The observation of persistent undersaturation of surface-water CO2 with respect to the atmosphere in this coastal upwelling system prompts two immediate questions. First, what about this location differs from open-ocean upwelling areas, typically strong sources of CO2 to the atmosphere? Second, is this sink significant in the global carbon budget?
4.1. Unique Biogeochemical Aspects of the Study Area
 In response to the first, we examined characteristics of the upwelled source water, which is easy to track at the northernmost site. Density anomaly (sigma-t; Figure 6, top panel) distributions show that dense water, with sigma-t ≥ 26.5, is drawn up and shoreward along the bottom. This water's salinity is nearly 34 (Figure 6, middle panel), and its temperature slightly less than 7°C (Figure 6, bottom panel). We chose to examine this water because it has been shown that the onshore upwelling transport originates primarily from this density range and proceeds onshore through the bottom boundary layer [e.g., Smith, 1974; Lentz, 1992; Perlin et al., 2005].
Figure 6. Cross-shelf sections of (a) density (expressed as density anomaly, sigma-t), (b) salinity, and (c) temperature at the Cascade head transect on 27 May 2001.
Download figure to PowerPoint
 Reconstruction of the source water's CO2 response to upwelling, warming, and photosynthesis requires some knowledge of its TCO2. Temperature, salinity, and TCO2 profiles show that the source water's depth is about 200 m; TCO2, measured by a discrete-sample modification of the continuous method of L. Bandstra et al. (High-frequency measurements of total CO2: Method development and first oceanographic observations, submitted to Marine Chemistry, 2004), at this depth is 2.30 mmol kg−1 (Figure 7). This, and an observed source-water XCO2 of 975 ppm, allows calculation of a “pseudo-alkalinity” or a “calculated alkalinity” of 2.33 meq kg−1. This calculation is based on acid-base equilibrium relationships and a simple model of alkalinity consisting of carbonate, bicarbonate, and borate ion concentrations (= [HCO3−] + 2[CO3−2] + [H2BO3−1]). Such a simple model may include errors associated with neglecting other species such as silicate, phosphate, and hydroxyl ions. The computed values are probably accurate to better than ±0.5% for estimation of total alkalinity, sufficient for this exercise. Both TCO2 and alkalinity values are consistent with those seen in similar-T/S waters on the WOCE P17N line [Lamb et al., 2002] (World Ocean Circulation Experiment, available at http://whpo.ucsd.edu/data/onetime/pacific/p17/p17n), lending confidence to the basic observations summarized above.
Figure 7. Vertical distributions of (a) temperature, (b) salinity, and (c) TCO2 at a 1000-m-deep station off Newport, Oregon. Arrows indicate the T, S, and corresponding TCO2 of the upwelled source water.
Download figure to PowerPoint
 We know from our own high-resolution measurements (Hales et al., submitted manuscript, 2004), and from historical data [Kokkinakis and Wheeler, 1987] (U.S. GLOBal Ocean ECosystems Dynamics North East Pacific program, available at http://globec.coas.oregonstate.edu/jg/serv/globec/nep) of nutrient chemistry in this area, that the nitrate concentration in this source water is about 34 μmol/kg and is drawn to zero by rapid photosynthesis. The N:C stoichiometry ratio of 16:106 for photosynthetic uptake [e.g., Redfield et al., 1963] implies that a decrease of 7 units of TCO2 and a 1 unit increase in alkalinity results from each 1 unit of nitrate decrease. This results in a biologically modified source water with TCO2 = 2.05 mmol kg−1(= 2.30–0.25 mmol kg−1) and alkalinity = 2.33 meq kg−1(= 2.30 + 0.03 meq kg−1). From these we calculate XCO2 = 165 ppm at the source-water temperature of 7°C. Warming, significant in determining XCO2 of any water mass, raises temperatures to only 11° or 12°C, increasing XCO2 to only 215 ppm. These values are well within the range of surface XCO2 summarized in Figures 123–4.
 This exercise identifies several key aspects of this upwelling system. The first, and most significant, is the large preformed nutrient concentration of the upwelled source waters. The preformed nutrient concentration is the nutrient concentration at the time of the formation of water mass, and may be approximated by subtracting the respired amount from the observed concentration. The respired amount is commonly estimated using the apparent oxygen utilization (AOU) at the potential temperature and the respiration stoichiometry. Hence waters with large preformed nutrient values should support greater amounts of photosynthesis and yield low XCO2 waters when all the nutrients are utilized by the photosynthesis. Our own calibrated in situ O2 data show a source-water O2 concentration of 70 μmol kg−1, corresponding to an apparent oxygen utilization (AOU) of 220 μmol kg−1. On the basis of the AOU:NO3− respiration ratio [e.g., Anderson and Sarmiento, 1994; Hedges et al., 2002; Takahashi et al., 1985], we estimate that about 22 μmol kg−1 of the source-water NO3− was respiration-produced. The remaining 12 μmol kg−1 are preformed, consistent with the high NO3− of the 26.5 sigma-t surface outcrop in the western North Pacific in late winter [Takahashi et al., 1993]. This will support an additional CO2 utilization of 0.084 mmol kg−1, when all the nutrients are consumed. This will reduce the XCO2 by an additional 50%.
 The second important aspect of this system is that productivity is limited only by available nitrate, evidenced by its complete exhaustion. There appears to be no micronutrient limitation, consistent with the upwelling path through the benthic boundary layer, which is exceptionally high in dissolved iron [Chase et al., 2002]. There also appears to be no significant grazing limitation keeping primary producers from completely consuming available nitrate. The lack of limitation allows photosynthesizers to consume TCO2 in stoichiometric proportion to the total available nitrate.
 Finally, warming of these upwelled waters is modest in the study area. Water temperature increases by 4°–5°C across the shelf (Figure 6), which increases XCO2 by about 30% (50 ppm), which is much less than the reducing effect of the photosynthesis supported by preformed nutrients.
 All of these features contrast with the conditions experienced by, for example, the open-ocean upwelling region of the central and eastern equatorial Pacific. Upwelled water supplied to the euphotic zone there originates at the top of the Equatorial Undercurrent at a depth of ∼100 m, and contains essentially no preformed nutrients [Archer et al., 1996; Chai et al., 1996; Radenac and Rodier, 1996]. Upwelled nutrients are not depleted to zero in surface waters [Chai et al., 1996; Murray et al., 1995; Archer et al., 1996; Radenac and Rodier, 1996]. This implies a non-nutrient limitation of phytoplankton growth, either due to grazing [e.g., Verity et al., 1996] or trace element limitation [e.g., Fitzwater et al., 1996]. In addition, surface waters warm significantly, by nearly 10°C, as they move westward along the equator. These factors combine to make the equatorial Pacific one of the largest natural sources of CO2 to the atmosphere [Takahashi et al., 2002; Tans et al., 1990].
4.2. Global Significance of the Upwelling Along the U. S. Pacific Coast
 To address the second question posed above, the net air-sea gas exchange flux (F) for the upwelling season must be quantified. F can be estimated from
where kCO2 is sea-air CO2 gas transfer coefficient (m d−1), KCO2 is CO2 solubility (mol m−3 atm−1), and ΔPCO2 is the sea-air PCO2 difference (atm). ΔPCO2 was calculated from the observed ΔXCO2, ambient atmospheric pressure, and water-vapor content. The atmospheric pressure and water-vapor corrections are small (order a few percent each), and largely cancel each other; thus the ΔPCO2 values are nearly equivalent to the ΔXCO2 values presented earlier. To obtain the mean air-sea CO2 flux for the study area and season, the mean values for each of the parameters in the equation above have been used. We determined a mean kCO2 for the May–August period by averaging kCO2 values calculated from wind speed measurements (hourly mean U10; m s−1) from the nearby NDBC buoy 46050 (National Data Buoy Center; http://www.ndbc.noaa.gov) and the following dependence on U10 [McGillis et al., 2001]:
 The average upwelling-season gas-exchange coefficient for CO2, kCO2, has been estimated to be 3 m d−1. The mean flux was estimated by multiplying this mean kCO2 by the spatially and temporally weighted average ΔXCO2 given in section 3, corrected for atmospheric pressure and water-vapor content at the time of each surface CO2 measurement. If correlation of ΔXCO2 with wind speed is weak, the season-average flux may be estimated from the product of the season-average kCO2 and our estimate of season-average ΔXCO2 given earlier. This assumption is supported by two observations. First, there is no correlation between wind speed and wind direction, as discussed earlier, arguing against any correlation between ΔXCO2 variability caused by upwelling/reversal cycles and kCO2. Second, it has long been known that ΔXCO2 and wind velocity do not correlate strongly in the open ocean as a result of the strong buffering of ocean CO2 with respect to gas exchange [e.g., Broecker and Peng, 1980]. We find that the average flux calculated in this way is 20 mmol m−2 d−1, which is about 15 times as large as the mean global ocean CO2 uptake flux of about 1.3 mmol m−2 d−1 (or 2 Pg-C yr−1). If upwelling prevails May–August [Allen et al., 1995; Federiuk and Allen, 1995; Lentz, 1992, Strub et al., 1987a, 1987b], CO2 uptake in this 120-day interval is 2 mol m−2.
 This is a large area-specific flux, and the nearshore regions were specifically excluded from the global compilations such as that of Takahashi et al. . Assessing its global significance requires estimation of the area that such conditions cover. The limited number of studies of CO2 chemistry in eastern boundary upwelling coastal waters makes such an extrapolation difficult at best. Ianson and Allen  and Ianson et al. , with a combination of modeling and measurements in the region immediately offshore of Vancouver Island at about 49°N, report a similar dynamic range to that reported here. Their reported XCO2 values range from 200 ppm to 1500 ppm, with the outer shelf representing a strong sink for atmospheric CO2, and a narrow inner-shelf band representing a strong source. These findings are in good qualitative agreement with those presented here. Friederich et al. , however, with a series of studies along a single cross-shelf transect off Monterey, California, concluded that that region represented a net source of CO2 to the atmosphere. More recently, Friederich observed XCO2 variability along that transect consistent with that reported here, including some values as low as 150 ppm (G. Friederich, personal communication, 2004). Hydrographic data suggest that the water upwelled along the Oregon coast is present along the entire eastern boundary of the North Pacific [e.g., Reid, 1965; Talley, 1993]. If this water is supplied to the coastal ocean by upwelling along the entire margin, then it is possible that the conditions seen off Oregon are representative of a much greater area. The multiyear seasonal observations for the surface-water pCO2 indicate that the low pCO2 waters (dark blue and purple colors in Figure 5) are present during the spring-summer months as far west as 127°W or 235 km offshore (150 km west of the shelf break). We recognize that some other aspects of the Oregon coast system may not be general, for example, the complete and rapid photosynthetic uptake of excess nitrate and moderate warming of upwelled waters. These caveats must be taken into account when considering extrapolations of the sort we make below.
 Nonetheless, if the conditions seen off Oregon are representative, and if eastern-boundary upwelling regions in the North Pacific cover 0.7 × 106 km2 (or 25% of the area of the contiguous Pacific shelf area with water depths between 0–200 m [Menard and Smith, 1966]), then upwelling-season uptake of atmospheric CO2 by such areas is 2 × 1012 mol (0.02 Pg- C). Although this area represents less than 2% of the North Pacific 14°N–50°N, and the upwelling season lasts only a third of the year, the seasonal flux is about 5% of the mean annual North Pacific CO2 uptake, estimated at about 40 × 1012 mol yr−1 (0.5 Pg C yr−1) [Takahashi et al., 2002; Gloor et al., 2003]. Further, the uptake of CO2 by this region is about half of the open North Pacific's CO2 uptake during the same May–August period (www.ldeo.columbia.edu/CO2). These comparisons suggest that the sum of ocean margin upwelling systems, including the west coasts of North America, South America, and South Africa, could contribute significantly to the global air-to-sea CO2 fluxes.
 The increased XCO2 of subsurface waters in the August sections of Figure 3 is caused by the respiration of locally produced biogenic debris settling through the water column, and hence is derived from photosynthetic production of organic carbon fueled by upwelled nutrients. Some portion of the atmospheric CO2 taken up by the surface seawater over the shelf area is thus transferred to subsurface waters. During winter, poleward along-shore winds cause downwelling, which results in bulk subduction of the dense isopycnals that had been on the shelf during summertime [Allen and Newberger, 1996] to their ≥200 m offshore depths. This process would thus move high-CO2 shelf water offshore to deep ocean-interior regimes, where the CO2 could mix to even greater depths via enhanced vertical mixing along the continental slope [Ledwell et al., 2000]. This represents an important pathway through which atmospheric CO2 is sequestered in deep ocean regimes via a seasonal biological pump in shelf environments.
 Assessing the process of transfer of CO2 to the ocean interior by the route suggested above requires observations during the fall transition from predominantly upwelling summer conditions to predominantly downwelling winter conditions to determine if impacted summertime waters are in fact moved offshore in bulk. We completed a series of studies in this region in January of 2003 (B. Hales unpublished results, 2003), and saw surface CO2 values near or slightly below saturation with respect to the atmosphere. There was no indication of the presence of the dense upwelled waters seen in summer, and no hint of degassing from the respiration-driven high-PCO2 waters seen in the late summer data presented here. These observations thus tentatively support the notion of wholesale movement of CO2-rich waters from the shelf to the ocean interior, but cannot definitively rule out offgassing processes in the fall transition period. Further field observations are required to determine the fate of the CO2 sequestered during the summer upwelling season.