For continental shelf regions, the long-term trend in sea surface carbon dioxide (CO₂) partial pressure (pCO₂) and rates of ocean acidification are not accurately known. Here, we investigate the decadal trend of observed wintertime pCO₂ as well as computed wintertime pH and aragonite saturation state (Ω_ar) in the northern North Sea, using the first decade long monthly underway data from a voluntary observing ship covering the period 2004–2015. We also evaluate how seawater CO₂ chemistry, in response to physical and biological processes, drives variations in the above parameters on seasonal and interannual timescales.

In the northern North Sea, pCO₂, pH, and Ω_ar are subject to strong seasonal variations with mean wintertime values of 375 ± 11 μatm, 8.17 ± 0.01, and 1.96 ± 0.05. Dissolved inorganic carbon is found to be the primary driver of both seasonal and interannual changes while total alkalinity and sea surface temperature have secondary effects that reduce the changes produced by dissolved inorganic carbon. Average interannual variations during winter are around 3%, 0.1%, and 2% for pCO₂, pH, and Ω_ar, respectively and slightly larger in the eastern part of the study area (Skagerrak region) than in the western part (North Atlantic Water region). Statistically significant long-term trends were found only in the North Atlantic Water region with mean annual rates of 2.39 ± 0.58 μatm/year, −0.0024 ± 0.001 year^-1, and −0.010 ± 0.003 year^-1 for pCO₂, pH, and Ω_ar, respectively. The drivers of the observed trends as well as reasons for the lack of statistically significant trends in the Skagerrak region are discussed.

Plain Language Summary

Temperate and high latitude marine shelf areas are generally net sinks of atmospheric carbon dioxide (CO₂), and they are experiencing ocean acidification. Decadal trends in the magnitude of the sinks and acidification occurring in these regions are not accurately known mainly due to limited time series and higher natural spatiotemporal variability compared to open oceans. Hence, an important question is whether the surface seawater CO₂ growth and acidification on the shelves can be predicted from atmospheric CO₂ increase as is the case for the open oceans? To contribute to the answer of this question, we compiled the first decade-long, monthly time series of surface seawater CO₂ and acidification parameters in the northern North Sea (2004–2015). Our analyses confirm that the area is a year-round CO₂ sink and further demonstrate its strong seasonal and interannual variations. In the western parts of the study area, we found wintertime trends that are statistically significant and similar to what is expected from atmospheric CO₂ increase and observed in the open ocean. In the eastern parts, seasonal and interannual changes were somewhat stronger, but wintertime trends were weaker and not statistically significant.

1 Introduction

The increase in the atmospheric carbon dioxide (CO₂) concentration drives anthropogenic net ocean CO₂ uptake and this leads to ocean acidification (OA, e.g., Royal Society, 2005). OA is currently receiving an intensive research attention because of its negative effects on many marine organisms including decreased survival, calcification, growth, development, and abundance (Kroeker et al., 2013).

The rates of OA vary between ocean regions. For the open ocean, where uptake of anthropogenic CO₂ from the atmosphere is the dominant driver, the OA rate over the last decades has been relatively well documented and understood (e.g., Bates et al., 2014). Observations from time series stations and volunteer observing ships (VOS) in different oceanic regions consistently show systematic changes in surface ocean chemistry resulting from OA. Specifically, long-term negative trends of pH and aragonite saturation state (Ω_ar) have been observed (e.g., Bates et al., 2014; Lauvset et al., 2015).

For coastal regions, observed rates of pH change differ from those expected from oceanic CO₂ uptake alone, as changes in other biogeochemical processes, related for example to changes in nutrient loading and eutrophication, are important as well (Burt et al., 2016; Cai et al., 2017; Clargo et al., 2015; Provoost et al., 2010; Wootton et al., 2008). Moreover, the long-term trend in atmosphere-ocean pCO₂ difference, the thermodynamic force of the CO₂ uptake from the atmosphere, is not accurately known for shelf regions. For instance, Wang et al. (2017) reported that in Northern Hemisphere marginal seas, the rate of change of surface ocean pCO₂ on decadal time scales (1.9 ± 1.6 μatm/yearr) closely follows the rate of atmospheric pCO₂ increase (1.90 ± 0.06 μatm/year), that is, the atmosphere-ocean difference is constant. Laruelle et al. (2018), on the other hand, used global data spanning a period of up to 35 years and found a tendency of increased uptake, that is, a seawater pCO₂ growth rate that lagged that in the atmosphere, in 80% of the investigated shelf regions for which a trend could be calculated. They also pointed out that trends appear highly variable both within the same shelf and across different shelf systems.

The North Sea is one of the best-studied shelf seas with respect to carbon cycle biogeochemistry (Bozec et al., 2005, 2006; Omar et al., 2010; Thomas et al., 2004, 2005, 2007). In the seasonally stratified northern North Sea, including the western Norwegian fjords, carbon is taken up by phytoplankton in the surface and respired in the subsurface layer (Thomas et al., 2004). The area is undersaturated with respect to atmospheric CO₂ throughout the year, driving a flux of CO₂ into the ocean (Omar et al., 2010; Omar et al., 2016). There are also large seasonal variations, driven by biology and mixing, temperature changes, and air-sea CO₂ exchange. The hydrography and biogeochemical properties of the northern North Sea are mainly influenced by two water inflows (e.g., Chierichi et al., 2017; Lee, 1980). In the eastern parts (east of 6°E), high input of less saline water from the Baltic Sea and from the Norwegian coast occurs and the low salinity (S) Skagerrak water (SW) flows northward along the Norwegian coast. In the western parts, the warm saline North Atlantic Water (NAW) enters from the north-west opening and extends as far south as about 55°N where during summer, it is overlain by a combination of fresher coastal water and/or SW. These regions will be henceforth referred to as the Skagerrak and NAW regions, respectively.

Interannual variations are significant in the North Sea, and their magnitude appears to be dependent on region and season, being generally enhanced in the southern North Sea and during summer/spring (Omar et al., 2010). Salt et al. (2013) reported that the North Atlantic Oscillation index influences the CO₂ system in the North Sea by regulating the inflow of water from North Atlantic and Baltic. During positive North Atlantic Oscillation index, higher rates of the above inflows lead to a limited mixing between the north and south, and this leads to a steeper gradient in pH and pCO₂ between south and north in the productive period.

The long-term trends are poorly quantified due to lack of proper long-term time series, but published low resolution repeated measurements indicate that the increase in atmospheric pCO₂ is driving a secular increase in surface water pCO₂ in the North Sea (Clargo et al., 2015; Omar et al., 2010; Thomas et al., 2007). However, observation-based OA estimates and studies considering decadal trends are still scarce.

Here, we investigate the decadal trend of wintertime pCO₂ as well as pH and Ω_ar in the northern North Sea for the period 2004–2015. We combine the underway data from a VOS with available multiyear, basin-wide station data from research cruises to facilitate a complete description of the seawater CO₂ chemistry as was described in Lauvset et al. (2015). We focus on analyses of pCO₂, Ω_ar, and pH values and evaluate their spatiotemporal variations and drivers. First, we present the mean distribution along the track of the VOS line to understand the seasonal and spatial patterns. Then we divide the data into NAW and Skagerrak regions to analyze interannual and long-term changes and their drivers. The long-term changes are determined during the winter season when the influence of biological activity is minimal and the computed trends are compared to those observed in the adjacent open ocean regions.

2 Material and Methods

2.1 Monthly Underway VOS Data

Underway measurements of sea surface pCO₂ and temperature (SST) were obtained on the containership M/V Nuka Arctica (henceforth Nuka; operated by Royal Arctic Line, www.ral.dk,nuka.uib.no). Nuka crosses the North Sea approximately every 3 weeks along a transect between 59.5–57.7°N heading southeast until 7°E, and then continues eastward until 10°E, where it turns southward before entering the port of Aalborg, Denmark (Figure 1). The measurement system was installed on Nuka in 2004. It was first installed in the bow thruster room, but was relocated in early 2006 to the engine room close to the middle of the ship, in order to reduce the amount of air bubbles in the seawater stream. The pCO₂ measurements on Nuka were first presented in Olsen et al. (2008) and are conducted using an instrument based on the design and recommendations of Pierrot et al. (2009). Briefly, the instrument uses a nondispersive infrared CO₂/H₂O gas analyzer (LI-COR 6262) to determine the CO₂ concentration in headspace air in equilibrium with a continuous stream of seawater. Headspace analyses are carried out every minute, and the instrument is calibrated roughly every 6 hr with three reference gasses, which are traceable to reference standards provided by National Oceanic and Atmospheric Administration/Earth System Research Laboratory. The LI-COR is calibrated approximately three times per day using N₂ (CO₂ = 0) and the standard gas with the highest CO₂ concentration. Data acquired between 57°N–61°N and 1.0°W–10.4°E are used for the analyses in this contribution. They were extracted from the SOCATv6 database (Bakker et al., 2014). The accuracy of the pCO₂ instrument on Nuka is approximately ±2 μatm.

**Figure 1**
Open in figure viewer PowerPoint

Map of the northern North Sea showing the tracks of the containership Nuka Arctica (red dots) overlaid by tracks for which satellite derived net primary productivity has been collocated with the ship measurements (gray dots). Also shown are stations covered during the cruises aboard R/V Pelagia (blue circles: 2005, 2008, and 2011) and R/V GO Sars (blue filled circles: 2015). The region enclosed by the dashed rectangle is referred to as the North Atlantic Water region whereas the region enclosed by the polygon is referred to as the Skagerrak region in the main text.

Sea surface S (SSS) data are measured on Nuka using a SBE-21 Seacat thermosalinograph. The data are calibrated following Alory et al. (2015), excluding data that are indicative of insufficient water flow rates. The remaining data are corrected based on in situ water samples, collected across the Atlantic and analyzed at the Grønland Naturinstitut, Nuuk, Greenland, and occasional close-by Argo data with quality flags “good.” Corrections are typically in the range of 0 to 0.6 psu.

2.2 Cruise Data

We supplemented the VOS data with station data acquired during four scientific cruises in the study area in the period 2005–2015. Three of these were late summer cruises conducted in the North Sea on R/V Pelagia on (17 August 2005–5 September 2005; 21 August 2008–7 September 2008; 1–25 September 2011). These data have been described in detail by Thomas et al. (2007), Bozec et al. (2005), Thomas et al. (2004), and Thomas (2002). Briefly, the Pelagia cruises covered the North Sea during all four seasons and obtained station data with a sampling resolution of approximately 1° by 1°. During each cruise, water column profile data for (but not limited to) dissolved inorganic carbon (DIC), total alkalinity (TA), S, and temperature were obtained. DIC and TA have been determined using the coulometric method according to Johnson et al. (1993) and a Gran potentiometric open cell titration, respectively. Uncertainty was 1–2 μmol/kg (≈0.8‰) for DIC and 2–3 μmol/kg (≈1‰) for TA. The 2005 and 2008 cruise data have been released as part of the GLODAPv2 data synthesis (Olsen et al., 2016). Similar water column carbonate measurements were also carried out in April 2015 on R/V GO Sars (Fröb et al., 2019). The 2015 cruise covered an east-west section perpendicular to the Norwegian west coast (59.3°N, 3–4°E; Figure 1, blue filled circles) and the data are available at https://cchdo.ucsd.edu/cruise/58GS20150410.

2.3 Satellite Derived Productivity Data

Monthly values for net primary production (NPP), based on the standard Vertical General Production Model which uses MODIS surface chlorophyll, SST, and photosynthetically active radiation (Behrenfeld & Falkowski, 1997) have been downloaded from the Oregon State University Ocean Productivity website. Negative values have been filtered out, and a subset of the data has been colocated with the data from Nuka (Figure 1).

2.4 Complete Description of the Seawater CO₂ System

To obtain a complete description of the seawater CO₂ system from the underway SST and pCO₂ data collected on Nuka, we used the frequently applied empirical approach that has been described in the literature (Lauvset et al., 2015; Nondal et al., 2009; Omar et al., 2010; Takahashi et al., 2014). The approach takes advantage of the strong correlation between TA and S in the ocean, which can be described with an empirical linear relationship of the form TA = a × S + b. Here, we used the cruise data sets to identify separate TA-S relationships, one from each cruise data set. The available cruise data were not sufficient to study seasonal changes in TA-S regressions in detail but was suitable to assess the interannual variability. Therefore, we derive separate TA-S relationships in order to be able to assess any interannual variations in the coefficients for the regressions and their significance for the computations (section 3.1). Furthermore, we have chosen to use full depth profile data (not only surface data) for the regressions because the North Sea is well mixed during winter (e.g., Thomas et al., 2004). Therefore we assume that data from deepest parts of the water column are representatives for winter conditions. The robustness of this assumption is tested during the validation of the TA-S regression (below). Moreover, the regression coefficients from the 2005 cruise were applied to the underway S data to compute TA values for 2004, 2005, and 2006; the coefficients from the 2008 cruise were used for computing TA values in 2007, 2008, and 2009; the coefficients from the 2011 cruise were used for TA values in 2010, 2011, and 2012; and finally, the coefficients from the 2015 cruise were used for TA values in 2013, 2014, and 2015.

The seawater CO₂ chemistry was fully described with the measured pCO₂ and the TA derived from S, using the dissociation constants of Millero et al. (2006) as implemented in CO2SYS (Lewis & Wallace, 1998; van Heuven et al., 2011). This choice of constants follows the recommendation of Salt et al. (2015), who investigated the internal consistency of the North Sea carbonate system. The KSO₄ dissociation constant of Dickson (1990) has been used and nutrient concentrations were set to zero during the above calculations. pH was determined on the total scale.

The TA-S relationship and the pH, DIC, and Ω_ar values estimated from calculated TA and pCO₂, were validated using TA measurements and pH and Ω_ar values computed from TA and DIC as measured on RV Pelagia cruises in November 2001 and February 2002, which has been described in Thomas et al. (2004). The 2001–2002 cruise data have not been used for deriving the TA-S relationships and so are fully independent. Furthermore, these winter data gave us a perfect opportunity to test the assumption that our TA-S relationship is valid for winter data since our regression coefficients were identified using profile data including subsurface samples with concentrations that resemble wintertime concentrations at the surface (2,214–2,339 μmol/kg). Thus, for the validation purpose, we disregarded the effect of interannual changes in the TA-S relationships (section 3.1) which is minimal during winter, and we used the mean values of the regression coefficients. The residuals (measured/computed-predicted) for TA, pH, DIC, and Ω_ar are shown in Figure S1 in the supporting information, and mean differences were −12.5 ± 9.3 μmol/kg for TA, −0.002 ± 0.002 for pH, −11 ± 8 μmol/kg for DIC, and −0.02 ± 0.02 for Ω_ar. Slightly higher residuals for midrange TA in the 2001 data (and corresponding increased residuals in lowest DIC and highest pH and Ω_ar values) were observed and probably were due to the aforementioned interannual variations in the TA-S relationship. By combining the maximum mean difference obtained from above residuals and typical uncertainties in pH values calculated from DIC-TA pairs (e.g., Omar et al., 2016), we estimated the total error associated with our computed pH and Ω_ar values to be ±0.01 and ±0.1, respectively. For TA and DIC, we will use a total error of ±16 μmol/kg and ±19 μmol/kg which we obtained by combining the mean offsets (13 and 11 μmol/kg) and the random errors (±9.3 and ±8 μmol/kg), respectively.

3 Results and Discussion

3.1 Identified Empirical Relationships

The relationships between TA and S obtained from the cruise data from 2005, 2008, 2011, and 2015 are presented in Table 1. Statistically significant (R² > 0.9 and p < 0.001) positive linear relationships were found with maximum root mean square error values that were within 3 times the state-of-the-art measurement uncertainty (2–3 μmol/kg, Dickson, 2010). The slopes and intercepts of the regression equations (y = ax + b) varied between the different years with mean and standard deviation values of a = 25.9 ± 1.4 μmol/kg and b = 1,414 ± 50 μmol/kg, respectively. The high intercept values indicate the presence of substantial inorganic carbon in the freshwater component (e.g., Olsson & Anderson, 1997) and are in agreement with earlier results from the area. Omar et al. (2010) reported an intercept of 1,817 ± 48 μmol/kg based on data from 2001 and 2002. Salt et al. (2013) investigated the effect of variable mixing between North Sea and SW by using data obtained in 2001, 2005, and 2008, and considering samples containing >3% of Baltic water mass fraction. They reported slopes and intercepts ranging between 20–27 μmol/kg and 1,390–1,590 μmol/kg, respectively, and argued that the interannual variability in the slopes and intercepts largely reflects changes in the water mass fractions rather than changes in the end-member concentrations. Despite the significant differences in the regression coefficients (Table 1), TA values predicted by the four regression equations were in agreement with each other to within total uncertainty of ±16 μmol/kg. Thus, sensitivity analyses showed that all the results presented in the subsequent sections could also be obtained by using the mean values of the coefficients. This means that the identified variability in the coefficients is not significant enough to impact our main findings.

Table 1. Coefficients for TA Versus S Regressions and Associated Statistics During Different Cruises

Cruise time (month, year)	Slope (μmol/kg)	Intercept (μmol/kg)	R²	p	#n	RMSE (μmol/kg)
Aug–Sep 2005	25.1 ± 0.4	1,438 ± 13	0.92	<0.005	392	7
Aug–Sep 2008	25.4 ± 0.4	1,432 ± 14	0.91	<0.005	379	8.9
Sep 2011	24.9 ± 0.4	1,447 ± 12	0.92	<0.005	425	7
May 2015	28.0 ± 0.8	1,339 ± 29	0.98	<0.005	30	3

Note. RMSE = root mean square error; S = salinity; TA = total alkalinity.

3.2 Seasonal and Spatial Patterns Along the Ship Track

Clear seasonal variations are observed in all variables; SST, NPP, pH, and Ω_ar all increase from winter to summer, while SSS and pCO₂ decrease (Figure 2). The patterns and magnitudes of the seasonal variations are similar to those reported earlier for the area (e.g., Omar et al., 2010; Salt et al., 2015; Thomas et al., 2005). The seasonal change in SST exceeds 10 °C, with coldest SST in February and March and warmest SST in August (Figure 2a). Furthermore, during winter and early spring, SST decreases eastward, but from late spring through fall the spatial SST gradient reverses as a result of decreased mixed layer throughout the study area combined with solar radiation input that increases eastward in the northern North Sea (Otto et al., 1990).

**Figure 2**
Open in figure viewer PowerPoint

(a) SST, (b) SSS, (c) pCO₂, (d) satellite derived NPP, (e) computed pH, (f) computed Ω_ar, as a function of longitude and month of the year. All data from 2004–2014 have been condensed into one virtual year and bin-averaged into 0.2 longitudes to resolve the spatial and seasonal variations. NPP = net primary production; SSS = sea surface salinity; SST = sea surface temperature.

The spatial variations of SSS show the main water masses in the study area: the saline NAW (SSS > = 35) is usually encountered in the western part of the transect (west of 5°E; i.e., in the NAW region; Figure 2b) and the fresher SW in the east (east of 6°E; i.e., in the Skagerrak region). Seasonal variations are stronger in the Skagerrak region, being fresher (SSS < 31) during summer (July) and more saline (SSS > 33) during winter (January). In the NAW region, SSS values of 34 appear between May and August.

pCO₂ is highest (360–400 μatm) during late fall (November) and early winter (December–January) and lowest (<280 μatm) during spring (March–May; Figure 2c). Furthermore, the low values during spring appear first in the eastern side of the study area and propagate westward while the increase of pCO₂ toward maximum winter values propagates eastward, except in the area east of 9°E, where pCO₂ values >300 μatm appear already in July. The high wintertime pCO₂ values stayed around the concurrent atmospheric pCO₂ values obtained at Mace Head, Ireland, (Bousquet et al., 1996) throughout the study period (Figure S2). This confirms that the area is a year-round sink for atmospheric CO₂ (Omar et al., 2010). The decrease during spring is due to biological carbon uptake by the spring phytoplankton bloom (Frankignoulle & Borges, 2001; Thomas et al., 2004). During the rest of the year, thermodynamic, remineralization, and mixing processes mainly control pCO₂ (these controls are quantified below).

The satellite derived NPP is high (>1,000 mg C·m⁻²·day⁻¹) between April and October with the most intense and longest bloom periods occurring in areas close to land (Figure 2d), that is, south of Norway (Figure 1). For instance, in the Skagerrak region, NPP > =2,000 mg C·m⁻²·day⁻¹ between April and October whereas further offshore, in the central northern North Sea, NPP reaches >1,000 mg C·m⁻²·day⁻¹ only in April–August. Furthermore, significant NPP occurs in the Skagerrak region even as early as March. If integrated over the whole year, the typically annual NPP values are around 200 g C/m² for the NAW region and 500 g C/m² for the Skagerrak region. However, for both regions, the estimated values are probably highly overestimated since the VGPM model has been shown to overestimate NPP in the North Atlantic by 100% due to high sensitivity to variations in temperature (e.g., Tilstone et al., 2015). In fact, model results of Moll (1997) reported annual NPP values of 100–125 g C·m²·year⁻¹ and 125–200 g C·m²·year⁻¹ in the NAW and Skagerrak regions, respectively (his Figure 3) which are about 100% lower than the above estimated annual NPP values.

**Figure 3**
Open in figure viewer PowerPoint

Monthly pH changes (∆pH) as observed (a, f) and expected due to sum of all derivers (b, g), SST changes (c, h), DIC changes (d, i), and by TA changes (e, j) in the North Atlantic Water region (left) and Skagerrak region (right). SST = sea surface temperature; DIC = dissolved inorganic carbon; TA = total alkalinity.

The spatial and seasonal patterns of the estimated pH are opposite to those observed in pCO_2, reflecting the close inverse relationship between these two variables. pH is lowest (8.05–8.08) during late fall and winter and highest (>8.15) during spring (Figure 2e).

The mean distribution of Ω_ar also shows a significant seasonal variation (Figure 2f). A notable feature is that the maximum Ω_ar (>2.5) occurs 1–2 months after the maximum pH and minimum pCO₂ (compare Figure 2c, 2e, and 2f). This despite the fact that the seasonal changes in all these variables (pCO₂, pH, and Ω_ar) are produced by the interplay of the phytoplankton spring bloom, summer warming/winter cooling, and freshwater input. The temperature sensitivity of Ω_ar is smaller compared to that for pH and pCO₂, and Ω_ar reaches its minimum values (≈1.5) during winter (January–March) when both pH and SST are low.

The Skagerrak region is characterized by stronger seasonal variations with much stronger spring-summer warming and freshening and higher NPP than the NAW region. The stronger seasonality in SST and SSS also contributes to stronger seasonality in Ω_ar in the Skagerrak region while for pCO₂ and pH, the seasonality is comparatively uniform across the longitudinal transect.

The controls on the seasonal pCO₂ variability in the study area were investigated previously by Omar et al. (2010) using a method that quantifies the effect of driving processes (Olsen et al., 2008). They showed that pCO₂ in the northern North Sea is controlled by (in a decreasing order) mixing and biology, changes in SST, and air-sea CO₂ exchange. Here, we use a different decomposition method that quantifies the effect of driving variables (DIC, TA, SST, and SSS) on the monthly changes of observed pCO₂ and computed pH and Ω_ar to gain more insight into the parameters governing the seasonal variations and their relative importance. Further, since the controls of pCO₂ and pH are identical (below), in the following, we will only discuss decomposition results for pH and Ω_ar. Furthermore, in order to accommodate for the observed different water masses and seasonality (section 3.1), we analyze results for the NAW and Skagerrak regions separately.

For pH, we used the decomposition method described in Lauvset et al. (2015) to quantify the importance of different drivers. This method first estimates the monthly pCO₂ changes expected from changes in SST, SSS, DIC, and TA as well as their sum. It then converts the pCO₂ changes to pH. The results are shown in Figures 3a–3j (SSS had a negligible effect on the seasonal pH variations and is therefore not shown). For the NAW region, DIC is the most important driver followed by SST and TA (Figures 3c–3e). The seasonal variability in DIC causes pH to increase during the first half of the year and to decrease during the second half of the year, whereas the SST effect primarily opposes that of DIC, that is, it causes pH to decrease during February–July and to increase during August–January. The influence of DIC during spring and fall outweighs the effect of SST, and pH increases due to decreasing SST in December–February and due to decreasing DIC in March–April. The increased importance of TA during spring and fall is evident from Figures 3e and 3j. During spring, larger TA decrease is produced by increased freshwater input from the Baltic and Scandinavian rivers. During fall, larger TA increase is produced by S increase (due to the erosion of the mixed layer). Both these effects are more pronounced in the Skagerrak where changes of freshwater input are biggest. The effects of SST and TA combined are nearly equal to but opposite to that of DIC (Figures 3c–3e). As a result, the sum of all computed effects is small (<0.05 pH units). Note also that the decomposition model is able to account for the observed seasonal changes, that is, the sum of the computed effects compares well with the observed amplitudes (Figure 3a).

The seasonal dynamics and controls on pH in the Skagerrak region are similar to those in the NAW region (described above), but the effect of SST, DIC, and TA on the monthly pH changes are enhanced in the Skagerrak region (Figures 3h–3j). This especially holds for TA, which has an up to five times stronger influence on pH in the Skagerrak region compared to the NAW region. Note also that the TA control is larger than that of SST during spring and fall seasons. Again, the effects of SST and TA combined are nearly equal to, but opposite to, that of DIC (Figures 3h–3j), and the sum of all effects is small (<0.06 pH units).

It must be noted that above quantified drivers of pCO₂ and pH also indicate the importance of different processes throughout the year. For instance, the strong DIC and SST controls during spring/summer confirm that photosynthesis, which takes up carbon from the surface water, is the dominant process followed by the effect of spring/summer warming, which has direct increasing/decreasing effect on pCO₂/pH, respectively. During fall, on the other hand, DIC and SST are dominant controls because upward mixing of colder CO₂-rich deeper water (e.g., Thomas et al., 2005) increases DIC while falling temperatures have direct decreasing/increasing effect on pCO₂/pH.

We decomposed the seasonal changes of Ω_ar according to:

$urn:x-wiley:21698953:media:jgrg21490:jgrg21490-math-0001$ (1)

where dΩ_ar/dDIC and dΩ _ar/dTA have been determined from the analytical expressions given by Egleston et al. (2010, see their Table 1), and dΩ_ar/dSST and dΩ_ar/dSSS have been estimated for the present data using the CO2SYS program (Lewis & Wallace, 1998). During these estimations, DIC and TA were held at their mean values, and either SST or SSS was varied between −2 and 25 °C or 30 and 36 °C, respectively. The values obtained from equation (1) are shown on Figure 4. The seasonal changes in DIC and TA are the most important drivers for changes in Ω_ar whereas the effects of variations of SST and SSS (not shown) on Ω_ar are negligible. This is in agreement with Omar et al. (2016), who analyzed data from the southwestern Norwegian coast using a different method and concluded that seasonal changes in DIC and TA are the most important drivers for changes in Ω_ar, and variations of SST and SSS are less relevant. Again, we note that the TA influence is enhanced in the Skagerrak region compared to the NAW region. We also note that the TA control outweighs that of DIC in July in the Skagerrak region (to be explained shortly). In both regions, Ω_ar increases between February and July primarily due to decreasing DIC driven by biological activity, while it decreases between September and December primarily due to increasing DIC driven by vertical mixing and respiration of organic matter. TA tends to have a decreasing effect on Ω_ar in the first half of the year due to decreasing TA that result from freshwater input throughout later winter (March) and midsummer (July). During the rest of the year, TA increases due to increasing SSS, and this has an increasing effect on Ω_ar. This also demonstrates how S variations indirectly impact Ω_ar through TA and DIC both of which directly covary with SSS. The effect of this indirect SSS control (through TA and DIC) on Ω_ar partially cancels out because DIC and TA have opposing effects on Ω_ar. Firstly, the sensitivity of Ω_ar to DIC and TA changes are opposite and nearly equal (i.e., dΩ_ar/dDIC and dΩ_ar/dTA in equiation (1) have equal magnitudes and opposite signs). Secondly, for any given change in SSS, the resulting TA change is about 30% larger than the corresponding DIC change so that a net impact on Ω_ar would be expected. However, monthly DIC changes result from several processes and are almost always larger than TA changes. Hence, the indirect net effect from SSS changes (through TA) is outweighed by larger DIC influence from processes other than SSS changes (compare Figures 4d and 4e and 4i and 4j). An exception is in July when both DIC and TA increase, but disproportionally; TA increases due to increasing S, but DIC does not keep up probably due to biological uptake neutralizing the effect of increasing SSS (Figure S4). Hence, Ω_ar increases due to increasing TA in July. This is particularly clear for the Skagerrak region (Figures 4i and 4j). In short, DIC is the primary driver of the seasonal changes in Ω_ar, but TA changes have secondary effects that reduce the amplitude of the seasonal Ω_ar changes produced by DIC changes except in July when TA exerts the strongest control on Ω_ar (see Figure 4 and Figure S4).

**Figure 4**
Open in figure viewer PowerPoint

As in Figure 3 but for monthly Ω_ar changes. Note different scales in y axis. SST = sea surface temperature; DIC = dissolved inorganic carbon; TA = total alkalinity.

3.3 Interannual Variations, Trends, and Controls

The interannual variations were determined as percentage standard deviations associated with each monthly mean over the study period; 2004–2015 for the NAW region and 2004–2012 for the Skagerrak. The reason for the different lengths is due to the fact that TA measurements of the cruise in 2015 were available only for the NAW region. Thus, for the Skagerrak region, complete CO₂-system parameters could not be computed from the underway data for the years 2013–2015. The possible effect of these different time series lengths on the computed trends is discussed at the end of the section. The monthly means were deseasonalized by subtracting the long-term mean of the respective month (e.g., Tjiputra et al., 2014) before the computation of the standard deviations. The computed interannual variations are generally between 3–15% for pCO₂ and Ω_ar and less than 10% for all other parameters (Figure 5) with the exception of late winter/spring SSTs in the Skagerrak, which show higher interannual variations of 15–35%. Generally, the Skagerrak region shows higher interannual variations except during spring when pCO₂, pH, and Ω_ar variations are 1.5–3 times larger in the NAW region. The lowest relative interannual changes are observed for pH (0.3–0.4%) and SSS (<5%) in the NAW region. Interannual SSS changes in the Skagerrak region (2–10%) are much larger than those for the NAW region. For all variables (except SST in the Skagerrak), interannual variability is lowest during winter and fall and highest during spring, a finding that is in good agreement with the results of Omar et al. (2010). Winter time pCO₂ interannual changes were around 3% both in the NAW and Skagerrak regions, and the decomposition analyses we have performed (not shown) revealed that the interannual changes in pCO₂, pH, and Ω_ar have the same drivers (in parameter and magnitude) as the seasonal changes. This means that interannual variations are primarily driven by DIC changes counteracted by the combined effects of TA and SST changes, while SSS has the least importance. Furthermore, interannual changes are highest during spring (April and May) mainly due to enhanced DIC and TA changes that are slightly phase shifted to each other, that is, counteracting each other to a smaller degree compared to the winter situation.

**Figure 5**
Open in figure viewer PowerPoint

Interannual changes (standard deviations relative to the long-term monthly averages) in SST, SSS, pCO₂, pH, and Ω_ar as a function of month of the year. Also shown are the number of monthly averages available in each month for all the above variables in the observation period (2004–2015 for North Atlantic Water region and 2004–2012 for the Skagerrak region). SSS = sea surface salinity; SST = sea surface temperature

For the determination of the trends, we used winter (December–February) data for the NAW region and (December–January) data for the Skagerrak region. We choose not to include March (and February for the Skagerrak) because significant primary production takes place in these months, especially in the Skagerrak region (Figure 2d). The computed trends superimposed on the long-term wintertime mean values (2003–2015 for the NAW region and 2003–2012 for the Skagerrak region) are shown in Figure 6. The superposition of the long-term mean values was convenient in order to give an idea about the temporal development of mean values in the study area. Also, note that data for winter 2003 were actually acquired in January/February 2004, that is, winter 2003 means winter 2003/2004 and so on.

**Figure 6**
Open in figure viewer PowerPoint

Long-term trends in winter-time pCO₂ (a and d), pH (b and e), and Ω_ar (c and e) superimposed on the long-term winter mean values as a function of time (years). For the computation of the trends monthly means were deseasonalized by subtracting the long-term mean of the respective months (see section 3.3 of the main text). Also shown are the best fit (solid) and 95% confidence lines (dotted). Note the different ranges in both axes. NAW = North Atlantic Water.

Statistically significant long-term trends, that is, with p < 0.05 are evident in the NAW region for pCO₂, pH, and Ω_ar (Figures 6a–6c and Table 2). The observed trends of increasing pCO₂ and decreasing pH and Ω_ar are consistent with the effects of anthropogenic CO₂ uptake by the surface ocean leading to OA. The trends are also within the range of the global atmospheric pCO₂ growth rate (2.0 μatm/year), and the OA expected from it (0.0018 ± 0.0002 pH unit per year (e.g., Lauvset et al., 2015), considering the uncertainties. Furthermore, decomposing the drivers of the observed trends indicates that the decreasing pH trend in the NAW originates almost entirely from increasing DIC while SSS has negligible influence (Figures 7a and 7b). TA and SST do have significant effects in driving the pH trend, but these effects are much smaller than that of DIC (Figures 7a and 7b). The same drivers control the pCO₂ growth rate with impact strengths equal to those shown for pH but with opposite sign. Similarly, the decomposition of the Ω_ar trends shows that DIC and TA are the most important drivers and the effect of TA is much smaller than that of DIC (Figures 7c and 7d). It must be emphasized that the sensitivity of Ω_ar to changes in SST, SSS, DIC, and TA, that is, the coefficients of equation (1) is of the same magnitude, but the trend in Ω_ar driven by DIC dominates, because the trend in DIC (1.2 ± 0.3 μmol/kg/year) is numerically up to an order of magnitude larger than those in SST, SSS, and TA (Table 2).

Table 2. Computed Trends in Wintertime SST, SSS, pCO₂, pH, and Ω_ar in 2004–2015 for the North Atlantic Water and for the Skagerrak Regions of the North Sea

Region	Parameter (unit)	Trend	Standard error (±)	R²	p	#Obs.
North Atlantic Water	SST (°C/year)	−0.004	0.040	<0.1	0.92	11
	SSS (year⁻¹)	0.007	0.011	<0.1	0.57	11
	pCO₂ (μatm/year⁻¹)	2.39	0.58	0.66	<0.01	11
	pH (year⁻¹)	−0.0024	0.001	0.70	<0.01	11
	Ω_ar (year⁻¹)	−0.010	0.003	0.59	<0.01	11
	DIC (μmol·kg⁻¹·year⁻¹)	1.2	0.3	0.59	<0.01	11
	TA (μmol·kg⁻¹·year⁻¹)	0.1	0.3	<0.1	0.68	11
Skagerrak	SST (°C/year)	−0.026	0.153	<0.1	0.87	9
	SSS (year⁻¹)	0.037	0.099	<0.1	0.72	9
	pCO₂ (μatm/year⁻¹)	1.2122	1.5440	<0.1	0.46	9
	pH (year⁻¹)	−0.0014	0.0014	0.1	0.34	9
	Ω_ar (year⁻¹)	−0.0025	0.0088	<0.1	0.79	9
	DIC (μmol·kg⁻¹·year⁻¹)	1.3	1.6	<0.1	0.44	9
	TA (μmol·kg⁻¹·year⁻¹)	1.0	2.4	<0.1	0.68	9

Note. The trends are computed from December–February data for the North Atlantic Water region and from December–January data for the Skagerrak (see section 3.3 of the main text). DIC = dissolved inorganic carbon; pCO₂ = carbon dioxide partial pressure; SSS = sea surface salinity; SST = sea surface temperature; TA = total alkalinity; Ω_ar = aragonite saturation state.

**Figure 7**
Open in figure viewer PowerPoint

Long-term pH (a) and Ω_ar (c) annual changes as observed (black bars) and as expected due to sum of all drivers (gray bars), SST changes, DIC changes, and by TA changes in the NAW region. (b) and (d) as above but for the Skagerrak region. Note that the scales in the y axis are different on the Ω_ar panels (c and d). The error bars indicate the uncertainties associated with the estimated changes. The individual errors are calculated by using the errors of the respective trends (see Table 2) and the total error was calculated as the square root of the sum of squares of the individual errors. NAW = North Atlantic Water; SST = sea surface temperature; DIC = dissolved inorganic carbon; TA = total alkalinity.

The pCO₂ trend in the NAW region is in general agreement with that reported for the adjacent subpolar North Atlantic where Lauvset and Gruber (2014) reported a seawater CO₂ fugacity (fCO₂) and pH trends of 2.0 ± 0.38 μatm/year and −0.0022 ± 0.0004 pH units per year, respectively, for the period 1981–2007. Newer wintertime fCO₂ data collected from the subpolar North Atlantic aboard Nuka in 2004–2016 show fCO₂ growth rates that increase from the Irminger Sea in the west (1.96 ± 0.27 μatm/year) to the Faeroe Bank Channel in the east (2.27 ± 0.14 μatm/year), mainly driven by DIC (Fröb et al., 2018). Thus, the computed wintertime pCO₂ trend in the NAW region is similar to that observed in the adjacent open ocean. It should be noted, however, that the summer trend (or annual mean trend) in the NAW region may be different from the value determined in this study.

Laruelle et al. (2018) summarized two main mechanisms which have been described in the literature to explain the evolution of the continental shelf CO₂ sink. One mechanism relies on the timescales of air–water and shelf–open ocean exchanges of CO₂, the other on the stimulation of the biological pump. According to the former mechanism, if shelf–open ocean exchanges of CO₂ are unable to keep up with the increasing air–sea flux of anthropogenic CO₂, CO₂ may accumulate in shelf waters so that seawater pCO₂ increase follows the atmosphere due to this bottleneck in offshore transport. In that respect, the fact that the pCO₂ trend in the NAW region closely tracks the atmosphere indicates that the accumulation of anthropogenic CO₂ in this region occurs faster than it is being transported to the open ocean.

In the Skagerrak region, the long-term trends are weaker (Table 2, second part) and statistically insignificant (R² < 0.2 and p > 0.3). Therefore, it is fair to report that the Skagerrak region does not show any clear trends in wintertime pCO₂ and OA variables during the period 2004–2012. Nevertheless, the decomposition shows that TA and DIC are the major drivers of the annual changes, and they have opposite and equal effects on pH (Figure 7b). Furthermore, SSS and SST have decreasing effects on pH, but these are much smaller effects compared to the controls of TA and DIC (Figure 7b). We also note that the magnitudes of drivers are larger in the Skagerrak compared to the NAW region. This is especially true for SSS and TA drivers which are six and nine times stronger than those in NAW. In short, SSS and TA increases counteract the effect of DIC increase, and the overall trends are weaker in the Skagerrak compared to the NAW region.

The reason for the absence of statistically significant trends in the Skagerrak region is not the main subject of this study and will not be treated comprehensively. However, few plausible explanations are discussed in the following. First, we analyzed shorter time series for the Skagerrak region since we focused on the computed OA parameters that were available only until 2012 (cruise TA data were available only until 2011 for this region, see Figure 1). However, this is not the sole reason for absence of significant trends in the Skagerrak region because the NAW region presents more clear trends even if time series of equal lengths are analyzed for the two regions. For instance, if the time series of the NAW region is shortened to 8 years, the resulting trends would still have p-values ranging from 0.03 to 0.09 which are much lower than those for the Skagerrak region (Table 2). On the other hand, including all wintertime pCO₂ data measured in the Skagerrak region between 2004 and 2015 result in a pCO₂ trend that is statistically insignificant (R² < 0.3 and p = 0.07) and highly uncertain (2.4 ± 1.2 μatm/year).

Second, both DIC and TA in the Skagerrak region showed increasing trends (although the TA trend was not statistically significant, p = 0.68), but the effect of the increasing DIC on pCO₂ growth and OA parameters has been effectively counteracted by the effect of increasing TA (Figure 7b). Thus it seems that the oceanic pCO₂ growth and OA trend expected in this region due to the increased atmospheric CO₂ are counteracted by TA increase. In fact, Wesslander et al. (2010) reported that TA in both Kattegat and the central Baltic Sea has increased over the 1993–2009 period probably due to internal change in the Baltic Sea circulation. It is therefore conceivable that increased TA in the Baltic outflow has affected the Skagerrak region as well. As mentioned above, the current TA data obtained between 2005 and 2012 also indicate an increasing trend in the Skagerrak region. However, this TA trend must be interpreted with care because it is not statistically significant, and the TA data have been estimated from S. This highlights the necessity of a measured TA time series in the area.

A third possible explanation is that the maximal pCO₂ (and lowest pH) values are controlled by different processes in the two regions. In the NAW region, the highest pCO₂ values are controlled by upward mixing of CO₂-rich deep water and uptake of atmospheric CO₂. This brings the wintertime surface seawater to near equilibration with the atmospheric CO₂ level (e.g., Omar et al., 2010; Thomas et al., 2005). The December–January disequilibrium (seawater pCO₂–atmospheric pCO₂) in the NAW region ranged between 2 and −27 μatm with a mean value of −12 ± 8 μatm. In the Skagerrak region, on the other hand, the average December–January pCO₂ disequilibrium was larger (−21 ± 8 μatm) compared to the NAW, ranging between −6 and −35 μatm. The larger disequilibrium in the Skagerrak region was specially enhanced in 2004 and 2007–2009 (Figure S3) when winter pCO₂ values were on average 27 μatm lower than atmospheric pCO₂. This indicates that wintertime pCO₂ in the Skagerrak region is probably governed by the combination of biological activity, sediment water interaction, and mixing (both lateral and vertical). In other words, our assumption that “the influence of biological activity is negligible” during wintertime is probably not fully met in the Skagerrak region. In particular, two ways biological activity in wintertime can conceal trends are either by introducing elevated interannual variability or by including trends that counteract the effect of C_ant. The latter is more plausible since wintertime interannual variability is only slightly higher in the Skagerrak compared to the NAW region (Figure 5). One particular mechanism, which involves all the aforementioned processes and which has been reported for the southern North Sea (Burt et al., 2016) is the denitrification of the allochthonous nitrate in near-coastal regions, which creates an increase in the standing stock of TA, strongly buffering the effect of OA (e.g., Borges & Gypens, 2010). However, denitrification is not a particularly dominant process of organic matter degradation in the Skagerrak area (Rysgaard et al., 2001).

In any case, a more complete understanding of the absence of statistically significant trends in the Skagerrak region is important since it is a prerequisite for predicting the future development of CO₂ sink and OA in the area. A first step toward this goal is to investigate the relative importance of the solubility and biological pumps in the development of the CO₂ system in the Skagerrak on decadal time scales. The resolution of this issue requires not only a sustained measurement program, but also measurements that integrate physical, biological, and chemical data sets. This needs to be taken into account by monitoring programs such as the Ocean Acidification monitoring project, funded by the Norwegian Environmental Agency, which covers areas including the Skagerrak region. In the 2016 report from this monitoring project, Chierichi et al. (2017) analyzed 5 years of data of the CO₂ system, temperature, and S, obtained from repeated stations in the Skagerrak. They found significant pH and Ω_ar trends only in the deepest part of the water column (below 600 m). In the surface water, on the other hand, no statistically significant trends were observed between 2011 and 2016 in agreement with our results. Furthermore, Chierichi et al. (2017) concluded that longer time series were needed to investigate the long-term trends in the area, and they recommended more comprehensive monitoring, that is, include integrated measurements and studies of primary production, ocean physics, freshwater supplies, and land-sea exchanges. Based on the results of our analyses, we support these recommendations and suggest that in addition to time series of sufficient length, the measured variables and analyses methods also need to be widened. This will enable accounting for the effects of physical and biological process on the long-term development of CO₂-system parameters in the Skagerrak region. Furthermore, such an extended data set would give us opportunity to determine trends for all seasons and not only during winter.

3.4 Conclusions and Further Comments

A new, decade-long, time series enabled us to investigate the variability in surface seawater pCO₂ and acidification parameters in the northern North Sea (2004–2015) on seasonal, interannual, and decadal time scales. The area is found to be a year-round CO₂ sink and subject to strong seasonal changes and intermediate interannual variations. Regarding the decadal development of pCO₂ and OA parameters during winter (the season with minimal interannual variability and biological activity), the study area shows a clear west-east division in accordance with the well-known high spatial heterogeneity in shelf regions. In the western part (the NAW region), wintertime trends are statistically significant and similar to what is expected from atmospheric CO₂ increase and observed in the adjacent open-ocean region. This indicates that the surface waters in this shelf region fully equilibrate with atmospheric CO₂ perturbation before they are transported to the open ocean.

In the eastern parts (the Skagerrak region), wintertime trends are weaker and statistically insignificant. The exact reasons for this are not known, but interannual variability, which could entail longer times of emergence for trends, seem to play a minor role. It is thus likely that compensating trends (e.g., in biological activity, sediment water interaction, mixing, circulation, etc.) are concealing the anthropogenic trends in CO₂-system variables in the Skagerrak region. In order to understand and disentangle these trends for all seasons of the year, there is a need for sustained measurements incorporating physical, biological, and chemical parameters.

Acknowledgments

This work has received funding from the European Union's Horizon 2020 research and innovation program (654462), the Research Council of Norway (245972), and the Norwegian Environment Agency (258608). We are grateful for the technical assistance provided by Tor de Lange, Kristin Jackson, and Sigve Naustdal. We are also grateful for two anonymous referees whose comments and suggestions improved the manuscript. This work would not have been possible without the generosity and help of the liner company Royal Arctic Line AS and the captains and crew of MS Nuka Arctica. We thank INSU for support to SNO SSS and LEGOS/OMP for validation and archiving of the data of the thermosalinograph. Helmuth Thomas acknowledges support by the German Academic Exchange Service DAAD (57429828) from funds of the German Federal Ministry of Education and Research (BMBF). Data used in this study are accessible at SOCAT (https://www.nodc.noaa.gov/ocads/oceans/glodap/GlopDV.html http://www.socat.info/ ), GLODAP v2 (https://www.nodc.noaa.gov/ocads/oceans/glodap/GlopDV.html ), and the Ocean Productivity website http://orca.science.oregonstate.edu/1080.by.2160.monthly.xyz.vgpm.m.chl.m.sst.php .

Supporting Information

References

Alory, G., Delcroix, T., Téchiné, P., Diverrés, D., Varillon, D., Cravatte, S., Gouriou, Y., Grelet, J., Jacquin, S., Kestenare, É., Maes, C., Morrow, R., Perrier, J., Reverdin, G., & Roubaud, F. (2015). The French contribution to the voluntary observing ships network of sea surface salinity. Deep Sea Research Part I: Oceanographic Research Papers, 105, 1– 18. https://doi.org/10.1016/j.dsr.2015.08.005
CrossrefADSWeb of Science®Google Scholar
Bakker, D. C. E., Pfeil, B., Smith, K., Hankin, S., Olsen, A., Alin, S. R., Cosca, C., Harasawa, S., Kozyr, A., Nojiri, Y., O'Brien, K. M., Schuster, U., Telszewski, M., Tilbrook, B., Wada, C., Akl, J., Barbero, L., Bates, N. R., Boutin, J., Bozec, Y., Cai, W. J., Castle, R. D., Chavez, F. P., Chen, L., Chierici, M., Currie, K., de Baar, H. J. W., Evans, W., Feely, R. A., Fransson, A., Gao, Z., Hales, B., Hardman-Mountford, N. J., Hoppema, M., Huang, W. J., Hunt, C. W., Huss, B., Ichikawa, T., Johannessen, T., Jones, E. M., Jones, S. D., Jutterström, S., Kitidis, V., Körtzinger, A., Landschützer, P., Lauvset, S. K., Lefèvre, N., Manke, A. B., Mathis, J. T., Merlivat, L., Metzl, N., Murata, A., Newberger, T., Omar, A. M., Ono, T., Park, G. H., Paterson, K., Pierrot, D., Ríos, A. F., Sabine, C. L., Saito, S., Salisbury, J., Sarma, V. V. S. S., Schlitzer, R., Sieger, R., Skjelvan, I., Steinhoff, T., Sullivan, K. F., Sun, H., Sutton, A. J., Suzuki, T., Sweeney, C., Takahashi, T., Tjiputra, J., Tsurushima, N., van Heuven, S. M. A. C., Vandemark, D., Vlahos, P., Wallace, D. W. R., Wanninkhof, R., & Watson, A. J. (2014). An update to the Surface Ocean CO2 Atlas (SOCAT version 2). Earth System Science Data, 6(1), 69– 90. https://doi.org/10.5194/essd-6-69-2014
CrossrefADSWeb of Science®Google Scholar
Bates, N. R., Asto, Y., Church, M., Currie, K., Dore, J., González-Dávila, M., Lorenzoni, L., Muller-Karger, F., Olafsson, J., & Santana-Casiano, J. (2014). A time-series view of changing ocean chemistry due to ocean uptake of anthropogenic CO2 and ocean acidification. Oceanography, 27(1), 126– 141. https://doi.org/10.5670/oceanog.2014.16
CrossrefWeb of Science®Google Scholar
Behrenfeld, M. J., & Falkowski, P. G. (1997). Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnology and Oceanography, 42, 1– 20.
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Borges, A. V., & Gypens, N. (2010). Carbonate chemistry responds more strongly to eutrophication than ocean acidification in the coastal zone. Limnology and Oceanography, 55(1), 346– 353. https://doi.org/10.4319/lo.2010.55.1.0346
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Bousquet, P., Gaudry, A., Ciais, P., Kazan, V., Monfray, P., Simmonds, P. G., Jennings, S. G., & O'Connor, T. C. (1996). Atmospheric CO2 concentration variations recorded at Mace Head, Ireland, from 1992 to 1994. Physics and Chemistry of the Earth, 21(5-6), 477– 481. https://doi.org/10.1016/S0079-1946(97)81145-7
CrossrefADSGoogle Scholar
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 observations. Marine Chemistry, 93(2-4), 131– 147. https://doi.org/10.1016/j.marchem.2004.07.006
CrossrefCASWeb of Science®Google Scholar
Bozec, Y., Thomas, H., Schiettecatte, L.-S., Borges, A. V., Elkalay, K., & de Baar, H. J. W. (2006). Assessment of the processes controlling the seasonal variations of dissolved inorganic carbon in the North Sea. Limnology and Oceanography, 51(6), 2746– 2762. https://doi.org/10.4319/lo.2006.51.6.2746
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Burt, W. J., Thomas, H., Hagens, M., Pätsch, J., Clargo, N. M., Salt, L. A., Winde, V., & Boettcher, M. E. (2016). Carbon sources in the North Sea evaluated by means of radium and stable carbon isotope tracers. Limnology and Oceanography, 61(2), 666– 683. https://doi.org/10.1002/lno.10243
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Cai, W. J., Huang, W. J., Luther, G. W., Pierrot, D., Li, M., Testa, J., Xue, M., Joesoef, A., Mann, R., Brodeur, J., Xu, Y. Y., Chen, B., Hussain, N., Waldbusser, G. G., Cornwell, J., & Kemp, W. M. (2017). Redox reactions and weak buffering capacity lead to acidification in the Chesapeake Bay. Nature Communications, 8(1), 369. https://doi.org/10.1038/s41467-017-00417-7
CrossrefADSPubMedWeb of Science®Google Scholar
Chierichi, M., Skjelvan, I., Norli, M., Jones, E., Børsheim, K. Y., Lauvset, S. K., Lødemel, H. H., Sørensen, K., & Johannessen, T.. (2017). Overvåking av havforsuring i norske farvann i 2016, Rapport, Miljødirektoratet, M-776. (in Norwegian).
Google Scholar
Clargo, N. M., Salt, L. A., Thomas, H., & de Baar, H. J. W. (2015). Rapid increase of observed DIC and pCO2 in the surface waters of the North Sea in the 2001–2011 decade ascribed to climate change superimposed by biological processes. Marine Chemistry, 177, 566– 581. https://doi.org/10.1016/j.marchem.2015.08.010
CrossrefCASWeb of Science®Google Scholar
Dickson, A. G. (1990). Standard potential of the reaction: AgCl(s) + 1/2H2(g) = Ag(s) + Cl (aq), and the standard acidity constant of the ion HSO4—In synthetic seawater from 273.15 to 318.15 K. The Journal of Chemical Thermodynamics, 22(2), 113– 127. https://doi.org/10.1016/0021-9614(90)90074-Z
CrossrefCASWeb of Science®Google Scholar
Dickson, A. G. (2010). The carbon dioxide system in seawater: Equilibrium chemistry and measurements. In U. Riebesell, V. J. Fabry, L. Hansson, & J.-P. Gattuso (Eds.), Guide to best practices for ocean acidification research and data reporting (Chap. 1, pp. 17– 40). Luxembourg: Publications Office of the European Union.
Google Scholar
Egleston, E. S., Sabine, C. L., & Morel, F. M. M. (2010). Revelle revisited: Buffer factors that quantify the response of ocean chemistry to changes in DIC and alkalinity. Global Biogeochemical Cycles, 24, GB1002. https://doi.org/10.1029/2008GB003407
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Frankignoulle, M., & Borges, A. V. (2001). The European continental shelf as a significant sink for atmospheric carbon dioxide. Global Biogeochemical Cycles, 15, 569– 576. https://doi.org/10.1029/2000GB001307
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Fröb, F., Olsen, A., Becker, M., Chafik, L., Johannessen, T., Reverdin, G., & Omar, A. (2019). Wintertime fCO₂ variability in the subpolar North Atlantic since 2004. Geophysical Research Letters, 6. https://doi.org/10.1029/2018GL080554
Google Scholar
Johnson, K. M., Wills, K. D., Butler, D. B., Johnson, W. K., & Wong, C. S. (1993). Coulometric total carbon dioxide analysis for marine studies: Maximizing the performance of an automated gas extraction system and coulometric detector. Marine Chemistry, 44(2-4), 167– 187. https://doi.org/10.1016/0304-4203(93)90201-X
CrossrefCASWeb of Science®Google Scholar
Kroeker, K. J., Kordas, R. L., Crim, R. N., Hendriks, I. E., Ramajo, L., Singh, G. G., Duarte, C. M., & Gattuso, J. P. (2013). Impacts of ocean acidification on marine organisms: Quantifying sensitivities and interaction with warming. Global Change Biology, 19(6), 1884– 1896. https://doi.org/10.1111/gcb.12179
Wiley Online LibraryADSPubMedWeb of Science®Google Scholar
Laruelle, G. G., Cai, W. J., Hu, X., Gruber, N., Mackenzie, F. T., & Regnier, P. (2018). Continental shelves as a variable but increasing global sink for atmospheric carbon dioxide. Nature Communications, 9(1), 11. https://hdl.handle.net/10.1038/s41467-017-02738-z
CrossrefADSPubMedWeb of Science®Google Scholar
Lauvset, S., Gruber, N., Landschützer, P., Olsen, A., & Tjiputra, J. (2015). Trends and drivers in global surface ocean pH over the past 3 decades. Biogeosciences, 12(5), 1285– 1298. https://doi.org/10.5194/bg-12-1285
CrossrefCASADSWeb of Science®Google Scholar
Lauvset, S. K., & Gruber, N. (2014). Long-term trends in surface ocean pH in the North Atlantic. Marine Chemistry, 162, 71– 76. https://doi.org/10.1016/j.marchem.2014.03.009
CrossrefCASWeb of Science®Google Scholar
Lee, A. J. (1980). North Sea: Physical Oceanography, The north-west European shelf seas the seabed and the sea in motion II. Physical and Chemical Oceanography. In F. T. Banner, M. B. Collins, & K. S. Massie (Eds.), Elsevier Oceanography Series, 24B (pp. 467– 493). Amsterdam: Elsevier.
CrossrefGoogle Scholar
Lewis, E., & Wallace, D. W. R. (1998). Program developed for CO2 system calculations, ORNL/CDIAC-105. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy.
Google Scholar
Millero, F. J., Graham, T. B., Huang, F., Bustos-Serrano, H., & Pierrot, D. (2006). Dissociation constants of carbonic acid in seawater as a function of salinity and temperature. Marine Chemistry, 100(1-2), 80– 94. https://doi.org/10.1016/j.marchem.2005.12.001
CrossrefCASWeb of Science®Google Scholar
Moll, A. (1997). Modelling primary production in the North Sea. Oceanography, 10(1), 24– 26. https://doi.org/10.5670/oceanog.1997.41
CrossrefGoogle Scholar
Nondal, G., Bellerby, R. G. J., Olsen, A., Johannessen, T., & Olafsson, J. (2009). Optimal evaluation of the surface ocean CO2 system in the northern North Atlantic using data from voluntary observing ships. Limnology and Oceanography: Methods, 7, 109– 118.
Wiley Online LibraryCASWeb of Science®Google Scholar
Olsen, A., Brown, K., Chierici, M., Johannessen, T., & Neill, G. C. (2008). Sea-surface CO2 fugacity in the subpolar North Atlantic. Biogeosciences, 5(2), 535– 547. https://doi.org/10.5194/bg-5-535-2008
CrossrefCASADSWeb of Science®Google Scholar
Olsen, A., Key, R. M., van Heuven, S., Lauvset, S. K., Velo, A., Lin, X., Schirnick, C., Kozyr, A., Tanhua, T., Hoppema, M., Jutterström, S., Steinfeldt, R., Jeansson, E., Ishii, M., Pérez, F. F., & Suzuki, T. (2016). The Global Ocean Data Analysis Project version 2 (GLODAPv2)—An internally consistent data product for the world ocean. Earth System Science Data, 8(2), 297– 323. https://doi.org/10.5194/essd-8-297
CrossrefADSWeb of Science®Google Scholar
Olsson, K., & Anderson, L. G. (1997). Input and biogeochemical transformation of dissolved carbon in the Siberian shelf seas. Continental Shelf Research, 17(7), 819– 833. https://doi.org/10.1016/S0278-4343(96)00059-3
CrossrefADSWeb of Science®Google Scholar
Omar, A., Olsen, A., Johannessen, T., Hoppema, M., Thomas, H., & Borges, A. (2010). Spatiotemporal variations of fCO2 in the North Sea. Ocean Science, 6(1), 77– 89. https://doi.org/10.5194/os-6-77-2010
CrossrefCASADSWeb of Science®Google Scholar
Omar, A. M., Skjelvan, I., Erga, S. R., & Olsen, A. (2016). Aragonite saturation states and pH in western Norwegian fjords: Seasonal cycles and controlling factors, 2005–2009. Ocean Science, 12(4), 937– 951. https://doi.org/10.5194/os-12-937-2016
CrossrefCASADSWeb of Science®Google Scholar
Otto, L., Zimmerman, J. T. F., Furnes, K., Mork, M., Saetre, R., & Becker, G. (1990). Review of the physical oceanography of the North Sea. Netherlands Journal of Sea Research, 26(2-4), 161– 238. https://doi.org/10.1016/0077-7579(90)90091-T
CrossrefADSWeb of Science®Google Scholar
Pierrot, D., Neill, C., Sullivan, K., Castle, R., Wanninkhof, R., Luger, H., Johannessen, T., Olsen, A., Feely, R. A., & Cosca, C. E. (2009). Recommendations for autonomous underway pCO2 measuring systems and data-reduction routines. Deep Sea Research, Part II, 56(8-10), 512– 522. https://doi.org/10.1016/j.dsr2.2008.12.005
CrossrefADSWeb of Science®Google Scholar
Provoost, P., van Heuven, S., Soetaert, K., Laane, R. W. P. M., & Middelburg, J. J. (2010). Seasonal and long-term changes in pH in the Dutch coastal zone. Biogeosciences, 7(11), 3869– 3878. https://doi.org/10.5194/bg-7-3869-2010
CrossrefCASADSWeb of Science®Google Scholar
Royal Society. (2005). Ocean acidification due to increasing atmospheric carbon dioxide. The Royal Society Policy Document 12/05. London. 68 pp.
Google Scholar
Rysgaard, S., Fossing, H., & Jensen, M. M. (2001). Organic matter degradation through oxygen respiration, denitrification, and manganese, iron, and sulfate reduction in marine sediments (the Kattegat and the Skagerrak). Ophelia, 55(2), 77– 91. https://doi.org/10.1080/00785236.2001.10409475
CrossrefGoogle Scholar
Salt, L. A., Thomas, H., Bozec, Y., Borges, A. V., & de Baar, H. J. W. (2015). The internal consistency of the North Sea carbonate system. Journal of Marine Systems, 157(2016), 52– 64.
ADSGoogle Scholar
Salt, L. A., Thomas, H., Prowe, A. E. F., Borges, A. V., Bozec, Y., & de Baar, H. J. W. (2013). Variability of North Sea pH and CO2 in response to North Atlantic Oscillation forcing. Journal of Geophysical Research: Biogeosciences, 118, 1584– 1592. https://doi.org/10.1002/2013JG002306
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Takahashi, T., Sutherland, S. C., Chipman, D. W., Goddard, J. G., Ho, C., Newberger, T., Sweeney, C., & Munro, D. R. (2014). Climatological distributions of pH, pCO(2), total CO2, alkalinity, and CaCO3 saturation in the global surface ocean, and temporal changes at selected locations. Marine Chemistry, 164, 95– 125. https://doi.org/10.1016/j.marchem.2014.06.004
CrossrefCASWeb of Science®Google Scholar
Thomas, H. (2002). Shipboard report of the RV Pelagia cruises 64PE184, 64PE187, 64PE190 and 64PE195. Den Burg, Texel, Netherlands: Royal Netherlands Institute for Sea Research.
Google Scholar
Thomas, H., Bozec, Y., Elkalay, K., & De Baar, H. (2004). Enhanced open ocean storage of CO2 from shelf sea pumping. Science, 304(5673), 1005– 1008. https://doi.org/10.1126/science.1095491
CrossrefCASADSPubMedWeb of Science®Google Scholar
Thomas, H., Bozec, Y., Elkalay, K., De Baar, H. J. W., Borges, A. V., & Schiettecatte, L.-S. (2005). Controls of the surface water partial pressure of CO2 in the North Sea. Biogeosciences, 2(4), 323– 334. https://doi.org/10.5194/bg-2-323-2005
CrossrefCASADSWeb of Science®Google Scholar
Thomas, H., Prowe, A. E. F., Van Heuven, S., Bozec, Y., De Baar, H. J. W., Schiettecatte, L.-S., Suykens, K., Koné, M., Borges, A. V., Lima, I. D., & Doney, S. C. (2007). Rapid decline of the CO2 buffering capacity in the North Sea and implications for the North Atlantic Ocean. Global Biogechemical Cycles, 21, GB4001. https://doi.org/10.1029/2006GB002825
Wiley Online LibraryADSWeb of Science®Google Scholar
Tilstone, G. H., Taylor, B., Blondeau-Patissier, D., Powell, T., Groom, S. B., Rees, A. P., & Lucas, M. (2015). Comparison of new and primary production models using SeaWiFS data in contrasting hydrographic zones of the northern North Atlantic. Remote Sensing of Environment, 156, 473– 489. https://doi.org/10.1016/j.rse.2014.10.013
CrossrefADSWeb of Science®Google Scholar
Tjiputra, J., Olsen, A., Bopp, L., Lenton, A., Pfeil, B., Roy, T., Segschneider, J., Totterdell, I., & Heinze, C. (2014). Long-term surface pCO₂ trends from observations and models. Tellus B, 66(1). https://doi.org/10.3402/tellusb.v66.23083
CrossrefGoogle Scholar
van Heuven, S., Pierrot, D., Rae, J. W. B., Lewis, E., & Wallace, D. W. R. (2011). MATLAB program developed for CO2 system calculations. ORNL/CDIAC-105b. Oak Ridge, Tennessee, U.S.A: CDIAC, Oak Ridge National Laboratory, Department of Energy. https://doi.org/10.3334/CDIAC/otg.CO2SYS_MATLAB_v1.1
Google Scholar
Wang, H., Hu, X., Cai, W.-J., & Sterba-Boatwright, B. (2017). Decadal fCO₂ trends in global ocean margins and adjacent boundary current-influenced areas. Geophysical Research Letters, 44, 8962– 8970. https://doi.org/10.1002/2017GL074724
Wiley Online LibraryCASADSWeb of Science®Google Scholar
Wesslander, K., Omstedt, A., & Schneider, B. (2010). Inter-annual and seasonal variations in the air–sea CO2 balance in the central Baltic Sea and the Kattegat. Continental Shelf Research, 30(14), 1511– 1521. https://doi.org/10.1016/j.csr.2010.05.014
CrossrefADSWeb of Science®Google Scholar
Wootton, J. T., Pfister, C. A., & Forester, J. D. (2008). Dynamic patterns and ecological impacts of declining ocean pH in a high-resolution multi-year dataset. PNAS, 105(48), 18848– 18853. https://doi.org/10.1073/pnas.0810079105
CrossrefCASADSPubMedWeb of Science®Google Scholar

Citing Literature

Volume124, Issue10

October 2019

Pages 3088-3103

Trends of Ocean Acidification and pCO₂ in the Northern North Sea, 2003–2015

Abstract

Plain Language Summary

1 Introduction