Biogeochemical processes as drivers of surface fCO2 in contrasting provinces in the subarctic North Pacific Ocean

Authors

  • Melissa Chierici,

    1. Climate Change Research Project, National Institute for Environmental Studies, Tsukuba, Japan
    2. Now at Marine Chemistry, Department of Chemistry, Göteburg University, Göteburg, Sweden.
    Search for more papers by this author
  • Agneta Fransson,

    1. Climate Change Research Project, National Institute for Environmental Studies, Tsukuba, Japan
    2. Now at Department of Oceanography, Earth Science Centre, Göteburg University, Göteburg, Sweden.
    Search for more papers by this author
  • Yukihiro Nojiri

    1. Climate Change Research Project, National Institute for Environmental Studies, Tsukuba, Japan
    Search for more papers by this author

Abstract

[1] The effect of temperature, biological processes, air-sea CO2 exchange and vertical mixing as drivers of the seasonality of the surface water fugacity of CO2 (fCO2sw) were studied for the year 2000 in the subarctic North Pacific Ocean. The regional and seasonal variability of the surface water chemistry was studied using an extensive data set on surface water fCO2 and nutrient concentrations in six contrasting provinces. We observed the largest seasonal amplitude for all parameters in the western provinces (Oyashio and Western Subarctic Gyre, WSG). Our study showed that biological processes and temperature were major controls for the monthly fCO2sw change in all provinces. The magnitude and strength of the processes showed large temporal and spatial variability. The WSG showed larger influence by biological processes and vertical mixing than the Alaska gyre (AG), where the effect caused by temperature was larger, implying different forcing of the fCO2 change in the two gyres. Biological activity, estimated from the monthly nitrate change corrected for addition induced by vertical mixing, resulted in a net annual CO2 loss. The net carbon loss out of the top 50 m driven by biological activity was 3 times higher in the WSG (64 g C m−2 yr−1) than in the AG (23 g C m−2 yr−1). The annual sum of the fCO2sw change based on all processes resulted in a CO2 buildup in the surface waters for all provinces. Although the air-sea CO2 exchange was of minor importance relative to the other considered processes (4 to 13%), all provinces showed a net annual uptake of atmospheric CO2 from 1 to 23 g C m−2 yr−1 and an average for the whole study area of about 12 (±9) g C m−2 yr−1.

1. Introduction

[2] The North Pacific Ocean (NP) is an important sink region for atmospheric CO2 [Honjo, 1997; Takahashi et al., 1993] and due to its large areas with presumably high biological production making it an important region for global carbon cycle studies [Longhurst et al., 1995; Honda, 2003; Schlitzer, 2004]. To a large extent, the oceans ability to act as a sink for atmospheric CO2 is based on the strength of the biological pump, where dissolved inorganic carbon is removed from the surface water as organic matter formed during photosynthesis and transported to deeper layers. The subarctic North Pacific Ocean is a well known high nutrient low chlorophyll a (HNLC) area, where primary production is limited by parameters other than macronutrients, and nitrate is rarely depleted in the surface water [Whitney et al., 1998; Wong et al., 2002a]. Several studies showed that iron is likely the limiting factor for phytoplankton growth [Boyd et al., 1998, 2004; Harrison et al., 1999]. A recent investigation performed in the Gulf of Alaska explained the observed enhanced biological production due to the presence of mesoscale iron-rich eddies [Johnson et al., 2005; Chierici et al., 2005; Crawford et al., 2005]. Other reports showed that the NW Pacific has a higher primary productivity than the NE Pacific, possibly attributed to a higher input of iron-rich dust from the Asian continent [Uematsu et al., 1983; Banse and English, 1999; Nishioka et al., 2003] and a deeper winter mixed layer depth [Glover et al., 1994; Harrison et al., 2004]. Deeper winter mixing of nutrient rich subsurface waters would lead to greater addition of nutrients to the surface waters thus affecting the nutrient availability. Liu et al. [2004], observed differences in phytoplankton dynamics and their mechanisms in the Oyashio region and the Western subarctic gyre, which was explained as being caused by different iron bioavailability.

[3] Several studies have shown large variability in physical, ecological and biogeochemical regimes in the North Pacific [Favorite et al., 1976: Longhurst et al., 1995] and a recent study, based on satellite-derived chlorophyll a and nitrate data, showed differences in primary production and the nutrient supply during winter between the east and the west as well as in subregions within these areas [Goes et al., 2004]. Another study by Zeng et al. [2002] observed east-west gradients in the seasonal amplitude of the partial pressure of CO2 (pCO2) in the surface water. Most comparisons of the variability in biogeochemical processes and ecosystem structure that have been performed in the subarctic Pacific were based on observations from two time series stations; the western Kyodo Northwestern Pacific Ocean time series station (KNOT, 44°N, 155°E) and the eastern time series station Ocean Station Papa (OSP, 50°N, 145°W) [Harrison et al., 2004]. KNOT is located in the Western subarctic gyre (WSG) and OSP in the Alaska gyre (AG). Both KNOT and OSP are located in the southern border of gyres and may not be entirely representative for the whole gyre system. An increased understanding of the temporal and spatial variability in both gyres is required [Harrison et al., 1999, 2004]. Occasional intrusions of coastal and subtropical water were observed both at OSP [Tabata, 1965; Whitney et al., 1998] and KNOT [Imai et al., 2002; Tsurushima et al., 2002] which complicate extrapolation of observations on the seasonal and interannual variability from a single station to a whole province or gyre [Harrison et al., 2004]. Besides being recognized as an important region for the uptake of atmospheric CO2 [e.g., Takahashi et al., 2002], the subarctic North Pacific exhibits strong seasonal to interannual variations in ocean biology and atmospheric CO2 uptake which are tightly coupled to changes in the physical environment [Stephens et al., 1995]. A major feature of the subarctic Pacific is the development of the intense atmospheric low pressure system, referred to as the Aleutian Low Pressure System, which strongly affects the climate-modulated variations in meteorological and physical oceanographic conditions [e.g., Overland et al., 1999; Niebauer, 1998]. Wong et al. [1998] reported an increased biological production as an effect of changing atmospheric circulation associated with El Niño–Southern Oscillation (ENSO) events and the North Pacific Decadal Oscillation [Minobe and Mantua, 1999]. Thus the variability in the atmospheric conditions (e.g., balance between low-high pressure systems) can affect the biological production as well as the changes in the upwelling of waters and the oceanic stratification. The close coupling between the atmosphere, ocean and ecosystems in the subarctic North Pacific makes it an ideal area to observe and gain knowledge of climate variability in both short and long timescales and its consequences for carbon cycling [McGowan et al., 1998].

[4] In this paper, we analyzed the seasonal evolution of surface ocean chemistry in six provinces in the subarctic North Pacific for a one year period, January 2000 to February 2001. We used measurements on the fugacity of carbon dioxide (fCO2) in the surface water and air, nitrate, sea surface salinity and temperature to estimate the effect of temperature, biological activity, vertical mixing and sea-air CO2 exchange on monthly fCO2 changes. We present the magnitude and strength in these processes in each of six contrasting provinces in an attempt to resolve the underlying mechanisms for the observed biogeochemical differences in the subarctic North Pacific.

2. Provinces in the Subarctic North Pacific Ocean

[5] Longhurst et al. [1995] defined ecological domains in the ocean on the basis of different characteristics in the seasonal nutrient supply and light availability. Further, these domains were separated into smaller subregions, biogeochemical provinces, on the basis of physical parameters such as topography, currents and fronts. The thorough hydrographical investigation by Favorite et al. [1976] still forms the basis to describe the biogeochemical provinces in the subarctic North Pacific Ocean. Recent studies in this area [e.g., Harrison et al., 2004; Wong et al., 2002a, 2002b], described 12 provinces with distinctively different hydrographic conditions in the area north of 35°N including the Bering Sea. Our study focuses on the area north of the Subarctic boundary (SAB > ∼45°N) and does not include the Bering Sea; thus we consider six areas. In Figure 1a we illustrate the general oceanographic conditions showing the main currents and fronts in our study area and the approximate location of these six provinces. The SAB divides the northern subarctic Pacific region from the southern subtropical Pacific region [e.g., Yuan and Talley, 1996], and is characterized by large temperature and salinity gradients which defines the Subarctic Front (SAF). North of the SAF (∼>45°N), the Subarctic Current (SAC) flows eastward, encircling the North Pacific Ocean with warm and salty water to depths reaching 2000 to 3000 m. The division of the SAC into two branches (AC and CC in Figure 1a) in the Gulf of Alaska forms the eastern boundary of the Alaska Gyre (AG). The strong Alaska Stream Current (ASC) forms its northern and western boundaries connecting with the Subarctic Current near 170°W resulting in a cyclonic, semi-enclosed gyre, centered around 52°N, 155°W. The SAC also forms the southern boundary for the western subarctic gyre (WSG), which is smaller than the AG and centered at 50°N, 165°E [Favorite et al., 1976]. The OY province is located where the Oyashio-Kuroshio Current system forms an intricate interfrontal region [Yasuda, 2003]. Between the two gyres lies an area referred to as the Central Subarctic Pacific (CSP) which is described as a highly variable area, strongly dependent on the temporal and spatial variability of the extension and mixing of the SAC and the westward ASC [Favorite et al., 1976; Qiu, 2002]. The dilute province (DIL) is highly influenced by large river runoff and precipitation.

Figure 1.

(a) General circulation in the subarctic North Pacific showing the subarctic boundary (SAB), which separates the subarctic Pacific from the subtropical south. The subarctic front (SAF) with the eastward flowing subarctic current (SAC) as well as the six relevant provinces for this study: OY, Oyashio; WSG, Western subarctic Gyre; CSP, central subarctic Pacific; AG, Alaska Gyre; SUB, subarctic province; and DIL, dilute province. Shown is the Bering Sea Gyre (BSG), although excluded in this work. The surface currents are shown in italic font and defined as follows: KC, Kuroshio current; OC, Oyashio current; SAC, subarctic current; AC, Alaska current; CC, California current; ASC, Alaska Stream Current; EKC, East Kamchatka current; and OKC, Okhotsk-Kuril current. The figure is adapted from Favorite et al. [1976], Wong et al. [2002a], and Harrison et al. [2004]. (b) The study area and the cruise tracks (black lines) from M/S Alligator Hope between January 2000 and February 2001. We also show the location of stations where we used subsurface data from the World Ocean Circulation Experiment, WOCE (solid boxes). For reference, the location of Ocean Station Papa (open circle) is at 50°N 145°W and station KNOT (K) is at 44°N, 155°E.

[6] Data from 145°E to 125°W was divided into the six described provinces (Figure 1a) using the gradients of sea surface salinity, SST and nitrate based on works by Favorite et al. [1976] and Wong et al. [2002a]. Since the SAB reaches farther south in areas close to the continents, we included data between 41°N and 45°N (Figure 1a). In Table 1 we have summarized the latitude and longitude borders representing the six provinces.

Table 1. Extent of Each Area of the Six Studied Provinces
ProvinceLongitudeLatitude, °N
OY145 to 155°E41–47
WSG155 to 165°E45–51
CSP170 to 180°E45–51
AG165 to 150°W45–52
SUB145 to 135°W47–51
DIL135 to 125°W48–51

3. Data Set and Analytical Methods

[7] A suite of analytical equipment was installed onboard the cargo ship M/S Alligator Hope for sampling and continuous underway measurements of fCO2 in surface water and air, sea surface salinity and temperature. Seawater was piped online from an approximate depth of 10 m, and air was pumped from the upper deck to the measurement unit located in the ships lower deck for the determination of atmospheric CO2. The ship traveled routinely between western North America and Japan performing underway measurements for the study of seasonality and interannual variability in the North Pacific Ocean. Figure 1b shows the cruise tracks of M/S Alligator Hope from January 2000 to February 2001, serving the same ports which resulted in a dense coverage in the subarctic North Pacific. In most cases the underway data coverage was monthly, or at least every 2 months.

[8] We briefly explain the instrument setup here; more information regarding the instrument setup, cruise dates and the measured parameters is available at the following web page http://ah.soop.jp as well as in the work by Lüger et al. [2004] which used a nearly identical system crossing the North Atlantic. Underway measurements of sea surface salinity and temperature were performed using a thermosalinograph SBE-21 (Seabird Electronics Inc.). After the seawater passed through the thermosalinograph, seawater flowed into the CO2 measurement unit at a rate of approximately 15 L min−1 into the equilibrator (tandem type equilibrator manufactured by Kimoto Electric Co., LTD, Japanese Patent P2001-83053A) which combines a static mixer type in addition to the tandem type [Kimoto and Harashima, 1993; Harashima et al., 1997] giving an improved exchange efficiency of more than 99.5% between air and seawater sample (Y. Nojiri, personal communication, 2004). Prior to analysis, the air was treated and dried in several steps which included aerosol filters, a Peltier cooling element, and Nafion® tubing. The mole fraction of CO2 (×CO2) in dry air extracted from seawater expressed by part per million (ppm) was continuously measured every minute by a non-dispersive infrared (NDIR, LiCOR®, model 6262, Lincoln, Nebraska) detector. A nearly identical measurement system was used to measure the xCO2 in atmospheric air (for details see http://ah.soop.jp). The analysis cycle for both xCO2 in seawater and air involved a calibration cycle every 12 hours using a suite of four standard gases supplied by Taiyo-Toyo Sanso Co., Japan, which agreed within 0.12 ppm with the NOAA/CMDL scale (Quality Assurance/Science Activity Centre; http://gaw.kishou.go.jp/qasac.html). The raw voltage readings of the NDIR were corrected for temperature and pressure effects [Dickson and Goyet, 1994] using a least squares procedure for the quadratic regression function. The xCO2 measurement system has a precision of about 0.5 ppm (Y. Nojiri, personal communication, 2004) based on a recent intercomparison experiment (http://www.ioc.unesco.org/ioccp/Tsukuba2004ws.htm).

[9] To convert the xCO2 values to fCO2 in seawater (μatm) we used measured temperature and pressure in the equilibrator at the time of the xCO2 measurement [Weiss, 1974; Beer, 1983]. The calculations were performed according to Dickson and Goyet [1994], described in detail by Körtzinger et al. [1996]. The atmospheric CO2 values (in dry air) were recalculated to fCO2 in wet air (μatm) using our ship data for sea surface temperature, salinity, air pressure and humidity [Weiss, 1974; Beer, 1983].

[10] Continuous measurements with a 1-minute reading using the rapid response system indicated dramatically more fCO2 variability in seawater than observed from hourly data. This result has important consequences when sampling in regions of high biological production and/or regions with large gradients in temperature and salinity as in the subarctic North Pacific.

[11] Discrete samples were collected every 12 hours for the analysis of nitrate and phosphate (PO4), which were measured at a shore-based laboratory (National Institute for Environmental Studies, Japan) by standard colorimetric methods using an auto-analyzer (ACCS-II, Bran Luebbe) with an estimated precision for nitrate of 0.5 μmol kg−1. The distribution of phosphate closely followed that of nitrate (PO4 = 0.068 × NO3 + 0.33, r2 = 0.969 for all stations at all depths); therefore it is not discussed further. Information regarding additional sampled parameters not included in this study are available at the above mentioned website.

[12] Nitrate concentrations in the water column (subsurface data) were obtained by using publicly available data from the cruises conducted during the World Ocean Circulation Experiment (WOCE) P01 (T. Ono and T. Watanabe, WOCE section P01 data and documentation, 1999, available from Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee at http://cdiac.esd.ornl.gov/oceans/woce_p01.html), P13 (C. D. Keeling et al., WOCE section P13 data and documentation, 1992, available from Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee, at http://cdiac.esd.ornl.gov/oceans/woce_p13.html), and P17N (A. Murata and M. Fukasawa, WOCE section P17N data and documentation, 2001, available from Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee, at http://cdiac.esd.ornl.gov/oceans/woce_p17n.html) available at the CDIAC website http://cdiac.esd.ornl.gov/oceans/RepeatSections/pacific_clivar.html. In Figure 1b, we show the approximate location of the stations for subsurface data. For our estimates of the sea-air CO2 flux we used monthly averaged wind speeds based on 6-hourly winds obtained from the NCEP Reanalysis data provided by the NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado, USA, from their web site at http://www.cdc.noaa.gov/. We used the estimates on mixed layer depths (MLD) from recently published data available at (http://www.lodyc.jussieu.fr/∼cdblod/mld.html), which is based on direct estimates from individual profiles of observed temperature and salinity data [de Boyer-Montegut et al., 2004].

4. Results and Discussion

4.1. Monthly Estimates and Seasonal Cycles of the Studied Parameters

[13] Although, we had dense data coverage for the parameters that were measured underway, the sampling for discrete samples (i.e., nutrients) was not as frequent, and resulted in an uneven data set. Earlier reports in the North Pacific showed that the seasonal cycle of fCO2, nutrients, salinity and temperature basically followed sigmoidal functions expressed by a harmonic equation [Zeng et al., 2002; Wong et al., 2002a]. We used their approach to estimate monthly values of fCO2 in surface water (fCO2sw) and air (fCO2air), NO3, S, SST, wind speed (ws) and MLD based on the harmonic fit of observed data (equation (1)). The five terms, c0–c4, are unique for each province and parameter, shown in Table 2 for surface water fCO2. In equation (1) the fCO2sw is the seasonally varying quantity and t is time (t = month),

equation image
Table 2. Values of the Seasonal Coefficients c0 to c4 in the Sigmoidal Functions From the Harmonic Fit (Equation (1)) of Observed Surface Water fCO2 Values in Each Province
Provincec0c1c2c3c4
OY344.6536.4815.3814.03−16.37
WSG362.8244.66−9.623.903.53
CSP363.531.320.148.723.46
AG359.9418.17−2.978.71−3.13
SUB351.739.02−2.20−2.07−0.61
DIL338.83−6.9520.984.19−0.30

[14] For complete seasonal coverage of sampling in each province, the maximum data gap should not exceed 3 months [Zeng et al., 2002]. To estimate the statistical error of this fitting procedure we compared observed and predicted surface water fCO2 data (Figure 2). The regression of observed and predicted values yielded the following:

equation image
Figure 2.

Correlation between the observed fCO2 data (x axis, fCO2 observed) and the predicted fCO2 data (y axis, fCO2 predicted) obtained from the harmonic function. The regression of the geometric mean of observed and predicted data yields fCO2 predicted = 0.9993 fCO2 observed (±11.0), r2 = 0.80 for all data used in this study. The mean of the observed minus predicted fCO2 is 0.04 μatm and the root mean square deviation is 11.0 μatm. The distribution of the residuals (predicted – observed data) shows no trend within the fitting method, neither in space or time (not shown).

[15] The mean of the observed minus predicted fCO2 was 0.0004 μatm and the root mean square deviation was 11.0 μatm which likely included the uncertainty introduced by the spatial averaging of data into provinces. The residuals (predicted – observed data) were randomly distributed and showed no trend within the fitting method, neither in space or time (not shown).

[16] Note that the hereafter, the denotations; fCO2sw, fCO2air, NO3, S, SST, refers to the predicted values of fCO2 in surface water and air, nitrate, salinity and temperature, respectively. By studying the seasonal evolution of SST and S (Figures 3a and 3b), the characteristics in our provinces agreed with that reported by Favorite et al. [1976]. We clearly observed a gradual decrease in salinity (Figure 3b) from the Oyashio province (OY) to the DIL in the east caused by the gradual mixing of saline, warm waters originating from the Kuroshio Current in the west with the surface water in the eastern provinces, where rainfall and river runoff was much more prominent than in the western provinces. The NO3 seasonality (Figure 3c) varied greatly between provinces and revealed the contrasting nutrient dynamics in the subarctic North Pacific. For example, in the AG, considered an HNLC area [Wong et al., 2002a], showed consistently high nitrate values throughout the year (Figure 3c), whereas the nitrate level was constantly low and showed little monthly change in the macronutrient limited DIL province.

Figure 3.

Seasonal cycle of (a) sea surface temperature (SST), (b) sea surface salinity (S), (c) nitrate (NO3) salinity normalized to S = 33, and (d) the mixed layer depth (MLD) in each province. The figure show the monthly means based on predicted values according to equation (1). The MLD is based on the climatology by de Boyer-Montegut et al. [2004].

[17] This was in clear contrast to the waters in the OY and the WSG province where we noted both the highest and lowest NO3 levels (Figure 3c), likely caused by high biological production and efficient nitrate supply [Harrison et al., 2004]. The evolution of the MLD used in this study [de Boyer-Montegut et al., 2004], showed similar seasonality in all provinces. Only in the near-coastal provinces, OY and DIL, we observed a shallower MLD, especially evident in the DIL province (Figure 3d). Observations in winter at OSP and KNOT showed similar MLD between the two stations which agreed with the data for AG and WSG used in our study [Whitney and Freeland, 1999; Harrison et al., 2004; de Boyer-Montegut et al., 2004]. However, Harrison et al. [2004] reported that the observed mixed layer depth in summer was twice as deep at OSP compared to KNOT. This was not found in the MLD used here (Figure 3d).

[18] In Figure 4, we observed the seasonal fCO2sw cycle for each province, starting in the west in the OY province following the eastward direction of the SAC to the DIL province off Vancouver Island at 125°W. Especially in the OY (Figure 4a) and the WSG (Figure 4b), we observed high fCO2sw values reaching more than 400 μatm in March decreasing to ∼315 μatm in early summer in the OY and in September in the WSG, likely caused by biological CO2 drawdown during production of organic matter. This sharp decline also coincided with a rapid increase in SST (Figure 3a) thus enabling a shallow and stable surface layer, preferable for the onset of the biological production by phytoplankton. In three provinces (OY, CSP, AG) our study showed a secondary fCO2sw minimum in fall, which was confirmed in the measured fCO2 data (marked + in Figures 4a, 4c and 4d, respectively). Especially obvious for the western provinces (OY and WSG) and notable in the AG was the effect of the deepening of the surface layer in fall (Figure 3c) which brought CO2 rich subsurface waters into the surface waters. The deepening lead to increasing fCO2sw values in fall and the subsequent CO2 oversaturation relative to the atmosphere by January (Figure 4). The surface water in the OY province was affected by the cold nutrient-rich boundary current referred to as the Oyashio Current (OC). The OY region, is relatively well studied [e.g., Saito et al., 2002] and reports showed large differences between winter and spring fCO2sw values. A large fCO2sw decrease from winter to spring was attributed to biologically mediated drawdown of CO2 during photosynthesis [Midorikawa et al., 2003a].

Figure 4.

Seasonal fCO2 cycle in the surface water (fCO2sw, solid line) and in the air (dashed line) for each province in year 2000 as well as the mean error for fCO2sw values based on the harmonic fit, mean −1σ, and the mean +1σ are shown as thin solid lines. The cross denotes the monthly mean fCO2 values in the surface water based on measurements.

Figure 4.

(continued)

[19] The CSP (Figure 4c) and the eastern provinces (AG and SUB, Figures 4d and 4e) showed moderate fCO2sw seasonality, and the fCO2sw did not reach the low or the high values observed in the OY and WSG. However, the annual mean fCO2sw values in the WSG, CSP and AG were similar, and estimated to 363, 364 and 360 μatm, respectively. We observed lower annual mean values in the near-coastal provinces (OY, 345 μatm; DIL, 339 μatm) and in the SUB (∼345 μatm).

[20] The surface waters in the DIL province were clearly undersaturated throughout the year with a fCO2sw minimum in May (∼310 μatm), a drop of around 50 μatm from the highest values of 358 μatm observed in December and January (Figure 4f).

[21] The peak-to-peak seasonal amplitude of SST, S, fCO2sw, and NO3 in each of the six provinces (Figure 5) showed a large amplitude difference for all parameters and provinces, in particular between the western and the eastern provinces. The amplitude was largest for all parameters in the western provinces. Apart from the CSP, the temperature amplitude showed a gradual decrease toward the east (Figure 5a). We found the most extreme temperature variability in the OY region, where the temperature in March was as low as 1.75°C, increasing to above 22°C by August, which resulted in an almost 3 times higher SST amplitude than the coastal province in the east (DIL). The large SST and S amplitude in the OY province (Figures 5a and 5b) was consistent with findings reported by Saito et al. [2002], which was explained as being caused by the presence of cold/warm currents. The OY is located in a region under the influence of the East Kamchatka Current (EKC, Figure 1a) carrying cold, and low-salinity water southward from the Bering Sea and the Arctic Ocean, and the warm saline subtropical waters in the Kuroshio Current (KC). The SST amplitude was also relatively large in the WSG (around 14°C, Figure 5a) which is comparable to station KNOT [Tsurushima et al., 2002], higher than the 10°C we observed in the AG (Figure 5a). The lowest seasonal variability was observed in the CSP (∼7.9°C, Figure 5a).

Figure 5.

Seasonal amplitude of the surface water properties in each province in the subarctic North Pacific, (a) sea surface temperature (SST, °C), (b) sea surface salinity (S), (c) fCO2sw (μatm), and (d) NO3 (μmol kg−1) for each province.

[22] The salinity amplitude decreased from 0.61 in the OY to 0.33 in the SUB province and increased approaching the DIL province (0.6, Figure 5b). The large salinity fluctuations in the DIL province was likely caused by the mix of a large freshwater addition from river runoff and heavy precipitation, with the occasional intrusions of subtropical waters from south [Favorite et al., 1976].

[23] Interestingly, a clear division in the seasonal amplitude of fCO2sw and NO3 was observed west and east of the CSP (approximately 170°E), where we observed the lowest seasonal amplitude for both parameters (Figures 5c and 5d). The fCO2sw and NO3 amplitude west of CSP was more than twice as high (average OY and WSG, fCO2sw = 95 μatm and NO3 = 21 μmol kg−1) as the amplitude within the CSP and further to the east (average CSP to DIL, fCO2sw = 40 μatm and NO3 = 8 μmol kg−1). This was probably a result of a larger upwelling of CO2 and nutrient rich subsurface water in the western North Pacific [e.g., Sasai and Ikeda, 2003]. The DIL had the highest fCO2sw amplitude (Figure 5c) and the lowest NO3 amplitude (Figure 5d) of the eastern provinces, which suggested that drivers other than biological processes were responsible for the seasonal fCO2sw change in the DIL (Figure 4f).

4.2. Estimates of the Drivers of the Seasonal fCO2sw Variability

[24] In surface waters, changes in the fCO2sw (ΔfCO2sw) over time may be due to different processes: changes in temperature (ΔfCO2T), biological processes, ΔfCO2bio, (a decrease of carbon dioxide during phytoplankton production and an increase by respiration), air-sea CO2 exchange (ΔfCO2as), the CO2 increase induced by vertical mixing and diffusion of CO2 and nutrient rich waters from below (sum of both; ΔfCO2mix), and a residual term (ΔfCO2res) required to close the budget. However, since our study was performed in one single year, the budget was almost certainly not balanced (this likely requires several years) and ΔfCO2res should be taken as indicative of the size of effects we have not accounted for in our relatively simplistic approach, e.g. the effect by horizontal transport, evaporation/precipitation effects, the use of annual mean C:N ratio, the use of a linear AT-S relationship as well as the uncertainty based on the difference between predicted and observed data. Thus if ΔfCO2sw represents the change between two consecutive months then the sum is simplified,

equation image

[25] Others used a similar approach to evaluate fCO2 controls in the equatorial Pacific [Lefèvre et al., 1994] and the North Atlantic [Lüger et al., 2004].

[26] In the following subsections we examine the magnitude and strength of each process (in μatm) driving the monthly change in fCO2 in the surface water in each biogeochemical province according to equations (3) to (6). In the equations, the formulation of C with the subscript m and pm denote the value of one of the parameters (C), SST, fCO2sw, NO3, the air-sea CO2 flux (F) for the considered month (m) or the previous month (pm), respectively. The NO3 concentrations were corrected for evaporation/dilution effects by normalization to a salinity of 33. We used the annual mean value of dissolved inorganic carbon (CTmean) and the Revelle factor, R, for each province calculated from the annual mean data of fCO2sw, total alkalinity (AT), salinity and temperature for each province using a CO2 system calculation program [Lewis and Wallace, 1998] together with the equilibrium constants from Mehrbach et al. [1973] refitted by Dickson and Millero [1987]. The AT was estimated using our annual mean salinity data for each province and the relationship between salinity and AT in the surface water in the North Pacific reported by Wong et al. [2002a] as follows: AT (μmol/kg) = 44.4 × salinity + 797.7. The buffer factor (R) ranged from 11.9 (DIL) to 13.4 (WSG) with an average of 12.7 ± 0.7 for all provinces. Calculated annual mean CT had a clear west-to-east gradient with decreasing values and ranged from 2096 (WSG) to 2027 μmol kg−1 in the DIL domain.

4.2.1. Temperature-Related Effects

[27] Generally, the temperature effect on fCO2sw is similar in strength to the biological effect but opposite in direction. For each 1°C degree change the fCO2sw changes by 4.23%, which was determined experimentally from a North Atlantic surface water sample by Takahashi et al. [1993]. In equation (3) we estimated the monthly change in fCO2sw caused by temperature effects (ΔfCO2T).

equation image

4.2.2. Air-Sea Exchange of CO2

[28] The disequilibrium between the fugacity of CO2 in the ocean (fCO2sw) and the atmosphere (fCO2air), ΔfCO2, drives the flux of carbon dioxide, F, thus the monthly air-sea CO2 flux (Fm) can be estimated using the following formula: Fm = Vp K0 (fCO2sw − fCO2air) where K0 is the solubility of CO2 at sea surface temperature (mol m−3 atm−1) [Weiss, 1974] and Vp is the gas transfer velocity (cm h−1), which has been described by a number of empirical equations [e.g., Liss and Merlivat, 1986; Wanninkhof, 1992]. In order to calculate the effect of the air-sea CO2 flux (ΔfCO2 as) on the observed change in fCO2, we used mixed layer depths (MLD, Figure 3c) from de Boyer-Montegut et al. [2004] and wind speed using the monthly mean averages from the NCEP/NCAR reanalysis based on 6-hour wind speed climatology and the short-term parameterization of Wanninkhof [1992]. Subsequently, the ΔfCO2 associated with gas exchange (ΔfCO2as) was calculated according to equation (4).

equation image

4.2.3. Vertical Mixing and Diffusion

[29] The processes that determine the addition of CO2 caused by the mixing of subsurface water (ΔfCO2mix) are the entrainment of water into the surface water whenever the MLD deepens together with the vertical diffusion across the lower boundary of the surface mixed layer. In this study, the addition of NO3 from subsurface waters by vertical mixing and diffusion (ΔNO3mix, equation (5a)) was converted to ΔfCO2mix using the C:N ratio, R, CTmean and fCO2pm as in equation (5b). We used the gradient between the nitrate concentration in the surface water (NO3sw) and the nitrate concentration at the base of the mixed layer (NO3ssw) (for data reference see section 3), followed in equation (5a) and similar to the approach used by Anderson et al. [2000].

equation image
equation image

where MLD is the mixed layer depth at time, t (month), and Kz is the vertical diffusivity at the base of the mixed layer, and NO3ssw and NO3sw are the nitrate concentration (μmol kg−1) below the mixed layer (subsurface layer) and in the surface water, respectively. We used the reported values for Kz within the thermocline of between 4 · 10−6 m2 s−1 in low winds and 2 · 10−5 m2 s−1 in high winds [Denman and Gargett, 1983]. Since only the deepening of the surface mixed layer induces mixing with underlying waters [e.g., Fasham et al., 1990], the function (∂MLD/∂t) is equal to ∂MLD/∂t when ∂MLD/∂t > 0, and equal to 0 when ∂MLD/∂t ≤ 0.

4.2.4. Biological Processes

[30] In subarctic oceans, the biological drawdown of carbon dioxide (CO2) during photosynthesis in spring and summer plays an important role in bringing the surface waters to CO2 undersaturation relative to the atmospheric CO2 concentration hence promoting oceanic CO2 uptake in this period. The biological pump removes carbon and nutrients from the ocean surface and exports them in the form of dissolved and particulate organic matter to deeper waters where they are sequestered. During decay of organic matter, carbon dioxide and nutrients are released, thus acting as a source of CO2 at those times. The intensity of the biological pump can be derived from the magnitude of seasonal fCO2 change as well as that of nitrate loss at the ocean surface [e.g., Bates et al., 1998; Wong et al., 2002a]. Here we used the monthly nitrate change (NO3m − NO3pm), corrected for the contribution from vertical mixing, (NO3mix, equation (5a), (5b)), to mirror the effect of biological activity on the monthly fCO2sw change (ΔfCO2bio). This implies that a nitrate loss was attributed to biological carbon consumption whereas a monthly nitrate gain was a decay of organic matter. Note that ΔfCO2bio approximates only net community production, not total primary production, because it does not account for recycled nutrients.

[31] The remaining nitrate change was converted into a fCO2 change (ΔfCO2bio, equation (6)) using CTmean, R, fCO2pm and the average carbon to nitrogen (C:N) drawdown ratio of 6.43, 6.82, 6.96, 6.55, 6.13 and 6.75, for OY, WSG, CSP, AG, SUB and DIL provinces, respectively, reported by Wong et al. [2002b] (note that their denotation for OY is HOK).

equation image

4.3. Magnitude and Variability of the fCO2sw Drivers

[32] In Figure 6, we have summarized the effects related to temperature (ΔfCO2T), biology (ΔfCO2bio), air-sea CO2 exchange (ΔfCO2as), and vertical mixing and diffusion (ΔfCO2mix), on the monthly change in the surface water fCO2 for each province. Positive values indicated an increase of fCO2sw in the surface water and negative values indicated that a loss of CO2 has occurred since the previous month. We observed largest regional difference in the magnitude of the controlling processes between the near coastal provinces, OY and DIL. For example, during spring/summer, the OY province (Figure 6a) showed an almost 5 times higher fCO2sw total loss due to biological activity (316 μatm) compared to the DIL province (65 μatm, Figure 6f). In the OY, most of this loss was canceled out by seasonal warming (323 μatm), while in the DIL, the temperature effect was almost twice as large as the loss driven by biological processes (105 μatm) and significantly exceeded the biological effect at this time of the year. It is obvious that AG differed significantly in the seasonal evolution of most of the effects compared to the other provinces. For example, in the AG (Figure 6d), the fCO2sw change was small for a 4-month-long period in the beginning of the year (February to May) and the maximum biological drawdown occurred in September (∼60 μatm) which was relatively late than observed in the other provinces (Figure 6), especially the western provinces (OY and WSG, Figures 6a and 6b). We could not explain this difference in timing by dissimilarities in the stabilization of the mixed layer depth (June, Figure 3c), or differences in the SST cycle (Figure 3a) since all provinces showed similar SST and MLD seasonality. Largest seasonal biological CO2 drawdown was observed in the WSG (363 μatm), which was twice as large as the corresponding value in the Alaska gyre (AG, 186 μatm). However, our estimates also suggested a small CO2 loss in March/April driven by biological processes in the AG which we did not observe in any other province. In the WSG, we noted a small fCO2sw loss due to air-sea CO2 exchange in the first six months of the year whereas in the other provinces we observed a gain due to the same process. The timing of the biological CO2 drawdown in the SUB (Figure 6e) was similar to that in OY and the WSG (Figures 6a and 6b) but the magnitude was 3 to 4 times lower.

Figure 6.

A summary of the monthly effect (μatm) of temperature (ΔfCO2T), biology (ΔfCO2bio), air-sea CO2 exchange (ΔfCO2a-s) and vertical mixing (ΔfCO2mix) on the surface water fCO2 in each domain: (a) the westernmost Oyashio region (OY), (b) the Western subarctic gyre (WSG), (c) the central subarctic Pacific (CSP), (d) the Alaska Gyre (AG), (e) the subarctic current system (SUB), and (f) the diluted domain (DIL). A positive value means that an increase of fCO2sw has occurred relative the previous month and a negative denotes a loss of fCO2sw from the surface water. A positive ΔfCO2as indicates an oceanic uptake of atmospheric CO2.

[33] During fall, the effect of vertical mixing and biological processes (i.e. remineralization/decay of organic matter), resulted in a CO2 gain in the surface mixed layer at the end of the year (September to December). To some extent the cooling of surface waters in the same period counteracted the observed fCO2 increase. Vertical mixing (ΔfCO2mix) added fCO2sw to the surface water with large seasonal and regional variability. The total annual gain was largest in the WSG (∼223 μatm, Figure 6b) which was 5 times larger than the gain in the DIL province (50 μatm, Figure 6f). In the WSG, the mixing effect was of the same magnitude as the loss of CO2 caused by the cooling of waters at the end of the year and dominated the change in the WSG in January and February (Figure 6b). Interestingly, vertical mixing was responsible for most of the fCO2sw gain in fall/winter in the WSG (Figure 6b), while the biological decay of organic matter contributed to most of the gain in the AG in the same period (Figure 6d). In the CSP, the effect by biological processes varied considerably and in a more complicated pattern for shorter timescales than any of the other provinces. Goes et al. [2004] found consistently low concentrations of satellite-derived chlorophyll-a in the CSP region and distinctively different biological features, likely related to water column stability.

[34] To compute the effect of each process relative to the total annual fCO2sw change (%) in each province, we calculated the annual sum of the absolute values of the fCO2sw change driven by each process and the total annual fCO2sw change from all processes in each province (Figure 7). It was obvious that temperature and biological processes dominated the fCO2sw change in all provinces. Generally, thermodynamics accounted for 31 to 42% of the total annual change and biological processes of 24 to 50%. It was only in the WSG (Figure 7b) and CSP (Figure 7c) where the biological effect clearly dominated and was responsible for 38% and 50% change on the total annual fCO2 change, respectively. In the OY province, the corresponding value was 33% and the effect driven by temperature dominated (40%, Figure 7a), this was also the case in the SUB province (42%). In the DIL, the warming-cooling effect was the major driver (40%) and only 24% was due to biological processes (Figure 7f), which agreed with the observation made by Zeng et al. [2002]. They attributed the effect of temperature as the responsible driver for the observed large seasonal pCO2sw amplitude, because NO3 levels were low throughout the year. The gas exchange effect was more significant in the DIL (13%) relative to the other provinces (Figure 7f), which was likely a result of the extensive CO2 undersaturation relative to the atmosphere (Figure 4f) throughout the year in combination with a more shallow MLD in winter than the other provinces (Figure 3d). Vertical mixing was generally responsible for less than 10% of the total annual fCO2sw change in all provinces, except in the WSG where the relative effect was 18% (Figure 7b). The annual contribution of the residual term (ΔfCO2res) was included in an attempt to evaluate the magnitude of the amount of carbon required to close the budget relative to the other processes in each province. We noted that the residual term was smallest in the two gyres (5% in both, Figures 7b and 7d) and the CSP (2%, Figure 7c) and largest in the near-coastal provinces (OY and DIL). The uncertainty from the harmonic analysis of ±11 μatm accounted for 1 to 2% of the fCO2sw change relative the total annual fCO2sw change depending on province. Both the OY and DIL are located in waters with complex surface current circulation, water mass mixing and significant freshwater input hence our simplistic approach cannot resolve all the intricacies in these areas. Perhaps horizontal advection and biological CO2 drawdown during primary production based on recycled nutrients are more important in these areas. Interestingly, the annual mean fCO2sw did not differ between the two gyres, even though we observed differences in the strength and magnitude of the controlling processes between the WSG than the AG. This was also the case in the North Atlantic, where Lüger et al. [2004] found the same annual mean pCO2 in two regions with different underlying mechanisms for the seasonal pCO2 cycle.

Figure 7.

Relative strength of each driver expressed as percent of the total annual fCO2 change in each province due to the effect of temperature (ΔfCO2T), biological processes (ΔfCO2bio), air-sea CO2 exchange (ΔfCO2as), vertical mixing (ΔfCO2mix), and the residual term to close the budget (ΔfCO2res). The residual term includes uncertainties in the calculations and the effect of processes not accounted for in our simplified approach, for example, the effect of horizontal advection and diffusion and the use of an annual mean C:N ratio.

4.4. Net Annual Carbon Fluxes

[35] A summary of the net annual fCO2 change in the top 50 m (i.e., average annual depth of euphotic zone in the North Pacific [Wong et al. 2002a]) caused by each process in each province is presented in Table 3. In all provinces, biological processes resulted in an annual net loss of carbon from the surface water and in the two gyres we estimated a net carbon loss of 22 g C m−2 yr−1 (∼1.9 mol C m−2 yr−1) in the AG and 64 g C m−2 yr−1 (∼5.4 mol C m−2 yr−1) in the WSG. Our estimate for the WSG was slightly lower than the 73 g C m−2 yr−1 reported by Sasai and Ikeda [2003] for the NW Pacific, but agreed well with the 67 g C m−2 yr−1 estimated by Wong et al. [2002a]. Our study confirmed earlier findings that the biological production was larger in the western subarctic North Pacific than the eastern [e.g., Wong et al., 2002a; Harrison et al., 2004]. Using an area of 1.58 · 1012 m2 for the AG and 1.13 · 1012 m2 for the WSG [Wong et al., 2002a], we obtained a net carbon loss by biological activity of 3.6 · 1013 and 7.2 · 1013 g C yr−1 in the AG and WSG, respectively. This gave lower values for both gyres compared to the study of Wong et al. [2002a] which found 5.8 and 8.6 g C 1013 yr−1, based on nitrate data obtained from 1995 to 1997. Also, in the OY province we found a large net carbon loss by biological activity (50 g C m−2 yr−1), consistent with both the findings by Sugimoto and Tadokoro [1997] and Wong et al. [2002a]. Saito et al. [2002] reported the OY region to have among the highest biological production based on nitrate loss in the North Pacific. In our study, the WSG had the largest annual carbon loss driven by biological activity. Anderson et al. [2000] used a similar approach studying the Greenland Sea in the subarctic North Atlantic, and estimated a carbon loss driven by biological activity of 34 g C m−2 yr−1 in the top 150 m. Integrated to the same depth as in our study (50 m), a comparison between these two areas implies larger carbon loss driven by biological activity in the subarctic North Pacific.

Table 3. Summary of the Net Annual Gain/Loss of Carbon to the Surface Water in the Upper 50 m Owing to Temperature (ΔfCO2T), Air-Sea CO2 Exchange (ΔfCO2as), Biological Processes (ΔfCO2bio), Vertical Mixing (ΔfCO2mix), and the Residual Closing the Budget (ΔfCO2res) in Each Provincea
ProvinceRΔfCO2TΔfCO2asΔfCO2mixΔfCO2bioΔfCO2Resb
  • a

    Temperature, ΔfCO2T; air-sea CO2 exchange: ΔfCO2as; biological processes: ΔfCO2bio; vertical mixing: ΔfCO2mix; residual closing the budget: ΔfCO2res. Units are g C m−2 yr−1. The denotation (ns) is used for nonsignificant values, and R is the Revelle factor (R) for each province.

  • b

    Calculated to close the budget and likely includes the effect from horizontal mixing and the use of an annual mean C:N ratio, as well as the uncertainty between observed and predicted fCO2 data from the harmonic fit of 11 μatm (∼2.8 g CO2 m−2 yr−1), which explains from 7 to 70% of the residual depending on province.

OY11.97172348−50−38
WSG13.4415758−64−16
CSP13.284111−13−4
AG13.037720−22−12
SUB12.3941414−15−17
DIL11.91−0.4 (ns)2114−13−22

[36] The air-sea CO2 exchange showed a net annual gain of 7 g C m−2 yr−1 (∼0.6 mol C m−2 yr−1) in the surface water of both gyres which agreed with the reported estimates at OSP of 7 g C m−2 yr−1 [Antoine and Morel, 1995], but was slightly lower than the 8.5 g C m−2 yr−1 reported by Wong and Chan [1991]. Midorikawa et al. [2003b] estimated a CO2 outgassing of 6 mmol m−2 d−1 for a period of 40 days from winter to spring in the year 2000 in the northwest Pacific (e.g., 48°N, 165°E). Applying a similar period (March to April, 60 days) in the WSG province, we estimated a similar outgassing of CO2 to the atmosphere of about 5 mmol m−2 d−1 estimated for the upper 50 m. The temperature-controlled near-coastal OY and DIL provinces, showed the largest net effect due to air-sea CO2 exchange and the mean oceanic CO2 uptake for both regions was 22 g C m−2 yr−1 (Table 3). Both areas were temperature-controlled but differed significantly with regard to the magnitude and strength of the effects driving the seasonal fCO2sw cycle. For the whole study area, we estimated a net annual CO2 sink for atmospheric CO2 of 0.23 Gt C yr−1 in the top 20 m for an area of 7.86 · 1012 m2 after Wong et al. [2002a]. This agreed well with the oceanic CO2 sink of 0.17 Gt C yr−1 reported by Takahashi et al. [1997] calculated for our study area which was about 25% of the total oceanic CO2 uptake of 0.64 Gt C yr−1 estimated for the whole North Pacific (14°N to 50°N) [Takahashi et al., 2002]. The subarctic North Atlantic (N >50°N) also acted as an annual oceanic CO2 sink although slightly lower at about 0.40 Gt C yr−1 [Takahashi et al., 2002].

[37] Vertical mixing from below contributed annually 58 g C m−2 yr−1 in the WSG which was 3 times higher than the addition in the AG (20 g C m−2 yr−1). In our study, the average contribution from vertical mixing for the whole study area was estimated to be 28 g C m−2 yr−1, to be compared to the 44 g C m−2 yr−1 estimated in the subarctic North Pacific using a three-dimensional model [Sasai and Ikeda, 2003]. In the top 150 m in the Greenland Sea, the contribution of vertical mixing was about 11 g C m−2 yr−1, which was less than what we found in any province in the subarctic North Pacific except the CSP (Table 3).

[38] For all provinces, we observed a net annual build up of fCO2 in the surface water, ranging from 4 g C m−2 yr−1 in the CSP to 38 g C m−2 yr−1 in OY (Table 3). Except for the CSP, the build up was significantly larger than the uncertainty rising from the harmonic fit (±11 μatm or 2.8 g C m−2 yr−1 in the upper 50 m). Horizontal advection and diffusion is a process likely to contribute to a carbon change. For the budget to be balanced, the effect of horizontal advection resulted in a net annual loss of carbon from the surface waters, the largest being in the near-coastal provinces, OY and DIL. Wong et al. [2002a] found that horizontal advection of nitrate was potentially important and it contributed a nitrate loss of 1.8 μmol kg−1 in the WSG, 5.7 μmol kg−1 in AG and as large as 15.7 μmol kg−1 in the OY province. Also, Sasai and Ikeda [2003] reported that the excess carbon caused by vertical mixing in the subarctic North Pacific was exported by the southward Ekman flow and varied regionally between 20 to 60 mg C m−2 d−1 (7 to 22 g C m−2 yr−1). The advection of water and its effect on the fCO2 is not easily described, and it is only by 3-D circulation models we can gain a complete understanding of the effect of horizontal advection and mixing on the fCO2 cycle [Lefèvre et al., 1994; Wanninkhof et al., 1995; Wong et al., 2002a].

4.5. Observations of Long-Term Variations in the North Pacific Ocean

[39] During the last 3 decades several studies have reported decreasing values in oxygen and nutrients, and a shallower mixed layer depth, possibly caused by the observed freshening of the surface water in both the Northwest and the Northeast Pacific [e.g., Whitney and Freeland, 1999; Emerson et al., 2001; Ono et al., 2001]. These changes are thought to be linked to the undulating modes of atmospheric pressure patterns, the North Pacific Oscillation (NPO) and/or El Niño/Southern Oscillation (ENSO) events [e.g., Wong et al., 1998; Minobe and Mantua, 1999; Goes et al., 2001]. In the western subarctic Pacific, phytoplankton biomass decreased from late 1970s to late 1990s and in the Oyashio region chlorophyll-a levels and the presence of diatoms was almost decreased by half from 1970 to late 1990s [Tadokoro et al., 2005; Chiba et al., 2004]. One possible reason for these observations was a decrease of the supply of carbon and nutrients from deeper layers to the surface water due to a shallower mixed layer. In an area where biological processes drive the carbon dioxide cycle this would likely have an affect on the total carbon flux. Thus the variability in the atmospheric conditions (e.g., balance between low-high pressure systems) can effect the biological production, hence the air-sea CO2 flux. In addition, Freeland et al. [1997] found increasing sea surface temperatures in the last six decades in the northeastern subarctic Pacific. Warmer surface water would lead to a decrease in the CO2 solubility, thus hampering the oceanic CO2 uptake, which could have a stronger effect on the more temperature controlled provinces, whereas a change in the surface layer stratification may have a greater impact in the more biologically controlled provinces.

5. Summary

[40] By using relatively simple means, we found substantial differences in the timing and magnitude of the processes driving the fCO2 in the surface water in six provinces in the subarctic North Pacific. We observed contrasting seasonal cycles and amplitudes of temperature, salinity, nitrate and fCO2 in the surface water of the studied provinces. Generally, the effect of temperature and biological processes were the major drivers for the fCO2 change in the subarctic North Pacific. Largest seasonal amplitude in all parameters where observed in the western provinces (OY and WSG) where we also observed the largest effect by biological activity. In the Alaska gyre the effect of temperature and biological processes was of similar significance with regard to the annual fCO2sw change. This was different from the observation in the WSG where biological processes dominated the annual fCO2 change and physical mixing had the largest influence of all provinces, which confirmed earlier comparisons between these areas [e.g., Takahashi et al., 2002; Wong et al., 2002a, 2002b; Sasai and Ikeda, 2003; Harrison et al., 2004]. We also noted a large difference in magnitude and timing of the investigated drivers both between the western and the eastern regions, and also within provinces, this was particularly significant in the eastern provinces. In the east, the Alaska gyre had the largest biological CO2 drawdown in spring and summer, which started later than the other eastern provinces. Interestingly, increasing surface water fCO2 in fall was mostly caused by biological processes (likely the decay of organic matter) in the AG while vertical mixing was responsible for most of the fCO2 increase in the WSG for the same period. Temperature dominated the annual fCO2 change in the SUB and the DIL provinces. This was also the case in the Oyashio province, which had among the strongest seasonal fCO2sw drawdown driven by biological processes in our study. The CSP province showed significantly different seasonal patterns and amplitudes than any other province. Goes et al. [2004] observed greatest north-south differences in the seasonal cycles of nitrate and SST in the area between the two gyres, which correspond to the CSP province in our study. This implies that our general definition of the location of the CSP area was insufficient. A better temporal and spatial resolution together with further knowledge about the seasonal and interannual variability of prevailing current systems is required to successfully resolve the variability in the processes affecting the fCO2sw in the CSP. In the other provinces, the use of a high-density data set as in this study, has given valuable insight and clarified contrasts in the seasonal fCO2sw cycle and its drivers for one year. Continuing high-frequency measurements of biogeochemical parameters in the surface water together with subsurface data, mixed layer depth and wind speed data on a seasonal and interannual timescale provide powerful tools to obtain valuable information to understand the variability and strength of the processes affecting the CO2 cycle in order to resolve the role of the North Pacific in the global carbon cycle.

Acknowledgments

[41] This work was gratefully supported by funding obtained from the Ministry of Environment, Japan within the Eco-Frontier Fellowship Program and from The Japan Society for the Promotion of Science (JSPS). Thanks to Jiye Zeng (CGER, Japan) for extracting the mixed layer depth and wind speed data and Junko Yamamura (NIES, Japan) who carried out the extensive work with analyzing nutrient samples. We are also grateful to the Captain, officers and the ship crew onboard M/S Alligator Hope for the superb support and collection of water samples. We thank Kasper Plattner (UCLA, USA) and anonymous reviewers for constructive comments on the manuscript.

Ancillary

Advertisement