increase in DIC was +1.3 to 2.3 μmoles kg−1 yr−1 from 1992 to 2001 [Wakita et al., 2005]. At both sites, the observed rate of change of surface ocean DIC, for example, was slightly higher than the expected oceanic equilibration with anthropogenic CO2 in the atmosphere. The causes of the trend variability is not certain, but is presently thought to relate to subdecadal basinwide changes in biological properties (e.g., primary production), physical properties (e.g., precipitation-evaporation balance and atmospheric annular mode influences such as Pacific decadal oscillation [e.g., Dore et al., 2003; Keeling et al., 2004; Brix et al., 2004, 2006]), and intermediate water formation changes [Wakita et al., 2005].

[36] Elsewhere, across much of the North Pacific, Equatorial Pacific and South Pacific to the Subantarctic Zone, seawater pCO2, and/or DIC increased at rates close to equilibrium [Peng et al., 2003; Takahashi et al., 2003; Sabine et al., 2004b; Yoshikawa-Inoue and Ishii, 2005]. From the Polar Frontal zone southward toward Antarctica, the rate of increase was lower than the expected oceanic equilibrium with anthropogenic CO2 in the atmosphere [Yoshikawa-Inoue and Ishii, 2005].

4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

4.1. Rates of Air-Sea CO2 Flux at the BATS and Hydrostation S Sites

[37] Daily, seasonal and annual rates of air-sea CO2 flux were computed using ΔpCO2 data from BATS/Hydrostation S sites and hourly wind speed observations (i.e., BWS; Figure 3) and 6-hourly data assimilation model data sets (i.e., ECMWF and NNR). Both negative (i.e., net air-to-sea CO2 influx) and positive (i.e., net sea-to-air CO2 effflux) daily values (Figure 4) were observed with a broad range of values (e.g., <−40 to >+25 mmol CO2 m−2 d−1).

image

Figure 3. Daily averaged wind speed (m s−1) and rates of air-sea CO2 flux (mmoles CO2 m−2 d−1) observed near Bermuda in the NW Atlantic Ocean from 1983–2005. (a) Hourly observations of wind speed (BWS) collected from the island of Bermuda. Hurricanes Felix and Fabian are shown. (b) Daily air-sea CO2 fluxes calculated using the gas transfer velocity–wind speed relationships for short-term wind conditions based on a quadratic dependency between wind speed and k [i.e., Wanninkhof, 1992]. Rates of air-sea CO2 flux for Hurricanes Felix and Fabian are shown.

Download figure to PowerPoint

image

Figure 4. Daily averaged rates of air-sea CO2 flux (mmoles CO2 m−2 d−1) calculated using BWS, ECMWF, and NNR wind speed data sets observed near Bermuda in the NW Atlantic Ocean from 1983–2005. Daily air-sea CO2 fluxes were calculated using the gas transfer velocity–wind speed relationships for short-term wind conditions based on a quadratic dependency between wind speed and k [i.e., Wanninkhof, 1992].

Download figure to PowerPoint

[38] On annual timescales, the North Atlantic subtropical gyre is an oceanic sink for atmospheric CO2. Air-sea CO2 fluxes for the 1984 to 2005 period were computed as the sum of daily fluxes computed from the daily mean of higher-frequency wind speed observations (hourly for BWS and 6-hourly for ECMWF and NNR), and the quadratic wind speed dependency (equation (2)). A net annual air-sea CO2 influx rate of −810 ± 249 mmol CO2 m−2 yr−1 (Tables 4a and 4b) was computed using the BWS wind speed data. The average net annual air-sea CO2 fluxes were higher using ECMWF (−1024 ± 282 mmol CO2 m−2 yr−1); note: fluxes only computed for 1984−2002 using ECMWF data sets) and NNR (−1295 ± 294 mmol CO2 m−2 yr−1) wind speed data sets (Tables 4a and 4b and Figure 5a). If higher-frequency wind speed observations were used to compute air-sea CO2 fluxes, the net annual air-sea CO2 fluxes for BWS (hourly meteorological observations), ECMWF and NNR (6-hourly data assimilation values) were higher by 3.3%, 4.4% and 5.5%, respectively.

image

Figure 5. Annual, wintertime (JFMAM), summertime (JJAS), and fall (OND) rates of air-sea CO2 flux (mmol CO2 m−2 d−1) calculated using BWS, ECMWF, and NNR wind speed data sets observed near Bermuda in the NW Atlantic Ocean from 1983–2005. Annual and seasonal averages were calculated from daily air-sea CO2 fluxes estimated using the gas transfer velocity–wind speed relationships for short-term wind conditions based on a quadratic dependency between wind speed and k [i.e., Wanninkhof, 1992].

Download figure to PowerPoint

Table 4a. Annual Mean and Standard Deviation of Annual Wind Speed and Air-Sea CO2 Fluxes for 1983 to 2005 in the North Atlantic Ocean Near Bermudaa
Wind Speed, m s−1Annual Air-Sea CO2 Flux, mmol CO2 m−2
YearBWSECMWFNNRBWSECMWFNNRBWSECMWFNNR
  • a

    Annual wind speed data and statistics were compiled from BWS, ECMWF, and NNR data sets. Air-sea CO2 fluxes were determined using seawater pCO2 from Hydrostation S/BATS, atmospheric pCO2 data, wind speed data, and U2 (quadratic) and U3 (cubic) parameterizations of k. BWS, ECMWF, and NNR wind speed data sets were averaged over each day. If hourly BWS or 6-hourly ECMWF or NNR data sets were used, air-sea CO2 fluxes were ∼3–5% higher using the U2 (quadratic) parameterizations of k. Negative CO2 flux values denote net air to sea CO2 flux, whereas positive CO2 flux values denote net sea to air CO2 flux.

19846.70 ± 1.886.64 ± 2.807.46 ± 3.32−547−821−1060−733−1208−1873
19856.67 ± 1.876.62 ± 3.027.55 ± 3.56−596−935−1190−831−1658−2548
19866.82 ± 1.796.48 ± 2.607.48 ± 3.17−967−1161−1570−1107−1584−2652
19877.18 ± 2.266.63 ± 3.287.63 ± 3.80−782−984−1294−1200−1684−2796
19887.07 ± 2.206.44 ± 2.947.30 ± 3.52−1222−1315−1788−1622−1990−3261
19896.74 ± 2.066.22 ± 2.726.97 ± 3.18−1058−1171−1474−1293−1606−2386
19907.11 ± 2.046.08 ± 2.756.90 ± 3.26−645−692−895−899−972−1488
19917.28 ± 2.146.51 ± 2.847.36 ± 3.40−992−1018−1321−1363−1519−2472
19927.31 ± 2.036.92 ± 2.307.87 ± 3.59−828−1029−1365−1292−1802−2967
19937.15 ± 1.976.53 ± 2.917.49 ± 3.48−571−809−1115−948−1418−2415
19946.86 ± 2.016.18 ± 2.987.15 ± 3.45−604−984−1316−956−1587−2646
19956.86 ± 2.016.79 ± 3.107.91 ± 3.85−732−940−1230−1082−1481−2456
19967.44 ± 2.276.69 ± 3.017.97 ± 3.68−694−840−1131−1151−1431−2391
19977.40 ± 2.136.90 ± 2.748.02 ± 3.31−586−790−1133−913−1303−2397
19987.24 ± 2.266.60 ± 2.907.71 ± 3.60−472−618−894−925−1173−2125
19997.36 ± 2.256.60 ± 2.807.55 ± 3.32−425−555−709−824−1011−1566
20007.42 ± 2.277.94 ± 3.478.01 ± 3.50−1039−1560−1580−1541−2915−2971
20017.36 ± 2.297.82 ± 3.397.90 ± 3.47−1047−1528−1494−1419−2714−2679
20027.29 ± 2.187.42 ± 3.337.86 ± 3.42−1291−1337−1891−1825−2318−3511
20037.58 ± 2.30 8.08 ± 3.29−939n/a−1504−1379n/a−2893
20048.09 ± 2.53 8.64 ±3.66−1074n/a−1459−1866n/a−3447
20057.84 ± 2.42 8.60 ± 3.42−689n/a−1024−1243n/a−2705
Range   86810051182   
Mean7.246.747.70−810−1024−1295−1184−1682−2525
Std. Dev±0.35±0.50±0.45±249±282±294±298±523±504
Table 4b. Trends of Annual Wind Speed and Air-Sea CO2 Fluxes for 1983 to 2005 in the North Atlantic Ocean Near Bermuda
TypePeriodChangeRate of Change per Yearr2p-value
BWS wind speed01/1984–12/2005+1.00 (+13%)+0.045 ± 0.007 m s−10.70<0.01
ECMWF wind speed01/1984–12/2002+1.10 (+16%)+0.050 ± 0.010 m s−10.53<0.01
NNR wind speed01/1984–12/2005+1.23 (+16%)+0.056 + 0.017 m s−10.40<0.01
 
U2 (Quadratic) Parameterization of k
BWS wind speed01/1984–12/2005−129 (+16%)−6 ± 9 mmol CO2 m−20.020.50
ECMWF wind speed01/1984–12/2002−166 (+16%)−8 ± 12 mmol CO2 m−20.020.79
NNR wind speed01/1984–12/2005−39 (+ 3%)−2 ± 10 mmol CO2 m−20.000.86
 
U3 (Cubic) Parameterization of k
BWS air-sea CO2 flux01/1984–12/2005−260 (+21%)−12 ± 10 mmol CO2 m−20.200.03
ECMWF air-sea CO2 flux01/1984–12/2002−325 (+19%)−14 ± 21 mmol CO2 m−20.140.11
NNR air-sea CO2 flux01/1984–12/2005−78 (+ 3%)−3 ± 17 mmol CO2 m−20.040.36

[39] If the cubic wind speed–gas exchange dependency (equation (3)) was used to compute air-sea CO2 fluxes, the mean net annual air-sea CO2 influxes were significantly higher (Tables 4a and 4b). For example, the net annual air-sea CO2 fluxes using BWS, ECMWF and NNR wind speed data sets were higher by 45–90% with mean fluxes for the 1983 to 2005 period of −1184 ± 298, −1682 ± 523, and −2497 ± 509 mmol CO2 m−2 yr−1, respectively (Tables 4a and 4b).

[40] As reported in previous studies, the North Atlantic subtropical gyre is supersaturated with respect to CO2 during the stratified summer months (i.e., oceanic source of CO2 to the atmosphere) and undersaturated during the deepened mixed layer period from fall (e.g., OND) and wintertime (JFMAM) (i.e., oceanic sink of CO2 from atmosphere) [Bates et al., 1996a, 1998a; Takahashi et al., 2002]. The average net wintertime (JFMAM) air-sea CO2 influx was −1038 ± 155 mmol CO2 m−2 yr−1 (Tables 5a and 5b and Figure 5c), using the quadratic gas exchange relationship and BWS wind speed data. During the summertime period (i.e., ∼June to September), there is an efflux of CO2 from the ocean to the atmosphere (Figure 5). The average net summertime (JJAS) air-sea CO2 efflux was +473 ± 147 mmol CO2 m−2 yr−1 (Table 6a, 6b and Figure 5c), using the quadratic gas exchange relationship and BWS wind speed data. The net summertime air-sea CO2 efflux computed with ECMWF and NNR wind speed data sets were slightly lower (Table 6a, 6b and Figure 5c). The summertime (i.e., June to September; JJAS) efflux of CO2 is seasonally counterbalanced by a stronger fall (i.e., October to December; OND) and wintertime (i.e., January to May; JFMAM) influx of CO2 to the ocean from the atmosphere. The average fall air-sea CO2 influx was −248 ± 95 mmol CO2 m−2 yr−1 (higher with ECMWF and NNR wind speed data sets; Table 7a and 7b and Figure 5b).

Table 5a. Wintertime (JFMAM) Mean Wind Speed and Air-Sea CO2 Fluxes for 1983–2005 in the North Atlantic Ocean Near Bermudaa
YearWind Speed, m s−1ΔpCO2Wintertime Air-Sea CO2 Flux, mmol CO2 m−2
BWSECMWFNNRQuadratic Dependency U2Cubic Dependency U3
MeanSDMeanSDMeanSDMeanBWS MeanECMWF MeanNNR MeanBWS MeanECMWF MeanNNR Mean
  • a

    Annual wind speed data and statistics were compiled from BWS, ECMWF, and NNR data sets. Air-sea CO2 fluxes were determined using seawater pCO2 from BATS/Hydrostation S, atmospheric pCO2 data, wind speed data, and U2 (quadratic) and U3 (cubic) parameterizations of k. BWS, ECMWF, and NNR wind speed data sets were averaged over each day. If hourly BWS or 6-hourly ECMWF or NNR data sets were used, air-sea CO2 fluxes were ∼3–5% higher using the U2 (quadratic) parameterizations of k. Negative CO2 flux values denote net air to sea CO2 flux, whereas positive CO2 flux values denote net sea to air CO2 flux.

19847.401.997.452.768.533.21−29.7−861−937−1232−1033−1256−1966
19857.132.157.593.508.704.07−29.3−819−1073−1403−1007−1722−2673
19867.291.777.512.778.723.43−34.3−920−1069−1465−1042−1449−2430
19877.752.267.473.238.553.78−28.3−942−996−1315−1255−1484−2367
19887.612.187.243.048.473.70−32.4−1028−1030−1425−1304−1459−2458
19896.861.926.532.577.422.98−42.8−1116−1143−1491−1298−1470−2234
19907.542.076.732.747.633.29−28.5−835−725−949−1015−910−1419
19917.712.177.063.068.153.63−37.4−1181−1082−1452−1475−1465−2381
19928.062.217.773.288.893.98−28.3−1015−1069−1423−1379−1649−2660
19937.901.937.372.898.573.45−30.4−1024−1028−1394−1294−1461−2381
19947.382.187.153.088.213.65−30.4−966−1015−1351−1212−1436−2294
19957.911.917.892.809.133.55−30.5−1004−1064−1447−1259−1517−2574
19968.272.247.803.019.213.60−25.6−1006−992−1388−1414−1507−2551
19977.652.177.402.798.573.36−30.2−957−950−1288−1212−1302−2145
19987.992.517.563.118.973.82−27.6−1004−986−1386−1422−1482−2598
19997.912.437.392.998.453.50−22.5−899−861−1123−1274−1256−1950
20008.182.249.153.359.213.32−30.6−1136−1538−1554−1534−2649−2672
20017.962.109.003.358.903.36−29.1−945−1318−1301−1224−2247−2227
20027.662.088.303.388.623.37−40.3−1278−1668−1750−1594−2696−2896
20038.012.25  8.923.44−36.0−1263 −1693−1696 −2947
20048.382.66  8.933.93−33.9−1407 −1793−2150 −3447
20058.162.56  8.883.59−31.2−1221 −1546−1745 −2705
Range       572 8451145 2028
Mean7.76 7.60 8.62 31.3103810811417135616012453
SD0.38 0.65 0.46 4.6155219190271454407
Table 5b. Wintertime (JFMAM) Trends of Wind Speed and Air-Sea CO2 Fluxes for 1983–2005 in the North Atlantic Ocean Near Bermuda
TypePeriodChangeRate of Change per Yearr2p-value
BWS wind speed01/1984–12/2005+0.93 (12%)+0.042 ± 0.009 m s−10.51<0.01
ECMWF wind speed01/1984–12/2002+1.51 (20%)+0.068 ± 0.023 m s−10.35<0.01
NNR wind speed01/1984–12/2005+0.69 ± 8%)+0.031 ± 0.016 m s−10.200.04
 
U2 (Quadratic) Parameterization of k
BWS wind speed01/1984–12/2005−344 (33%)−16 ± 4 mmol CO2 m−20.43<0.01
ECMWF wind speed01/1984–12/2002−414 (38%)−19 ± 8 mmol CO2 m−20.230.04
NNR wind speed01/1984–12/2005−297 (21%)−14 ± 6 mmol CO2 m−20.210.03
 
U3 (Cubic) Parameterization of k
BWS air-sea CO2 flux01/1984–12/2005−695 (51%)−32 ± 6 mmol CO2 m−20.57<0.01
ECMWF air-sea CO2 flux01/1984–12/2002−977 (61%)−44 ± 16 mmol CO2 m−20.300.01
NNR air-sea CO2 flux01/1984–12/2005−661 (27%)−30 ± 12 mmol CO2 m−20.230.02
Table 6a. Summertime (JJAS) Mean Wind Speed and Standard Deviation Air-Sea CO2 Fluxes for 1983–2005 in the North Atlantic Ocean Near Bermudaa
YearWind Speed, m s−1ΔpCO2Summertime Air-Sea CO2 Flux, mmol CO2 m−2
BWSECMWFNNRQuadratic Dependency U2Cubic Dependency U3
BWS MeanECMWF MeanNNR MeanBWS MeanECMWF MeanNNR Mean
MeanSDMeanSDMeanSDMean
  • a

    Annual wind speed data and statistics were compiled from BWS, ECMWF, and NNR data sets. Air-sea CO2 fluxes were determined using seawater pCO2 from BATS/Hydrostation S, atmospheric pCO2 data, wind speed data, and U2 (quadratic) and U3 (cubic) parameterizations of k. BWS, ECMWF, and NNR wind speed data sets were averaged over each day. If hourly BWS or 6-hourly ECMWF or NNR data sets were used, air-sea CO2 fluxes were ∼3–5% higher using the U2 (quadratic) parameterizations of k. Negative CO2 flux values denote net air to sea CO2 flux, whereas positive CO2 flux values denote net sea to air CO2 flux.

19846.411.525.222.335.792.8029.7430297378405274405
19856.131.475.402.286.222.8127.1388330453367333557
19866.371.655.371.996.192.4512.4167123176166123220
19876.071.684.742.075.502.3723.8329226303327209323
19886.181.995.082.185.532.5719.6264178209267167220
19896.061.915.082.335.692.7419.0276206263279200296
19906.351.624.712.065.382.5733.5448275366435266442
19916.581.865.412.176.152.5534.7491349460487327497
19926.251.445.402.086.122.4533.7496383510464359562
19936.341.405.012.045.772.4540.4586408551543370586
19945.891.424.281.844.982.1349.6664362490599294440
19956.542.215.172.676.213.5533.3555395597662471959
19966.331.925.092.406.193.1734.9532377574556395781
19976.651.685.542.036.422.3133.6591396512626376558
19986.451.795.452.266.292.8748.0724566768723573966
19996.341.875.072.185.862.7037.9611438594650433731
20006.061.675.682.555.652.6229.0406402398392444446
20016.391.946.162.816.383.0917.5285297324302348426
20026.311.76  6.152.7626.8402 417400417488
20036.562.25  6.452.6540.6636 617749 695
20046.381.70  6.412.5536.7533 598528 703
20055.981.66  5.772.7731.2592 617566 753
Range       469 592557 746
Mean6.30 5.21 5.96 28.7473334463477336548
SD0.20 0.41 0.39 16.3147106150158113210
Table 6b. Summertime (JJAS) Trends of Wind Speed and Air-Sea CO2 Fluxes for 1983–2005 in the North Atlantic Ocean Near Bermuda
TypePeriodChangeRate of Change per Yearr2p-value
BWS wind speed01/1984–12/2005+0.08 ± 1%)+0.004 ± 0.007 m s−10.010.58
ECMWF wind speed01/1984–12/2002+0.60 (11%)+0.028 ± 0.018 m s−10.130.14
NNR wind speed01/1984–12/2005+0.46 ± 8%)+0.020 ± 0.012 m s−10.120.11
 
U2 (Quadratic) Parameterization of k
BWS wind speed01/1984–12/2005+251 (53%)+11 ± 4 mmol CO2 m−20.260.02
ECMWF wind speed01/1984–12/2002+289 (87%)+13 ± 4 mmol CO2 m−20.44<0.01
NNR wind speed01/1984–12/2005+302 (65%)+14 ± 4 mmol CO2 m−20.35<0.01
 
U3 (Cubic) Parameterization of k
BWS air-sea CO2 flux01/1984–12/2005+293 (61%)+13 ± 5 mmol CO2 m−20.30<0.01
ECMWF air-sea CO2 flux01/1984–12/2002+344 (102%)+16 ± 4 mmol CO2 m−20.53<0.01
NNR air-sea CO2 flux01/1984–12/2005+396 (72%)+18 ± 6 mmol CO2 m−20.31<0.01
Table 7a. Fall (OND) Wind Speed and Air-Sea CO2 Fluxes for 1983–2005 in the North Atlantic Ocean Near Bermudaa
YearWind Speed, m s−1ΔpCO2Fall Air-Sea CO2 Flux, mmol CO2 m−2
BWSECMWFNNRQuadratic Dependency U2Cubic Dependency U3
MeanSDMeanSDMeanSDMeanBWS MeanECMWF MeanNNR MeanBWS MeanECMWF MeanNNR Mean
  • a

    Annual wind speed data and statistics were compiled from BWS, ECMWF, and NNR datasets. Air-sea CO2 fluxes were determined using seawater pCO2 from BATS/Hydrostation S, atmospheric pCO2 data, wind speed data, and U2 (quadratic) and U3 (cubic) parameterizations of k. BWS, ECMWF, and NNR wind speed data sets were averaged over each day. If hourly BWS or 6-hourly ECMWF or NNR data sets were used, air-sea CO2 fluxes were ∼3–5% higher using the U2 (quadratic) parameterizations of k. Negative CO2 flux values denote net air to sea CO2 flux, whereas positive CO2 flux values denote net sea to air CO2 flux.

19845.931.757.192.757.913.29−12.3−116−182−206−105−226−273
19856.631.696.652.417.432.90−9.4−166−193−239−192−270−389
19866.651.826.282.407.132.81−15.5−214−215−282−232−259−400
19877.712.427.753.598.934.18−4.2−169−215−283−271−408−661
19887.352.166.923.027.703.40−26.8−458−464−573−586−698−966
19897.442.207.212.937.913.50−13.6−218−234−292−275−338−492
19907.422.216.852.917.723.35−17.6−259−242−313−319−329−492
19917.492.247.062.837.673.57−19.9−303−285−329−376−382−522
19927.461.817.522.788.503.33−17.1−309−343−453−378−513−815
19937.002.217.163.168.003.84−4.0−133−189−272−198−327−575
19947.311.917.102.868.303.19−21.8−302−332−455−343−446−740
19957.702.797.133.218.153.90−9.9−283−271−380−486−436−745
19967.562.216.982.868.283.56−12.1−221−225−316−293−319−573
19977.992.347.882.799.243.56−10.0−220−237−357−327−378−718
19987.042.016.542.767.553.42−16.1−193−199−276−226−264−452
19997.781.977.342.398.322.90−7.8−137−132−181−200−188−319
20007.992.228.953.259.163.25−17.6−309−424−424−400−710−696
20017.632.598.103.288.273.44−21.9−388−507−518−498−814−827
20027.952.42  8.913.44−18.7−416 −559−632 −1078
20038.192.01  8.853.01−15.2−313 −428−433 −747
20047.612.31  8.163.21−13.8−201 −263−244 −400
20057.302.18  8.123.13−7.9−128 −190−171 −319
Range       342 379526 804
Mean7.41 7.26 8.19 −14.2−248−272−345−327−406−600
SD0.52 0.62 0.57 5.995104115137176217
Table 7b. Fall (OND) Trends of Wind Speed and Air-Sea CO2 Fluxes for 1983–2005 in the North Atlantic Ocean Near Bermuda
TypePeriodChangeRate of Change per Yearr2p-value
BWS wind speed01/1984–12/2005+1.14 (15%)+0.052 ± 0.014 m s−10.42<0.01
ECMWF wind speed01/1984–12/2002+1.33 (18%)+0.061 ± 0.024 m s−10.270.03
NNR wind speed01/1984–12/2005+1.14 (14%)+0.044 ± 0.017 m s−10.250.02
 
U2 (Quadratic) Parameterization of k
BWS wind speed01/1984–12/2005−54 (22%)−2 ± 3 mmol CO2 m−20.030.46
ECMWF wind speed01/1984–12/2002−130 (48%)−6 ± 5 mmol CO2 m−20.090.22
NNR wind speed01/1984–12/2005−69 (20%)−3 ± 4 mmol CO2 m−20.030.43
 
U3 (Cubic) Parameterization of k
BWS air-sea CO2 flux01/1984–12/2005−117 (36%)−5 ± 5 mmol CO2 m−20.060.25
ECMWF air-sea CO2 flux01/1984–12/2002−264 (65%)−12 ± 8 mmol CO2 m−20.130.14
NNR air-sea CO2 flux01/1984–12/2005−139 (23%)−6 ± 7 mmol CO2 m−20.040.40

4.2. Comparison of Air-Sea CO2 Fluxes Using Different Wind Speed Data Sets (i.e., BWS, ECMWF, and NNR)

[41] Comparison of air-sea CO2 fluxes computed using the three different wind speed data sets showed considerable differences. The differences in air-sea CO2 fluxes relate primarily to the type, distribution and frequency of wind speed data, and the nonlinear nature of the relationship between wind speed and CO2 gas transfer coefficient [e.g., Wanninkhof, 1992; Wanninkhof and McGillis, 1999] rather than ΔpCO2 variability (note: the same ΔpCO2 data set is used for all computations of air-sea CO2 flux).

[42] Distinct differences were found in the daily averaged wind speed of the BWS, ECMWF and NNR data. Wind speed data from the data assimilation models (i.e., ECMWF and NNR) had a broader distribution than the synoptic meteorological observations (i.e., BWS; Figure 6). In general terms, the ECMWF and NNR data had a greater frequency of wind speeds in the 0–2 m s−1 and >8 m s−1 ranges (wind distributions for BWS, ECMWF and NNR are plotted each year in the accompanying auxiliary material). For example, during the relatively calm summertime, (i.e., when CO2 effluxes from the ocean to the atmosphere and wind speeds typically range from 0−<7 m s−1), the mean wind speed for the BWS data (6.30 ± 0.20 m s−1) was higher compared to ECMWF (5.21 ± 0.41 m s−1) and NNR (5.96 ± 0.39 m s−1) data (Table 6a and 6b). During the windier fall and wintertime, (i.e., when there is an influx of CO2 from the atmosphere to the ocean and wind speeds typically range from >3−<10 m s−1), the mean wind speed for the JFMAM NNR data (8.62 ± 0.46 m s−1) was higher compared to BWS (7.76 ± 0.38 m s−1) and ECMWF (7.60 ± 0.65 m s−1) JFMAM data (Tables 5a and 5b). In both seasons, the larger standard deviations of ECMWF and NNR data compared to BWS data (Tables 57 and Figure 6) reflect the broader wind speed distribution of the data assimilation models.

image

Figure 6. Wind speed distributions for BWS data from the island of Bermuda, and ECMWF and NNR data assimilation model data sets for the 2.5° box centered on the island of Bermuda. Wind speeds have been binned into 1 m s−1 intervals.

Download figure to PowerPoint

[43] The local differences in wind speed distributions between synoptic meteorological observations (i.e., BWS) and data assimilation model data sets in the vicinity of Bermuda is reflective of more general difficulties in comparing and applying different wind speed data products. For example, over land, the NNR surface wind speed data sets have to be adjusted since the NNR data assimilation model is heavily weighted to upper air observations and does not use surface observations [Kalnay and Cai, 2003]. Over the ocean, large-scale deficiencies in NCEP surface heat, freshwater and momentum fluxes are adjusted by temperature and salinity conditions so that the NNR model simulation becomes reasonably consistent with ocean observations [e.g., Stammer et al., 2004]. Island meteorological observations such as those from Bermuda are treated like land observations, and only surface pressure is assimilated into the NNR, with the NNR wind stress model typically calculated from surface and profile pressure and adjusted to inferred satellite wind stress measurements. Recent studies suggest that NNR surface winds over the tropical and midlatitude ocean tend to be overestimated compared to observations [Stammer et al., 2004; Menemenlis et al., 2005].

[44] The differences in wind speed distributions and type imparted considerable differences in computed air-sea CO2 fluxes at the BATS/Hydrostation S site. For example, during the summertime (JJAS), average wind speed and air-sea CO2 effluxes are higher for BWS (+473 ± 147 mmol CO2 m−2 yr−1) compared to ECMWF (+334 ± 106 mmol CO2 m−2 yr−1) and NNR data sets (+463 ± 150 mmol CO2 m−2 yr−1) (Table 6a and 6b). In contrast, during wintertime, computed air-sea CO2 influxes using NNR (−1417 ± 190 mmol CO2 m−2 yr−1) data are much higher than using BWS (−1038 ± 155 mmol CO2 m−2 yr−1) or ECMWF (−1081 ± 219 mmol CO2 m−2 yr−1) (Tables 5a and 5b). The NNR data also has a greater tendency to fall more heavily toward the higher wind speeds. Given the nonlinear nature of the relationship between wind speed and CO2 gas transfer coefficient [e.g., Wanninkhof, 1992; Wanninkhof and McGillis, 1999], it not surprising that the air-sea CO2 fluxes are higher when computed using NNR data.

[45] The above discussion on air-sea CO2 fluxes has assumed a quadratic dependency between wind speed and gas CO2 gas transfer coefficient. However, if the air-sea CO2 fluxes computed using BWS, ECMWF and NNR data sets and the cubic wind speed–gas transfer relationship [Wanninkhof and McGillis, 1999], the rates were significantly different by ∼110% (Tables 4a and 4b).

[46] The above analysis indicates that, in addition to known uncertainties (i.e., a factor of 2) in the relationship between wind speed and gas CO2 gas transfer coefficient, the different types of wind speed data used to compute air-sea CO2 fluxes imparts additional uncertainty. Determining and comparing local, regional and global rates of air-sea CO2 exchange thus has to be undertaken with care. In previous studies, net annual air-sea CO2 influxes at the BATS site were estimated at −531 ± 99 mmol CO2 m−2 yr−1 for the period 1989–1995 [Bates et al., 1998a] using daily averaged wind speed data. In contrast, Gruber et al [2002] reported an average net air-sea CO2 influx of −1900 ± 200 mmol CO2 m−2 yr−1 for the 1983–2001 period, using monthly averaged wind speed derived from an earlier version of the NCEP/NCAR reanalysis data assimilation model (using the 2.5° model output centered on the island of Bermuda). Both studies used the same quadratic wind speed–gas exchange parameterization. Compared to the range of annual air-sea CO2 fluxes (−810 to −1295 mmol CO2 m−2 yr−1) computed here, the earlier estimates of Bates et al. [1998a] likely underestimate annual air-sea CO2 fluxes while the study of Gruber et al. [2002] likely overestimates CO2 fluxes.

5. The Influence of Hurricanes on Air-Sea CO2 Fluxes

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

[47] Previous studies have shown that episodic wind events such as hurricanes and midlatitude storms have the potential for significantly impacting air-sea CO2 gas exchange [Bates et al., 1998b; Kawabata et al., 2003; Kuss et al., 2004; Perrie et al., 2004]. For example, in 1995, three hurricanes (Felix, Luis and Marilyn) passing near the island of Bermuda, increased the air-sea CO2 efflux during the summertime period [Bates et al., 1998a, 1998b, Figure 3b].

[48] The potential for significant impact of hurricanes on air-sea CO2 gas exchange is based on several major assumptions and unknowns. Experimental parameterizations for the gas transfer velocity k have been typically derived for wind speeds up to ∼15 m s−1 only [e.g., Wanninkhof, 1992; Wanninkhof and McGillis, 1999], with k linearly extrapolated at higher wind speeds. At hurricane strength winds, boundary layer atmospheric physics and the presence of a mixed layer entrained bubble field beneath the hurricane could reduce k from its extrapolated values with a reduction of CO2 gas exchange from expected prediction (R. Wanninkhof, personal communication, 2005). Recent experiments suggest that k values are similar to those predicted by Wanninkhof [1992] for winds in the ∼35−>50 m s−1 range, but ∼20−40% lower for winds in the ∼20−>45 m s−1 [McNeil and D'Asaro, 2006; D'Asaro and McNeil, 2006]. Accurate determination of air-sea CO2 flux, and the impact of decreased atmospheric pressure and vertical mixing on CO2 flux under high wind speed remains to be experimentally quantified.

[49] If the k parameterizations of Wanninkhof [1992] reasonably predict CO2 fluxes, major hurricanes can significantly impact air-sea CO2 fluxes. Over the 1983 to 2005 period, the passage of hurricanes close to the island of Bermuda and the BATS/Hydrostation S increased the summertime efflux of CO2 from the ocean by <+10 to >+150 mmol CO2 m−2 or <3–29% (assuming the quadratic dependency holds at wind speeds >15 m s−1; Table 8). Strong winds associated with two notable major hurricanes, Hurricanes Felix (1995) and Fabian (2003) represented a potential air-sea CO2 flux of ∼102 and 81 mmol CO2 m−2, respectively. Hurricane Fabian (a major category 3 hurricane on the Saffir-Simpson scale; National Weather Service, National Hurricane Centre, Tropical Prediction Center; http://www.nhc.noaa.gov) with sustained winds of >45 m s−1 passed directly over the BATS/Hydrostation S sites and the island of Bermuda in September 2003. Hurricane Felix gave the island of Bermuda two glancing blows (due to a doubling back of the hurricane) and strong winds over a 7-day period.

Table 8. Wind Speed and Air-Sea CO2 Fluxes for Hurricanes and Tropical Storms Observed Near the Island of Bermuda Over the 1983–2005 Perioda
YearHurricaneDateMaximum Wind, m s−1Flux due to Hurricane EventSummertime Flux, BWS, mmol CO2 m−2TotalContribution to Total Flux, %
BWS, mmol CO2 m−2ECMWF, mmol CO2 m−2NNR, mmol CO2 m−2
  • a

    Air-sea CO2 fluxes were determined using seawater pCO2 from BATS/Hydrostation S, atmospheric pCO2 data, wind speed data, and U2 (quadratic) and U3 (cubic) parameterizations of k. BWS, ECMWF, and NNR wind speed data sets were averaged over each day. If hourly BWS or 6-hourly ECMWF or NNR data sets were used, air-sea CO2 fluxes were typically less than 5% higher using the U2 (quadratic) parameterization of k, and <10% higher using the U3 (cubic) parameterization of k. Only for Hurricane Fabian were fluxes higher using hourly rather than daily wind speeds (shown in parentheses). Positive air-sea CO2 flux values denote net sea-to-air CO2 flux.

  • b

    Hurricane Felix has a +188(365) mmol CO2 m−2 flux if calculated using the cubic wind speed-gas transfer relationship.

  • c

    Hurricane Fabian has a +214(480) if calculated using the cubic wind speed-gas transfer relationship.

  • d

    Calculated with hourly wind speed data.

1985TS Ana16 Jul 198515.1201725+410+4511
 TS Claudette11–14 Aug 198514.2252436   
1987Arlene13–15 Aug 198715.8331422+389+4111
 EmilySep 25 198728.6865   
1989DeanAug 6–7 198924.23422 +361+349
1990Bertha29 Jul to 1 Aug 199015.4565083+500+6913
 Lili11–12 Oct 199014.1131216   
1993Floyd8–9 Sep 199311.2201725+633+203
1994TS Chris21–23 Aug 19949.4332632+679+335
1995Felix14–18 Aug 199526.2102b96148+611+17729
 Luis9–10 Sep 199519.1403460   
 Marilyn18–21 Sep 199519.5352647   
1996Eduoard31Aug to Sep 2 199612.1353458+589+6010
 Hortense13–14 Sep 199613.9252344   
1998Danielle1–2 Sep 199816.5475274+752+476
1999Gert20–23 Sep 199919.8514372+678+518
2000Florence16–17 Sep 200016.2101114+449+102
2001TS Dean25 Aug 200113.7181919+296+5820
 Erin9–11 Sep 200114.3221728   
 Gabrielle17–18 Sep 200116.4182428   
2003Fabian4–6 Sep 200346.881c (102)dn/a51+660+81(102)12(16)%

[50] The influence of hurricanes on air-sea CO2 fluxes is predicted to be much greater if the cubic wind speed–gas exchange dependency (equation (3)) and hourly meteorological observations are used. Although the cubic wind speed–gas exchange dependency is likely not to hold a very high wind speeds (e.g., >25 m s−1), Hurricanes Felix and Fabian represented a potential air-sea CO2 efflux of ∼+365 and +480 mmol CO2 m−2, respectively, almost equivalent to the total summertime efflux of CO2 from the ocean each year (Table 6a and 6b). These high CO2 flux values are unlikely given the recent results of McNeil and D'Asaro [2006] and D'Asaro and McNeil [2006].

[51] As shown in earlier studies of Bates et al. [1998a], hurricanes are likely to have a regional impact on air-sea CO2 fluxes. Air-sea CO2 effluxes may be higher (assuming a quadratic wind speed–gas transfer velocity) in the tropical regions due to higher ΔpCO2 (e.g., Takahashi et al., 2002). For example, air-sea CO2 effluxes observed from Hurricane Fabian (∼+100–400 mmol CO2 m−2 event−1) scaled to approximately ∼500,000−900,000 km2 of ocean surface transited over the 6-day lifetime history of this hurricane across the North Atlantic (assuming climatological ΔpCO2 conditions across the region [Takahashi et al., 2002) yields a potential range of CO2 efflux of ∼0.6–4 Tg C (1012 g C). The total number of major hurric]anes (i.e., category 3 or higher on the Saffir-Simpson scale) occurring each year in the North Atlantic Ocean basin has typically varied from 0 to 6 over the last 3 decades [e.g., Elsner et al., 2004; Molinari and Mestas-Nuñez, 2003]. If hurricane Fabian is representative of major hurricanes transiting across the warm, supersaturated tropics and subtropical regions of the North Atlantic, year-to-year variability in the number of hurricanes could potentially impart a year-to-year variability in the North Atlantic Ocean CO2 efflux of ∼3–24 Tg C (or 0.003–0.024 Pg C). It should also be noted that there has been a stepwise increase in the number of major hurricanes occurring in the North Atlantic basin each year since 1995 (0–2 pre-1995; 2–6 post-1995 [Elsner et al., 2004; Molinari and Mestas-Nuñez, 2003]). If the impact of hurricanes like Fabian (with a lifetime of 6 days) are scaled globally with the consideration that there is typically ∼130–250 hurricane days per year (1983–1992 [Bates et al., 1998b]), tropical cyclones could potentially contribute to an air-sea CO2 efflux of 40–80 Tg C yr−1 (0.040–0.080 Pg C yr−1), a smaller range than previously estimated (0.04–0.51 Pg C yr−1 [Bates et al., 1998b]).

6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

6.1. Interannual Variability of Air-Sea CO2 Fluxes

[52] Over the 1984 to 2005 period, there were standout years for air-sea CO2 fluxes. The highest net annual air-sea CO2 influxes occurred in 1988 and 2002 (also 2000 and 2001 for ECMWF wind data) depending on type of wind used (Tables 4a and 4b). The lowest net annual air-sea CO2 fluxes occurred in 1998 and 1999 (Tables 4a and 4b).

[53] The range of annual air-sea CO2 influxes was ∼−868 mmol CO2 m−2 yr−1, ∼−1005 mmol CO2 m−2 yr−1, and ∼−1182 mmol CO2 m−2 yr−1 when computed with BWS, ECMWF and NNR wind speed data, respectively. Gruber et al. [2002] suggested that the peak-to-peak variability in air-sea CO2 fluxes (i.e., 800 mmol CO2 m−2) observed near Bermuda, when extrapolated to the entire subtropical gyre (15°N to 40°N; assuming an area of ∼1.7 × 107 km2), represented an interannual variability of 0.2 Pg C yr−1 and sufficient to contribute measurably to Northern Hemisphere-scale atmospheric CO2 anomalies. In this study, the peak-to-peak variability of ∼800−1200 mmol CO2 m−2 yr−1, if scaled to the entire subtropical gyre, constitutes an interannual variability of ∼0.2−0.3 Pg C yr−1 in air-sea CO2 fluxes in the midlatitude North Atlantic Ocean.

6.2. Long-Term Trends of Wind speed and Air-Sea CO2 Fluxes

[54] During the 1984 to 2005 period, the mean annual and seasonal wind speed has increased considerably. For example, synoptic meteorological data (i.e., BWS) and COADS data in the vicinity of the island of Bermuda exhibited an increase of wind speed by 1.00 and 1.10 m s−1, respectively (Tables 4a and 4b). The data assimilation models of ECMWF and NNR) also showed an increase in wind speed of 1.10 and 1.23 m s−1, respectively, over the last 22 years for the 2.5° box overlying Bermuda (Tables 4a and 4b). These changes represented a ∼10–17% change in wind speed over the period of this study.

[55] The long-term increase in wind speed was, however, not uniformly distributed seasonally. The smallest changes in wind speed occurred during the summertime (i.e., during air-sea CO2 efflux). The BWS, ECMWF and NNR data sets show modest increases in wind speed of 0.08, 0.60 and 0.46 m s−1, respectively, over the last 22 years (Table 6a and 6b; p values were not significant though). In contrast, the largest seasonal increases occurred during the fall and wintertime periods. For example, the BWS, ECMWF and NNR data sets exhibited statistically significant increases in wind speed of 0.93, 1.51 and 0.69 m s−1, respectively, during wintertime (Tables 5a and 5b), but also during the fall (Table 7a and 7b). This represents a 13–21% change in wintertime wind speed over the period of this study.

[56] The increase in wind speed observed in Bermuda has also observed elsewhere in the marginal seas of the North Atlantic [e.g., Pirazzoli and Tomasin, 2003; Pirazzoli, 2005]. Elsewhere, for example, the frequency of winter storms and wave heights occurring in the subpolar region of the North Atlantic, enhanced during the predominantly NAO positive period of 1980 to 1995 [Beersma et al., 1997; WASA group, 1998; Bijl et al., 1999; Alexandersson et al., 2000; Alexander et al., 2005] appears to be in decline since the mid-1990s [Weisse et al., 2005]. Large spatiotemporal changes in wind speed are evident between 1984 and 2005 are evident from both NNR and ECMWF data (Figure 7). For example, increases of +0.5−>1.0 m s−1 in wind speed have been observed near Bermuda in the western North Atlantic Ocean, but also in the marginal seas of Europe (e.g., North Sea, Mediterranean Sea, Baltic Sea, Bay of Biscaye), Greenland and Labrador Seas. In other regions, decreases in wind speed of up to 1 m s−1 have been observed in the tropical North Atlantic, off Nova Scotia and Newfoundland, and in the NE Atlantic Ocean. Such changes not only have potential influence for gas exchange rates, but, for heat and freshwater fluxes, mixing and stratification, circulation, biological production and ocean ecosystem structure, for example.

image

Figure 7. Spatiotemporal changes in wind speed (m s−1) over the North Atlantic Ocean over the 1984–2005 period. Although the plotted wind speed data are NNR data, ECMWF data (not shown) exhibit similar spatiotemporal patterns.

Download figure to PowerPoint

[57] Rates of air-sea CO2 flux (as with wind speed) also changed over the 1983 to 2005 period. Seasonally, the summertime oceanic CO2 source to the atmosphere in the subtropical gyre has substantially increased (Table 6a and 6b). For example, summertime air-sea CO2 effluxes computed with the BWS, ECMWF and NNR data sets exhibited an increase of ∼+11−14 mmol CO2 m−2 per year or +251−302 mmol CO2 m−2, respectively, over the 1983−2005 period (Table 6a and 6b). The trends in summertime air-sea CO2 fluxes were statistically significant for fluxes computed from BWS, ECMWF and NNR data sets (Table 6a and 6b; p values 0.02 to <0.01). The change in the summertime CO2 source status was however compensated for by a larger increase in the fall and wintertime air-sea CO2 influx. Wintertime air-sea CO2 influxes computed using BWS, ECMWF and NNR data sets increased by ∼−14−19 mmol CO2 m−2 per year or by −297 to −414 mmol CO2 m−2 period−1, respectively, from 1984 to 2005 (Tables 5a and 5b; p values <0.01). The trends in wintertime air-sea CO2 influxes were statistically significant for fluxes computed from BWS and NNR data sets (Tables 5a and 5b; p values <0.01), but not for ECMWF data sets (note: from 1984 to 2001 only). The increase in seasonal rates of air-sea CO2 flux appears primarily associated with long-term changes in wind speed rather than changes in ΔpCO2 conditions. As shown earlier, both surface seawater and atmospheric pCO2 have increased at similar rates over the period of this study (Table 1).

[58] Although both fall/wintertime and summertime air-sea CO2 fluxes have increased, annual rates of air-sea CO2 influx exhibited relatively small increases of ∼−2–6 mmol CO2 m−2 per year, or 39–166 mmol CO2 m−2, respectively, over the 1984 to 2005 period (Tables 4a and 4b; computed from BWS, ECMWF and NNR data sets). If scaled to the entire subtropical gyre, these changes constitute a longer-term increase of ∼0.01–0.02 Pg C in the air-sea CO2 flux or oceanic CO2 sink in the midlatitude North Atlantic Ocean from 1984 to 2005. However, the long-term trends in annual air-sea CO2 fluxes were statistically insignificant for fluxes when computed from all wind speed data sets (Tables 4a and 4b; p values >0.05).

6.3. Potential Causes for Long-Term Trends of Wind Speed and Air-Sea CO2 Fluxes?

[59] What are the underlying factors that influence the large year-to-year variability and long-term trends of air-sea CO2 fluxes? The fall (i.e., OND) period is characterized by rapidly cooling and deepening mixed layers [Bates et al., 1996a], a transition to undersaturated seawater CO2 conditions and a period of net air-sea CO2 influx. The wintertime (i.e., JFMAM) period is typically characterized by higher wind speeds, deeper mixed layers (i.e., 100−>150 m [e.g., Steinberg et al., 2001]), seawater CO2 conditions undersaturated with respect to the atmosphere (i.e., negative ΔpCO2), and net air-sea CO2 influx. The following summertime June to September (i.e., JJAS) period is characterized by lower wind speeds, shallow mixed layers (i.e., ∼10–>50 m [e.g., Steinberg et al., 2001]), seawater CO2 conditions oversaturated with respect to the atmosphere, and net air-sea CO2 fluxes from the ocean. Since air-sea CO2 fluxes are directly related to ΔpCO2 and wind speed (equation (2)), air-sea CO2 fluxes each month were strongly correlated with both parameters. In addition, air-sea CO2 fluxes each month were correlated with surface temperature each season, and, with the exception of fall, weakly correlated with variability of other parameters such as salinity, DIC or total alkalinity.

[60] Previous studies have shown that interannual variability of hydrographic, and biogeochemical properties in the North Atlantic subtropical gyre are coupled to modes of low-frequency climate variability such as NAO, AO and ENSO [e.g., Oschlies, 2001; Bates, 2001; Gruber et al., 2002; Bates and Hansell, 2004]. The NAO/AO exerts influence throughout the year [Marshall et al., 2001], but is the dominant leading mode in the wintertime JFMAM period. Here low-frequency modes of climate variability can be viewed as the expression of physical changes in the North Atlantic Ocean that can in turn potentially influence air-sea CO2 fluxes by altering atmospheric forcing (e.g., wind speed distributions) and oceanic properties (e.g., surface temperature and salinity distributions, water-column stratification, and primary production) that in turn influences ΔpCO2 conditions.

[61] In the North Atlantic Ocean, the NAO is the dominant mode of low-frequency climate variability [Marshall et al., 2001]. The NAO can be viewed as the regional expression of the AO [e.g., Thompson and Wallace, 1998; Wallace, 2000] or Northern Annular Mode (NAM) [Thompson and Lorenz, 2004; Quadrelli and Wallace, 2004], with the AO being the leading wintertime Northern Hemispheric low-frequency mode of sea level pressure (SLP) variability. In the North Atlantic, the NAO manifests as a dipole SLP oscillation between the Icelandic low-pressure and Azores high atmospheric pressure centers [e.g., Rogers, 1990; Hurrell, 1995; Jones et al., 1997; Osborn et al., 1999; Hurrell and Van Loon, 1997; Hurrell et al., 2001, 2002], with variability of parameters influenced by the NAO exhibiting a tripole pattern in the North Atlantic Ocean [Marshall et al., 2001]. Maps of the AO and NAO modes of variability are nearly indistinguishable [Marshall et al., 2001].

[62] The NAO/AO has a pronounced effect on the Northern Hemisphere [e.g., Visbeck et al., 2001; Hurrell et al., 2002], exhibiting considerable seasonal and interannual variability. In the subpolar region, positive (negative) states of the NAO index tend to result in enhanced (reduced) precipitation, storminess and wave heights [e.g., Beersma et al., 1997; WASA group, 1998; Bijl et al., 1999; Alexandersson et al., 2000; Alexander et al., 2005]. During a NAO positive state, the mean position of Gulf Stream tends to shift northward [Taylor et al., 1998; Taylor and Stephens, 1998; Weisse et al., 2005], and baroclinic mass transport of the Gulf Stream increases [Curry and McCartney, 2001; Molinari, 2004].

[63] In the midlatitudes of the North Atlantic Ocean, positive and (negative) states of the NAO index result in reduced (enhanced) midlatitude westerlies, reduced (enhanced) wind stress and heat exchange [Bjerknes, 1964; Cayan, 1992a, 1992b], enhanced (reduced) sea surface temperature [e.g., Davies et al., 1997; Kapala et al., 1998; Rodwell et al., 1999] recorded in ocean temperatures [Bates, 2001] and corals from Bermuda [e.g., Kuhnert et al., 2005]. Interannual anomalies of mixed layer depths and temperature, integrated primary and new production, phytoplankton community structure, and seawater CO2 properties in the North Atlantic Ocean are correlated to the NAO [Oschlies, 2001; Bates, 2001; Bates and Hansell, 2004; Lomas and Bates, 2004].

[64] The NAO/AO may also be related to the Tropical Atlantic Variability (TAV; a covarying fluctuation of SST and trade winds straddling the Intertropical Convergence Zone, ITCZ), with both the TAV and NAO/AO exerting influence on air-sea interaction in the tropics of the North Atlantic [Marshall et al., 2001]. The Pacific Ocean ENSO, however, exerts its influence on the tropical Caribbean Sea and western Atlantic Ocean with a typical lag time of 6–9 months [e.g., Zhang et al., 1996; Bojariu, 1997; Penland and Matrosova, 1998]. For example, sea surface salinity and temperature anomalies observed at BATS and Hydrostation S in the 1990s have been shown to be correlated to ENSO [Bates, 2001]; a pattern also observed in the Caribbean Sea and tropical Atlantic Ocean [e.g., Zhang et al., 1996; Bojariu, 1997; Penland and Matrosova, 1998].

[65] Do air-sea CO2 fluxes in the subtropical gyre of the western North Atlantic Ocean correlate with the low-frequency modes? On monthly timescales, monthly averaged air-sea CO2 fluxes, ΔpCO2, and wind speed were not significantly correlated with monthly values of climate indices such as NAO, AO and ENSO (i.e., Southern Oscillation Index, SOI, with and without a 6-month lag). This is perhaps not surprising as the low-frequency modes in the Northern Hemisphere exhibit considerable month-to-month variability. In addition, the Northern Hemisphere modes including the NAO and other patterns such as the East Atlantic (EA) [e.g., Wallace and Gutzler, 1981; Quadrelli et al., 2001; Ferreira and Frankignoul, 2005] and Pacific/North American (PNA) [e.g., Barnston and Livezey, 1987; Wallace et al., 1990; Honda and Nakamura, 2001; Wallace and Thompson, 2002] typically explains ∼50−70% (and often much less) of the SLP variance (http://www.cpc.ncep.noaa.gov). On seasonal timescales, air-sea CO2 fluxes were correlated with NAO, but not generally with the AO and SOI (or other Northern Hemisphere modes).

[66] During the wintertime period (i.e., JFMAM), the correlations between air-sea CO2 influxes, and NAO, AO and SIO were generally poor (r2 < 0.2; p values >0.05) for fluxes calculated over the 1984–2005 period from the different types of wind speed data (i.e., BWS, NNR, and ECMWF) and using quadratic and cubic wind speed–flux relationships (Table 9). Weak correlations (r2 values of ∼0.17 to 0.40; p values of 0.44 to 0.06) existed between air-sea CO2 fluxes calculated from NNR wind speed data, and NAO, AO and SOI (with a 6-month lag). However, some generalizations can be made about wintertime air-sea CO2 fluxes. The highest air-sea CO2 influxes (calculated with NNR wind speed data and cubic wind speed–flux relationships) tended to occur during years with strongly negative wintertime AO values (r2 value of ∼0.40; p value of 0.06; Figure 8). Furthermore, higher air-sea CO2 fluxes (calculated with NNR wind speed data and cubic wind speed–flux relationships) also tended to occur during years with negative wintertime SOI values (r2 value of ∼0.35; p value of 0.10). In winters with strongly negative NAO and AO values (i.e., 1985, 1996, 2001, 2005), air-sea CO2 fluxes were 9–43% higher in El Niño years (i.e., winters with negative SIO with 6-month lag). In winters with negative NAO and positive AO values (i.e., 1984, 1986–1988, 1998, 1999, 2004), air-sea CO2 fluxes were also higher by 15–28% in El Niño years compared to La Niña years. Last, in winters with strongly positive NAO and AO values (i.e., 1989–1995, 1997, 2000, 2002, 2003), air-sea CO2 fluxes were higher by 11–22% in El Niño years.

image

Figure 8. Seasonal relationships between air-sea CO2 fluxes and low-frequency climate modes, and wind speed and ΔpCO2. Regression statistics are detailed in Table 9. (a) wintertime (JFMAM) cubic air-sea CO2 fluxes (NNR data, solid diamonds; BWS data, grey circle) and mean JFMAM Arctic Oscillation (AO) index. (b) Mean wintertime (JFMAM) ΔpCO2 and mean wintertime (JFMAM) wind speed (m s−1) (NNR data, solid diamonds; BWS data, grey circle). (c) Summertime (JJAS) cubic air-sea CO2 fluxes (NNR data, solid diamonds; BWS data, grey circle) and mean JJAS North Atlantic Oscillation (NAO) index. (d) Fall cubic (OND) air-sea CO2 fluxes (NNR data, solid diamonds; BWS data, grey circle) and mean OND North Atlantic Oscillation (NAO) index.

Download figure to PowerPoint

Table 9. Correlations of r2 and p-Values Between Seasonal Mean Air-Sea CO2 Fluxes and Mean Seasonal Low-Frequency Climate Modesa
 NAOAOSOISOI With 6 Month Lag
r2p-Valuer2p-Valuer2p-Valuer2p-Value
  • a

    Modes are NAO, AO, SOI and SOI with 6-month lag. Asterisks indicate r values < 0.15. Correlations having statistical significance are indicated with bold.

Quadratic Wind Speed–k Relationship
JFMAM air-sea CO2 fluxes
   BWS********
   NNR**0.170.43**0.200.38
   ECMWF****0.290.220.180.45
JJAS air-sea CO2 fluxes
   BWS0.310.16****0.340.13
   NNR0.400.07****0.330.12
   ECMWF0.390.10****0.220.37
OND air-sea CO2 fluxes
   BWS0.410.06******
   NNR0.480.02******
   ECMWF0.350.16******
 
Cubic Wind Speed–k Relationship
JFMAM air-sea CO2 fluxes
   BWS**0.180.43****
   NNR0.290.190.400.06**0.350.10
   ECMWF****0.280.240.180.45
JJAS air-sea CO2 fluxes
   BWS0.310.16****0.270.22
   NNR0.420.050.200.37**0.280.21
   ECMWF0.420.09******
OND air-sea CO2 fluxes
   BWS0.520.01******
   NNR0.560.000.230.29**0.300.17
   ECMWF0.390.12******

[67] Wintertime air-sea CO2 influxes were higher in the 2000s (i.e., 2000–2005) compared to the 1980s (i.e., 1984–1989) and 1990s (Table 10). During the 2000s, there were proportionately more winters with higher air-sea CO2 influx cooccurring with winters of strongly negative AO and/or negative SIO values (i.e., El Niño years). Thus the increase in wintertime air-sea CO2 influx described earlier (in section 6.2) presumably reflects this interannual variability.

Table 10. Mean Seasonal Air-Sea CO2 Fluxes and Mean Seasonal Low-Frequency Climate Modes for the 1980s, 1990s and 2000sa
 Mean Air-Sea CO2 FluxMean Climate Index
BWSNNRNAOAO
  • a

    Air-sea CO2 fluxes are computed using the quadratic windspeed-k relationship.

JFMAM Air-Sea CO2Fluxes
1980s (1983–1989)−948−1389+0.30−0.20
1990s−989−1320+0.58+0.31
2000s (2000–2005)−1208−1606+0.32−0.07
 
JJAS Air-Sea CO2Fluxes
1980s (1983–1989)+309+296+0.15+0.00
1990s+570+542−0.02+0.04
2000s (2000–2005)+476+495−0.02+0.07
 
OND Air-Sea CO2Fluxes
1980s (1983–1989)−222−141+0.12+0.10
1990s−235−161−0.06+0.05
2000s (2000–2005)−293−196−0.25−0.30

[68] Previous studies [Gruber et al., 2002; McKinley et al., 2004] have suggested that wintertime air-sea CO2 influxes in the subtropical gyre of the North Atlantic are coupled to the wintertime state of the NAO. However, the lack of good correlations between wintertime air-sea CO2 influxes and low-frequency climate modes observed at the BATS site (1984–2005) suggests that climate patterns such as the NAO only partially explain the variance observed in SLP (and other factors) that in turn can influence gas exchange. In addition, wintertime wind speed and ΔpCO2 tend to be anticorrelated (Figure 8b) potentially suppressing large interannual changes in wintertime air-sea CO2 fluxes due to atmospheric forcing associated with low-frequency modes or other factors. For example, the enhancement of westerlies in the midlatitudes of the North Atlantic Ocean during a negative NAO (or AO) state [Bjerknes, 1964; Cayan, 1992a, 1992b; Davies et al., 1997; Kapala et al., 1998; Rodwell et al., 1999] should presumably increase wind speed and enhance air-sea CO2 flux. During a negative NAO (or AO) state, sea surface temperatures cool, the mixed layer deepens and integrated primary and new production increases due to enhanced vertical supply of nutrients [e.g., Oschlies, 2001; Bates, 2001; Lomas and Bates, 2004]. However, deeper vertical mixing tends to bring excess DIC relative to Redfield stoichiometric ratios [e.g., Redfield et al., 1963] of DIC and nitrate resulting in positive surface DIC anomalies [Bates, 2001] that in turn reduce ΔpCO2 values (and reduce air-sea CO2 flux). Thus interannual changes in atmospheric forcing (as reflected by changes in the NAO, AO or SIO indices) may not result in large year-to-year variability in wintertime air-sea CO2 fluxes.

[69] In contrast to the wintertime, summertime (JJAS) and fall (OND) air-sea CO2 effluxes were correlated with NAO, but not significantly correlated with AO or SIO (with or without 6-month lag) (Table 9). In summertime, air-sea CO2 effluxes were correlated with summertime NAO values (r2 values of ∼0.30–0.41; p values of 0.16–0.06). The highest air-sea CO2 effluxes tended to occur during years with negative summertime NAO values (Figure 8c). In the fall, air-sea CO2 influxes were correlated with fall NAO values (r2 values of ∼0.48–0.56; p values of 0.02–<0.01). Higher air-sea CO2 fluxes tended to occur during years with negative NAO values for the fall. In both seasons, irrespective of the direction of gas exchange (and ΔpCO2 values), the highest air-sea CO2 fluxes cooccurred with periods of negative NAO values (Figure 8d). Although NAO cannot directly influence gas exchange, NAO variability appears to characterize changes in atmospheric forcing that influence air-sea CO2 fluxes. Previous studies have also shown that summertime SST and SLP anomalies in the midlatitudes of the North Atlantic Ocean exert a preconditioning influence on the following October to December fall period [e.g., Czaja et al., 2002; Czaja and Frankignoul, 2002; Cassou et al., 2004; Frankignoul and Kestenare, 2005; Wu and Liu, 2005].

[70] Summertime and fall air-sea CO2 fluxes were higher in the 1990s and 2000s (i.e., 2000–2005) compared to the 1980s (i.e., 1984–1989). For example, the average air-sea CO2 fluxes were lower during the 1980s when summertime and fall NAO values had mean positive values (+0.15 and +0.12). In contrast, in the 1990s and 2000s average air-sea CO2 fluxes were higher when summertime and fall NAO values had mean negative values (Table 10). Thus the increase in summertime and fall air-sea CO2 flux described earlier (in section 6.2) presumably reflects interannual changes in atmospheric forcing between the 1980s and 1990s. Indeed, the NAO (and AO) transitioned to a period of positive states in 1989 [e.g., Curry and McCartney, 2001], cooccurring with an increase in sea surface salinity in the subtropical gyre of the North Atlantic Ocean [Hakkinen, 2001] and BATS site (Figure 1a) [Bates, 2001], presumably as a consequence of weakening of overturning circulation [e.g., Wu and Liu, 2005], and changing heat and freshwater fluxes. Since 1989, the NAO has remained generally positive, though it has trended toward a negative state, with fluctuations between strongly negative years (e.g., 1996 and 1998) and positive years (e.g., 1999, 2000, 2002). Coincident with this change is a reduction in baroclinic mass transport of the Gulf Stream [Curry and McCartney, 2001; Molinari, 2004] and freshening (particularly during the summer) of the surface layer (Figure 1a). Beginning in 1995, there has also been a stepwise increase in the number of major hurricanes occurring in the North Atlantic basin each year (0–2 pre-1995; 2–6 post-1995 [Molinari and Mestas-Nuñez, 2003; Elsner et al., 2004]), which may have also contributed to the enhancement of summertime air-sea CO2 fluxes in the midlatitudes of the North Atlantic Ocean.

7. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

[71] Two decades of monthly oceanic CO2 observations in the North Atlantic Ocean near Bermuda at Hydrostation S (32°50′N, 64°10′W; 1983–1988) and BATS (Bermuda Atlantic Time series Study; 31°43′N, 64°30′W; 1988–2005) sites are examined for long-term trends, changes in the oceanic sink of CO2, and the influence of atmospheric changes and short-term hurricane wind events. At the BATS/Hydrostation S site near Bermuda, seasonally detrended surface DIC and total alkalinity increased at a rate of +1.27 ± 0.08 μmoles kg−1 yr−1 and +0.58 ± 0.09 μmoles kg−1 yr−1, respectively, over the 1983–2003 period. The observed rate of surface ocean salinity normalized DIC (nDIC) was +0.80 ± 0.06 μmoles kg−1 yr−1 and oceanic pCO2 (+1.80 ± 0.13 μatm yr−1) was close to that expected from oceanic equilibration with increasing CO2 in the atmosphere. The ocean near Bermuda has also become more acidic, with a decrease in seawater pH of 0.0017 ± 0.0003 pH units yr−1, decrease in CO32− ion concentrations by ∼0.47 μmoles kg−1 yr−1, and saturations states of calcite and aragonite. The decline of pH by ∼0.037 pH units over the last 20 years, represents about one third of the ocean acidity increase observed since pre-industrial times [Sabine et al., 2004a]. The salinity normalized total alkalinity (nTA) remained generally constant (mean of 2389 μmoles kg−1) within a typically small range of values (∼2387−2393 μmoles kg−1), reflecting the conservative nature of total alkalinity with respect to salinity in the subtropical gyre of the North Atlantic. However, several significant short-term decreases (lower by >20 μmoles kg−1 than the mean nTA value) in nTA were observed in October 1985, February 1992, and February 2001, and attributed to calcification.

[72] During the 1984 to 2005 period, the mean net annual rate of air-sea CO2 influx was −810 ± 249 mol CO2 m−2 yr−1 (BWS), and −1182 ± 294 mol CO2 m−2 yr−1 (NNR), with an interannual range of −850 to −1200 mmol CO2 m−2 yr−1 when computed with BWS and NNR wind speed data, respectively. The peak-to-peak variability in air-sea CO2 flux observed at BATS/Hydrostation S, if extrapolated to the North Atlantic subtropical gyre, represents an interannual variability of oceanic CO2 sink of ∼0.2 Pg C yr−1. The annual air-sea CO2 fluxes were higher if ECMWF and NNR data assimilation model observations were used. The occurrence of hurricanes near Bermuda enhanced the summertime efflux of CO2 from the ocean by 3–29%. For example, during the lifetime of Hurricane Fabian as a major hurricane (6 days), the air-sea CO2 efflux may have been enhanced by ∼2 Tg C (1012 g C). The stepwise increase in the frequency of hurricanes in 1995 could potentially imparts an interannual variability in the North Atlantic Ocean CO2 efflux of ∼40 Tg C (or 0.04 Pg C yr−1).

[73] Over the 1984 to 2005 period, the annual rate of net air-sea CO2 influx slightly increased by ∼5–17%. Wintertime (JFMAM), summertime (JJAS) and fall (OND) air-sea CO2 fluxes all increased significantly associated primarily with an increase in wind speed in the North Atlantic Ocean near Bermuda. Statistical analyses indicated that summertime and fall air-sea CO2 fluxes were correlated with NAO variability. However, wintertime (JFMAM) air-sea CO2 influxes were poorly correlated with the NAO or AO, although air-sea CO2 fluxes were higher during El Niño years (i.e., negative SOI index with a 6 month lag) compared to La Niña years. The lack of strong correlations between wintertime air-sea CO2 fluxes and the NAO or AO appears to result from the wintertime anticorrelation of wind speed and ΔpCO2, the two dominant controls on gas exchange.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

[74] Anthony H. Knap, Rodney J. Johnson, Anthony F. Michaels, Dennis A. Hansell, Deborah K. Steinberg, Craig A. Carlson, Michael W. Lomas, numerous research technicians and graduate students, marine superintendent, and captains and crew of R/V Weatherbird II are thanked for their efforts and contributions to the BATS and Hydrostation S programs. Frances Howse, Julian Mitchell, and Margaret H. P. Best are thanked for their efforts with DIC and total alkalinity analyses. Marlene Jeffries is thanked for her help analyzing the synoptic and data assimilation meteorological data and air-sea CO2 fluxes. Andrew G. Dickson is thanked for his efforts providing CRM standards for DIC analysis. Kenneth M. Johnson, Frank J. Millero, James N. Butler, Ludger Mintrop, Marilyn Roberts-Lamb, Richard A. Feely, Rik Wannikhof, Chris L. Sabine, C. D. Keeling, David W. Chipman, and Taro Takahashi are thanked for their helpful advice over the years regarding the oceanic CO2 time series at BATS. Pieter Tans and Tom Conway are thanked for the use of atmospheric pCO2 data from the island of Bermuda and Terçeira Island, Açores (NOAA CMDL; http://www.cmdl.noaa.gov). Mick Follows, Eugenia Kalnay, Dimitris Menemenlis, Robert Kistler, Niki Gruber, Roger Williams, Rik Wanninkhof, Rod Johnson, and David Holland are thanked for their helpful discussions. Two anonymous reviewers are thanked for their detailed and thorough reviews of the manuscript. This research was supported by the National Science Foundation grant NSF-0326885 and NOAA.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Long-Term Trends of Seawater CO2 in the Subtropical Gyre of the North Atlantic Ocean
  6. 4. Air-Sea CO2 Fluxes in the Subtropical Gyre of the North Atlantic Ocean
  7. 5. The Influence of Hurricanes on Air-Sea CO2 Fluxes
  8. 6. Interannual Variability and Long-Term Trends in Air-Sea CO2 Fluxes
  9. 7. Conclusions
  10. Acknowledgments
  11. References
  12. Supporting Information

Auxiliary material for this article contains 21 additional figures.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

See Plugins for a list of applications and supported file formats.

Additional file information is provided in the readme.txt.

FilenameFormatSizeDescription
jgrc10431-sup-0001-readme.txtplain text document6Kreadme.txt
jgrc10431-sup-0002-fs01.tifTIFF image2725KFigure S1. Wind direction for uncorrected BWS (a), and corrected BWS (b) meteorological data sets and histogram of wind speed bins (m s−1) for 1984 (c).
jgrc10431-sup-0003-fs02.jpgimage/pjpeg356KFigure S2. Wind speed (m s−1), air-sea CO2 flux (mmoles CO2 m2 d−1) using the quadratic and cubic relationships for uncorrected BWS meteorological data sets.
jgrc10431-sup-0004-fs03.jpgimage/pjpeg361KFigure S3. Wind speed (m s−1), air-sea CO2 flux (mmoles CO2 m2 d−1) using the quadratic and cubic relationships for corrected BWS meteorological data sets.
jgrc10431-sup-0005-fs04.jpgimage/pjpeg344KFigure S4. Wind speed (m s−1), air-sea CO2 flux (mmoles CO2 m2 d−1) using the quadratic and cubic relationships for ECMWF data assimilation data sets.
jgrc10431-sup-0006-fs05.jpgimage/pjpeg371KFigure S5. Wind speed (m s−1), air-sea CO2 flux (mmoles CO2 m2 d−1) using the quadratic and cubic relationships for NNR data assimilation data sets.
jgrc10431-sup-0007-fs06.jpgimage/pjpeg205KFigure S6. Annual mean wind speed (m s−1) using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0008-fs07.jpgimage/pjpeg280KFigure S7. Seasonal mean wind speed (m s−1) using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0009-fs08.jpgimage/pjpeg279KFigure S8. Mean wind speed (m s−1) for the undersaturated seawater pCO2 period (i.e., ∼fall to spring period) and over saturated seawater pCO2 period (i.e., summertime) using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0010-fs09.jpgimage/pjpeg233KFigure S9. Histogram of wind speed bins (m s−1) for BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0011-fs10.jpgimage/pjpeg337KFigure S10. Histogram of wind speed bins (m s−1) for 1984–1987 using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0012-fs11.jpgimage/pjpeg336KFigure S11. Histogram of wind speed bins (m s−1) for 1988–1991 using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0013-fs12.jpgimage/pjpeg337KFigure S12. Histogram of wind speed bins (m s−1) for 1992–1995 using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0014-fs13.jpgimage/pjpeg335KFigure S13. Histogram of wind speed bins (m s−1) for 1996–1999 using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0015-fs14.jpgimage/pjpeg333KFigure S14. Histogram of wind speed bins (m s−1) for 2000-2003 using BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0016-fs15.jpgimage/pjpeg360KFigure S15. Comparison of air-sea CO2 flux (mmoles CO2 m2 d−1) computed using the quadratic wind speed–gas transfer relationship and using corrected (and uncorrected) BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0017-fs16.jpgimage/pjpeg354KFigure S16. Comparison of air-sea CO2 flux (mmoles CO2 m2 d−1) computed using the cubic wind speed–gas transfer relationship and using corrected (and uncorrected) BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0018-fs17.jpgimage/pjpeg623KFigure S17. Comparison of air-sea CO2 flux (mmoles CO2 m2 d−1) computed using the cubic wind speed–gas transfer relationship and using corrected (and uncorrected) BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0019-fs18.jpgimage/pjpeg318KFigure S18. Sea-to-air CO2 flux (mmoles CO2 m2 d−1) during Hurricane Emily (1987) computed using the quadratic transfer relationship and using corrected BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0020-fs19.jpgimage/pjpeg415KFigure S19. Sea-to-air CO2 flux (mmoles CO2 m2 d−1) during Hurricane Bertha (1990) computed using the quadratic transfer relationship and using corrected BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0021-fs20.jpgimage/pjpeg381KFigure S20. Sea-to-air CO2 flux (mmoles CO2 m2 d−1) during Hurricane Felix (1995) computed using the quadratic transfer relationship and using corrected BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0022-fs21.jpgimage/pjpeg378KFigure S21. Sea-to-air CO2 flux (mmoles CO2 m2 d−1) during Hurricane Fabian (2003) computed using the quadratic transfer relationship and using corrected BWS (BDA), ECMWF and NCEP-NCAR (NNR) meteorological data sets.
jgrc10431-sup-0023-t01.txtplain text document2KTab-delimited Table 1.
jgrc10431-sup-0024-t02.txtplain text document1KTab-delimited Table 2.
jgrc10431-sup-0025-t03.txtplain text document3KTab-delimited Table 3.
jgrc10431-sup-0026-t04a.txtplain text document3KTab-delimited Table 4a.
jgrc10431-sup-0027-t04b.txtplain text document1KTab-delimited Table 4b.
jgrc10431-sup-0028-t05a.txtplain text document3KTab-delimited Table 5a.
jgrc10431-sup-0029-t05b.txtplain text document1KTab-delimited Table 5b.
jgrc10431-sup-0030-t06a.txtplain text document2KTab-delimited Table 6a.
jgrc10431-sup-0031-t06b.txtplain text document1KTab-delimited Table 6b.
jgrc10431-sup-0032-t07a.txtplain text document3KTab-delimited Table 7a.
jgrc10431-sup-0033-t07b.txtplain text document1KTab-delimited Table 7b.
jgrc10431-sup-0035-t09.txtplain text document1KTab-delimited Table 9.
jgrc10431-sup-0036-t10.txtplain text document1KTab-delimited Table 10.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.