Journal of Marine Systems 82 (2010) 245–251
Contents lists available at ScienceDirect
Journal of Marine Systems j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j m a r s y s
Carbon dynamics in a productive coastal region—The Skagerrak Sofia Hjalmarsson, Melissa Chierici, Leif G. Anderson ⁎ Department of Chemistry, University of Gothenburg, Sweden
a r t i c l e
i n f o
Article history: Received 30 November 2009 Received in revised form 21 May 2010 Accepted 25 May 2010 Available online 4 June 2010 Keywords: Carbon dioxide Skagerrak Biogeochemical processes Sea–air CO2 flux Seasonality
a b s t r a c t The importance of the coastal seas as areas of CO2 uptake from the atmosphere has gained more attention during recent years. This study utilizes dissolved inorganic carbon and hydrographic data collected in the Skagerrak for 10 months in 2006 to assess the carbon dynamics over the year. The surface water is under-saturated in CO2 relative to the atmosphere during the first half of the year and stays close to equilibrium at least until November. Consequently primary production compensates for the increase in pCO2 caused by the temperature increase from ∼2 to ∼10 °C in spring. Integrating the annual air–sea CO2 flux as computed using the Wanninkhof (1992) parameterization gives a net uptake of 1.2 mol m− 2 year− 1 which, if representative for the whole Skagerrak area, equals 3.7 ∙ 1010 mol year− 1 or 0.45 TgC year− 1. Converting the nitrate consumption in the surface mixed layer from January to May to carbon units through the RKR ratio (Redfield et al., 1963) gives a drawdown of 6 g C m− 2. This number increases by a factor of two if primary productivity also occurs in the waters below the surface mixed layer, i.e. an increase in depth from 10 to 25 m as a seasonal average.We estimated the effect of salinity, biological processes and air–sea CO2 exchange on the monthly DIC change. We found that salinity was one of the major drivers for the DIC change. © 2010 Elsevier B.V. All rights reserved.
1. Introduction The coastal seas have often been overlooked when discussing the global carbon cycle. However, during recent years this has changed and studies have shown that the coastal area due to large biological production plays a large role in the carbon cycle of the oceans despite its relatively small surface area (e.g. Borges et al., 2005; Bozec et al., 2005). The intense input of nutrients from land as well as from upwelling (Chen and Borges, 2009) leads to high biological productivity in coastal seas promoting a flux of CO2 from the atmosphere into the surface water as a result of primary production. The produced particulate organic carbon is transported out of the surface water to deeper layers where it is decomposed and may then be exported to the deep oceans as DIC (dissolved inorganic carbon). This export of DIC to the deep ocean is known as the continental shelf pump. The coastal and marginal seas take up 27–30% of the CO2 taken up by the open ocean (Chen and Borges, 2009). This work describes the processes that control the carbon dioxide system in Skagerrak using monthly data on Total Alkalinity (TA), DIC, and pH in 2006. From the partial pressure of CO2 (pCO2) in the air and calculated pCO2 in the water we estimate the air–sea CO2 flux and calculate the net annual CO2 flux, as well as assess the major drivers for the air–sea CO2 exchange. The Skagerrak is a coastal sea in Northern Europe; it borders the North Sea to the west and the Kattegat to the south. The latter is connected to the
⁎ Corresponding author. E-mail address:
[email protected] (L.G. Anderson). 0924-7963/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jmarsys.2010.05.013
Baltic Sea through the Danish Belts and the Öresund in the south (Fig. 1). Consequently, the water in the Skagerrak is a mixture of water masses from these two bordering seas. The large scale circulation and hydrography of the Skagerrak have been described by e.g. Rodhe (1996) and a schematic view of the general circulation is shown in Fig. 1. Water from the Northern Jutland Current (NJC) flows along the Danish coast and enters the Skagerrak in the southwest. The NJC consists of water masses from the North Sea as well as Atlantic water from the Southern Jutland Current which flows along the Danish west coast. The salinity of this water usually varies between 31 and 34 (Rydberg et al., 1996). In the southeast the brackish Baltic Current (BC), consisting of surface waters from the Baltic Sea and the Kattegat, enters the Skagerrak. The BC is then mixed with Skagerrak surface water, forming the Norwegian Coastal Current (NCC) which leaves the Skagerrak in the northwest. A cyclonic current dominates the circulation in Skagerrak and it extends from the surface to the bottom or down to 400–500 m (Rodhe, 1996). The mixing of high saline water and low-saline surface water from the BC occurs in fronts. The most persistent front is the Skagerrak front from the northern tip of Denmark towards the northeast. The front forces the low-saline BC water to contract along the Swedish coast. However, due to the wind driven advection, the low-saline water can at times be found in most of the Skagerrak. The advection of low-saline water away from the coast is the likely cause for the efficient mixing in the Skagerrak (Rodhe, 1989). The daily variability in the distribution of the low-saline water is much larger than seasonal changes. The different water masses in the Skagerrak are distinguishable by their salinities. The brackish waters of the BC have salinities between 20 and 30, while intermediate and deep water masses originating in the North Sea and
246
S. Hjalmarsson et al. / Journal of Marine Systems 82 (2010) 245–251
Fig. 1. Map of the research area. To the left a schematic illustration of the general circulation in Skagerrak, the surface currents shown are the Baltic Current (BC) and the Norwegian Coastal Current (NCC) while the Northern Jutland Current (NJC) is a subsurface current. As indicated by filled and dashed arrows representing subsurface and surface currents, respectively. To the right the location of the 5 stations sampled in 2006.
the Atlantic are characterized by salinities of 31 to 35 and N35, respectively. The freshwater input to the Skagerrak is dominated by the inflow of surface water from the Kattegat, this water contains about 1.5 · 104 m3 s− 1 of freshwater mainly from Baltic rivers (Rodhe, 1996). An additional freshwater supply of about 0.2 · 104 m3 s− 1 enters the Skagerrak originating from major Norwegian rivers emptying into the Oslo Fjord area in the northeastern corner of Skagerrak (Gustafsson and Stigebrandt, 1996). The flushing time of the Skagerrak is about 100 days if the stagnant deep water is excluded (Rodhe, 1996). In 2006, five stations (Fig. 1) were visited 10 times. The stratification in the region is fairly stable during the year with a surface layer of ∼ 10 m having a salinity around 24–30 that consists of a mixture with low salinity Baltic Sea and Kattegat water. The deeper layer has a salinity of ∼ 35 and is a mix of North Sea water and Atlantic water. The onset of the biological primary production in the Skagerrak generally starts in March–April. The spring bloom is dominated by diatoms and is followed by an autumn bloom in September; dominated by dinoflagellates. Blooms of calcium carbonate forming organisms which affect TA, such as Coccolitophorids (Emiliania Huxleyi) are not recurrent in the Skagerrak, but do occur. In 2006, the spring bloom started in February, earlier than normal according to measurements done by the Swedish Meteorological and Hydrological Institute (SMHI). SMHI further reports that the fall bloom began in September and that there were no major blooms of Coccolitophorids (Skjevik, 2007). 2. Methods The study is based on data collected during 10 cruises in the year 2006. Five stations along an east to west transect in the Skagerrak (Fig. 1) were visited each month with the exceptions of July and December. The station names, location and bottom depths are shown in Table 1. Samples were collected from a 12-bottle Niskin-Rosette water sampler attached to a Conductivity–Temperature–Depth (CTD) sensor.
Seawater samples were taken throughout the water column for the determination of DIC, TA, pH, oxygen (O2) and nutrients (phosphate, nitrate, and silicate). Standard sampling depths were at 5 m, 10 m, 15 m and 20 m, while the interval between the deeper samples ranged from 10 to 50 m. The samples for DIC, TA and pH were stored cold and dark and were analysed within 2 days after the cruise. DIC was determined using coulometric titration of an acidified sample with photometric detection (Johnson et al., 1987) and TA was determined by potentiometric titration with hydrochloric acid (0.05 M) in an open cell according to Haraldsson et al. (1997). pH was determined spectrophotometrically on the total scale using sulphonephthalein dye, m-cresol purple, as indicator (Clayton and Byrne, 1993; Lee and Millero, 1995). Prior to analysis the samples were thermostated to a temperature of 15 °C. The accuracy in pH was approximately ±0.002 pH units (Dickson, 1993). The accuracy is determined by the accuracy of the temperature measurements and the accuracy in the determination of the equilibrium constants of the dye. The magnitude of the perturbation of seawater pH caused by addition of the indicator solution was calculated and corrected for by the use of the method described in Chierici et al. (1999). The precision of the methods was estimated by calculating the standard deviation for sample replicates, giving about ± 1 µmol kg− 1 for TA, ± 2 µmol kg− 1 for DIC and ± 0.002 units for pH. The accuracy of
Table 1 Station locations and bottom depth. Station no.
Latitude °N
Longitude °E
Bottom depth M
A13 A14 A15 A16 A17
58.33 53.32 58.3 58.27 58.28
11.03 10.93 10.85 10.72 10.52
85 120 140 210 350
S. Hjalmarsson et al. / Journal of Marine Systems 82 (2010) 245–251
247
DIC and TA was determined by analysing Certified Reference Material (CRM), supplied by A. Dickson, Scripps Institution of Oceanography (USA). The O2 concentration was determined onboard using a modified Winkler titration with visual endpoint detection, as described by Hansen (1999). The samples for determining nutrients were stored cold and dark before being analysed at an accredited laboratory (at the SMHI), typically within a few days after sampling. The Apparent Oxygen Utilization (AOU) reflects oxygen production or consumption which indicates production or degradation of organic matter and is calculated as follows: AOU= [O2]saturated − [O2]measured, where the [O2]saturation was calculated based on the [O2]saturated equation of Weiss (1970). 2.1. Quality control of CO2 system parameters and calculation of pCO2 The TA data for January could not be used due to instrumental failure. Hence the chemical speciation computer program CO2SYS (Pierrot et al., 2006) was used to calculate the TA concentrations from DIC and pH using the carbonate system constants of Mehrbach et al. (1973) as refitted by Dickson and Millero (1987). CO2SYS was also used to calculate the pCO2 from DIC and pH using the same carbonate system constants. DIC and pH were chosen since they were measured accurately on all cruises. The depth of the surface mixed layer (SML) was calculated using the high resolution CTD record at the depth where the largest salinity gradient between surface and subsurface waters was found according to Lorbacher et al. (2006). Sigma (σ) was calculated from the density (ρ [kg dm− 3]) according to: σ = (ρ − 1) 1000. 2.2. Sea–air CO2 flux To calculate the flux of CO2 across the air–sea interface, the following equation was used: w
a
FCO2 = K0 k pCO2 –pCO2
ð1Þ
where K0 is the salinity and temperature dependent solubility of CO2, k is a the gas transfer velocity and pCOw 2 and pCO2 are the partial pressure of CO2 in water and atmosphere respectively. k depends on the wind speed and the Schmidt number (a function of the temperature) and for this study the parameterization k = 0.31 u2 (Sc/660)− 1/2 from Wanninkhof (1992) (W92) was used to facilitate comparisons with flux values given by (Borges et al., 2005). Also the parameterizations from Weiss et al. (2007), W07, and Nightingale et al. (2000), N00, were used for comparison since they have been developed in areas close to Skagerrak; W07 in the Arkona Basin in the southern Baltic Sea and N00 in the North Sea. Data for the daily sea level pressure and wind speed at 10 m altitude was obtained from the NCEP Daily Global Analyses data provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their website at http://www.esrl.noaa.gov/psd/. The pCOw 2 was calculated from DIC and pH at 5 m depth and was linearly interpolated between the monthly data points to get daily values. In December, January values were used as reference for this interpolation. The same was done with the observed temperature and salinity for calculating K0. Monthly data of the atmospheric mole fraction (xCO2) were provided by the NOAA/ESRL Global Monitoring Division, Boulder, Colorado, USA from their website at http://www. esrl.noaa.gov/gmd/dv/site/site_table.html#ccg_surface. The xCO2 were recalculated to be representative for the Skagerrak as a function of latitude between the station Mace Head, Ireland (53.33 N, 9.90 W) and the Baltic Sea station, Poland (55.35 N, 17.22 E). The xCO2 was converted to pCO2 according to; a SL pCO2 = xCO2 p –pH2 O
ð2Þ
where pSL is the atmospheric pressure at sea level also obtained from NOAA and pH2O is the water vapour pressure (Weiss and Price, 1980).
Fig. 2. Monthly mean sigma (σ) profiles in 2006 at the five stations. The σ-profiles for February, March, May and June are comparable to April and are not shown. The standard deviation for the profiles ranges from ±3 in the surface to about 0 in the bottom water.
3. Results and discussion 3.1. Seasonal variability 3.1.1. Hydrography The seasonal evolution of sigma during 2006 is shown in Fig. 2 as a spatial mean for each month (all five stations) of sigma (σ) versus depth. The σ in the surface water increases gradually down to ∼20– 30 m until autumn. In October a decrease in σ reaching to ∼100 m was observed due to a larger contribution of low salinity water from the BC in the study area which was probably mixed down due to an increase in windspeed. This water is mixed down deeper in November and reaches down to ∼250 m and a much smaller quantity of the densest water was present in the section. In Fig. 3 the seasonal evolution of some parameters is shown at 5 m depth and at the station bottom depth. The salinity in the two layers is shown in Fig. 3a and in Fig. 3g. Large horizontal gradients and seasonal changes are observed for all parameters in the surface layer. The salinity gradient in the surface water is particularly large in the beginning of the year and varies between S N 33 in the west to S b 25 in the east. This is a signal of the influence by Baltic Sea water, which is present in the surface water during the beginning and the end of the year. From April to September the salinity increases due to decreasing inflow of Baltic Sea water and/ or the Baltic Sea water being closer to the Swedish coast. In April the salinity and temperature in the deep water (not shown) is lower at station A13 and at the same time there is no change in the surface salinity which indicates that surface water is mixed down at this station. The temperature in the surface varies seasonally, but is relatively constant along the transect for each month (Fig. 3b). The temperature in the deep water is on average 7 ± 2 °C (not shown) and shows less variation than in the surface water. 3.1.2. The CO2 system and ancillary parameters Before the onset of biological CO2 drawdown, the surface concentrations of DIC and NO 3 (Fig. 3c, d), are in average 2090 µmol kg− 1 and 4 µmol kg− 1, respectively. This is higher than during the rest of the year, even though the salinity is lower at this time, which illustrates the contribution of DIC (and TA) from Baltic Sea river runoff. The rivers draining into the Baltic Sea has a relatively high TA with the weight averaged concentration being about 1350 µmol/kg (Hjalmarsson et al., 2008). At the same time pH (Fig. 3e) is relatively constant at 7.93. The first sign of biological CO2 drawdown from primary production is noticed at the easternmost station (A13) in February when NO3 and pCOw 2 decreases (Fig. 3d and f). Also DIC decreases in the surface water during summer, from an average of 2060 µmol kg− 1 in April to 2000–2010 µmol kg− 1
248
S. Hjalmarsson et al. / Journal of Marine Systems 82 (2010) 245–251
Fig. 3. Annual distribution at 5 m for salinity (a), temperature in °C (b), DIC in µmol kg− 1 (c), NO3 in µmol kg− 1(d) pHtot at 15 °C (e) and pCOw 2 in µatm (f), and at bottom depth for salinity (g), NO3 in µmol kg− 1 (h) and pHtot at 15 °C (i).
throughout the year and NO3 is depleted from May. As indicated by pH, the production seems to be active over the whole section i.e. also in the open Skagerrak. The somewhat lower production at station A13 seems to be the result of the mixing events that occurs closer to the coast. The most prominent mixing event is the one starting in April, mentioned in Section 3.1.1, when surface water is mixed down to the deep water. The primary production signal from the surface is seen at the bottom of station A13. For NO3 the concentration decreases to
below 1 µmol kg− 1 while pH increases (Fig. 3h and i). The TA has a strong correlation with salinity in both layers and varies in the surface from 2120 µmol kg− 1 in the low-saline water to 2270 µmol kg− 1 in the high saline water in the western part in October. In the deep water both TA and DIC is stable over the year with an average of 2319 ± 11 µmol kg− 1 and 2151 ± 21 µmol kg− 1, respectively. The pCOw 2 pattern in the surface (Fig. 3f) shows a 50 µatm decrease from spring to summer. Part of the change in pCOw 2 as a result of
S. Hjalmarsson et al. / Journal of Marine Systems 82 (2010) 245–251
249
Table 2 Mean values of the surface mixed layer for the individual months. Sample dates
SML m
Salinity
Nitrate µmol kg− 1
DIC µmol kg− 1
January 25 February 22 March 21 April 25 May 23 June 27 August 30 September 26 October 24 November 21
10 ± 0 12 ± 5 10 ± 0 10 ± 0 18 ± 7 13 ± 3 10 ± 0 13 ± 2 10 ± 1 26 ± 12
30.72 ± 3.86 31.39 ± 3.32 29.70 ± 3.20 31.46 ± 0.67 31.13 ± 1.57 30.35 ± 1.45 31.53 ± 2.83 32.38 ± 1.02 28.83 ± 3.46 33.91 ± 0.27
5.27 ± 1.31 4.64 ± 2.15 4.04 ± 2.23 0.38 ± 0.93 0.12 ± 0.26 0.06 ± 0.03 0.04 ± 0.01 0.02 ± 0.01 0.07 ± 0.05 2.29 ± 0.19
2114 ± 29 2105 ± 28 2092 ± 50 2072 ± 22 2031 ± 32 2023 ± 42 1998 ± 27 2019 ± 17 1990 ± 67 2095 ± 4
gives a net uptake of 5.3 ∙ 1010 mol year− 1 (0.64 Tg C year− 1) and 3.0∙1010 mol year− 1 (0.36TgC year− 1) respectively. Fig. 4. pCOw 2 at 5 m (open circles and solid line), error bars are the standard deviation between the 5 stations, pCOw 2 at 5 m corrected to 15 °C (open diamonds and dotted line) and pCO2 atm (filled circles).
primary production is compensated for by warming during this period. In the east pCOw 2 decreases even more, about 75 µatm, by primary production. 3.2. Sea–air exchange of CO2 and net annual CO2 uptake The change in pCOw 2 at 5 m depth over the year is shown in Fig. 4. Each data point is the average for the five stations in that month. Temperature normalized pCOw 2 (to 15 °C, Fig. 4) is ∼500 µatm during the first three months of the year and then decreases by ∼ 100 µatm to ∼ 380 µatm. This illustrates the effect of biological drawdown in spring, a signal that is not seen in the in situ pCOw 2 . Hence the seasonal temperature variability is as important as primary productivity in driving the air–sea flux, at least during the first half of the year. The calculated sea–air flux of CO2 using W92 is given in Fig. 5 and it is evident that the Skagerrak acts as a strong sink during about the first half of 2006, while the flux is small and variable during the next 4–5 months, returning to a sink in the end of the year. The flux computed with W92 gives a net uptake of 1.2 mol m− 2 year− 1 or 14gC m− 2 year− 1. Application of the W07 gives 1.7molC m− 2 year− 1 (20 gC m− 2 year− 1) and the N00 flux formulation gives an uptake of 0.9mol C m− 2 year− 1 (11 gC m− 2 year− 1). Assuming these values as representative for the whole Skagerrak area the net uptake using W92 equals 3.7∙1010 mol year− 1 (0.45TgC year− 1), while using W07 and N00
3.3. Evaluation of major controls on DIC in the surface water DIC is affected by several processes, such as biological processes, physical mixing and sea–air CO2 exchange. The seasonal variability is strongest in the surface water and as the signal is distributed over the depth of SML its thickness is highly relevant. In the Skagerrak the physical mixing often implies changes in salinity since the Skagerrak is a mixing area between the Baltic Sea and the North Sea and the input of these source waters varies over time. The average depth of the SML, mean concentrations of salinity, DIC and nitrate used for the evaluation are given in Table 2. The SML average is based on the five stations and is for the largest part of the year as shallow as 10 m. The SML shows large vertical and horizontal gradients in all study parameters, which is reflected in the standard deviation of the parameters in Table 2. The large standard deviations observed for nitrate in February and March, are due to the initiation of the spring bloom, which has developed to a different degree across the section. To evaluate the processes that affect the changes between two sampling occasions of DIC in the SML (ΔDIC) we used a similar approach as (Bégovic and Copin-Montégut, 2002) and the following equation is used; ΔDICSW = ΔDICB + ΔDICSSS + ΔDICAS + ΔDICR ;
where ΔDICSW is the observed change in DIC between two sampling occasions, ΔDICB is the change due to biological processes, ΔDICSSS is the change in DIC due to changes in salinity, which reflects changes related to freshening and mixing, and ΔDICAS represents changes due to sea–air CO2 gas exchange. The residual, ΔDICR, is the residual calculated as the balancing term. ΔDICSSS is calculated from TA and pH and the change in salinity between two sampling occasions (ΔS) using the CO2SYS program (Pierrot et al., 2006). ΔDICB is calculated by using the equation: ΔDICB = rC:N · ΔNO3
Fig. 5. The daily sea–air flux of CO2 computed using the Wanninkhof 1992 parameterization and wind data from 10 m above sea level provided by NOAA NCEP. The surface water pCOw 2 is calculated from measurements of DIC and pH at 5 m water depth and the atmospheric pCOw 2 is calculated based on data provided by NOAA.
ð3Þ
ð4Þ
where ΔNO3 is the difference in the nitrate concentration between the data of two sampling occasions and rC:N is the Redfield et al. (1963) ratio of 106:16. We use the nitrate concentration to calculate the organic carbon, since it is considered to limit the primary production in the Skagerrak (Fonselius, 1996; Pettersson, 1991). The C:N ratio appropriate for our data set could be estimated to range between 4.5 and 7.9, introducing an error of 16 to 32% in ΔDICB. We note that the C:N ratios are strictly valid over large time and space scales and we have not included the C:N in the produced dissolved organic matter. This means that our estimates are only indicative of the fraction of the carbon change that is caused by biological processes and are not true quantitative calculations of that fraction.
250
S. Hjalmarsson et al. / Journal of Marine Systems 82 (2010) 245–251
Fig. 6. The changes in DIC in the MLD for each month by biological effects (ΔDICB, diagonal striped bars), by salinity (ΔDICSSS, striped bars), by air–sea flux (ΔDICAS, black bars), and the residual (ΔDICR, white bars).
It should be noted that the change in nitrate is also impacted by mixing. However, the effect of mixed layer deepening and vertical diffusion was negligible since the change was in the range of 1 µmol C kg− 1. This is due to the small changes in the depth of the SML and that the vertical gradient in nitrate is below the SML at 30–50 m. Phosphate is also depleted in the Skagerrak and the concentration is below 0.1 µmol kg− 1 from April to August. The ΔDICAS is calculated from the flux of CO2 (Fig. 4). The contribution to DIC of the three processes and the residual is shown in Fig. 6. Positive values indicate that the process caused an increase in the DIC concentration relative to the previous measurement and negative values represent a decrease in DIC. In January, changes in salinity cause a DIC loss of about 15 µmol kg− 1 using the November data as the reference value for winter. This loss is counteracted by the gain from CO2 due to biological processes (decomposition of organic matter). The decrease in DIC due to biological primary production is initiated in February and continues to a maximum DIC loss of about 25 µmol kg− 1 in April. The sea–air CO2 exchange gives a noticeable increase in DIC during the first 5 months of the year while enduring the rest of the year gas exchange exhibits only a small effect. In November there is an increase in ΔDICB of 15 µmol C kg− 1, a possible reason for this is biological remineralisation in the SML or addition from below during the deepening of the SML by about 15 m. This is in agreement with the nitrate concentration that increases with 2.2 µmol kg− 1 (Table 2). The relative contribution from the three processes showed that changes in salinity were the major driver for changes in DIC. It explained about 36% of the changes in DIC, while biological processes and air–sea exchange explained 24% and 13%, respectively. The carbon consumed by biological primary production as estimated from the nitrate maximum in January to the nitrate depletion in May (Table 2) was 35 µmol C kg− 1 or 6 g C m− 2 if distributed evenly in the surface mixed layer and increases to 82 µmol C kg− 1 or 15 g C m− 2 if the ΔDICR is included (see section Uncertainties below). However, subsurface phytoplankton blooms has previously been reported in the Skagerrak (Richardson et al., 2003) implying a deeper photic zone than the SML. In the Skagerrak the secchi depth has been estimated to be around 10 m (Aarup, 2002) and if the photic zone is 2.5 times the secchi depth (LeBrasseur et al., 1979), the carbon consumed corresponds to 11 g C m− 2 (or 26 g C m− 2 if including ΔDICR). The assumption of a deeper photic zone is supported by a decrease in pCO2 at 20 m depth (not shown) from March to June. Also, nitrate and oxygen data from the same period shows low concentration of NO3, b0.6 µmol kg− 1 and AOU is less than −20 down to 30 m. The above estimates of primary production correspond to a concentration of particulate organic carbon between 35 (82) and 65 (152) µmol kg− 1 (the latter being if the photic zone is 2.5 the secchi
depth). This can be compared with reports from the Skagerrak of 74 C µmol l− 1 in the beginning of March and decreasing to 26 µmol C l− 1 in May (Pettersson, 1991). However, their initial concentration of nitrate is about twice as high. It should be noted that their data are from Kosterfjorden, which is closer to the Swedish coast and is therefore more exposed to eutrophication from land. Aure and Dahl (1994) estimated a supply of particulate carbon (PC) to the bottom water (below 350 m) in the Skagerrak of 43gC m− 2 year− 1. A large fraction of this PC is suggested to be imported into the Skagerrak region from the North Sea. On the other hand not all locally produced organic matter is expected to reach below 350 m as much of it would be rematerialized within the water column before reaching this depth. Stigebrandt (1991) used a box model constrained by oxygen and computed an export production in the Skagerrak of 49 g C m− 2 year− 1. This is significantly higher than our maximum new production from January to May, 29 g C m− 2, when integrating the nitrate consumption in a depth of 2.5 times the secchi depth and also including the ΔDICR. However, the computation of Stigebrandt (1991) estimated that at least half of the export production computed from May to October, while or new production is mainly to April, making our data comparable. 3.4. Uncertainties The residual between the observed change in measured DIC and the sum of the calculated changes, ΔDICR, shows positive values in the winter (November and January) and mostly negative values during the rest of the year. One explanation for this is that the change by primary production is underestimated by neglecting external sources of nitrate. These could include atmospheric deposition of nitrate as well as nitrate, ammonium and organic nitrogen advected from surrounding seas, or input by rivers. Fixation of nitrogen by cyanobacteria should however not be a major source as no substantial blooms has been reported. If primary production is the only contribution to ΔDICR this value should be added to the ΔDICB concentration. However, there is also an uncertainty in the gas transfer velocity that could add to the ΔDICR term, as well as uncertainties in the ΔDICSSS term. The large residual term likely also included the uncertainty due to the fact that we use mean values with large standard deviations due to large horizontal gradients (Fig. 3). The large horizontal gradient likely introduced large uncertainty in the ΔDICsss term since salinity has large differences in the surface water from west to east. This work is based on monthly cruises and in most cases an average is used which is considered to be the integrated situation of the water. For a highly dynamic area as the Skagerrak monthly measurements may be too sparse, with the effect that occasional events like strong winds or an enlarged inflow of source waters during the cruise has a large impact on the results. However, this
S. Hjalmarsson et al. / Journal of Marine Systems 82 (2010) 245–251
presentation illustrates the dynamics that impact the carbon cycle of the region and this caveat will not impact the conclusions. 4. Summary and conclusions The concept of the continental shelf pump requires uptake of CO2 from the atmosphere during the productive season followed by sedimentation of the organic matter out of the SML and with an export of the decay products to the subsurface waters of the deep connecting ocean. In the Skagerrak, we observe an uptake of CO2 from the atmosphere during the productive first half of the year with little exchange during the second half. The total carbon consumption equals 0.9 to 1.7 mol m− 2 year− 1, which for W92 and N00 is in agreement with the value of 1.05 mol m− 2 year− 1 given by Chen and Borges (2009) for global shelves of 40−200 m. Our W92 estimate also agrees well with estimates for the Baltic Sea of 0.8 mol m− 2 and for the North Sea 1.4 mol m− 2 (Borges et al., 2005). We found little out-gassing in the fall at the time of increase in pCO2 from decay of organic matter. This implies that most of the produced DIC remains in the deeper water layers and that vertical mixing of deep water to the surface plays a minor role. Thus the excess DIC may be exported to the North Sea and further out into the North Atlantic, making the Skagerrak a sink of atmospheric CO2. However, this cannot be concluded from a single year investigation. The primary production as computed from the nitrate consumption from January to May was estimated to between 35 and 65 µmol C kg− 1 increasing to between 93 and 170 µmol C kg− 1 including the residual change when the effect of mixing and air–sea flux is subtracted. This is comparable to reports from the Skagerrak of 74 µmol C l− 1 in the beginning of March (Pettersson, 1991). When these values are integrated over the depth of the photic zone they correspond to 6 g C m− 2 (15 g C m− 2 if including ΔDICR), SML of 10 m or 11 g C m− 2 (26 g C m− 2 if including ΔDICR) when the photic zone corresponds to 2.5 times the SML thickness. Acknowledgments This research was supported by grants from the Swedish Research Council Formas, Swedish Research Council, and from the EU project CARBOOCEAN, project no. 511176. The authors would like to thank the crew on R/S Skagerak, Sara Jutterström and Ludger Mintrop for valuable help during the cruises. We also thank two reviewers for constructive comments that have significantly improved the manuscript. This is a contribution from Tellus, the Centre of Earth Systems Science at the University of Gothenburg. References Aarup, T., 2002. Transparency of the North Sea and Baltic Sea—a Secchi depth data mining study. Oceanologia 44 (3), 323–337. Aure, J., Dahl, E., 1994. Oxygen, nutrients, carbon and water exchange in the Skagerrak Basin. Cont. Shelf Res. 14 (9), 965–977. Bégovic, M., Copin-Montégut, C., 2002. Processes controlling annual variations in the partial pressure of CO2 in surface waters of the central northwestern Mediterranean Sea (Dyfamed site). Deep Sea Res. Part II: Top. Stud. Oceanogr. 49 (11), 2031–2047. Borges, A.V., Delille, B., Frankignoulle, M., 2005. Budgeting sinks and sources of CO2 in the coastal ocean: diversity of ecosystems counts. Geophys. Res. Lett. 32.
251
Bozec, Y., Thomas, H., Elkalay, K., de Baar, H.J.W., 2005. The continental shelf pump for CO2 in the North Sea—evidence from summer observation. Mar. Chem. 93, 131–147. Chen, C-T. A, Borges, A.V., 2009. Reconciling opposing views on carbon cycling in the coastal ocean: Continental shelves as sinks and near-shore ecosystems as sources of atmospheric CO2. Deep-Sea Res. Part II 56, 578–590. Chierici, M., Fransson, A., Anderson, L.G., 1999. Influence of m-cresol purple indicator additions on the pH of seawater samples: correction factors evaluated from a chemical speciation model. Mar. Chem. 65, 281–290. Clayton, T.D., Byrne, R.H., 1993. Spectrophotometric seawater pH measurements: total hydrogen ion concentration scale calibration of m-cresol purple and at-sea results. Deep-Sea Res. 40, 2115–2129. Dickson, A.G., 1993. The measurement of seawater pH. Mar. Chem. 44, 131–142. Dickson, A.G., Millero, F.J., 1987. A comparison of the equilibrium constants for the dissociation of carbonic acid in seawater media. Deep-Sea Res. 34, 1733–1743. Fonselius, S., 1996. The upwelling of nutrients in the central Skagerrak. Deep Sea Res. Part II: Top. Stud. Oceanogr. 43 (1), 57–71. Gustafsson, B., Stigebrandt, A., 1996. Dynamics of the freshwater-influenced surface layers in the Skagerrak. J. Sea Res. 35, 39–53. Hansen, H.P., 1999. Determination of oxygen In: K.K.a.M.E. eds. K. Grasshoff (Editor), Methods of Seawater Analysis, third edition. Wiley–VCH. Germany Haraldsson, C., Anderson, L.G., Hassellöv, M., Hulth, S., Olsson, K., 1997. Rapid, highprecision potentiometric titration of alkalinity in the ocean and sediment pore waters. Deep-Sea Research., Part I 44, 2031–2044. Hjalmarsson, S., Wesslander, K., Anderson, L.G., Omstedt, A., Perttilä, M., Mintrop, L., 2008. Distribution, long-term development and mass balance calculation of total alkalinity in the Baltic Sea. Cont. Shelf Res. 28, 593–601. Johnson, K.M., Sieburth, J.M., Williams, P.J., Brandstrom, L., 1987. Coulometric total carbon dioxide analysis for marine studies: automation and calibration. Mar. Chem. 21, 117–133. LeBrasseur, R.J., McAllister, C.D., Parsons, T.R., 1979. Addition of nutrients to a lake leads to greatly increased catch of salmon. Environmental Conservation 6, 187–190. Lee, K., Millero, F.J., 1995. Thermodynamic studies of the carbonate system in seawater. Deep Sea Res., Part I 42, 2035–2061. Lorbacher, K., Dommenget, D., Niiler, P.P., Köhl, A., 2006. Ocean mixed layer depth: a subsurface proxy of ocean–atmosphere variability. J. Geophys. Res. 111. Mehrbach, C., Culberson, C.H., Hawley, J.E., Pytkowicz, R.M., 1973. Measurement of apparent dissociation constants of carbonic acid in seawater at atmospheric pressure. Limnol. Oceanogr. 18, 897–907. Nightingale, P.D., et al., 2000. In situ evaluation of air–sea gas exchange parameterizations using novel conservative and volatile tracers. Glob. Biogeochem. Cycles 14 (1), 373–387. Pettersson, K., 1991. Seasonal uptake of carbon and nitrogen and intracellular storage of nitrate in planktonic organisms in the Skagerrak. J. Exp. Mar. Biol. Ecol. 151 (1), 121–137. Pierrot, D., Lewis, E., Wallace, D.W.R., 2006. MS Excel Program developed for CO2 system calculations, ORNL/CDIAC-105. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tennessee. Redfield, A.C., Ketchum, B.H., Richards, F.A., 1963. The influence of organisms on the composition of seawater. In: Hill, M.N. (Ed.), The Sea, vol. 2. John Wiley, New York, pp. 26–77. Richardson, K., Rasmussen, B., Bunk, T., Mouritsen, L.T., 2003. Multiple subsurface phytoplankton blooms occurring simultaneously in the Skagerrak. J. Plankton Res. 25, 799–813. Rodhe, J., 1989. The large scale mixing and estuarine circulation in the Skagerrak; calculations from observations of the salinity and velocity fields. Tellus 41A, 436–446. Rodhe, J., 1996. On the dynamics of the large-scale circulation of the Skagerrak. J. Sea Res. 35 (1–3), 9–21. Rydberg, L., Haamer, J., Liungman, O., 1996. Fluxes of water and nutrients within and into the Skagerrak. J. Sea Res. 35, 23–38. Skjevik, A.-T., 2007. Växtplankton 2006, SMHI. Stigebrandt, A., 1991. Computations of oxygen fluxes through the sea surface and the net production of organic matter with application to The Baltic and adjacent seas. Limnol. Oceanogr. 36, 444–454. Wanninkhof, R., 1992. Relationship between gas exchange and wind speed over the ocean. J. Geophys. Res. 97, 7373–7381. Weiss, A., Kuss, J., Peters, G., Schneider, B., 2007. Evaluating transfer velocity–wind speed relationship using a long-term series of direct eddy correlation CO2 flux measurements. J. Mar. Syst. 66 (1–4), 130–139. Weiss, R., 1970. The solubility of nitrogen, oxygen, and argon in water and seawater. Deep-Sea Res. 17, 721–735. Weiss, R., Price, B.A., 1980. Nitrous oxide solubility in water and seawater. Mar. Chem. 8, 347–359.