Upwelling velocity and ventilation in the Mauritanian upwelling system estimated by CFC-12 and SF6 observations

Journal of Marine Systems 151 (2015) 57–70

Contents lists available at ScienceDirect

Journal of Marine Systems journal homepage: www.elsevier.com/locate/jmarsys

Upwelling velocity and ventilation in the Mauritanian upwelling system estimated by CFC-12 and SF6 observations Toste Tanhua ⁎, Mian Liu 1 GEOMAR Helmholtz Centre for Ocean Research Kiel, Düsternbrooker Weg 20, 241 05 Kiel, Germany

a r t i c l e

i n f o

Article history: Received 5 February 2015 Received in revised form 8 July 2015 Accepted 10 July 2015 Available online 17 July 2015 Keywords: Mauritanian coast Atlantic Upwelling Tracers Ventilation Hydrography

a b s t r a c t Transient tracer data (CFC-12 and SF6) from three oceanographic field campaigns to the Mauritanian Upwelling area conducted during winter, spring and summer from 2005 to 2007 is presented. The transient tracers are used to constrain a possible solution to the transient time distribution (TTD) along 18°N and to quantify the mean ages in vertical sections perpendicular to the coast. We found that an Inverse Gaussian distribution where the ratio of the moments Δ and Γ equals 1.2 is a possible solution (Δ/Γ = 1.2) of the TTD. The transient tracers further show considerable under-saturation in the mixed layer during the winter and spring cruises that can only be maintained by mixing or upwelling by tracer-poor water from below the mixed layer. We use dissipation data from microstructure measurements and the tracer depth distribution to quantify the flux of tracers to the mixed layer by vertical diffusivity and wind data from the ship to quantify the air–sea flux. We then use the magnitude of the under-saturation in the mixed layer to estimate the advective upwelling velocity which is the balance the first two processes, in a steady state assumption. We find that the upwelling velocities range from less than 1 to 5.6 × 10-5 m s-1 (b 0.8–4.8 m d-1), with generally higher values close to the coast, but with comparable upwelling velocities during spring and winter. During the summer cruise the transient tracers were close to equilibrium with the atmosphere, suggesting no upwelling. We have shown the use of CFC-12 and SF6 transient tracer data for calculating upwelling velocity, and found an overall uncertainty of roughly ±50%. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Coastal upwelling along the eastern boundaries of the world ocean is an important process that brings cold, low-oxygen and high-nutrient sub-surface seawater to the surface ocean, where it fuels some of the most productive areas in the world ocean. It is driven by alongshore winds causing an offshore Ekman transport within the surface layer of the ocean. The Mauritanian upwelling system is an Eastern Boundary Upwelling System (EBUS) that stretches from the Iberian Peninsula to about 10°N along the Northwest African coast. Due to changes in wind forcing associated with the migration of the Intertropical Convergence Zone, coastal upwelling off Mauritania exhibits a pronounced seasonal cycle. Winds favorable to upwelling prevail primarily from December to April (e.g. Barton et al., 1998; Pelegri et al., 2005; Schafstall, 2010; Schafstall et al., 2010). The Mauritanian upwelling system is the most productive branch of the Canary Current upwelling system (e.g. Cropper et al., 2014), and the extension cold surface waters can be observed by sea surface temperature observations from, for instance, satellite (e.g. Pelegri et al., 2005).

⁎ Corresponding author. Tel.: +49 431 6004219. E-mail addresses: [email protected] (T. Tanhua), [email protected] (M. Liu). 1 Tel.: +49 431 6004203.

http://dx.doi.org/10.1016/j.jmarsys.2015.07.002 0924-7963/© 2015 Elsevier B.V. All rights reserved.

Upwelling also brings climate relevant trace gases, such as N2O and CO2, to the surface where they can outgas to the atmosphere. For instance, Rees et al. (2011) estimate an annual flux of N2O to the atmosphere by following upwelling filaments with a deliberately released tracer (SF6) and without separating supply to the mixed layer by vertical advection from vertical diapycnal flux of 1.3 to 2.1 Gg N per filament. Similarly, Kock et al. (2012) calculate the diapycnal flux of N2O for the Mauritanian upwelling area to 0.019 (0.007 to 0.048) nmol m- 2 s- 1 and a corresponding average air–sea flux of the same magnitude (0.020 nmol m-2 s-1), although they estimate the vertical advection to be an order of magnitude lower. This would correspond to an annual N2O air–sea flux of ~ 90 Gt N over an area of 104 km2, roughly corresponding to the Mauritanian upwelling area. Steinhoff (2010) report on high concentrations of CO2 from the Mauritanian upwelling region during the upwelling season, and as a result high fluxes of CO2 into the atmosphere. During the weak upwelling season (summer), the partial pressure of CO2 is close equilibration with the atmosphere. This is a result of the combined effect of physical gas exchange of air–sea interface, biological activity and upwelling intensity. A study by Loucaides et al. (2012) from the same experiment as Rees et al. (2011) explores the effect of high CO2 concentration waters on the calcium carbonate saturation state in the upwelling filaments. Understanding of the intensity and duration of upwelling, as well as the circulation and ventilation of the area is thus of importance for

58

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

regional climate, primary production, fisheries, and for the flux of trace gases to the atmosphere. This is often achieved with measurements of sea surface temperature as a proxy for upwelling (e.g. Benazzouz et al., 2014), or surface ocean measurements of, for instance, CO2. Direct measurements of the upwelling velocity, however, are virtually impossible due to the low velocities; in the order of 10- 5 m s- 1. Indirect methods of estimating upwelling velocities must be used. Upwelling velocities have been estimated from arrays of current moorings in the tropical Atlantic and Pacific Oceans with values in the range of 10-5 m s-1, although with large uncertainty (e.g. Weingartner and Weisberg, 1991; Gouriou and Reverdin, 1992; Weisberg and Qiao, 2000). The Mauritanian upwelling system is carefully examined in a recent study by Cropper et al. (2014) using upwelling indices based on seas surface temperature of wind speed/direction characterizing the integrated effect of diapycnal mixing and advective upwelling. Schafstall (2010) calculated positive Ekman transports for the Mauritanian coast up to 10-5 m s-1 during the January to June period based on climatological winds. In a different approach Klein and Rhein (2004) laid out the principles for using 3He in the upper ocean for estimates of upwelling velocities, a concept that was later applied to a larger data-set of 3He measurements in the equatorial Atlantic where an upwelling velocity in the order of 10-5 m s-1 could be quantified with an uncertainty of 42% (Rhein et al., 2010). Helium surface disequilibrium measurements have also been used for determining upwelling velocities off Mauritania and Peru (Steinfeldt et al., 2014). An alternative method using the short lived 7 Be isotope found comparable upwelling velocities (calculated as the combination of diapycnal flux and advective transport) in the tropical Atlantic and Pacific oceans (Kadko and Johns, 2011; Haskell Ii et al., 2015). Here we report on transient tracer (SF6 and CFC-12) and oxygen data from the Mauritanian upwelling area collected during three field campaigns from 2005 to 2007. The data are used for characterization of the ventilation of the area and for inferring upwelling velocities. An introduction to the general oceanography of the area and specifics about these field campaigns can be found in Schafstall (2010) and Schafstall et al. (2010). 2. Data and methods The data presented here were collected during three cruises conducted between 2005 and 2007 on the German research vessels Poseidon and Meteor, see Table 1 for details, and Fig. 1 for a graphic representation of the station network. These cruises were part of the field program for the German BMBF-funded project SOPRAN (Surface Ocean Processes in the Anthropocene) and the German DFG-funded project MUMP (Mauritanian upwelling and mixing process study). The cruises were carried out during different times of the year, and thus, potentially survey the area during variable strength of upwelling. We use stations sampled for transient tracers on short zonal sections along 18°N or 17°30′N. For more details on the cruises, please see references in Table 1. Seabird CTD systems were used to measure the conductivity, temperature and pressure. The concentration of oxygen was measured by the Winkler titration during the cruises, for more details see the cruise reports (Table 1). The continuous under-way measurement of salinity and temperature (SST and SSS) were recorded by ship-based thermosalinographs (TSG).

2.1. Transient tracers During the cruises samples for the determination of SF6 and CFC-12 were taken for later analysis in the lab in Kiel. All transient tracer samples were taken in 300 ml glass ampoules that were flame sealed under a flow of tracer free nitrogen gas in a process similar to that described by Vollmer and Weiss (2002). The ampoules were stored for several years in the lab prior to analysis that were performed mainly during 2011. The integrity of the samples is conserved since the tracers are stable in seawater, and a correctly performed flame-seal is gas tight so that no exchange with the ambient atmosphere can take place. Incorrectly flame sealed ampoules can mostly be identified by abnormally high tracer concentrations, particularly for SF6—the least soluble of the two tracers. CFC-12 and SF6 have been shown to be stable in seawater, also at low oxygen concentrations, so we do not expect any degradation of these tracers in the ampoules during storage (e.g. Lee et al., 1999; Holtermann et al., 2012). Other potentially useful halogenated transient tracers that could have been used here include CFC-11, CFC-113 and CCl4. These were not measured for mainly two reasons: 1) CFC-11 data provide only limited additional information to that of CFC-12 when it comes to mean ages (similar input function), although the higher solubility of CFC-11 compared to that of CFC-12 could potentially have provided additional information, and 2) CFC-113 and CCl4 are known to be unstable in low oxygen environments (e.g. Tanhua et al., 1996; Tanhua and Olsson, 2005) and in warm sea-water (e.g. Huhn et al., 2001; Roether et al., 2001), introducing additional uncertainties in the calculations. The procedure to measure the samples in the lab is based on purgeand-trap followed by chromatographic separation and detection on an electron capture detector (ECD), with the process similar to the one described by Vollmer and Weiss (2002). The ampoules were placed in an ampoule cracking device—the cracker—that allows for quantitative measurements of all the tracer in the ampoules including the headspace (Vollmer and Weiss, 2002). This is necessary since most of the tracer will be in the headspace of the ampoule due to the low solubility of the tracer. The purge efficiency is thus mostly an effect of how efficiently the headspace and cracker volumes are flushed with the purge gas. Each sample was purged at least twice for 10 min each with N2-gas at ~ 80 ml min- 1; with this method the purge efficiency was close to 100%. A detailed description of the analytical methods and techniques used for the calibration is described in Schneider (2011), no correction for sampling blanks was made. The samples from the 2005 cruise were measured on a capillary gaschromatographic system with a dual-trap injection procedure. In this system the tracers are trapped at − 100 °C on a 1 m, 1/16″ stainless steel trap filled with Haysep D. This back-flushed during desorption onto a precolumn (50 cm Porasil C and 50 cm molecular sieve 5A in a 1/8″ steel tubing at 35 °C) and to the second trap, which was a 1 m 1/32″ stainless steel tubing where 5 cm is filled with Carboxen 1000 kept at −70 °C. This trap is desorbed onto the main column (75 m DB 624 followed by 30 m RT-molecular sieve 5A). This system proved to be sensitive and somewhat fragile, so the samples from the 2006 and 2007 cruises were measured on a packed column gaschromatographic system with a single trap (70 cm Haysep D in 1/16″ SS tubing kept at -70 °C). After desorption at 130 °C the sample was separated on a 30 cm 1/8″ Porasil C column and the SF6 and CFC-12 fraction was passed on to the main column (180 cm Carbograph and 20 cm Molecular Sieve 5A tail-end). A thorough description of the analytical

Table 1 Chief scientists, cruise report reference and data repository for the three oceanographic cruises to the Mauritanian Upwelling region considered in this study. Cruise

Chief scientist

Time

Cruise reports

Data repository

Poseidon 320 Meteor 68/3 Poseidon 347

H. Bange A. Körtzinger M. Dengler

March 21 – April 7, 2005 July 12 – August 6, 2006 January 18 – February 5, 2007

(Bange, 2008) (Körtzinger, 2009) (Dengler et al., 2008)

http://doi.pangaea.de/10.1594/PANGAEA.817256 http://cdiac.ornl.gov/ftp/oceans/CLIVAR/Met_68_3.data/ http://doi.pangaea.de/10.1594/PANGAEA.833885

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

59

a

bo

Surface Saturation of SF6

Surface Saturation of CFC−12

1.15

19 N 40’

1.1

20’

18oN 1.05

40’

2005

20’

2005

o

1

17 N 19oN

0.95

40’ 20’ o

0.9

18 N 40’

2006

20’

2006 0.85

17oN 0.8

o

19 N 40’ 20’

0.75

o

18 N 40’

2007

20’

2007

0.7

o

17 N

21oW

20oW

19oW

18oW

17oW

16oW

15oW

o

21 W

o

20 W

o

19 W

o

18 W

o

17 W

o

16 W

o

15 W

0.65

Fig. 1. Panel a: Maps showing the location of the working area. Panel b: Circles denote position of CTD stations carried out during the three cruises discussed in this paper. The stations with transient tracer samples are color-coded based on the near-surface saturation of CFC-12 (left column) and SF6 (right column).

60

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

techniques can be found in Schneider (2011) and Stöven (2011). Unfortunately no samples were run on both systems for comparative studies of the two systems. No double samples were taken during these cruises, so we do not have any direct data for estimating the precision of the analysis. However, based on measurements from other campaigns the precision is typically up to 5% for SF6 and 2% for CFC-12 (e.g. Stöven, 2011) for measurements from ampoules. The SF6 data from 2005 that were measured with the capillary system have a higher, but not quantified, uncertainty.

advective to diffusive transport mechanisms, where a Δ/Γ ratio of 0 would indicate purely advective flow. In large parts of the ocean a Δ/Γ ratio of 1.0 seems appropriate (Waugh et al., 2004) and particularly for the eastern tropical north Atlantic Schneider et al. (2012) found a Δ/Γ ratio of 1.0. The ratio can be empirically determined by applying the TTD concept on measured tracer concentrations from tracers with sufficiently different input function.

2.2. Microstructure measurements

The mixed layer concentration of the volatile transient tracers SF6 and CFC-12 is determined by the balance of the flux with the atmosphere by air–sea exchange, the diapycnal flux from below and the upwelling of water from below the mixed layer. The air–sea flux alone will tend to drive the mixed layer concentration to be in equilibrium with the atmospheric concentration (that is assumed to vary only on weekly time-scales), although rapid cooling/heating of the mixed layer can lead to deviations from saturation. Any deviation from saturation of the tracers in the mixed layer is interpreted as a result of upward flux of water from below the mixed layer. The air–sea exchange and diapycnal flux can be calculated based on measurable quantities such the diapycnal mixing coefficient, concentration gradients below the mixed layer and wind-speed, leaving the flux from upwelling as the balance. In Fig. 2 we show a schematic picture of processes determining the concentration of inert trace gases, such as CFC-12 and SF6, in the mixed layer. Fluxes of tracer to the mixed layer through air–sea gas exchange, diapycnal mixing and upwelling velocity must balance, assuming steady state conditions (i.e. neglecting rapid cooling/heating and fast advective processes):

Microstructure measurements by a tethered Micro Structure Sonde (MSS) to measure local shear for the calculation of turbulent mixing processes were used during all three cruises. Three different, but similar, MMS profiling systems, all with 16 channels and data transmission rate of 1024 Hz, were used during the cruises (Schafstall et al., 2010). Temperature, conductivity, pressure and oxygen gradients in the micro-scale (0.6 mm) were detected by the sensors and turbulent fluctuations can be inferred with an accuracy of 0.6 m s- 1. The MSS data has been used to calculate the local diapycnal diffusivity for the tracer budget. At all the MSS stations, 3 to 5 MSS profiles were measured between the sea surface and a depth of 150–250 m, or a few meters above the bottom in the shallow region. Additionally, two (in year 2006) and three (in year 2007) long-duration stations near the continental slope were sampled as reference stations. 2.3. Wind speed In this study the 24 and 48 hour average wind speeds of both ship recorder and satellite data from CERAST (http://products.cersat.fr/ details/?id=CER-L2A-WND-GLO-012-L2A-QS) were considered and compared. The ship data is continuously recorded providing high temporal resolution while the satellite data are only available as mean values every 6 h. While the wind recording from the ship was not collocated with the sampling position for the full 48 h, the satellite data was, albeit with a low spatial and temporal resolution (0.5 × 0.5°), see discussion in section 3.4.1. 2.4. Transit time distribution The transient tracer data can be used to constrain ventilation time scales using the well-known atmospheric histories (e.g. Walker et al., 2000; Bullister, 2011) and solubility functions for seawater (Warner and Weiss, 1985; Bullister et al., 2002). We have assumed 100% saturation of the tracers at the time of water mass formation, although this probably over-estimates the mean ages slightly since late winter surface water is often under-saturated with regard to CFCs and SF6 (e.g. Shao et al., 2013). Ventilation ages were calculated using the transit time distribution (TTD) concept (e.g. Waugh et al., 2003) where the interior tracer concentration c(t) is given by: Z cðt Þ ¼

0

∞

c0 ðt−t 0 Þ  Gðt 0 Þdt 0

ð1Þ

where c0(t − t′) is surface water tracer concentration. For steady transport an inverse Gaussian function is a possible solution for the TTD, as is commonly applied to oceanographic data. rffiffiffiffiffiffiffiffiffiffiffiffiffiffi Gðt 0 Þ ¼

Γ3

4πΔ2 t 0 3

2 −Γ ðt 0 −Γ Þ 4Δ2 t 0

e

ð2Þ

where Δ is the “width” of the TTD and Γ is the mean age. The two moments Γ (mean age) and Δ (width) define the inverse Gaussian TTD and the ratio of these two moments (Δ/Γ) indicates the ratio of

2.5. Calculation of upwelling velocity from transient tracer data

Fg þ Fd þ Fu ¼ 0

ð3Þ

where the first term (Fg) is the air–sea gas exchange, the second term (Fd) is the diapycnal mixing flux and the third term (Fu) is the flux from upwelling. In the following we have formulated Eq. (3) so that positive values indicate flux of tracer into the mixed layer. The air–sea gas exchange between the mixed layer and the atmosphere is defined as F g ¼ vg ΔC

ð4Þ

where vg is the gas transfer velocity, and ΔC the concentration difference between the mixed layer and the atmospheric concentrations ΔC ¼ C eq −C 1

ð5Þ

where C1 is the concentration in the mixed layer (calculated as the average of all tracer measurements within the mixed layer on a station per station basis) and Ceq is the equilibrium concentration in the sea water calculated from the known atmospheric concentration, salinity and temperature. The gas transfer velocity vg is a function of wind speed and the Schmidt number (e.g. Ho et al., 2011).  vg ¼ k600 

 Sc −0:5 600

ð6Þ

where Sc is the Schmidt number of CFC-12, which is a function of temperature (King and Saltzman, 1995) and k600 is the gas transfer velocity when the Schmidt number is 600, as for CO2 (Ho et al., 2011). k600 is calculated from wind-speed 10 m above the sea surface (U10) based on the formulation in (Ho et al., 2011), Eq. (7). k600 ¼ ð0:262  0:022Þ  U 10 2

ð7Þ

The second term in Eq. (3) shows the flux due to diapycnal mixing between the mixed layer (ML) and the thermocline (Oakey, 1982) where the flux depends on the mixing coefficient (Kz) and the gradient

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

61

1980), where Γ, the mixing efficiency, is set to be 0.2 (e.g. Oakey, 1982), and N2 is the buoyancy frequency. Kz ¼ Γ

ð9Þ

Both the local buoyancy frequency N2 and the turbulent kinetic energy ε are based on the measurements carried out during the cruises. For the year 2005, the values of Kz had to be interpolated from nearby MSS stations for two tracer stations that did not have MSS measurements. Finally, the third term in Eq. (3), the flux due to upwelling is calculated as the residual in Eq. (3), and the upwelling velocity ω can be calculated if the concentrations in and under the mixed layer (C1 and C2) are known.

MLD

MLD + 20m

Fig. 2. Schematic of processes determining flux of properties between the interior ocean and the atmosphere to the mixed layer in a coastal upwelling region, see main text for abbreviations.

of the tracer concentration (dC/dz). The gradient is calculated from the base of the mixed layer (using the average concentration in the mixed layer as starting point) linearly interpolated through all measurements b100 m below the mixed layer on each station, forcing the fit through the MLD concentration. The definition of ML depth (MLD) is based on the threshold method of potential density since the gradient in potential densities is the important unit for diapycnal mixing. For this study a threshold value (ΔσΘ) for air–sea interaction studies of 0.10 kg m-3 was chosen due to the strong gradients in the area (Thomson and Fine, 2003); lower threshold values lead to unrealistic shallow MLD for several stations. Fd ¼ Kz

ε N2

C 2 ¼ C 1 −20 

dC dz

ð10Þ

The concentration gradient of the tracer below the mixed layer is calculated from the base of the mixed layer (using the average concentration in the mixed layer) as starting point and linearly interpolated through all measurements b 100 m below the mixed layer on each station, forcing the fit through the MLD concentration at the base of the mixed layer. The concentration C2 is strongly dependent on how C2 is defined; in this study we use the CFC-12 concentration 20 m below the base of the mixed layer, see Fig. 2 and discussion in section 3.4.1. We therefore label the upwelling velocity as ω20 in the following. F u ¼ ω20  ðC 2 −C 1 Þ

ð11Þ

3. Results

dC dz

ð8Þ

The local diapycnal mixing coefficient Kz is calculated from the dissipation (ε) as determined from microstructure measurements (Osborn,

In this section we start with a short general description of the oceanography in the Mauritanian upwelling area along the 18°N sections observed during the three field campaigns from 2005 to 2007. This provides background for the discussion of transient tracer distribution

T−S distribution of Water Masses in Summer, 2006 30

4000

3500

Potential Temperature (°C)

σΘ = 23.5 3000 20 SACW

2500 NACW 2000

1500 10 1000 AAIW

NADW

σΘ = 27.3

σΘ = 27.5 0 34.5

500

AABW

σΘ = 27.88 35

35.5

36

Salinity

36.5

37

37.5

0

pressure

Fig. 3. Salinity – temperature diagram of all data along 18°N from the three cruises discussed here. Major water masses are distinguished by their densities (σθ); NADW—North Atlantic Deep Water, AABW—Antarctic Bottom Water, AAIW—Antarctic Intermediate Water, SACW—South Atlantic Central Water, and NACW—North Atlantic Central Water.

62

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

and the derived ventilation ages. In Section 3.3 we use the observed transient tracer distribution, along with wind-speed data and measurements of turbulent diffusivity to calculate upwelling velocities. 3.1. Hydrography and oxygen The Mauritanian Upwelling Region is dominated by the presence of 5 water masses that can be easily identified by their salinity and temperature characteristics (e.g. Stramma et al., 2005) (Figs. 3 and 4). Below the surface mixed layer, Central Water is dominating the water column. As indicated by two linear temperature–salinity relationships in the θ/S diagram (Fig. 3) two different Central Waters can be found in the study region: North Atlantic Central Water (NACW) and South Atlantic Central Water (SACW). The study region is located in the Cape Verde Frontal Zone that separates the NACW from the SACW, so

that both variations can be found in the area. The SACW has lower oxygen concentration and higher nutrient concentration compared to the NACW due to a longer pathway from its formation regions in the South Atlantic. The tracer data from these cruises are mostly in the mixing zone between the SACW and NACW and no significant differences in age can be detected from this data set. Below the central water the Antarctic Intermediate Water (AAIW) can be easily recognized as low salinity layer that is situated at about 700 m depth. The water below the AAIW layer is dominated by the overflow component of the NADW with some influence of AABW close to the bottom, although this water mass is mainly found west of 21°W, so that we do not find AABW on the sections in Figs. 4, 5 and 7. We here choose to show sections from the M68/3 cruise for displaying the general oceanographic settings since it covers a larger area; the section with transient tracer measurements extends to 36.4

a

36.3

0

36 35.8

36

35.8

35.6

36.2

35.6

36.1

35.4

35.

35.4

4

36 35.9 35.8

35.2

500

35.7 35.6

35

35.5

5

34.95

34.92

35.4 35.3

1000

35.2

35

34.95

35

Pressure (db)

35

35

2000 34.95

5

.9 34 Salinity

3000

34.92

Summer, 2006 4000 21

20.5

20

19.5

19

18.5

18

17.5

17

35.1 35 34.995 34.99 34.985 34.98 34.975 34.97 34.965 34.96 34.955 34.95 34.945 34.94 34.935 34.93 34.925 34.92 34.915 34.91 34.905 34.9 34.895 34.89

1500

16.5

Longitude (°W)

b

26

0

15

15 22

12

12

10

500

10 18

8 8

14

1000

6

6

6

4

4

2000 3

3

Pressure (db)

1500

Temperature (°C)

3000

3

Summer, 2006 20.5

20

19.5

19

18.5

18

17.5

3

4000 21

17

10 9.5 9 8.5 8 7.5 7 6.5 6 5.5 5 4.5 4 3.5 3 2.5 2

16.5

Longitude (°W) Fig. 4. Salinity (upper panel) and potential temperature (lower panel) sections along 18°N occupied during the M68/3 cruise in July 2006. The white lines show the density defined boundaries of different water masses as defined in Fig. 3, black vertical lines locate the CTD stations. Note the non-linear color scale. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

63 240

0

180

60

45

220

60

60

200

60

45

45

60

180

45

500

160

60 140

120

1000

100

180

80 78 76 74 72 70 68 66 64 62 60 58 56 54 52 50 48 46 44 42 40

2000

Oxygen (μmol/kg)

23

0

230

Pressure (db)

1500

3000

Summer, 2006 4000 21

20.5

20

19.5

19

18.5

18

17.5

17

16.5

Longitude (°W) Fig. 5. Oxygen section along 18°N occupied during the M68/3 cruise in July 2006. The white lines show the density defined boundaries of different water masses as defined in Fig. 3, black vertical lines locate the CTD stations. Note the non-linear color scale. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

measure of upwelling intensity. The figure clearly indicates that the near-coast SST is several degrees colder than the offshore area during the 2005 and 2007 cruises, indicating active upwelling. The situation is different for the 2006 cruise that was conducted during summer, i.e. during the non-upwelling season; here the SST is generally higher in the near-coast area compared to further offshore, indicating that no upwelling takes place. We will discuss the upwelling strength based on measurements of transient tracers in Section 3.3.

21°W. Fig. 4a shows the salinity section from the shelf-break to about 21°W. The section for oxygen concentration (Fig. 5) clearly shows the low oxygen values in the upper water column, with a deep minimum at around 500 m depth and a shallow minimum around 100 m depth. The lowest oxygen concentration in the deep oxygen minimum is around 40 μmol kg-1, whereas in the upper oxygen minima concentrations lower than 20 μmol kg-1 is found. There is a very sharp gradient in oxygen concentration below the mixed layer, so that upwelling has the potential to transport low oxygen water into the mixed layer. In the following we will focus on the upper few hundred meters of the water column that is more directly involved in the upwelling along the coast. A commonly used, and easy to measure, proxy of upwelling is the difference in sea-surface temperature (SST) between the coastal region and the off-shore region. SST is regularly measured with satellite although cloudy conditions, typical for upwelling conditions, tend to reduce the coverage. Here we use the SST as determined from the ship's thermosalinograph (TSG) during the cruises (Fig. 6) to get a qualitative

3.2. Transient tracers The CFC-12 section (Fig. 7) shows the relatively well ventilated upper water column and decreasing concentrations with depth until at about 500 m where the concentrations are lower than 0.1 pmol kg-1, only towards the bottom an increase in CFC-12 can be seen again, influenced by Antarctic Bottom Water. We now use the concept of saturation rather than the concentration since this unit does not only compensate for the various salinities and temperatures (that influence the solubility of CFC-12 and SF6), but also provides a simple means to compare

o

20oN

20oN

o

19oN

19oN

N

N

18oN

o

N

Spring 2005

o

N o

21 W

16

o

20 W

o

19 W

o

18 W

o

17 W

18

o

16 W

o

15 W

o

17 N o

14 W

o

21 W

20

18oN

Summer 2006 o

20 W

o

19 W

o

18 W

22

o

17 W

o

16 W

o

15 W

Winter 2007

o

17 N o

14 W

24

21oW

20oW

19oW

18oW

26

17oW

16oW

15oW

14oW

28

Fig. 6. Sea surface temperature (SST, °C) as observed by the shipboard thermosalinograph during the three cruises to the Mauritanian coast. Note that the temperature is lower along the coast than in the offshore region during the winter and spring cruises (indication of upwelling), but not for the summer cruise.

64

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70 1.2

0

1.1

1

500

1

1

0.7

0.9

0.5

0.8

0.3

0.7 0.6

0.15

0.15 0.5

0.07

0.07

0.4 0.3

1000

0.2

Pressure (db)

1500

0.1 0.095 0.09 0.085 0.08 0.075 0.07 0.065 0.06 0.055 0.05 0.045 0.04 0.035 0.03 0.025 0.02 0.015 0.01

0.03 0.03

2000

0.07

CFC−12 (pmol/kg)

3000

Summer, 2006 4000 21

20.5

20

19.5

19

18.5

18

17.5

17

16.5

Longitude (°W) Fig. 7. CFC-12 section along 18°N occupied during the M68/3 cruise in July 2006. Note the non-linear color scale. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

transient tracer measurements from different years. Focusing on the upper water column saturation levels it can be seen that the saturation in the mixed layer is frequently lower than 100%. Fig. 1 shows the surface saturation values which, particularly for the cruises in 2005 and 2007 in the area close to the shelf, show undersaturation, possibly reflecting the strength of upwelling. Also the vertical sections of CFC12 and SF6 saturations show that the under-saturated water reaches close to the surface close to the coast during the 2005 and 2007 cruises that were both conducted during upwelling season (Figs. 8 and 9). The degree of under-saturation is slightly larger for SF6 compared to CFC-12, possibly reflecting the continuous increase of SF6 in the atmosphere as opposed to the CFC-12 concentration that has remained essentially steady a decade and half prior to our sampling, leading to subsurface waters with an age of up to a couple of decades being close to saturated with respect to the surface, see discussion in Tanhua et al. (2008) and Shao et al. (2013).

450 ppt, and SF6 measurements for the other data-points (Fig. 11). The reason for this is to use the continuous increase of SF6 during recent years for recently ventilated waters, but to utilize the better precision of the CFC-12 measurements in older water where the SF6 concentration becomes small (Tanhua et al., 2008). The mean age increases rapidly with depth, a situation typical for the oxygen minimum zone of the eastern tropical Atlantic (e.g. Schneider et al., 2012), and is very different from the situation at a similar latitude in the western Atlantic (e.g. Stanley et al., 2012). We typically find nonzero mean-ages in the mixed layer (for the winter and spring cruises) which is a reflection of upwelling bringing water from the interior, with higher age, toward the surface. Older water are typically enriched in nutrients and CO2, and depleted in oxygen influencing the mixed layer properties in the upwelling area. There is a distinct difference in Saturation of CFC−12 0

0.6 0.5

0.6 0.5

100

0.5

0.9

200 2005

Pressure (db)

Using the SF6 and CFC-12 data along 18°N we now quantify the transit time distribution, or age-spectra, of the upper water column. As a first step we use the SF6 and CFC-12 data from the upper 300 m of the water column (below 300 m depth the SF6 concentration is small and the relative uncertainties in the SF6 determination are large) to constrain the shape of the TTD, assuming an inverse Gaussian form. The tracer pair cannot resolve Δ/Γ ratios much higher than about unity well, so that a higher Δ/Γ ratio could be conceivable. In Fig. 10 we plot the mean age as calculated from the CFC-12 data vs. those calculated with SF6 data for different choices of the Δ/Γ ratio. The scatter in the data around the 1:1 line can reflect: 1) differences in saturation in the tracer data during water formation, 2) deviations from an inverse Gaussian shape of the TTD, or 3) uncertainties in the measurement of the tracers. Based on these figures we found that the TTD can be represented by an Inverse Gaussian distribution with a Δ/Γ ratio of 1.2, although with some uncertainty. We apply the TTD concept to the CFC-12 and SF6 data to calculate the mean-age of the upper 300 m of the water column, using the mean age calculated from CFC-12 if the partial pressure of CFC-12 is less than

1

0.8 0.8

3.3. Ventilation ages

300 0

0.3

0.8

1

0.8

0.7

0.6

100

0.5

200 2006

300 0 0.8

100

0.6

200

0.8

0

0.6

0.5

0.4

0.5

2007

300 19

18.5

18

17.5

17

16.5

0.3

Longitude (°W) Fig. 8. Sections of CFC-12 saturation for the upper 300 m of the water column along 18°N (17°30′N for the 2005 cruise).

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

Saturation of SF 0

0.6 0.5 0.3

0.3

Pressure (db)

60

200

300 0

0.8 0.6 0.5

100

0.6

0.3

200

0.5

2006

300 0

0.6 0.5

100

0.3

200 18.5

18

17.5

Longitude (°W)

17

300 0 100

50

300 19

18.5

18

mean age based on SF6

Δ / Γ = 1.2

70

70

60

60

50

50 2005 2006 2007

90

2005 2006 2007

40 30 30

40

50

Δ / Γ = 1.4 90

80

80

70

70

60

60

50

50 2005 2006 2007

40 60

70

60

70

80

90

Δ / Γ = 1.6

90

50

16.5

80

20

place during summer, i.e. not during the upwelling season (see SST for instance, Fig. 6). Due to the higher analytical uncertainty of the SF6 measurements, particularly below the mixed layer, we will focus our analysis on the CFC-12 data in the following analysis, but still report on the SF6-based upwelling velocities for comparison. We estimate the diapycnal flux of tracer based on the diapycnal mixing coefficient calculated from dissipation rates obtained from MSS measurements and the gradient of the tracer below the mixed layer. The air–sea exchange is calculated based on the mixed layer tracer concentration, the known atmospheric concentration and the air–sea exchange coefficient calculated from the solubility of the tracer and the wind-speed. The residual

80

40

17

Fig. 11. Sections of mean age for three years for the upper 300 m of the water column along 18°N (17°30′N for the 2005 cruise).

80

30 30

17.5

Longitude (°W)

90

80

40 30

80

Δ / Γ = 1.0

70

40

2007

90

40

60

20

60

The upwelling velocity could only be determined for the 2005 and 2007 cruises since for the 2006 cruise we found CFC-12 and SF6 values either close to saturation or slightly over-saturated, indicating no active upwelling (Fig. 1). This is not surprising since the 2006 cruise did take

60

2006

100

200

3.4. Upwelling velocity

50

70

80

0.2

mean age at 300 m depth between the three field campaigns, reflecting temporal variability, possibly with some seasonality although this cannot be determined from these three campaigns alone.

40

60

200

40

Fig. 9. Sections of SF6 saturation for the upper 300 m of the water column along 18°N (17°30′N for the 2005 cruise).

30 30

90 80

100

0.1

16.5

2005

20 40

0.3

2007

0.3

100

80

100

20

0.6 0.5

0.8

80 100

300 0

0.4 0.8

110

60

12

0.7

1

40 60

40

100

0.8

2005

120

20

20

0.9

200

300 19

0

1

Pressure (db)

100

Mean age [year]

6

0.8

0.6 0.5

0.8

65

90

2005 2006 2007

40 30 30

40

50

60

70

80

mean age based on CFC−12 Fig. 10. Mean ages as calculated from CFC-12 and SF6 data assuming different Δ/Γ ratios, the black line denotes equal values.

90

66

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

necessary to balance the air–sea and diapycnal fluxes is the flux due to upwelling, which we will attempt to quantify for the Mauritanian upwelling system based on tracer data (Eq. (3)). Since the concentrations of CFC-12 and SF6 are, generally, lower in the ocean interior than in the mixed layer, upwelling and diapycnal mixing tend to decrease the concentration of the tracer in the mixed layer, whereas air–sea exchange will tend to increase the mixed-layer concentration in an active upwelling area, see Eq. (3). However, the air–sea flux of the tracer can also be out of the ocean during periods of rapid temperature increase in a well isolated mixed layer (e.g. Shao et al., 2013), which can also be seen in the 2006 data set discussed here. The data used to calculate the upwelling velocity (ω20), which was calculated for each station with transient tracer data in 2005 and 2007, are listed in Table 2. The upwelling velocities calculated from CFC-12 data range from less than 1 to ~ 5 × 10- 5 m s- 1, although the range of ω20 is somewhat larger when calculated from SF6 data. The upwelling velocity is generally higher close to the coast than in the open ocean (Fig. 12 and Table 2). This geographical distribution of upwelling

can be expected from coastal upwelling systems and is roughly consistent with the distribution of sea surface temperature (Fig. 5). The upwelling strength (velocity) is roughly equal for the spring (March/ April) 2005 cruise and the winter (January/February) 2007 cruise close to the coast, although the slight difference in station position makes a direct comparison difficult. There is a reasonable agreement between ω20 estimates from CFC-12 and SF6 measurements for the 2007 cruise. The comparison is not quite as good for the 2005 data, reflecting analytical difficulties in the measurement of SF6 concentrations. 3.4.1. Uncertainties Analytical uncertainties of the transient tracer concentrations are relatively straightforward to quantify. Even though no replicate samples were taken during any of these cruises, the total uncertainty of tracer measurements from flamed sealed ampoules is usually about 2% for CFC-12 and up to 5% for SF6. These numbers, based on off-line measurements of flame sealed ampoules in the lab, are about twice the values one would expect when the transient tracers are measured on the

Table 2 Upwelling velocity (ω20) based on CFC-12 and SF6 data during spring 2005 and winter 2007. The data going into Eqs. (3)–(6) are listed in the table, see main text for abbreviations. The atmospheric concentration of CFC-12 was assumed to be 546.5 ppt during the 2005 cruise and 543.0 ppt during the 2007 cruise, for SF6 the values are 5.74 and 6.17 ppt, respectively. The tracer concentration in the mixed layer (C1) is given both in volumetric units and as partial pressure since both units are relevant for different processes. The wind speed is the average wind speed recorded from the ship 48 h before each station. Upwelling velocity during spring of 2005 (POS320) Station

68

67

56

55

54

Longitude (°W) Wind speed (m s−1) MLD (m) Kz ×10−5 (m2 s−1)

19 8.3 17.9 3.0

18.5 8.4 14.8 3.0

17.5 7.8 21.8 2.7

17 7.9 24.1 3.4

16.5 8.2 28.3 3.0

CFC-12 C1 (nmol m−3) C1 (ppt) C2 (nmol m−3) dC/dz (nmol m−4) Fg ×10−6 (nmol m−2 s−1) Fd ×10−6 (nmol m−2 s−1) ω20 ×10−5 (m s−1)

1.23 541.3 1.12 −0.0053 0.44 −0.15 0.26

1.24 533.3 1.16 −0.0038 1.15 −0.12 1.3

1.25 526.0 1.04 −0.0106 1.55 −0.29 0.60

1.26 520.0 1.16 −0.0052 1.8 −0.18 1.2

1.17 436.7 0.96 −0.0104 9.4 −0.31 4.3

SF6 C1 (pmol m−3) C1 (ppt) C2 (pmol m−3) dC/dz (pmol m−4) Fg ×10−6 (pmol m−2 s−1) Fd ×10−6 (pmol m−2 s−1) ω20 ×10−5 (m s−1)

0.92 5.2 0.85 −0.0035 3.79 −0.10 5.3

0.94 5.2 0.84 −0.0052 3.58 −0.15 3.4

1.04 5.7 0.62 −0.0208 0.18 −0.56 0

0.81 4.4 0.69 −0.0061 8.88 −0.20 7.2

0.79 3.9 0.68 −0.0053 13.05 −0.16 11.7

Upwelling velocity during winter of 2007 (POS347) Station

113

115

117

119

120

121

122

Longitude (°W) Wind speed (m s−1) MLD (m) Kz ×10−5 (m2 s−1)

17.5 6.1 73.5 5.3

17 5.5 24.9 1.4

16.75 5.6 24.9 4.0

16.6 5.6 22.5 5.8

16.5 5.5 34.8 7.9

16.42 5.5 30.3 2.3

16.33 5.6 26.3 1.3

CFC-12 C1 (nmol m−3) C1 (ppt) C2 (nmol m−3) dC/dz (nmol m−4) Fg ×10−6 (nmol m−2 s−1) Fd ×10−6 (nmol m−2 s−1) ω20 (10−5 m s−1)

1.21 496.9 1.02 −0.0094 2.3 −0.49 0.96

1.20 488.8 1.12 −0.0040 2.66 −0.056 2.7

1.21 490.8 1.16 −0.0027 2.19 −0.11 4.2

1.22 497.7 1.14 −0.0039 1.92 −0.23 2.1

1.19 464.3 1.10 −0.0046 3.12 −0.36 3.0

1.24 474.7 1.11 −0.0067 2.75 −0.15 2.0

1.21 447.7 1.14 −0.0030 3.98 −0.039 5.6

SF6 C1 (pmol m−3) C1 (ppt) C2 (nmol m−3) dC/dz (nmol m−4) Fg (10−6 nmol m−2 s−1) Fd ×10−6 (nmol m−2 s−1) ω20 (10−5 m s−1)

1.04 5.60 0.52 −0.026 2.26 −1.36 0.2

0.98 5.25 0.86 −0.006 2.96 −0.08 2.4

1.06 5.66 0.92 −0.007 1.72 −0.28 1.0

1.1 5.90 0.93 −0.009 1.02 −0.50 0.3

0.93 4.83 0.77 −0.008 4.29 −0.63 2.3

0.84 4.30 0.65 −0.010 6.02 −0.23 3.1

0.91 4.53 0.83 −0.004 5.49 −0.05 6.8

Upwelling velocities are printed in bold letters.

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

ship during the cruise. The relative uncertainty is higher below the mixed layer as the concentrations are lower, so that particularly the SF6 values are more uncertain for the deeper water samples. Also, as discussed above, the SF6 data for the 2005 cruise has larger uncertainty due to analytical difficulties. As a consequence we are focusing our discussion based on CFC-12 data, except for the calculation of mean ages close to the surface where the higher uncertainty of SF6 data is compensated for by the fast changing atmospheric concentration in comparison to the near-steady-state CFC-12 concentrations. This uncertainty directly influences the calculation of mean ages, although the TTD concept is heavily dependent on the assumption of an inverse Gaussian distribution of the TTD, and the ratio of Δ/Γ used (see Section 3.3). Also, the inverse Gaussian solution to the TTD is generally valid for isopycnal, flows only. In the slowly ventilated Eastern Tropical North Atlantic it is has been shown that diapycnal fluxes are important (Banyte et al., 2012) thus increasing the uncertainties of the IG solution to the TTDs. The mean ages should therefore be seen as relative ages, and as upper limits. 3.4.2. Uncertainty of the upwelling velocity There are several sources of uncertainty to the calculations used to quantify the upwelling velocity (ω). In this section we will discuss and, if possible, quantify the most significant sources of uncertainty. The largest source of uncertainty (in the form of a systematic bias) in this approach stems from assumptions in the calculation of C2, the concentration of tracer below the mixed layer (Eqs. (10) and (11)). Below the mixed layer the tracer concentration decreases roughly linearly, and there is no obvious way to calculate “the tracer concentration below the mixed layer” (i.e. C2); defining C2 at shallower/deeper depths will lead to higher/smaller upwelling velocity. In the He3 method by Rhein et al. (2010), the average of all data between the 45 and 145 m was used to define C2, since this was considered to provide an average concentration of the Equatorial Undercurrent (EUC), the main water mass contributing to equatorial upwelling. Here we chose to followed a method that was used in the heat budget calculation of the mixed layer in the Eastern Tropical Pacific Ocean by Hayes et al. (1991) and defined C2 as the CFC-12 concentration 20 m below the mixed layer. We think that approach is more appropriate for the Mauritanian upwelling region, although difficult to quantify. We therefore use ω20 as symbol for the upwelling velocity calculated this way, analogous to the use of U10 for the commonly used definition of wind speed normalized to 10 m above the surface. The second largest source of uncertainty is in the measurement of CFC-12, which we assume have a 2% uncertainty. The CFC-12 concentrations are used in several of the steps to calculate ω and the average error

o

67

in ω is 41%, or 0.6 × 10-5 m s-1, although the relative error is, obviously, larger in stations with low upwelling velocities, and vice versa. Since the uncertainty in SF6 measurements is larger, so is the average error in ω larger (60%). The third most significant source of uncertainty is in the estimates of the air–sea flux; either from using different wind-speed products or different formulation of the air–sea exchange coefficient. In this study, wind speeds based on ship records and CERSAT dataset were considered and compared with each other (Figs. 13 and 14). Using the satellite derived wind speed from the CERSAT product has the advantage that the wind speed over the sampled station could be determined for several days/weeks before the measurements, but has the disadvantage that only 6 hour averages are available which could lead to biases since the air–sea exchange does not scale linearly to wind-speed (Eq. (7)). The ship-based wind speed has high temporal resolution but the ship was not in the region of the tracer measurements for several days, thus being susceptible for spatial variations in wind-speed. In Fig. 13 we show the ship-track with each 24 hour period in either gray or black, illustrating the potential spatial bias in the ship-based wind-speed. Fig. 14 shows the temporal evolution of wind-speed and the calculated k600 from ship and from satellite observations. There are obvious differences, particularly in k600 since short duration bursts in wind speed tend to be accentuated due to the non-linearity of the k600 formulation. Both estimates are, by default, erroneous due to issues described above, and it is hard to judge which one is most representative for the wind conditions prior to sampling. The difference in ω20 using the k600 estimates from either ship or satellite is 28% on average for the two cruises. The second element is the definition of k600. According to Ho et al. (2011) the coefficient of k600 has an uncertainty of 0.022 (see Eq. (7)). However, it has been shown that surfactants can affect the air–sea gas transfer velocities, and that the productive Mauritanian upwelling area provides favorable conditions for production of surfactants (e.g. Kock et al., 2012). We therefore assume that the uncertainty of k600 in the Mauritanian upwelling area is twice that of the 0.022 stated by (Ho et al., 2011), which for this study translated to an uncertainty of 18% on ω20. We also consider the uncertainties of flux of tracers due to diapycnal mixing, Fd, including MLD and local Kz. As discussed above, MLD can be defined differently, which will influence the tracer gradient below the mixed layer and estimates of the local Kz values and thus the diapycnal flux of tracers directly. However, the influence on the upwelling velocities due to MLD definition is small since Fd is significantly smaller than Fg, increasing/decreasing the change in density to define the MLD by 20% leads to 6% change in ω20. Table 3 summarizes the contribution of

Velocity based on SF6 (10−5 m/s)

Velocity based on CFC−12 (10−5 m/s)

12

19 N 40’

8

20’

5

18oN 40’

2005

4

2005

20’

17oN

3

19oN

2

40’

1.5 20’ o

18 N

1

40’

2007

2007

0.5

20’ o

17 N o 20 W

19oW

18oW

17oW

16oW

15oW 20oW

19oW

18oW

17oW

16oW

15oW

0

Fig. 12. Velocity of upwelling for the 2005 and 2007 cruises calculated based on the CFC-12 data (left column) and on SF6 data (right column), note the non-linear color scale. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

68

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

20oN 21oN 19oN 20oN 18oN 19oN 17oN 18oN 16oN o 20 W

19oW

18oW

17oW

16oW

15oW

20oW

19oW

18oW

2005

17oW

16oW

15oW

2007

Fig. 13. Maps of the spring 2005 (left) and winter 2007 (right) cruises. The thick gray and black lines represent 24 hour time periods, crosses show the location of each tracer station and dashed lines show the ship track after the last tracer station. This figure visualizes the potential spatial bias in ship-based wind-speed measurements.

the different processes and assumptions discussed above to the overall uncertainty in determining the upwelling velocity using CFC-12 data. These numbers are from this particular data set; relative uncertainties decrease during strong upwelling conditions, obviously. The way the tracer concentration below the MLD (C2) is defined constitutes the largest contribution to the uncertainty in calculation of upwelling velocity (ω), and so far no consensus on how to define the properties of the upwelled water from “below the base of the mixed layer” has been reached. Analytical uncertainties of the tracer concentrations in the mixed layer and calculation of the air–sea exchange are the most significant terms of uncertainty of upwelling velocity. Another issue is potential biasing in the spatial sampling that cannot be easily calculated. For instance, the 2007 cruise sampled closer to the coast than the 2005 and 2006 cruises did, and similarly, the 2005 section was sampled on 17°30′ N whereas during 2006 and 2007 sampling was conducted on 18°N. However, the seasonal variability is expected to dominate any observed variability

variables such as anthropogenic carbon content, few studies have been published on transient tracer measurements and the derived ventilation of Eastern Boundary Upwelling System (EBUS). Here we present CFC-12 and SF6 data from the Mauritanian Upwelling from three cruises conducted during three different years (2005–2007) and seasons (winter, spring and summer). The tracer distribution confirms the shallow mixed layer and the poorly ventilated water below the mixed layer, a situation commonly found in the tropical oceans. The transient tracer pair can be used to empirically constrain the shape of the transit time distribution (TTD), and we found that at least for the upper 400 m of the water column the TTD can be described with an inverse Gaussian distribution with the first two moments Δ/Γ = 1.2. A ratio of 1.0 is a commonly found relation in the ocean, although the uncertainties in the determining the Δ/Γ ratio from this data-set do not exclude a unity ratio, or other forms of the TTD. The transient tracer data show that relatively old water is entrained into the mixed layer due to upwelling and vertical diffusive processes. In the deeper part of the water column a weak tracer maximum is found in the AABW, co-located with a maximum in oxygen concentration. A tracer minimum is found in the deeper part of the NADW. Tracers have been used previously to determine the strength of upwelling, notable the use of helium isotopes in the tropical Atlantic (Klein

4. Discussion and conclusion Although transient tracers are commonly used in oceanography for quantification of transport-times and ventilation and for derived

Windspeed

15

15 Minute Hourly mean Satellite

10

10

5

5

0

28

29

30

31

1

2

3

4

5

0

18

19

k600

Date (March/April, 2005) 50

50

40

40

30

30

20

20

10

10

0

28

29

30

31

1

2

Date (March/April, 2005)

20

21

22

23

24

Date (January, 2007)

3

4

5

0

Minute Hourly mean Satellite

18

19

20

21

22

23

24

Date (January, 2007)

Fig. 14. Temporal evolution of wind speed (upper panels) and k600 (lower panels) during the 2005 (left) and 2007 (right) cruises. Gray points show the wind speed and k600 using wind data measured from the ship and the thick black lines show the hourly mean values. Horizontal thin black lines show 6 hourly mean value of wind speed (upper panels) and k600 (lower panels) based on satellite data. The vertical dashed lines show the timing of the tracer stations.

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

69

Table 3 Absolute (in m s-1) and relative (in %) uncertainties in upwelling velocity for each of the processes and assumptions discussed in the text as determined from this data set. The far right column, the total uncertainty, does not include the systematic bias from C2 definition. C2 definition

Analytical errors

Wind speed

k600 parameterization

MLD definition

Kz calculations

Total uncertainty

0.86 × 10−5 ms−1

0.6 × 10−5 ms−1 42%

0.42 × 10−5 ms−1 28%

0.40 × 10−5 ms−1 18%

0.1 × 10−5 ms−1 6%

0.02 × 10−5 ms−1 2%

0.8 × 10−5 ms−1 52%

and Rhein, 2004; Rhein et al., 2010). Those studies utilize the fact that the ratio of 3He/4He should be close to equilibrium to the atmosphere in the mixed layer, whereas there is an interior ocean source of the light isotope due to primordial emissions through hydrothermal activity, and use the observed dis-equilibrium of helium isotopes in the mixed layer to calculate the strength of upwelling. In this study, we calculated local upwelling velocities based on observations of CFC-12 and SF6 disequilibrium on 12 stations occupied during winter or spring. In a steady state assumption, the tracer flux to the mixed layer due to upwelling is the balance between the fluxes due to air–sea exchange and diapycnal mixing. We found that the tracer flux due to air–sea exchange was significantly larger than the flux by diapycnal mixing, and thus that the flux due to upwelling dominated over the flux due to diapycnal mixing. We have shown that CFC-12 and SF6 data can be used to estimate upwelling velocities, and we estimate the uncertainty of these estimates is roughly 50%, although the relative uncertainty is larger for weak upwelling conditions, and vice versa. We estimate that of the 12 stations occupied during the upwelling season, 3 stations have velocities less than 1 × 10- 5 m s-1, 6 were between 1 and 3 × 10- 5 m s- 1, and 3 stations had upwelling velocities larger than 3 × 10-5 m s-1, whereas the data from the summer season indicate no upwelling. Our data clearly show large spatial and temporal differences in the strength of the upwelling, with higher upwelling velocities close to the coast and during winter/spring. These results are similar to those by Rhein et al. (2010) for the equatorial Atlantic, although methodological differences in, primarily, the definition of C2 (the concentration of tracer below the mixed layer) bias a direct comparison. The upwelling velocities in this study would become significant lower than those of Rhein et al. (2010) if we were to use the same definition of C2 as they did. Upwelling velocities for the Peruvian and Mauritanian upwelling regions based on helium isotope measurements have also been presented by Steinfeldt et al. (2014), that found upwelling velocities generally consistent with those presented in this study. In a study by (Schafstall, 2010), the concentrations of NO− 3 and N2O in the mixed layer was balanced with the diapycnal flux (calculated from the vertical Ekman transport) and upwelling from below, and the loss rate in the mixed layer (air–sea exchange for N2O and biological consumption for NO3). This calculation gave a large residual for both N2O and NO3 which is attributed to horizontal advection, a process that is neglected in our study. Our study shows that deviations from saturation of transient tracers, such as CFC-12 or SF6, can be used to quantify upwelling velocities in upwelling systems, although with large uncertainties. These estimates of upwelling velocity can, for instance, be used to estimate flux of the climate relevant gases CO2 and N2O, and might be important to consider for local climate. Our study suggests upwelling velocities in the lower 10-5 m s-1 range during the upwelling season off Mauritania. Acknowledgments We acknowledge support from the German SOPRAN project (BMBF grant FKZ 03F0462A) for logistics in connection for the field campaigns. We thank the captain and crew of the Poseidon and Meteor for cooperation during the cruises. Special thanks are due to Dr. Hermann Bange for carrying out the tedious work of flame sealing on ampoules on the P320 and M68/3 cruises, and to Dr. Anke Schneider for careful measurements of samples in the lab in Kiel. M. Liu received funding from CSC (Chinese Scholarship Council, grant number 201408080087) during preparation of this manuscript.

References Bange, H.W., 2008. FS Poseidon Fahrtbericht/Cruise Report P320/1: Wo die Sahara den Atlantik trifft. IFM-GEOMAR, Kiel, Germany, Technischer Bericht. Banyte, D., Tanhua, T., Visbeck, M., Wallace, D.W.R., Karstensen, J., Krahmann, G., Schneider, A., Stramma, L., Dengler, M., 2012. Diapycnal diffusivity at the upper boundary of the tropical North Atlantic oxygen minimum zone. J. Geophys. Res. 117, C09016. http://dx.doi.org/10.1029/2011jc007762. Barton, E.D., Arı́stegui, J., Tett, P., Cantón, M., Garcı́a-Braun, J., Hernández-León, S., Nykjaer, L., Almeida, C., Almunia, J., Ballesteros, S., Basterretxea, G., Escánez, J., Garcı́a-Weill, L., Hernández-Guerra, A., López-Laatzen, F., Molina, R., Montero, M.F., Navarro-Pérez, E., Rodrı́guez, J.M., van Lenning, K., Vélez, H., Wild, K., 1998. The transition zone of the Canary Current upwelling region. Prog. Oceanogr. 41, 455–504. http://dx.doi.org/10. 1016/S0079-6611(98)00023-8. Benazzouz, A., Mordane, S., Orbi, A., Chagdali, M., Hilmi, K., Atillah, A., Lluís Pelegrí, J., Hervé, D., 2014. An improved coastal upwelling index from sea surface temperature using satellite-based approach—the case of the Canary Current upwelling system. Cont. Shelf Res. 81, 38–54. http://dx.doi.org/10.1016/j.csr.2014.03.012. Bullister, J.L., 2011. Atmospheric CFC-11, CFC-12, CFC-113, CCl4 and SF6 Histories. Carbon Dioxide Information Analysis Center (http://cdiac.ornl.gov/ftp/oceans/CFC_ATM_ Hist/CFC_ATM_Hist_2011). Bullister, J.L., Wisegarver, D.P., Menzia, F.A., 2002. The solubility of sulfur hexafluoride in water and seawater. Deep-Sea Res. I 49, 175–187. Cropper, T.E., Hanna, E., Bigg, G.R., 2014. Spatial and temporal seasonal trends in coastal upwelling off Northwest Africa, 1981–2012. Deep-Sea Res. I Oceanogr. Res. Pap. 86, 94–111. http://dx.doi.org/10.1016/j.dsr.2014.01.007. Dengler, M., Schafstall, J., Tanhua, T., Fiedler, B., Krahmann, G., Löptin, U.E., 2008. FS Poseidon Fahrtbericht/Cruise Report P347: Mauritanian Upwelling and Mixing Process Study. IFM-GEOMAR, Kiel, Germany. Gouriou, Y., Reverdin, G., 1992. Isopycnal and diapycnal circulation of the upper equatorial Atlantic Ocean in 1983–1984. J. Geophys. Res. Oceans 97, 3543–3572. http://dx. doi.org/10.1029/91JC02935. Haskell Ii, W.Z., Kadko, D., Hammond, D.E., Knapp, A.N., Prokopenko, M.G., Berelson, W.M., Capone, D.G., 2015. Upwelling velocity and eddy diffusivity from 7Be measurements used to compare vertical nutrient flux to export POC flux in the Eastern Tropical South Pacific. Mar. Chem. 168, 140–150. http://dx.doi.org/10.1016/j.marchem.2014. 10.004. Hayes, S.P., Chang, P., McPhaden, M.J., 1991. Variability of the sea surface temperature in the eastern equatorial Pacific during 1986–1988. J. Geophys. Res. Oceans 96, 10553–10566. http://dx.doi.org/10.1029/91JC00942. Ho, D.T., Wanninkhof, R., Schlosser, P., Ullman, D.S., Hebert, D., Sullivan, K.F., 2011. Toward a universal relationship between wind speed and gas exchange: gas transfer velocities measured with 3He/SF6 during the Southern Ocean Gas Exchange Experiment. J. Geophys. Res. Oceans 116, C00F04. http://dx.doi.org/10.1029/2010JC006854. Holtermann, P.L., Umlauf, L., Tanhua, T., Schmale, O., Rehder, G., Waniek, J.J., 2012. The Baltic Sea tracer release experiment: 1. Mixing rates. J. Geophys. Res. 117, C01021. http://dx.doi.org/10.1029/2011jc007439. Huhn, O., Roether, W., Beining, P., Rose, H., 2001. Validity limits of carbon tetrachloride as an ocean tracer. Deep-Sea Res. I 48, 2025–2049. Kadko, D., Johns, W., 2011. Inferring upwelling rates in the equatorial Atlantic using 7 Be measurements in the upper ocean. Deep-Sea Res. I Oceanogr. Res. Pap. 58, 647–657. http://dx.doi.org/10.1016/j.dsr.2011.03.004. King, D.B., Saltzman, E.S., 1995. Measurement of the diffusion coefficient of sulfur hexafluoride in water. J. Geophys. Res. 100, 7083–7088. Klein, B., Rhein, M., 2004. Equatorial upwelling rates inferred from helium isotope data: a novel approach. Geophys. Res. Lett. 31. Kock, A., Schafstall, J., Dengler, M., Brandt, P., Bange, H.W., 2012. Sea-to-air and diapycnal nitrous oxide fluxes in the eastern tropical North Atlantic Ocean. Biogeosciences 9, 957–964. http://dx.doi.org/10.5194/bg-9-957-2012. Körtzinger, A., 2009. FS Meteor Fahrtbericht/Cruise Report M68/3: Circulation and Variability in the Tropical Atlantic Part3. Leitstelle METEOR/MERIAN, Hamburg, Germany. Lee, B.-S., Bullister, J.L., Whitney, F.A., 1999. Chlorofluorocarbon CFC-11 and carbon tetrachloride removal in Saanich Inlet, an intermittently anoxic basin. Mar. Chem. 66, 171–185. Loucaides, S., Tyrrell, T., Achterberg, E.P., Torres, R., Nightingale, P.D., Kitidis, V., Serret, P., Woodward, M., Robinson, C., 2012. Biological and physical forcing of carbonate chemistry in an upwelling filament off northwest Africa: results from a Lagrangian study. Glob. Biogeochem. Cycles 26, GB3008. http://dx.doi.org/10.1029/2011GB004216. Oakey, N.S., 1982. Determination of the rate of dissipation of turbulent energy from simultaneous temperature and velocity shear microstructure measurements. J. Phys. Oceanogr. 12, 256–271. http://dx.doi.org/10.1175/1520[-‐]0485(1982)012b0256: DOTRODN2.0.CO;2. Osborn, T.S., 1980. Estimates of local rate of vertical diffusion from dissipation measurements. J. Phys. Oceanogr. 10, 83–89. Pelegri, J.L., Marrero-Diaz, A., Ratsimandresy, A., Antoranz, A., Cisneros-Aguirre, J., Gordo, C., Grisolia, D., Hernandez-Guerra, A., Laiz, I., Martinez, A., Parrilla, G., Perez-

70

T. Tanhua, M. Liu / Journal of Marine Systems 151 (2015) 57–70

Rodriguez, P., Rodriguez-Santana, A., Sangra, P., 2005. Hydrographic cruises off northwest Africa: the Canary Current and the Cape Ghir region. J. Mar. Syst. 54, 39–63. Rees, A.P., Brown, I.J., Clark, D.R., Torres, R., 2011. The Lagrangian progression of nitrous oxide within filaments formed in the Mauritanian upwelling. Geophys. Res. Lett. 38, L21606. http://dx.doi.org/10.1029/2011GL049322. Rhein, M., Dengler, M., Sültenfuß, J., Hummels, R., Hüttl-Kabus, S., Bourles, B., 2010. Upwelling and associated heat flux in the equatorial Atlantic inferred from helium isotope disequilibrium. J. Geophys. Res. Oceans 115, C08021. http://dx.doi.org/10. 1029/2009JC005772. Roether, W., Klein, B., Bulsiewicz, K., 2001. Apparent loss of CFC-113 in the upper ocean. J. Geophys. Res. 106, 2679–2688. Schafstall, J., 2010. The transition zone of the Canary Current upwelling region PhD, Physical Oceanography. Kiel University (219 pp.). Schafstall, J., Dengler, M., Brandt, P., Bange, H., 2010. Tidal‐induced mixing and diapycnal nutrient fluxes in the Mauritanian upwelling region. J. Geophys. Res. 115, C10014. http://dx.doi.org/10.1029/2009JC005940. Schneider, A., 2011. Tracers and Anthropogenic Carbon in the Tropical Atlantic Ocean and the Meiterranean Sea. Mathematisch-Naturwissenschaftlichen Fakultät, ChristianAlbrecht-Universität zu Kiel (100 pp.). Schneider, A., Tanhua, T., Körtzinger, A., Wallace, D.W.R., 2012. An evaluation of tracer fields and anthropogenic carbon in the equatorial and the tropical North Atlantic. Deep-Sea Res. I Oceanogr. Res. Pap. 67, 85–97. http://dx.doi.org/10.1016/j.dsr.2012. 05.007. Shao, A.E., Mecking, S., Thompson, L., Sonnerup, R.E., 2013. Mixed layer saturations of CFC11, CFC-12, and SF6 in a global isopycnal model. J. Geophys. Res. Oceans 118, 4978–4988. http://dx.doi.org/10.1002/jgrc.20370. Stanley, R.H.R., Doney, S.C., Jenkins, W.J., Lott Iii, D.E., 2012. Apparent oxygen utilization rates calculated from tritium and helium-3 profiles at the Bermuda Atlantic Timeseries Study site. Biogeosciences 9, 1969–1983. http://dx.doi.org/10.5194/bg-91969-2012. Steinfeldt, R., Sültenfuß, J., Dengler, M., Fischer, T., Rhein, M., 2014. Upwelling velocities inferred from helium isotope disequilibrium. Ocean Science Meeting, Honolulu. Steinhoff, T., 2010. Carbon and nutrient fuxes in the North Atlantic Ocean PhD Mathematisch-Naturwissenschaftlichen Fakultät, Christian-Albrechts-Universität zu Kiel.

Stöven, T., 2011. Ventilation Processes of the Mediterranean Sea Based on CFC-12 and SF6 Measurements. Diploma, Leibniz-Institut für Meereswissenschaften der Mathematisch-naturwissenschaftlichen Fakultät, Christian-Albrecht-Universität zu Kiel, Kiel. Stramma, L., Huttl, S., Schafstall, J., 2005. Water masses and currents in the upper tropical northeast Atlantic off northwest Africa. J. Geophys. Res. 110. Tanhua, T., Olsson, K.A., 2005. Removal and bioaccumulation of anthropogenic, halogenated transient tracers in an anoxic fjord. Mar. Chem. 94, 27–41. http://dx.doi.org/10. 1016/j.marchem.2004.07.009. Tanhua, T., Fogelqvist, E., Bastürk, Ö., 1996. Reduction of volatile halocarbons in anoxic seawater, results from a study in the Black Sea. Mar. Chem. 54, 159–170. Tanhua, T., Waugh, D.W., Wallace, D.W.R., 2008. Use of SF6 to estimate anthropogenic carbon in the upper ocean. J. Geophys. Res. Oceans 113, C04037. Thomson, R.E., Fine, I.V., 2003. Estimating mixed layer depth from oceanic profile data. J. Atmos. Ocean. Technol. 2, 319–329. http://dx.doi.org/10.1175/15200426(2003)020b0319:EMLDFON2.0.CO. Vollmer, M.K., Weiss, R.F., 2002. Simultaneous determination of sulfur hexafluoride and three chlorofluorocarbons in water and air. Mar. Chem. 78, 137–148. Walker, S.J., Weiss, R.F., Salameh, P.K., 2000. Reconstructed histories of the annual mean atmospheric mole fractions for the halocarbons CFC-11, CFC-12, CFC-113 and carbon tetrachloride. J. Geophys. Res. 105, 14285–14296. Warner, M.J., Weiss, R.F., 1985. Solubilities of chlorofluorocarbons 11 and 12 in water and sea water. Deep-Sea Res. 32, 1485–1497. Waugh, D.W., Hall, T.M., Haine, T.W.N., 2003. Relationship among tracer ages. J. Geophys. Res. 108. http://dx.doi.org/10.1029/2002JC001325. Waugh, D.W., Hall, T.M., Haine, T.W.N., 2004. Transport times and anthropogenic carbon in the subpolar North Atlantic Ocean. Deep Sea Res. I 51, 1471–1491. http://dx.doi. org/10.1016/j.dsr.2004.06.011. Weingartner, T.J., Weisberg, R.H., 1991. On the annual cycle of equatorial upwelling in the Central Atlantic Ocean. J. Phys. Oceanogr. 21, 68–82. http://dx.doi.org/10.1175/ 1520[-‐]0485(1991)021b0068:OTACOEN2.0.CO;2. Weisberg, R.H., Qiao, L., 2000. Equatorial upwelling in the central Pacific estimated from moored velocity profilers. J. Phys. Oceanogr. 30, 105–124. http://dx.doi.org/10.1175/ 1520[-‐]0485(2000)030b0105:EUITCPN2.0.CO;2.