DOE/OR-857 Prepared for: United States Department of Energy March 1 985 Office of Energy Research Office of Basic Energy Sciences Carbon Dioxide Research Division Under Contract No. DE-AC05-840R21400 TR020 Reconstruction of Past Atmospheric C02 Contents from the Chemistry of the Comtemporary Ocean: An Evaluation DISCLAIMER “This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.” This report has been reproduced directly from the best available copy. Available from the National Technical Information Service, U. S. Department of Commerce, Springfield, Virginia 22161. Price: Printed Copy A05 Microfiche A01 Codes are used for pricing all publications. The code is determined by the number of pages in the publication. Information pertaining to the pricing codes can be found in the current issues of the following publications, which are generally available in most libraries: Energy Research Abstracts, (ERA); Government Reports Announcements and Index ( GRA and I); Scientific and Technical Abstract Reports (STAR); and publication, NTIS-PR-360 available from (NTIS) at the above address. March 1985 Prepared for: United States Department of Energy DOE/OR-857 Dist. Category UC-11 Office of Energy Research Office of Basic Energy Sciences Carbon Dioxide Research Division Washington, D.C. 20545 TR020 Reconstruction of Past Atmospheric CO 2 Contents from the Chemistry of the Comtemporary Ocean: An Evaluation Prepared by: W.S. Broecker and Taro Takahashi Lamont-Doherty Geological Observatory Columbia University Palisades, NY 10964 and Tsung-Hung Peng Environmental Sciences Division Oak Ridge National Laboratory Oak Ridge, TN 37839 Under Contract No. DE-AC05-840R21400 Digitized by the Internet Archive in 2020 with funding from Columbia University Libraries https://archive.org/details/reconstructionofOObroe ABSTRACT Previous attempts to reconstruct the preanthropogenic CO2 content of the atmosphere from ZCO2, alkalinity, and CO2 partial pressure (pC02) data for the contemporary ocean are evaluated. Also, new attempts based on the more extensive Transient Tracers in the Ocean (TTO) data for the North At¬ lantic are made. The conclusion is that such quests are quite likely des¬ tined to failure in that the uncertainties in the reconstructions will al¬ ways be larger than the uncertainties ultimately available from other ap- proaches (i.e., ice cores and C record deconvolutions). The difficulty lies in the fact that successful deconvolution requires the solution of a host of basic problems in the ocean sciences. To solve these problems, we must ( 1 ) develop satisfactory ocean mixing models; ( 2 ) understand the fac¬ tors controlling pC02 in surface waters (especially those in the winter for high latitudes), ( 3 ) determine the chemical composition of material being oxidized at various places and depths in the ocean and ( 4 ) determine the factors controlling the pattern of alkalinity within the sea. We con¬ clude that the value of the ocean data set lies in understanding processes taking place in the sea rather than in reconstructing past atmospheric CO2 contents. Once the history of atmospheric CO2 has been reconstructed from 13 ice core CO2 and C deconvolutions, the ocean ECO2 and pC 02 data will be¬ come a useful tracer in studies of ocean chemistry and ventilation. i ACKNOWLEDGEMENTS The TTO titrator measurements of EC02 and alkalinity on which the examples at the end of the paper are based were made largely by Dave Bos of the Scripps Physical and Chemical Oceanographic Data Facility (PCDOF) group. Because of his skill and diligence the precision of these measure¬ ments is outstanding. Our presentation benefited from detailed comments by Alan Shiller (from MIT), Financial aid was provided by the Department of Energy CO 2 program grant No, 79EV10038 and subcontract No, 19X-22237C with Lamont-Doherty Geological Observatory of Columbia University with Martin Marietta Energy System, Inc,, under contract No. DE-AC05-840R21400 with the U.S. Department of Energy to W.S, Broecker and grant No, DEAC02- 79EV10229 to T, Takahashi, Research at ORNL (T.-H. Peng) was supported jointly by the National Science Foundation's Ecosystem Studies Program un¬ der Interagency Agreement BSR 811536, A03 and the Carbon Dioxide Research Division, Office of Energy Research, U.S. Department of Energy, under con¬ tract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc. Publica¬ tion No. 2449, Environmental Sciences Division ORNL. Lamont-Doherty Geo¬ logical Observatory Technical Report No. LDGO-85-1. ii TABLE OF CONTENTS PAGE ABSTRACT - i ACKNOWLEDGEMENTS-ii LIST OF TABLES-iv LIST OF FIGURES- v 1. INTRODUCTION - 1 2. THE CHEN APPROACH- 4 3. THE IDEAL OCEAN- 7 4. CORRECTION OF SUBSURFACE WATER COMPOSITIONS FOR RESPIRATION-10 5. C0 2 PARTIAL PRESSURES FOR WATERS CURRENTLY DESCENDING INTO THE DEEP SEA-18 6. DEPENDENCE OF ANTHROPOGENIC EFFECT ON VENTILATION TIME-22 7. RECONSTRUCTION BASED ON MEASUREMENTS ON THERMOCLINE WATERS - 25 8. RECONSTRUCTION BASED ON MEASUREMENTS ON DEEP WATERS-50 9. CONCLUSIONS-75 10. REFERENCES-77 • • • in LIST OF TABLES TABLE Page 1 Results of two-box model calculation for the anthro¬ pogenic CO 2 scenarios shown in Fig. 2----------- 26 2 Comparison of GEOSECS and TTO ICO 2 and alkalinity results for deep water samples -------------- 27 3 Surface water data for the North Atlantic, obtained as part of TTO program ------------------ 30 4 Alkalinity and ZCO 2 data for TTO thermocline samples from the North Atlantic ------------------ 31 5 Model results for the 30° to 45° region as of 1981 ----- 49 6 Properties along the 02 = 36.95°/oo isopycnal horizon, based on measurementss made during the TTO program -------------------------- 52 7 Properties along the 04 = 45.730°/oo isopycnal horizon, based on data from the GEOSECS program ------ 54 8 Radioisotope data in the vicinity of the 02 = 36.95 °/00 isopycnal horizon ------------- 65 9 Initial pC 02 values for northern Atlantic water types ------------------------ 72 10 Summary of initial pC 02 values for key northern Atlantic water types -------------------- 74 iv LIST OF FIGURES Figure Page 1 Reconstruction of the CO 2 content of the atmosphere prior to 1958 (i.e., to the onset of Keeling's pre¬ cise atmospheric monitoring) derived from ocean- atmosphere models - -- -- -- -- -- -- -- -- -- - 3 2 Hypothetical salinity-normalised EC 02 versus poten¬ tial temperature plot showing present-day surface waters, preanthropogenic surface waters, and respir- atlon-and dissolution-corrected contemporary subsur¬ face waters - -- -- -- -- -- -- -- -- -- -- -- 5 3 The idealized model of ocean ventilation described in text --------------------------- 8 4 Histogram of excess dissolved oxygen contents found in the surface ocean during the GEOSECS, FGGE, and TTO programs ----------------------- H 5 Sections of AOU (apparent oxygen utilization) for the western Atlantic and western Pacific oceans ------ 14 6 Plots of salinity-normalized alkalinity and of nitrate against temperature for surface waters from high latitudes - 17 7 Map showing the geographic patterns for the surface ocean-atmosphere pC0 2 difference -------------- 19 8 PCO 2 versus temperature for surface water samples analyzed as part of the GEOSECS and TTO programs ------ 21 9 The annual cycle of pC 02 calculated for the ocean mixed layer of constant thickness, alkalinity, and salinity --------------------------23 10 Comparison of the anthropogenic change for two hypothetical ocean situations --------------- 24 11 Map showing the location of TTO stations in the North Atlantic ----------------------- 29 12 Plot of salinity-normalized ZCO 2 versus temperature as obtained during the TTO expedition in the North Atlantic --------------------------36 13 Histogram of surface water pC 02 values obtained during the March 1973 traverse of the North Atlantic made as part of the GEOSECS program ------------ 37 v 14 15 16 17 18 19 20 21 22 23 24 25 26 Page Plots of LCO 2 versus O 2 along individual isopycnal horizons - -- -- -- -- -- -- -- -- -- -- -- - 38 Tritium versus AOU for the North Atlantic temperate thermocline as measured during the GEOSECS program - - - 40 Mean ZCO 2 excesses as a function of AOU for the thermocline of the North Atlantic ----------- 41 Plot of salinity-normalized alkalinity for surface water and thermocline waters in the North Atlantic measured as part of the TTO program ---------------- 42 Plot of salinity-normalized alkalinity against AOU for thermocline waters ----------------- 44 ZCO 2 versus density for thermocline waters forming in the north temperate Atlantic ------------- 45 Models for the ventilation of the North Atlantic temperate thermocline ------------------ 47 Comparison of model prediction with observation, for the difference between contemporary surface outcrop water ZCO 2 and contemporary thermocline respiration and CaC 03 dissolution-corrected ZCO 2 concentration, as a function of AOU for the thermocline waters ----- 48 Section of dissolved silica content along the western basin of the western Atlantic Ocean ------- 51 Salinity-normalized and AOU-corrected ZCO 2 concentrations versus oxygen content along the 36.95°/oo isopycnal horizon in the Atlantic Ocean - - - 55 Dissolved oxygen versus potential temperature for the 02 = 36.95°/oo isopycnal horizon north of 20°S - - - 56 "NO" versus potential temperature plots for deep waters in the Atlantic ------------------ 57 Alkalinity versus dissolved silica as measured during the TTO program along the o 2 = 36.95°/oo isopycnal horizon, and dissolved silica versus dissolved oxygen on the same horizon as measured during both GEOSECS programs --------------- 60 vi Figure Page 27 Box model for the ventilation of the 0£ = 36.95°/oo isopycnal horizon -------------------- 62 28 Anthropogenic ECO 2 excesses in the five deep water boxes of the model shown in Fig. 27 for the year 1980 ---------------------- 63 14 3 29 AC and H values for samples close to the 02 = 36.93°/oo isopycnal horizon as a function of latitude ------------------------ 66 14 14 30 Plot of bomb- C-corrected A C values against dissolved O 2 concentrations for samples from close to the 02 = 36.95°/oo isopycnal horizon ----- 67 31 1<+ C distribution as yielded by the five-box NADW model for the case of advection only ----------- 68 32 Comparison of the atmospheric equilibrium ZCO 2 value (as of 1980) and the respiration-and dissolution-corrected ZCO 2 content of water of O 2 = 243 ymol/kg on the a 2 = 36.95 °/00 isopycnal horizon ------------ 70 vii Y7 ■ ' - ■ .i.r n. ■!<' ■ * _ $■ ■’ ' 4 • ft.* ' 1. INTRODUCTION The very interesting idea that the preanthropogenic CO 2 content of the atmosphere might be reconstructed from the ocean's contemporary chem¬ istry was first published by Brewer (1978). He calculated respiration- corrected CO 2 partial pressures (pC 02 values) for a series of water sam¬ ples taken along the axis of the Antarctic Intermediate Water mass and found a decrease away from the source region toward a nearly constant value for the temperate South Atlantic. Assuming that the constant DCO 2 found well away from the source region represented waters formed during the preanthropogenic era, Brewer used the difference between this value and that for waters closer to the source region as a measure of the anthropogenic-induced increase in the CO 2 content of the atmosphere. He concluded that the preanthropogenic C0 2 content was about 50 ppm (volume) lower than that in 1972 (i.e., it was about 280 ppm). # Chen and Millero (1979) adopted a different approach to the same problem. They compared respiration and calcium carbonate dissolution- corrected ECO 2 contents for subsurface water with observations on contem¬ porary surface waters. The differences were attributed to the anthropo¬ genic changes in the atmospheric CO 2 content. Based on this analysis Chen and Millero (1979) obtain an estimate of 265 ± 30 ppm for the CO 2 content of the pre-1800 atmosphere. Chen pursued this idea in a series of subse¬ quent papers (Chen and Pytkowicz 1979, Chen 1980, Chen 1982a, Chen 1982b, Chen et al. 1982). These claims have not gone unchallenged. Shiller (1981) points out a number of serious problems associated with the approaches of Brewer and of Chen and Millero. In the case of Brewer's calculation, he points out that mixing effects along the path of the Antarctic Intermediate Water mass obscure the CO 2 message. In the case of Chen's calculations, he pro- l poses that uncertainties in the preformed ZCO 2 and alkalinity values, in the AZCO 2 /AO 2 ratio for respiration and in the assumption regarding O 2 and CO 2 equilibrium for surface waters, lead to large uncertainties in the results. Interest in these reconstructions stems from the controversy regard¬ ing the extent to which the CO 2 content of the atmosphere has been modi¬ fied by releases of the forest- and soil-derived C0 2 associated with land- use modifications. A wide range of estimates have been made. Some re¬ searchers claimed that the amount of forest and soil CO 2 produced over the last 20 years exceeds that generated by the burning of fossil fuel CO 2 (Woodwell et al. 1978). Others conclude that the amount of forest and soil CO 2 produced during this period was small compared to that from fos¬ sil fuel burning (Bolin 1977, Seiler and Crutzen 1980). To date, no mutu¬ ally satisfactory resolution of this question has been achieved. Consid¬ erable progress toward this end could be made if the CO 2 content of the atmosphere before large-scale land modification and large-scale fossil fuel combustion could be established. Because waters exist within the ocean which have been isolated from contact with the atmosphere since the advent of these anthropogenic effects, it is tempting to look to these waters for the answer. The purpose of this paper is to explore the likeli¬ hood that this approach will yield a definitive answer to the forest and soil CO 2 controversy. An examination of Fig. 1 helps to define this quest. This figure places the 6cean-derived preanthropogenic pCC >2 estimates of Brewer (1978) and of Chen and Millero (1979) in context with estimates based on ice core studies (Neftel et al. 1982), early direct measurements (Munch and Aubin 1886), and reconstructions based on fuel consumption data and on carbon isotope data from tree rings. The latter both employ a version of the 2 Fig. 1. Reconstructions of the CO 2 content of the atmosphere prior to the onset in 1958 of Keeling's precise atmospheric monitoring derived from ocean-atmosphere models. The upper curve is based on the inputs of fossil-fuel-derived CO 2 . The lower curve is based on inputs of both fossil fuel COg and forest and soil CO 2 (as de¬ convolved from the tree-ring-based atmospheric record, Peng et al., 1983). The ocean-chemistry-based estimate of the prean- thropogenic atmospheric content (Brewer 1978 and Chen and Millero 1979) apply to some unspecified time prior to 1800 (see crosses). The estimate based on measurements of the C02 - to-air ratio for bubbles in ice (Neftel et al. 1982) is for about 600 years ago. The range for direct measurements on pre-1900 nonurban air by Munch and Aubin (1886) is also shown. The circles represent the modern record of Keeling and his colleagues at Mauna Loa Observa¬ tory, Hawaii (Keeling et al. 1982). (1 atm * 1.01 x 10 5 Pa.) 3 Oeschger-type model of the ocean-atmosphere-biosphere system (Peng et al. 1983). The upper model curve is obtained by assuming that the forest-soil contribution to the rise in atmospheric CO 2 content is insignificant com¬ pared to the fossil fuel contribution. The lower curve deconvolved from the tree-ring-based atmospheric C/ C record of Freyer and Belacy (1983). Further carbon isotope studies by Stuiver et al. (in press) and Freyer (1984) indicate that the Freyer and Belacy record yields too high an estimate of the forest-soil contribution. Thus, the lower curve pro¬ bably represents a lower limit for the actual CO 2 content trend. As can be seen, the two scenarios in Fig. 1 straddle the reconstructions made by Brewer (1978) and by Chen and Millero (1979) and the ice core value. If a clear resolution is to be achieved, paleo-pC02 estimates firm to 10 x 10~ 6 atmospheres must be obtained. As we will show, to achieve this accuracy using ocean chemistry data is almost certainly an insurmountable task. 2. THE CHEN APPROACH Fig. 2 shows an idealized version of the Chen concept. Measurements on contemporary surface water samples are used to establish a relationship between salinity-normalized ECO 2 concentration and temperature (see upper band in Fig. 2). Respiration- and dissolution-corrected measurements on subsurface water are used to establish the relationship between salinity- normalized ECO 2 concentration and temperature for waters which contacted the atmosphere at times in the past (see curved band in Fig. 2). However, as all waters warmer than about 6°C are likely ventilated fast enough to have been significantly contaminated with anthropogenic CO 2 , only the coldest of these waters are representative of preanthropogenic condi¬ tions. Thus, in this diagram only for the coldest waters does the contem¬ porary subsurface band approach the sought-for preanthropogenic curve. I CO 2 (/xmol/kg) •"V e (°c) Fig. 2. Hypothetical salinity-normalized ECO 2 versus potential tempera¬ ture plot showing present-day surface waters (upper band), prean- thropogenic surface waters (lower band), and respiration- and dissolution-corrected contemporary subsurface waters (curved intermediate band). 5 For warmest waters, the contemporary subsurface band approaches the curve for contemporary surface waters. Because of this the Chen approach de¬ pends on the existence of universal ZC02”temperature relationships encom¬ passing all surface waters at any given time. The trend for preanthropo- genic waters is based on an extrapolation of respiration- and dissolution- corrected results obtained on samples from the cold deep ocean waters to the entire temperature range. The extrapolated portion of the lower band in Fig. 2 serves as the reference for the Chen-type estimates of anthropo¬ genic CO 2 excesses in various waters. The graphs of excess ECO 2 versus depth in the Chen papers are the differences between the respiration- and dissolution-corrected subsurface water results (middle band in Fig. 2) and the reconstructed trend for preanthropogenic surface waters (lower band in Fig. 2). In this regard it is important to keep in mind that the anthro¬ pogenic increase attributed to warm surface waters is actually based entirely on measurements on cold waters in the sea! As will be discussed in detail below, the major problem with the Chen approach is that the universal relationships he envisions almost certainly do not hold. The cold deep waters of the ocean all contain contributions from both the northern Atlantic and the Antarctic. Thus, the surface waters in these two regions are of major interest. It should be pointed out that in the calculations of Chen and his colleagues no northern Atlan¬ tic surface waters were included; none were available. The coldest sur¬ face water in the northern Atlantic, published as part of the GEOSECS pro¬ gram, had a temperature of 18°C. Further, none of the high southern-lati¬ tude surface water samples on which the Chen et al. equations are based were collected during the winter period when deep waters form. We will focus our major criticism on this point. 6 3. THE IDEAL OCEAN The best way to appreciate the problems associated with the Chen approach is to start with the idealized ocean depicted in Fig. 3. This ocean consists of discrete reservoirs bounded by isopycnal horizons. Within the body of this ideal ocean, no mixing occurs between these isopycnal units. Each unit is divided into an outcrop and interior por¬ tion. These subreservoirs are well mixed. The outcrop reservoirs main¬ tain pC0 2 values equal to those for the atmosphere. Water is exchanged between the outcrop and interior reservoirs, leading to gradual ventila¬ tion of the interior. The ventilation times for the interior reservoirs increase with depth (i.e., the shallow reservoirs more nearly keep pace with the rising atmospheric C0 2 content than do the deep reservoirs). All the reservoirs are assumed to have the same salinity and alkalinity. No organisms inhabit this idealized ocean. In such an ocean, one would have to compare only the pC0 2 values of outcrop and interior waters to get the desired answer. However, it must be kept in mind that this difference would vary from layer to layer because of the increase in ventilation time with depth. For the deepest layers, the ventilation time is so great that little anthropogenic altera¬ tion would have occurred. Hence, for these layers the outcrop-interior pC0 2 difference would reflect the full anthropogenic increase. For the shallowest layers where ventilation is rapid, very little difference in pC0 2 would exist between the outcrop and interior reservoirs. Provided the ventilation times for each interior reservoir are known, the temporal record of atmospheric C0 2 content could be deconvolved from this set of differences. However, were it known only that the ventilation time for the deepest layers is sufficiently large that no significant buildup of 7 Fig. 3. The idealized model of ocean ventilation described in the text. Each subsurface reservoir is linked to the atmosphere through an isopycnal outcrop. 8 excess CO 2 has occurred, then only the preanthropogenic atmospheric CO 2 content could be obtained. The time history for the increase would remain unknown. The real ocean differs from this ideal ocean in several ways impor¬ tant to the reconstruction of preanthropogenic CO 2 contents. (1) Respiration and CaC 03 dissolution raise the 2C0 2 content (and hence change the pC0 2 ) of waters after they leave the surface. Uncertain¬ ties associated with the procedures used to correct for these additions will lead to errors in the reconstruction. (2) Waters at the sea surface do not have CO 2 contents at equilibrium with the atmosphere. They range from 30% higher (in the equatorial Pacif¬ ic surface water) to 30% lower (in summer surface waters of the Norwegian Sea). They vary seasonally. Because of this, the answer one gets depends strongly on the manner in which the reference surface water values are established. (3) As shown from radiocarbon measurements, surface waters in the Antarctic achieve only a small fraction of isotopic equilibrium with the atmosphere (Weiss et al. 1979). Although chemical equilibrium is achieved in surface waters more rapidly than is isotopic equilibrium (Broecker et al. 1980), it is unlikely that these waters remain at the surface long enough to achieve C0 2 equilibrium with the atmosphere. Likewise, they will show only a partial response to the increased atmospheric levels. Unless this deficiency in response is taken into account then as pointed out by Shiller (1981), the Chen scheme will underestimate the anthropo¬ genic increase. This is a subtle but exceedingly important point. 9 (4) All subsurface waters old enough to be of interest consist of contributions from more than one source. Because of regional differences in the properties of surface water, it is necessary to quantify the frac¬ tional contribution of each source. Uncertainties in the procedures used for this separation will lead to errors in the reconstruction. (5) The water in any given isopycnal horizon is not well mixed, nor does the water leaving the surface move along these surfaces in freight- train fashion. The real situation is complex. The action of eddies of all sizes smears the identity of the water produced during any given win¬ ter as it moves through the system. Thus, all subsurface waters are mix¬ tures of contributions covering a wide range of time. Because of this, reconstructions can only be made within the context of a mixing model. Again, inadequacies in the mixing model adopted will lead to errors in the reconstruction. (6) Diapycnal mixing occurs in the real ocean, further complicating the situation. 4. CORRECTION OF SUBSURFACE WATER COMPOSITIONS FOR RESPIRATION The real ocean is not sterile. Organic matter formed in the photic layer is in part consumed by animals and bacteria living within the body of the sea. This process leads to an increase in the ECO 2 content of sub¬ surface waters. Brewer (1978) and Chen and Millero (1979) correct for this alteration by assuming that: (1) the difference between the saturation-oxygen content and the ob¬ served oxygen content is a measure of the amount of oxygen consumed via respiration; (2) the ratio of CO 2 released to O 2 utilized during respiration is constant throughout the sea and equal to that obtained for marine organic matter with the ideal (i.e. Redfield) composition adopted by marine 10 NO. OF SAMPLES Of s - of 1 (/* mol /kg) Fig. 4. Histogram of excess dissolved oxygen contents found in the sur¬ face ocean during the GEOSECS, NORPAX and TTO programs (Bain- bridge 1981, Broecker et al. 1982 , Weiss et al. 1983 , Williams 1981a and 1981b). As the solubility of oxygen ranges from 220 pmol/kg for the warmest surface waters to 350 ymol/kg for the coldest surface waters, the mean excess of 7 pmol/kg corresponds to a 2 to 3 % supersaturation. 11 biologists and chemists (i.e., ACO 2 /AO 2 = -0.768), While no one would deny that such a correction is appropriate, nei¬ ther assumption is perfect. These imperfections lead to significant errors in the reconstructed ZCO 2 values. The problem with the first as¬ sumption is that surface ocean waters are almost always found to be super¬ saturated with dissolved oxygen. As shown in Fig. 4, this supersaturation averages about 7 ymol/kg. If this average supersaturation were taken as universal, the SCO 2 correction would have to be increased by 5 ymol/kg (7 x 0.76). A 5-ymol/kg ECO 2 content increase corresponds to an 8 x 10 -& atm pC02 increase (for cold water). It must be kept in mind that winter dissolved-oxygen concentrations may not show the oceanwide average O 2 supersaturation. In cases of rapid deep convection, the water brought to the surface might not have time to equilibrate with the atmosphere, resulting in undersaturation. The second assumption leads to comparable errors. The composition of the average organic matter undergoing respiration in the sea does not necessarily have the ideal Redfield oxygen demand. One factor of impor¬ tance is the ratio of nitrogen to carbon in organic matter. Another is the nature of the carbon-bearing compounds (fats, for example, require more O 2 per unit of carbon than do carbohydrates). Unfortunately, the composition of marine organic detritus is not well documented. Measure¬ ments of the C/N ratio in the bulk organic matter filtered from subsurface water yield values in the range 7.0 to 7.5 (J.K.B. Bishop, personal com¬ munication). If the average C/N ratio is taken to be 7.25 and the oxygen demand per carbon atom is taken to be 1 and per nitrogen atom to be 2, then the ratio of £C02 production to O 2 utilization is 0.78. Because the O 2 demand per carbon atom is likely to be somewhat greater than unity, the 12 coefficient of 0.78 must be considered a maximum. To be more conservative we will take the C/N ratio in marine organic material to be equal to or less than 7.7 and an O 2 demand per carbon atom equal to or greater than unity. This yields an upper limit of 0.80 for the ZCO 2 production to oxygen utilization ratio. Takahashi et al. (in press) analyzed the chemical variations along individual isopycnal horizons in the thermocline of the Atlantic and Indian oceans. In these analyses they show the importance of accounting for the mixing between waters generated in the north and south. They obtain estimates of the ratio (AECO 2 )-(ACa )/02 and of the ratio (A 02 )“( 2 AN 03 )/ 02 * Because of the temporal increase in the amount of fossil fuel CO 2 in surface waters descending into the thermocline, the former of these ratios can be shown to provide only a lower limit on the amount of carbon released per unit of oxygen consumed. Because the O 2 demand per carbon atom is greater than unity, the latter of these ratios provides only an upper limit on the amount of carbon released per unit of oxygen consumed. For the 27.0°/oo isopycnal in the Atlantic, these two limits require that the ratio of ECO 2 released to oxygen consumed lie between 0.61 and 0.79 and for the 27.0°/oo isopycnal in the Indian Ocean between 0.66 and 0.82. Based on these limits, the actual AC or &/-A02 might differ signifi¬ cantly from the value of 0.77 adopted by previous workers. It could range from the provisional upper limit of 0.80 given above down to 0.64. The best correct estimate for its value is 0.72±0.08. As the oxygen deficien¬ cies for subsurface waters range up to 300 ymol/kg (see Fig. 5), the uncertainty ECO 2 correction likely ranges up to 24 ymol/kg. For waters of 13 DEPTH (km) DEPTH (km) Fig. 5. Sections of AOU (apparent oxygen utilization; i.e., saturation O 2 content at the potential temperature of the water minus the ob¬ served O 2 content) for the western Atlantic and western Pacific oceans. These sections are based on the GEOSECS data set (Bain- bridge 1981, Broecker et al. 1982). The dotted lines show the topography of selected isopycnal horizons. 14 the deep Atlantic which have, for the most part, AOU values less than 120 ymol/kg, the error associated with the choice of a respiration coefficient would be less than 10 ymol/kg. For waters in the deep Pacific which have generally suffered O 2 depletions between 140 and 220 ymol/kg (see Fig. 5), the error due to this source is in the range of 10 to 16 ymol/kg. The ECO 2 contents of subsurface waters are also altered by the dis¬ solution of calcium carbonate. As the solution of CaC 03 changes the alkalinity of the water, this process leaves its independent mark. In the same way that respiration effects can be assessed from the decrease in oxygen content, calcium carbonate dissolution can be assessed by an in¬ crease in alkalinity. There are two complications. First, the generation of nitric acid during respiration also changes the alkalinity of sea water. Second, unlike dissolved oxygen, alkalinity has no thermodynamic reference state in surface water. Following Brewer et al. (1975) the nitrate complication is handled by using the equation: AC CaC0 3 = 1/2(AALK + ANO 3 ), where A refers to the difference between the concentrations in subsurface and surface water. As the invasion of anthropogenic CO 2 into the sea has probably not as yet significantly altered the alkalinity of any of its waters (through excess CaC 03 dissolution), the alkalinity measured for contemporary surface water can, in principle, be used as a measure of the initial alkalinity for subsurface waters. The same is true for nitrate. The problem, of course, is to sample the appropriate surface waters in the source regions during the winter episodes of active ventilation. Little such work has been done. For some subsurface waters the initial 15 alkalinity and nitrate contents can be estimated by extrapolating regres¬ sions of these parameters against dissolved-oxygen along isopycnal hori¬ zons to the saturation dissolved oxygen value (see Takahashi et al. , in press). However, as deep water isopycnal horizons generally do not out¬ crop, this procedure is confined to intermediate and thermocline waters. Surface waters in today’s ocean show a considerable range in salin¬ ity-normalized alkalinity and in nitrate content (see Fig. 6). Hence, the choice of the appropriate surface equivalent is important. This job is simplified by plotting salinity-normalized alkalinity and nitrate content against temperature. The range of the salinity-normalized alkalinity val¬ ues at a given temperature is smaller than the overall range in alkalin¬ ity. Even so, for a given temperature, significant differences in the salinity-normalized alkalinities are found between the cold summer surface waters of the northern Atlantic, northern Pacific, and Antarctic. As can be seen from Fig. 6, cold waters in the northern Atlantic have alkalini¬ ties about 40 peq/kg lower than those in the Antarctic. Thus, it is im¬ portant to assess the fraction of any deep water with an origin in the northern Atlantic and the fraction with an origin in the Antarctic. This currently cannot be done in the deep Pacific and Indian oceans to better than ±25%. Hence, the preformed alkalinity of any deep water from these oceans cannot be determined to better than ±10 peq/kg. This is equivalent to an uncertainty of ±5 Praol/kg of ECO 2 [i.e., to about ±8 x 10 -6 atm PCO 2 ]• It should be emphasized that the data in Fig. 6 are all for summer surface waters. The pattern for winter surface waters is likely somewhat different. 16 Fig. 6. Plots of salinity normalized alkalinity and of nitrate versus temperature for surface waters from high latitudes. Note the marked differences in the trends for the Antarctic, northern Pac¬ ific, and the northern Atlantic waters! The Antarctic measure¬ ments are from the GEOSECS expeditions to the southern Atlantic, Pacific, and Indian oceans (Bainbridge 1981, Broecker et al. 1981, Weiss et al. in press, Takahashi et al. 1980). The North Atlantic measurements are from the TTO expedition (Williams 1981b). 17 5. pC0 2 FOR WATERS CURRENTLY DESCENDING INTO THE DEEP SEA The respiration- and dissolution-corrected LCO 2 and alkalinity values obtained for subsurface waters using the procedures outlined above can be used to calculate pCO£ for the water when it left the sea surface. If waters leaving the surface were always to have pC 02 values equal to those in the atmosphere, then the difference between these respiration- and dissolution-corrected pC 02 values and the current atmospheric value would provide a direct estimate as to how much the atmosphere had increased in pC 02 since the time the water making up the sample left at the sea surface. Unfortunately, waters descending into the deep sea very likely do not carry the same pC02 as the atmosphere. Herein lies the greatest problem with the Chen approach. Were the errors associated with the respiration and dissolution corrections the only ones,then an uncertainty —6 of 10 Pmol/kg for the excess EC02 contents [equivalent to -15 x 10 atm in PCO 2 ] might be achievable. Unfortunately, there are comparable or even larger uncertainties associated with the fact that the waters currently descending into the deep sea are not at equilibrium with the pC 02 of the atmosphere. Fig. 7 shows a world map of pC02 anomalies for surface water. As can be seen, deviations from the atmospheric pC 02 exist which are larger in magnitude than the difference between the extreme estimates for the preanthropogenic atmospheric CO 2 content. As can be seen in Fig. 8, a rather large range of pC 02 values are found even at a single temperature. Furthermore, as GEOSECS and TTO expeditions went to high latitudes during the peak summer months, none of the results from high latitudes were obtained during the times when deep waters form. The importance of season 18 Fig. 7. Map showing the geographic pattern for the surface ocean- atmosphere pC02 difference. The anomalies are in 10 -6 atm (1 atm = 1.013 x 10 Pa). 19 is clearly apparent for the northern Atlantic. Summer surface waters —6 colder than 3°C have pC02 values less than 200 x 10 atm. As we will show below, observations on deep water samples in the northern Atlantic show that the pC02 for the winter waters less than 3°C in temperature are almost certainly greater than 260 x 10 -6 atm. Thus, in the region of the ocean where roughly one-half of the ocean's deep water forms, summer pC 02 values are totally misleading, so much so that the strict application of the Chen procedure would lead to the conclusion that the atmosphere's pC02 has been falling rather than rising! There is no reason to believe that use of summer values for Antarctic surface waters is likely to be any less misleading. As the regions sam¬ pled have strong submixed layer haloclines, it is difficult to believe that the available pCC >2 values have anything to do with the winter values in the regions like the Weddell Sea where deep water is formed. In this regard it is important to keep in mind that the time required for the pCC >2 in the mixed layer of the ocean to achieve equilibrium with the atmosphere is one half year or more [see Broecker and Peng (1982) for a discussion of this point]. Thus, gas exchange between the surface ocean and the atmosphere cannot keep pace with the seasonal temperature changes or algal blooms that drive its pC02 away from the atmospheric value. Fur¬ ther, in areas where strong upwelling or deep convection occurs, the resi¬ dence time of water at the surface may not be long enough to permit equilibrium with the atmosphere to be achieved. Putting aside biological effects, the winter pC02 for surface waters would be expected to be smaller than the annual average. The reason is that for a given salinity, alkalinity, and total dissolved inorganic 20 400 300 6 < 0 ° 'o W 200 0 400 U CL 200 «-r-" 1 • 400 -,-1-r: • • • • • M • • • • • • •• • • . * •« .. \ • . •• • • . • . • . • •% • • * . t - • • •*. 1. •• • •• •• • . • . GEOSECS 300 • ' • A. *- . • . * •; *\ • •• • * • # • . • • . • CENTRAL AND SOUTHERN ATLANTIC SURFACE WATERS _ 200 GEOSECS * PACIFIC OCEAN SURFACE WATERS —■■ ■ 1 . .. L. 20 30 10 20 30 -1- I -1- -,- . • 400 - m • — • • • • • • • • .V * •* ■» . * * V . ?•* • . • •* * * • . •. • « # « • • • • • 300 - . • • m GEOSECS .. < - - TTO INDIAN OCEAN • • • • NORTHERN ATLANTIC SURFACE WATERS • • * • • • • SURFACE WATERS 1 1 • * • 1 •• ..—. 1 . 1 10 20 30 10 20 30 TEMPERATURE (°C) Fig, 8, pC02 versus temperature for surface water samples analyzed as part of the GEOSECS and TTO programs. The values are calculated from the titrator ZCO 2 and alkalinity results (Takahashi et al. , 1980) (1 atm = 1.013 x 10 5 Pa). 21 carbon content, the pC0 2 drops about 4% per °C the water is cooled. The example in Fig. 9 shows the response expected for a mixed layer thickness of 80m and a C0 2 piston velocity of 3 m/d. As can be seen, the winter water has a pC 02 significantly lower than the mean for the year. Biological effects will work in the opposite direction. Photosyn¬ thesis during the polar summer will draw down the surface ECO 2 and hence PCO 2 values. Fig. 10 helps illustrate what in our estimation is the major flaw in the Chen approach. Shown in the left-hand panel of this figure is the ideal hypothetical situation where the atmosphere and surface ocean every¬ where maintain PCO 2 equilibrium. In this case the difference between pC 02 , reconstructed for deep waters and that of the current atmosphere provide information regarding the magnitude of the anthropogenic effect. Shown in the right-hand panel is the more realistic hypothetical situation where winter surface water in the high latitudes has pC 02 values lower than that for the atmosphere. In this case, the respiration- and dissolution-corrected pC 02 for deep waters is lower than the atmospheric value, in part because the source waters supplying the deep sea leave the surface with a pC 02 lower than the atmospheric value and in part because the atmosphere had a lower pCC >2 when these waters descended. Were this the case, the difference between the contemporary atmospheric PCO 2 and that reconstructed for deep water would be larger than the anthropogenic increase for the atmosphere. 6 . DEPENDENCE OF ANTHROPOGENIC EFFECT ON VENTILATION TIME All subsurface waters represent an integration of inputs of widely different ages. The concept that a single atmospheric contact time can 22 MONTH Fig. 9. The annual cycle of pC02 calculated for the ocean mixed layer of constant thickness, alkalinity, and salinity. A piston velocity of 3 m/d was used for CO 2 exchange (1 atm = 1.013 x 10 Pa). 23 TEMPERATURE SURF. H 2 0(°C) 350 E o LD I o (M 300- Q- 250 - 350 30 0 30 LATITUDE 6 o iO I O CM O O CL 300 —1— —1— —1— —,- T- 2 50 - ATM i - SURF J 980 ” sJrf } 1800 200 L_ 90 S _L X 60 30 0 30 LATITUDE 60 90 N Fig. 10. Comparison of the anthropogenic change for two hypothetical ocean situations. In the one on the left the pC02 for the waters ven¬ tilating the deep sea is taken to be equal to the atmospheric value. In the one on the right the pC02 for the waters venti¬ lating the deep sea is assumed to be less than atmospheric. 24 be assigned to a given sample is simply not a valid one. The deep water injection from any given winter is mixed with those from many previous winters. The nature of this problem is easily shown using the two-box mod¬ el shown in Fig. 3. The outcrop reservoir is assumed to remain at equi¬ librium with the atmosphere. The interior reservoir is well mixed and ex¬ changes water with the outcrop reservoir. Table 1 shows the difference be¬ tween the anthropogenic CO 2 excess in these two reservoirs (as of 1980) for various ventilation times (i.e., volume/exchange flux). These calcula¬ tions were carried out for both the fossil-fuel-C02~only scenario and for the C deconvolution scenario (see Fig. 1). As shown in the table, even waters ventilated on the time scale of 300 years show 9 to 19% of the full effect. Because the entire Atlantic is thought to be ventilated on this time scale, virtually all of its waters likely contain anthropogenic CO 2 ! 7. RECONSTRUCTION BASED ON MEASUREMENTS ON THERMOCLINE WATERS To make the difficulties outlined above more specific, we will make use of the extensive results obtained during the TTO program in the North Atlantic. Over the period April to November 1981 a survey of the Atlantic Ocean north of 20°N was carried out on the research vessel. A map showing the track followed during this expedition is shown in Fig. 11. Measure¬ ments of ZCO 2 and total alkalinity were carried out (using the Bainbridge automated titrator) by the members of the Scripps PCODF group with stan¬ dards and data reduction procedures prepared by Brewer and Bradshaw of Woods Hole. Comparison of these results with those obtained by the GEOSECS pro¬ gram in the same region (see Table 2) suggests that the TTO alkalinities 25 Table 1. Results of two-box model calculations for anthropogenic CO 2 scenarios 3 ZCO 2 difference^ Ventilation time (yr) 13 C Deconvolution scenario (pmol/kg) Fossil-fuel-C02 scenario (pmol/kg) 0 0 (100%) c 0 (100%) 3 2 (97%) 2 (93%) 10 7 (89%) 7 (76%) 30 19 (74%) 14 (51%) 100 41 (43%) 22 (24%) 300 59 (19%) 26 (9%) OO 73 (0%) 29 (0%) a See Fig. 2 for anthropogenic CO 2 scenarios. b lC02 Surface - £C0 2 deep as of 1980. c The numbers in parentheses represent the fraction of the equilibrium anthropogenic uptake calculated for the subsurface reservoir. 26 Table 2. Comparison of GEOSECS and TTO ZCO 2 and alkalinity results for deep water samples (Takahashi et al. 1980, Williams 1981b) Station No. Alkalinity a (peq/kg) ZC02 b (ymol/kg) GEOSECS TTO GEOSECS TTO A d GEOSECS TTO A d 3 l c 40 2342 2335 7 2204 2182 22 32 c 38 2343 2335 8 2202 2182 20 33 c 36 2353 2332 21 2206 2178 28 34 c 34 2344 2341 3 2203 2178 15 36 32 2338 2342 -4 2190 2187 3 113 71 2348 2341 7 2194 2185 9 Mean (all stations) 7 16 Mean (36-32 and 115- 71) 2 6 formalized to H 4 Si0 4 content of 30 ymol/kg using a regression slope of AAlk/AH4Si04 = 1 . 0 . ^Normalized to H 4 Si 04 content of 30 ymol/kg using a regression slope of AZC 02 /AH 4 Si 04 = 0.8. c Leg 3; titration system subject to problems corrected on subsequent legs. d A = GEOSECS - TTO. 27 are about 7 yeq/kg lower than those for GEOSECS. A similar comparison for ZCO 2 suggests that the TTO results are 16 ymol/kg lower than those for GEOSECS. However, If the GEOSECS leg 3 results are disregarded, because of difficulties with the titration system experienced during the early part of the GEOSECS expedition, the alkalinity difference drops to 2 peq/kg and the ZC02 difference drops to 6 Mmol/kg. As we do make minor use of the GEOSECS data, these differences should be kept in mind. Comparisons between Keeling’s ZCO 2 results on TTO samples and the titrator ECO 2 results from the TTO program show an unexplained depth de¬ pendent difference. The difference for surface waters averages 16 pmol/kg (titrator results higher). This difference decreases with depth, reach¬ ing close to zero at 1000 m. Below 1000 m it rises again to about 4 pmol/kg, remaining at this value throughout the deep water column. As ICO 2 values calculated from pC02 measurements of Takahashi and Chipman and the titrator alkalinity measurements yield ZCO 2 values which agree well with the Keeling ZCO 2 values, the error must lie in the titrator ECO 2 results. This error should also be kept in mind in connection with the following analysis. In Fig. 12 we show a plot of the salinity-normalized ECO 2 concentra¬ tions for surface waters and the respiration-corrected and salinity- normalized ZCO 2 concentrations for thermocline waters against potential temperature for the region north of 18°N in the Atlantic. The data used to construct this diagram are listed in Table 3 (surface waters) and Table 4 (thermocline waters). A respiration coefficient of 0.80 pmol ZC02/kg per ymol/AOU/kg (i.e., the upper limit given earlier in this paper) was used 28 o Large Volume Transient Tracers in the Ocean 1981 40W 20W a Small Volume I 80W 60W Fig. 11. Map showing the location of Transient Tracers in the Ocean (TTO) stations in the North Atlantic* 29 Table 3. Surface water data for the North Atlantic, obtained as part of TTO program (Williams 1981b) a Stn Lat • Long. T S EC0 2 EC0 2 N Aik AlkN No. (°N) (°w) (°c) (°/oo) (ymol/kg) (ymol/kg) (yeq/kg) (yeq/kg) 115 46.0 11.0 15.6 35.56 2059 2027 2333 2296 115 46.0 11.0 15.5 35.55 2062 2030 2332 2296 116 49.1 12.9 14.7 35.57 2069 2036 2357 2319 117 49.5 16.7 14.8 35.49 2043 2015 2330 2298 118 49.7 22.0 15.5 35.59 2048 2015 2335 2297 118 49.7 22.0 15.5 25.58 2059 2025 2342 2304 120 51.8 26 8 12.6 35.11 2061 2054 2326 2318 121 52.6 27.0 12.3 35.20 2073 2061 2328 2315 122 52.2 30.9 10.7 34.81 2072 2083 2300 2313 123 52.7 33.6 10.0 34.69 2066 2084 2296 2317 125 55.0 34.7 10.1 34.69 2050 2068 2286 2306 126 57.0 32.3 10.2 34.89 2075 2082 2295 2302 130 57.7 26.2 11.3 35.00 2374 2074 2309 2309 132 57.4 22.4 12.4 35.23 2067 2054 2333 2318 136 56.3 13.3 13.1 35.31 2077 2059 2328 2308 137 56.1 12.8 12.7 35.29 2066 2049 2319 2300 138 55.8 12.0 12.7 35.28 2066 2049 2322 2303 141 58.6 11.5 12.3 35.31 2079 2061 2332 2311 141 58.6 11.5 12.2 35.31 2082 2064 2328 2308 142 61.3 8.0 10.5 35.24 2084 2070 2321 2305 142 61.3 8.0 10.5 35.24 2089 2075 2322 2306 142 61.3 8.0 10.2 35.24 2098 2084 2323 2307 160 61.5 18.3 11.4 35.07 2071 2067 2313 2308 161 61.7 19.1 11.3 35.10 2071 2065 2313 2306 161 61.7 19.1 11.3 35.10 2067 2061 2315 2308 162 61.8 19.9 11.0 35.12 2061 2054 2318 2310 163 61.8 28.5 9.9 34.83 2069 2079 2307 2318 164 62.5 30.4 10.1 34.85 2064 2073 2313 2322 164 62.5 30.4 10.1 34.85 2062 2071 2314 2323 164 62.5 30.4 10.1 34.85 2062 2071 2317 2326 167 54.1 33.3 10.1 34.95 2057 2060 2323 2326 167 64.1 33.3 10.1 34.95 2058 2061 2322 2325 168 64.3 33.7 10.0 34.98 2063 2064 2324 2325 171 63.7 33.3 10.4 34.83 2062 2072 2316 2327 172 58.5 30.0 11.1 34.92 2067 2072 2306 2311 174 57.6 35.5 10.2 34.68 2076 2095 2309 2330 206 53.9 39.9 10.9 34.46 2056 2088 2280 2316 207 53.1 37.8 12.3 34.47 2038 2069 2285 2320 208 51.5 37.9 13.2 34.46 2032 2064 2288 2324 209 50.1 37.8 15.2 34.68 2032 2051 2298 2319 aZC0 2 N = eco 2 (35.0/S) ; Alk N = Aik (35.0/S). 30 Table 4. Alkalinity and EC0 2 data for TTO thermocline samples from the North Atlantic (Williams 1981b) a Stn 0 S AOU Aik AN0 3 + Alko EC0 2 EC0 2 s ZC0 2 ° AEC0 2 No. (°C) (°/oo)(ymol/kg)(yeq/kg)(ymol/kg)(yeq/kg)(ymol/kg)(ymol/kg) (ymol/kg) (ymol/kg) 15 to 16°C 16 15.9 36.16 50 2369 6 2299 2126 2017 2019 -2 36 15.9 36.20 54 2361 5 2289 2116 2017 2004 13 39 15.8 36.18 63 2367 5 2294 2114 2018 1996 22 229 15.8 36.17 45 2371 5 2299 2114 2018 2010 8 233 15.8 36.16 47 2368 4 2296 2103 2018 1999 19 47 15.7 36.10 21 2368 2 2298 2094 2019 2014 5 26 15.8 36.17 68 2367 6 2296 2132 2020 2010 10 52 15.4 36.11 40 2363 4 2294 2112 2022 2016 6 96 15.3 36.08 37 2363 4 2296 2111 2024 2019 5 227 15.3 36.08 46 2364 5 2298 2121 2024 2022 2 230 15.3 36.08 40 2367 5 2301 2117 2024 2023 1 15 15.2 36.05 70 2373 8 2312 2142 2025 2025 0 20 15.2 36.05 66 2365 8 2304 2134 2025 2021 4 38 15.2 36.07 64 2365 7 2302 2128 2025 2015 10 111 15.0 36.02 9 2363 1 2297 2088 2027 2022 5 14 to 15°C 16 14.6 35.97 78 2359 8 2303 2136 2032 2018 14 22 14.7 35.98 74 2357 8 2301 2137 2031 2021 10 24 14.9 36.01 74 2366 6 2306 2129 2029 2012 17 26 14.8 36.00 80 2357 8 2299 2135 2030 2013 17 40 14.7 35.97 68 2364 8 2308 2137 2031 2026 5 41 14.5 35.94 69 2355 9 2302 2133 2035 2023 12 42 14.7 35.99 57 2353 8 2296 2114 2031 2012 19 43 14.6 35.96 59 2364 8 2309 2125 2032 2022 10 44 14.2 35.91 53 2354 8 2302 2126 2036 2031 5 45 14.6 35.97 43 2356 6 2298 2113 2032 2023 9 45 14.6 35.97 43 2357 6 2299 2122 2032 2031 1 47 14.7 35.99 34 2367 4 2306 2106 2031 2022 9 49 14.1 35.90 46 2354 7 2302 2123 2038 2034 4 68 14.9 36.06 48 2366 5 2301 2120 2028 2020 8 71 14.7 36.01 45 2361 5 2300 2114 2031 2020 11 76 14.5 35.99 49 2364 5 2304 2122 2033 2026 7 80 14.2 35.94 47 2355 6 2299 2124 2036 2032 4 83 14.4 35.97 46 2359 5 2300 2118 2034 2025 9 109 14.8 36.00 31 2362 3 2299 2109 2030 2026 4 110 14.6 35.91 14 2361 2 2303 2098 2032 2034 -2 226 14.6 35.93 53 2356 6 2301 2128 2032 2032 0 229 14.4 35.92 51 2357 7 2304 2123 2034 2029 5 231 14.7 36.00 39 2364 5 2303 2124 2031 2035 -4 233 14.4 35.92 61 2355 8 2303 2129 2034 2027 7 13 to 14°C 15 13.6 35.81 88 2356 11 2314 2140 2044 2023 21 16 13.5 35.78 93 2342 10 2301 2147 2046 2027 19 20 13.9 35.86 79 2360 10 2313 2142 2039 2029 10 22 13.0 35.73 92 2349 10 2311 2153 2050 2037 13 31 Table 4 Continued Stn 0 S AOU Aik AN0 3 + Alk° ZC0 2 ZC0 2 s ZC0 2 ° AZC0 2 No. (°C) (°/oo)(ymol/kg)(yeq/kg)(ymol/kg)(yeq/kg)(ymol/kg)(ymol/kg) (ymol/kg) (pmol/kg) 13 to 14°C 26 13.9 35.88 88 2359 10 2311 2145 2039 2024 15 36 13.9 35.88 83 2354 9 2305 2143 2040 2026 14 38 13.1 35.77 90 2347 10 2306 2156 2050 2039 17 39 13.9 35.86 66 2350 10 2304 2128 2040 2025 15 42 13.1 35.72 72 2345 9 2307 2135 2049 2036 13 45 13.5 35.80 55 2357 8 2312 2125 2045 2034 11 46 13.6 35.82 55 2353 6 2305 2127 2044 2035 9 47 13.7 35.84 39 2348 6 2299 2126 2042 2046 -4 47 13.1 35.76 43 2354 5 2309 2119 2049 2040 9 68 13.8 35.87 53 2350 (7) 2300 2125 2041 2032 9 71 13.1 35.77 58 2358 8 2315 2134 2050 2043 7 76 13.1 35.79 62 2348 7 2303 2137 2049 2041 8 83 13.2 35.79 52 2350 7 2305 2131 2048 2043 5 107 13.1 35.76 50 2350 6 2306 2133 2049 2049 0 109 13.3 35.80 42 2345 5 2298 2122 2046 2042 4 110 13.0 35.74 39 2347 5 2303 2119 2051 2045 6 111 13.5 35.82 32 2345 4 2295 2110 2045 2037 8 112 13.6 35.91 29 2349 3 2292 2118 2044 2042 2 227 13.6 35.82 34 2349 5 2300 2107 2043 2032 11 228 13.4 35.77 50 2349 7 2305 2122 2046 2037 9 231 13.6 35.82 56 2349 7 2302 2138 2043 2045 -2 12 to 13°C 16 12.6 35.66 108 2339 12 2308 2162 2055 2037 18 20 12.9 35.71 87 2346 10 2309 2153 2051 2042 9 26 12.8 35.71 101 2353 11 2317 2158 2053 2036 17 40 12.5 35.65 92 2345 10 2312 2158 2057 2046 11 41 12.5 35.64 82 2341 11 2310 2156 2057 2053 4 43 12.9 35.70 78 2350 9 2313 2140 2051 2037 14 44 12.9 35.71 59 2351 8 2312 2131 2052 2042 10 45 12.2 35.62 72 2340 9 2308 2138 2060 2044 16 45 12.4 35.66 63 2342 9 2308 2140 2057 2051 6 47 12.7 35.70 51 2345 6 2305 2126 2054 2044 10 47 12.2 35.64 53 2352 7 2317 2130 2060 2050 10 49 12.7 35.71 56 2344 7 2304 2130 2054 2044 10 68 12.9 35.71 76 2342 ( 8 ) 2303 2140 2052 2038 14 68 12.4 35.70 67 2348 8 2310 2145 2057 2050 7 71 12.3 35.68 63 2341 8 2304 2137 2059 2047 12 80 12.8 35.74 59 2351 7 2309 2147 2053 2056 -3 83 12.0 35.64 58 2336 8 2302 2132 2062 2048 14 96 12.8 35.74 53 2348 6 2305 2126 2052 2040 12 99 13.0 35.75 52 2352 6 2309 2130 2051 2045 6 99 12.1 35.64 58 2336 8 2302 2132 2061 2048 13 107 12.0 35.63 59 2338 8 2305 2139 2062 2055 7 109 12.3 35.67 49 2344 6 2306 2132 2058 2053 5 111 12.5 35.69 37 2336 5 2296 2120 2054 2050 4 112 12.3 35.70 43 2334 5 2293 2124 2058 2049 9 32 Table 4. Continued Stn 9 S AOU Aik AN0 3 + Alk° ZC0 2 ZC0 2 s ZC0 2 ° AZC0 2 No. (°C) (°/oo)(pmol/kg)(yeq/kg)(ymol/kg)(jjeq/kg)(ymol/kg)(ymol/kg) (ymol/kg) (ymol/kg) 12 to 13°C (continued) 114 12.6 35.63 18 2334 1 2294 2103 2053 2052 1 114 12.3 35.62 18 2333 2 2294 2107 2059 2056 3 115 12.6 35.59 6 2331 0 2292 2095 2054 2055 -1 227 12.2 35.60 80 2338 11 2310 2150 2059 2051 8 231 12.4 35.65 76 2347 9 2313 2147 2058 2048 10 233 12.9 35.69 79 2342 9 2306 2132 2051 2029 22 11 to 12°C 15 11.5 35.50 118 2337 16 2320 2176 2068 2052 16 16 11.1 35.45 122 2333 16 2319 2182 2073 2058 15 20 11.8 35.54 101 2341 13 2318 2170 2065 2057 8 22 11.0 35.46 114 2344 16 2330 2177 2074 2059 15 24 11.0 35.46 118 2333 16 2319 2170 2073 2049 24 43 11.0 35.45 101 2338 15 2323 2162 2074 2055 19 44 11.2 35.50 83 2337 13 2318 2155 2071 2059 12 45 11.1 35.48 88 2333 13 2314 2163 2073 2064 9 46 11.5 35.57 74 2348 10 2320 2147 2068 2054 14 47 11.8 35.59 68 2336 9 2306 2141 2064 2052 12 47 11.5 35.57 58 2341 10 2313 2140 2068 2060 8 47 11.1 35.53 68 2334 11 2310 2144 2073 2058 15 49 11.7 35.59 62 2335 10 2306 2135 2066 2051 15 71 11.4 35.58 75 2347 10 2319 2155 2069 2061 8 76 12.0 35.65 67 2345 9 2311 2145 2062 2053 9 76 11.1 35.56 87 2334 12 2309 2154 2072 2052 20 80 11.9 35.64 71 2339 10 2307 2148 2064 2054 10 83 11.2 35.56 70 2340 11 2314 2149 2071 2060 11 96 11.9 35.62 62 2339 9 2307 2141 2063 2055 8 96 11.1 35.53 67 2330 11 2306 2141 2073 2056 17 99 11.1 35.52 53 2340 9 2315 2141 2073 2068 5 107 11.3 35.55 51 2336 9 2309 2140 2071 2067 4 109 11.4 35.56 53 2332 8 2340 2139 2069 2064 5 110 11.1 35.54 58 2341 10 2315 2148 2072 2070 2 111 12.0 35.62 39 2334 6 2299 2121 2062 2053 9 111 11.5 35.57 48 2336 8 2307 2132 2068 2060 8 112 11.8 35.64 45 2334 8 2300 2130 2064 2056 8 112 11.6 35.64 55 2328 8 2294 2139 2067 2057 10 115 11.8 35.60 15 2330 3 2294 2108 2064 2061 3 115 11.0 35.52 36 2323 8 2297 2129 2073 2069 4 226 11.9 35.55 84 2344 12 2320 2151 2063 2052 11 228 11.9 35.56 85 2340 12 2315 2156 2063 2055 8 229 11.7 35.53 81 2337 12 2314 2149 2065 2053 12 230 11.5 35.51 88 2337 11 2314 2162 2068 2062 6 231 11.4 35.50 91 2336 13 2316 2157 2059 2055 4 10 to 11°C 16 10.4 35.35 131 2329 19 2325 2188 2081 2063 18 20 10.8 35.43 114 2338 16 2326 2172 2076 2056 20 26 10.9 35.35 137 2339 18 2321 2185 2075 2055 20 33 Table 4 Continued Stn. 6 S AOU Aik AN0 3 + Alk° ZCO 2 ZC02 S ZCO 2 0 AECO 2 No. (°C) (°/oo)(ymol/kg)(peq/kg)(ymol/kg)(p eq/kg)(pmol/kg)(ymol/kg) (ymol/kg)(pmol/kg 10 to 11°C (Continued) 26 10.1 35.29 139 2331 20 2332 2191 2084 2063 21 38 10.8 35.46 119 2333 17 2320 2175 2076 2053 23 40 10.4 35.38 121 2338 17 2330 2184 2080 2065 15 41 10.8 35.39 115 2332 17 2323 2175 2075 2060 15 42 10.9 35.43 105 2327 15 2314 2166 2075 2057 18 46 10.4 35.50 82 2338 12 2317 2159 2081 2064 17 47 10.7 35.50 70 2337 12 2316 2147 2077 2062 15 47 10.6 35.50 73 2335 12 2314 2150 2078 2062 16 47 10.1 35.48 84 2342 13 2323 2159 2084 2064 20 49 11.6 35.54 72 2334 10 2308 2149 2074 2060 14 49 10.4 35.53 88 2337 13 2315 2163 2081 2061 20 52 10.7 35.51 82 2342 12 2320 2161 2077 2065 12 71 10.5 35.52 93 2336 15 2317 2164 2079 2059 20 107 10.8 35.53 66 2335 11 2311 2154 2076 2070 6 109 10.7 35.51 72 2336 11 2313 2151 2076 2063 13 114 10.9 35.54 48 2330 8 2303 2136 2075 2066 9 230 10.7 35.42 99 2335 15 2322 2171 2077 2067 10 233 10.9 35.42 107 2334 15 2321 2170 2075 2060 15 a Alk° = (Aik + AN0 3 )(35.0/S); ZC0 z ° = (£C02 - 0.80 AOU)(35.0/S); EC 02 ^ = salinity-normalized current winter outcrop value; AZCO 2 = £C 02 ^ - ECO 2 0 . ANO 3 : Values in parentheses are estimated from nearby stations. 34 to correct the thermocline waters for respiration effects We will show the impact of the value chosen for this respiration coefficient below. It can be seen that while the surface samples have, on the average, higher ECO 2 contents than do the respiration-corrected subsurface samples, there is a large overlap between the two data sets. One of the problems associated with the use of TTO surface water measurements for the problem of interest is that none were collected dur¬ ing the winter season. During the GEOSECS cross North Atlantic transect in March 1973, the pC02 values for late winter surface ocean water aver¬ aged well below the atmospheric value (see Fig. 13). Thus, it is not appropriate to assume that these waters leave the surface at equilibrium with the atmosphere. One way in which to obtain estimates of these winter outcrop values is to make plots of ECO 2 versus O 2 along isopycnal horizons and extrapolate the regression lines to O 2 saturation. Neglecting the uncertainty in the O 2 content of winter outcrop water, this extrapolated ECO 2 content should be a close approximation to the contemporary winter water value. Isopycnal analyses of this type are shown in Fig. 14. The estimates obtained in this way are shown along with the discrete sample data in Fig. 12. The scatter of the respiration-corrected thermocline data is more than that expected from the analytical error. The reason for this exces¬ sive scatter is likely that the data encompass a wide range of water ages (and hence also of anthropogenic CO 2 contents). A measure of the relative age of a sample is given by its AOU value. As shown in Fig. 15, there is a good inverse correlation between the AOU value and tritium content for GEOSECS samples in the 10 to 16°C temperature range in the temperate North 35 o 2080- 2060 8 O' o e 3 CVJ o O 1X1 2040- 2020 - 2000 12 14 TEMPERATURE (°C) Fig. 12. Plot of salinity-normalized ZCO 2 versus temperature as obtained during the TT0 expedition in the North Atlantic (Williams 1981b). The open circles are for surface waters, the solid circles are for respiration corrected thermocline samples, and the large crossed circles are for extrapolations of the 02 ~^C 02 trends along isopycnal horizons to oxygen saturation (see Fig. 14). All the ECO 2 values are normalized to 35.0°/oo salinity. 36 NO. OF SAMPLES 15 10 5 0 260 280 300 320 340 pC02 MIXED LAYER (ppm) 1 1 1 1 1 1 I GEOSECS STNS II5-121 (8= I8± 1 °j0) 1 1 1 i 1 1 1 1 1 1 1 1 _1_ Fig. 13. Histogram of surface water pC02 values obtained during the March 1973 traverse of the North Atlantic made as part of the GEOSECS program. These results are based on pC0£ measurements made on air equilibrated on shipboard with surface water (Takahashi 1979). The dashed line shows the partial pressure expected if the waters were at equilibrium with the atmosphere. 37 2C0 2 (/.imol/kg) IC0 2 (/xmol/kg) IC0 2 [f± mol/kg) Fig. 14. Plots of ZC0 2 versus 0 2 along individual isopycnal horizons (the values were obtained by interpolation of the results for samples immediateley above and below this horizon). Also shown is the saturation 0 2 content for water of this density. The intersec¬ tion of the ZC0 2 /0 2 trend with the saturation 0 2 content defines the average winter outcrop ZC0 2 content. This value and its salinity-normalized equivalent are given for each of the six iso¬ pycnal horizons along which this analysis was made. Only TTO data were used (Williams 1981b). 38 Atlantic We use GEOSECS rather than TTO data for this purpose because at the time this paper was written the TTO tritium analyses had not yet been completed. In Fig. 16 the difference between the surface ZC0 2 (as esti¬ mated by isopycnal extrapolation) and the average respiration-corrected thermocline ZC0 2 values is plotted against the oxygen deficit (i.e., AOU). To construct this figure, differences between the respiration- corrected ZCO 2 value for each thermocline sample and the surface water value obtained from the along-isopycnal analyses were used (see last column in Table 4). Because these differences are small, they are quite sensitive to the choice of the respiration coefficient. Had a respiration coefficient of 0.72 been chosen (rather than 0.80), the ZC0 2 excess at any given AOU value would be 50% smaller. Had a coefficient of 0.64 been used, then the difference would be zero. In this analysis no correction for CaC 03 dissolution was made. As this region of the ocean lies well above the lysoclines for calcite (-4400 m) and for aragonite (-2700 m), one might be tempted to dismiss dissolu¬ tion of these mineral phases as a significant contributor to the EC0 2 dis¬ tribution. As shown in Fig. 17, the alkalinities for surface waters (salinity normalized) and for thermocline waters (salinity-normalized and nitrate-corrected) show so much scatter that no firm conclusion in this regard can be drawn through this mode of comparison. However, when plotted against AOU the corrected alkalinity values for thermocline waters show a distinct trend (see Fig. 18). Such a trend might well be the result of CaC 03 dissolution in the guts of organisms or in organic-rich continental margin sediment. To explain the slope in Fig. 18, about 1 mole of CaC 03 would have to be dissolved for each 6 moles of organic 39 7 -,-,-1- -1- -,-,- -,-,— -1— G O O 8 G G — G G O G G G G G G — G G G — ® o G O G G - GEOSECS %■ NORTH ATLANTIC THERMOCLINE ° G 1972-1973 G O (10 TO I6°C) G _1_1_1_ _1_ _1_1 _1_1_ _i® 20 40 60 80 100 AOU (^mol/kg) Fig. 15. Tritium versus AOU for the North Atlantic temperate thermocline as measured during the GEOSECS program (Bainbridge 1981, Ostlund and Brescher 1982). 40 SC02 -SCO^^mol/kg) AOU (/i.mol/kg) Fig. 16. Mean ECO 2 differences (winter surface water concentration as ob¬ tained by extrapolation along isopycnal horizons minus respira¬ tion-corrected thermocline water concentration) as a function of AOU. The number of samples averaged in each AOU range is indi¬ cated for each point. The coefficient used for the respiration correction is 0.80 (i.e., the upper limit given in the text). 41 SALINITY NORM. ALK (/imol/kg) 2340 2320 i SURFACE * THERMOCLINE O O ** 2300 2280 ** * * %e>® @o° 0 o\ lie ° o*°a$£ o wg o o o ©■ °om> o o o 10 12 14 16 e (°c) Fig. 17. Plots of salinity-normalized alkalinity for surface water and thermocline waters measured as part of the TT0 program (Williams 1981b) against potential temperature. The thermocline values were corrected for the difference between the in situ and pre¬ formed nitrate content. 42 carbon oxidized. If this trend is accepted as a measure of CaC 03 dissolu¬ tion, a substantial correction has to be applied to the slope of ZCO 2 ver¬ sus AOU in Fig. 16. For an AOU of 100 pmol/kg the alkalinity is about 25 peq/kg greater than that for an AOU of zero. The corresponding NO 3 difference is about 11 pmol/kg. Hence, the amount of ECO 2 released by CaC 03 dissolution is about 25 +11 „ ---- or 18 pmol/kg. Using a respiration coefficient of 0.80 as shown in Fig. 16, the ECO 2 deficiency in thermocline waters with an AOU of 100 pmol/kg relative to contemporary surface water is about 16 pmol/kg. Hence, when the CaC 03 dissolution correction is made, this deficiency more than doubles, becom¬ ing 34 pmol/kg. Another potential contributor to the £C 02 _ A 0 U slope in Fig. 16 is mixing between waters of northern and southern hemisphere origin. Al¬ though the thermocline north of 18°N is dominated by water generated in the temperate North Atlantic, up to 20% of the water has an origin in the South Atlantic (Takahashi et al., in press). In Fig. 19 are shown plots of initial ECO 2 content for waters forming in the temperate North and temperate South Atlantic as a function of density. Those for the southern source were obtained by isopycnal analyses of the GEOSECS South Atlantic data (Bainbridge 1981, Takahashi et al. 1980). As can be seen, the initial ECO 2 differences at any given density are quite small. This is a fortunate circumstance in that in no case will the mixing of water of South Atlantic origin change the initial ECO 2 concentration by more than a few micromoles per kilogram. Hence no mixing correction is needed. ALK° (^.eq/kg) 2330 2320- 23 IQ- 2300 2290 80 AOU (^mol/kg) Fig. 18. Plot of salinity-normalized alkalinity against AOU for thermo- cline waters. As in Fig. 17, the alkalinity values were nitrate corrected. ZCO| (/imol/kg) 19. ZCO 2 versus density for thermocline waters forming in the temperate North and temperate South Atlantic. The solid lines are for the actual data. One dashed line shows the relationship for the South Atlantic if a correction is made to bring the GEOSECS data set in line with the TTO data set. The other dashed line corrects the TTO data for the atmospheric CO 2 increase between the time of the GEOSECS-1972 and TTO-1981 surveys. To interpret the initial EC0 2 versus AOU trends, a ventilation model for the thermocline is needed. In Fig. 20 are shown four different two- dimensional models for the ventilation of the temperate thermocline of the North Atlantic (Peng and Broecker, in press). All four of these models involve ventilation along isopycnal horizons. All four are calibrated by forcing a fit to a H distribution similar to that observed during the GEOSECS program. Shown in Table 5 are the IC0 2 differences between sur¬ face water and thermocline water obtained for these models, using the atmospheric pC0 2 versus time curve based on the assumption that fossil fuels are the sole contributor of excess C0 2 and using the atmospheric 1 3 pC0 2 versus time curve obtained by deconvolving the tree-ring-based C record for atmospheric C0 2 . In Fig. 21 we compare the model predictions with the observations (corrected for CaC 03 dissolution as well as for respiration). The model data are tied to AOU through the AOU- H relation¬ ship observed by GEOSECS in 1973 (i.e., each model box is assigned an AOU based on its 1973 H concentration). If the errors associated with the observations and the range of model predictions are considered, then it can only be said that the observations are broadly consistent with the entire range between these two limiting scenarios. Thus we gain no new insight into the C0 2 history of the atmosphere from the TTO observations in the Atlantic thermocline. The analysis of Atlantic surface waters shows the futility of using summer surface water data to estimate composition of waters descending into the oceans interior. Instead, by using surface water values obtained by the extrapolation of ZC0 2 -0 2 and ALK-0 2 trends the situation is considerably improved. However, the analysis still falls short of a 46 SIDEWELLING UPWELLING Okm 27.9 / km 27.9 <= 11.6 5.8 2.2 27.9 20.0 11.6 -s ■^ 8.0 5.8 <- 8.0 2.2 ■Ti 4.9 11.6 16.3 5.8 '' 10.5 2.2 8.3 BACKWELLING O km 15.3 / km 15.3 <*■ 13.9 12.2 <: 6.8 15.3 -i)-n X 68.2 19.5 X 5.8 ^> A 122 v ±> 15.3 AI22 i . V 15.3 15 *N 30°N 45 °N 15.3 8.3 24.9 * < AA 16.6 A 7.3 5.4 A 9.7 1.3 <*“ 5.8 41.5 IL2 >16.6 4.1 ^5.4 1.3 1.3 A 22.4 )f 19.1 8.3 10.8 2.6 V 8.3 DOWNWELLING Fig. 20. Models for the ventilation of the temperate North Atlantic thermocline (Peng and Broecker, in press). The water fluxes are in Sverdrups (i.e., 10^ m/s). The three boxes on the right-hand side of each model are the atmospheric outcrops for the upper, middle, and lower thermocline horizons. 47 IC0 2 (SURF H 2 0) - IC0 2 (THERM. H 2 0) ( F mol/kg) Fig. 21. Comparison of model prediction with observation for the difference be¬ tween contemporary surface outcrop water ECO 2 and contemporary thermocline res¬ piration and CaC 03 dissolution-corrected ECO 2 concentration as a function of AOU for the thermocline waters. The solid lines represent the observations using three different values for the respiration coefficient (i.e., 0.80, 0.72, and 0.64). The left-hand panel shows the model-based difference obtained for the fossil-fuel-C02-only scenario (see Fig. 1). The right-hand panel shows the model-based differei^es ^btained using the fossil-C02“plus-forest-soil scenario obtained from the C/ C record in tree rings (see Fig. 1). The individual models are designated as S (sidewelling), U (upwelling), B (backwelling), and D (downwelling). The model results are translated to the equivalent AOU values using the tritium-AOU relationship in Fig. 15. 48 Table 5. Model results for the 30° to 45° region as of 1981 a AZCO 2 excess (ymol/kg) Model Outcrop surface water minus upper thermocline Surface outcrop water minus middle thermocline Surface outcrop minus lower thermocline Fossil-fuel- -C02~only-scenario (see Fig. 1) Sidewelling 7.7 12.4 19.0 Upwel1ing 6.0 11.7 18.6 Backwelling 2.1 5.9 12.1 Downwelling 6.0 10.1 13.9 13 12 C/ C deconvolution scenario (see Fig. 1) Sidewelling 8.7 15.1 27.5 Upwel1ing 7.1 14.9 26.4 Backwelling 2.3 6.8 16.0 Downwelling 6.8 11.6 16.3 a See Peng and Broecker (in press) for full description of the models from which these differences were calculated. 49 definitive answer The reasons include: (1) analytical uncertainties in the ZC0 2 data, ( 2 ) uncertainties in the respiration coefficient, (3) uncertainties in the CaC 03 correction, and (4) the inability to come up with a unique model for the ventilation of the thermocline. 8 . RECONSTRUCTION BASED ON MEASUREMENTS ON DEEP WATERS For our analysis of deep waters we again select the North Atlantic. One reason is the superior TTO data set. Another is that the low AOU values found for waters of the North Atlantic Deep Water (NADW) mass re¬ duce the magnitude of the respiration correction and hence also of the error associated with its uncertainty. A third is that the upper NADW is relatively free of contributions from waters of Antarctic origin. For this analysis we chose the o 2 = 36.95°/oo isopynal horizon which has a depth of about 2000 m. As can be seen in Fig. 22, this hori¬ zon lies in the region of lowest F^SiOi* content (i.e., at that depth where the contributions of Antarctic Bottom Water (AABW) and Antarctic Intermed¬ iate Water (AAIW) are the smallest). We used both TTO (see Table 6 ) and corrected GEOSECS (see Table 7) data. The salinity-normalized and respira¬ tion-corrected ZC0 2 values are shown as a function of AOU value in Fig. 23. Despite the very broad geographic coverage of the horizon (i.e., con¬ tinent to continent and 50°N to 20°S), the range of these ZC0 2 values is quite small (i.e., -16 ymol/kg). As also shown in Fig. 23, the change in the respiration-corrected EC0 2 content along this horizon is in part coun¬ tered by CaC 03 dissolution. The alkalinity difference for the 0 2 range of 283 to 243 ymol/kg (AOU range 37 to 77 ymol/kg) is about 16 yeq/kg. To this we must add about 4 yeq/kg for the nitrate difference in order to 50 51_I_ 1—1 _I_i Vn i_ I—I _v Li i _ i _ i _ | _ | 60 50 40 30 20 10 0 10 20 30 40 50 60 S LATI TUDE(deg) N Fig# 22# Section of dissolved silica content along the western basin of the western Atlantic Ocean (based on GEOSECS data; Bainbridge 1981). 51 Table 6. Properties along a 2 = 36.95°/oo isopycnal horizon, based on measurements made during the TTO program (Williams 1981b) a Stn No. 9 (°c) S (°/oo) 0 2 AOU Si0 2 PO 4 N0 3 zco 2 zco 2 ° Aik Alk° (pmol/kg) (peq/kg) 15 3.41 34.977 265 55 16.3 1.20 18.3 2168 2128 2319 2321 16 3.43 34.981 265 56 16.2 1.22 18.2 2160 2119 2315 2316 20 3.44 34.984 262 58 16.5 1.20 18.4 2166 2123 2324 2325 22 3.45 34.987 262 58 17.4 1.25 18.7 2168 2125 2324 2325 24 3.52 34.998 260 59 18.2 1.24 19.0 2158 2114 2323 2323 26 3.43 34.982 260 60 17.0 1.24 18.5 2168 2124 2319 2320 28 3.41 34.979 263 56 16.7 1.22 18.5 2168 2127 2325 2326 29 3.41 34.979 263 58 17.6 1.23 18.5 2173 2131 2327 2328 31 3.39 34.979 256 65 20.2 1.27 19.5 2173 2126 2331 2332 32 3.43 34.982 259 62 18.5 1.26 19.1 2173 2128 2331 2332 34 3.41 34.981 255 65 19.8 1.27 19.4 2178 2130 2332 2333 36 3.49 34.994 251 68 20.9 1.30 19.7 2170 2119 2324 2324 38 3.58 35.013 248 71 21.8 1.30 19.6 2175 2121 2332 2331 39 3.66 35.028 250 68 20.6 1.29 19.5 2174 2125 2325 2323 40 3.64 35.024 250 68 20.3 1.27 19.3 2177 2124 2329 2327 41 3.62 35.018 255 64 19.1 1.24 18.8 2169 2120 2323 2322 42 3.56 35.008 259 61 18.3 1.25 18.8 2163 2117 2319 2319 43 3.61 35.017 258 61 19.4 1.27 18.8 2166 2119 2328 2327 44 3.59 35.010 259 59 19.5 1.25 18.5 2165 2120 2323 2323 45 3.63 35.020 259 60 17.7 1.26 18.5 2169 2123 2320 2319 46 3.69 35.032 252 66 18.7 1.26 18.6 2168 2117 2325 2323 49 3.65 35.026 258 60 17.4 1.21 18.2 2162 2115 2322 2320 52 3.90 35.071 249 68 19.2 1.28 19.5 2170 2115 2327 2322 66 3.94 35.083 243 73 20.5 1.30 19.6 2176 2116 2334 2328 68 3.90 25.075 251 65 18.1 1.24 19.0 2166 2113 2328 2323 71 3.94 35.081 242 74 22.1 1.34 20.2 2174 2114 2334 2329 76 3.89 35.071 229 87 25.4 1.41 21.1 2189 2119 2336 2331 80 4.01 35.097 239 77 25.2 1.37 20.8 2186 2122 2342 2336 83 3.98 35.087 252 63 18.0 1.23 18.8 2171 2118 2327 2321 99 3.75 35.044 257 60 17.5 1.22 18.6 2167 2119 2326 2323 107 3.55 35.006 262 67 17.2 1.22 18.2 2163 2120 2319 2319 109 3.48 34.992 265 55 16.5 1.19 18.3 2161 2120 2319 2319 110 3.36 34.969 267 54 16.6 1.18 18.1 2166 2127 2322 2324 111 3.79 35.053 256 62 17.6 1.22 18.6 2167 2117 2323 2319 112 4.01 35.097 250 67 19.3 1.23 18.8 2164 2108 2321 2315 113 4.21 35.136 245 69 20.5 1.25 19.2 2170 2110 2328 2319 114 3.59 35.012 258 61 19.1 1.20 18.8 2161 2115 2313 2312 115 3.46 34.988 262 58 19.3 1.20 18.4 2159 2116 2310 2311 52 Table 6. Continued Stn No. o O CD 'w' S (°/oo) 0 2 AOU Si0 2 P0 4 N0 3 ZC0 2 EC0 2 ° Aik Alk° (qmol/kg) (yeq/kg) 117 3.30 34.958 267 54 19.7 1.20 18.5 2153 2115 2304 2307 118 3.24 34.947 272 50 14.5 1.13 17.2 2154 2120 2309 2313 120 3.22 34.942 276 46 13.7 1.11 17.1 2159 2128 2310 2314 121 3.21 34.943 275 47 13.1 1.10 17.0 2165 2133 2312 2316 122 3.22 34.942 276 46 13.2 1.10 16.8 2152 2121 2305 2309 123 3.29 34.955 278 44 12.5 1.08 16.7 2152 2122 2306 2309 124 3.31 34.957 275 46 12.2 1.10 16.5 2147 2115 2396 2299 125 3.35 34.965 278 42 11.5 1.06 16.5 2153 2124 2300 2303 126 3.42 34.980 276 44 10.8 1.07 16.3 2159 2127 2307 2308 127 3.27 34.974 277 4 11.1 1.09 16.3 2155 2124 2307 2309 130 3.32 34.961 274 47 12.3 1.07 16.4 2156 2121 2306 2309 132 3.27 34.955 273 48 13.1 1.09 16.8 2154 2121 2306 2309 136 3.43 34.983 270 50 15.4 1.11 17.1 2162 2126 2315 2316 138 3.28 34.954 269 52 16.8 1.18 17.8 2160 2124 2311 2314 164 3.26 34.950 276 46 11.9 1.09 16.6 2155 2124 2312 2315 167 3.24 34.945 276 46 11.6 1.07 16.5 2153 2122 2312 2316 173 3.31 34.960 275 47 11.5 1.07 16.2 2167 2134 2326 2329 174 3.24 34.944 276 46 11.7 1.10 16.7 2163 2132 2319 2323 175 3.30 34.958 277 44 11.5 1.09 16.5 2162 2132 2318 2321 177 3.21 34.941 277 45 11.5 1.09 16.3 2163 2133 2318 2322 179 3.15 34.929 279 43 11.3 1.09 16.4 2162 2134 2317 2321 181 3.15 34.930 281 41 10.8 1.07 16.5 2160 2134 2318 2323 183 3.17 34.923 285 39 10.1 1.05 15.5 2157 2132 2315 2320 189 3.13 34.924 283 39 10.5 1.06 16.0 2156 2131 2313 2318 191 3.10 34.920 281 40 10.5 1.07 16.4 2157 2132 2313 2318 193 3.12 34.922 280 43 10.9 1.07 15.9 2155 2127 2317 2322 195 3.11 34.920 280 43 10.7 1.06 16.0 2154 2127 2310 2315 197 3.19 34.919 282 41 10.8 1.07 15.6 2157 2131 2315 2320 201 3.08 34.917 282 41 11.2 1.10 16.7 2155 2129 2311 2316 202 3.17 34.932 278 44 11.6 1.09 16.8 2159 2130 2315 2320 203 3.17 34.933 278 44 11.6 1.09 16.8 2155 2126 2310 2314 204 3.21 34.939 278 45 11.7 1.12 16.8 2154 2123 2308 2312 205 3.24 34.945 275 46 11.8 1.15 17.4 2157 2126 2311 2315 206 3.23 34.944 275 47 11.6 1.12 16.8 2152 2120 2306 2310 207 3.23 34.943 276 45 11.8 1.12 16.9 2157 2127 2308 2313 208 3.16 34.930 276 47 12.1 1.12 17.1 2151 2120 2305 2310 209 3.17 34.933 277 46 12.2 1.12 16.9 2149 2119 2305 2309 53 Table 7. Properties along the 04 = 45.730°/oo isopycnal horizon 3 based on data from the GEOSECS program (Bainbridge 1981, Takahashi et al. 1980) b S tn No. 0 O CD s (°/oo) °2 AOU Si0 2 P0 4 N0 3 IC0 2 ZC0 2 ° Aik J.Alk° 'M* (ymol/kg) (VJ eq/kg) 31 3.54 35.018 253 67 21.8 1 .28 19.2 2194 2143 2341 2339 32 3.50 35.009 251 59 23.7 1.27 19.3 2192 2140 2340 2340 33 3.45 34.996 252 68 21.6 1 .29 19.6 2201 2150 2351 2351 34 3.43 34.991 255 65 19.8 1.27 19.3 2189 2141 2328 2337 36 3.37 34.974 259 62 18.3 1.23 18.5 2180 2135 2327 2325 37 3.35 34.968 262 59 17.8 1.21 18.7 2176 2134 2335 2337 39 3.33 34.963 256 65 19.6 1.26 19.3 2178 2132 2333 2335 40 3.32 34.961 253 68 20.6 1.27 19.3 2170 2121 2328 2331 42 3.34 34.966 259 62 19.0 1.24 19.0 2173 2129 2128 2330 46 3.32 34.961 256 65 19.7 1.20 19.3 2170 2124 2329 2332 48 3.28 34.951 256 66 21.7 1.23 19.5 2177 2130 2336 2339 49 3.25 34.944 250 72 24.1 1.33 20.5 2176 2125 2332 2336 53 3.21 34.934 248 75 24.9 1.38 20.7 2172 2120 2332 2336 54 3.19 34.927 245 78 26.9 1.41 21.2 2173 2119 2333 2338 55 3.14 34.915 244 79 28.3 1.43 21.5 2177 2123 2332 2338 56 3.27 34.950 255 67 21.9 1.29 19.7 2171 2114 2329 2332 57 3.27 34.949 253 69 23.9 1.31 20.1 2175 2126 2337 2340 58 3.17 34.923 249 74 25.7 1.36 21.1 2176 2125 2334 2339 103 3.03 34.995 228 96 34.7 1.58 23.5 2194 2129 2338 2346 105 3.04 34.890 234 90 32.7 1.54 22.8 2190 2129 2332 2339 107 3.16 34.920 229 94 30.6 1.50 22.6 2194 2128 2333 2338 109 3.32 34.959 255 66 21.4 1.29 19.9 2170 2123 2327 2330 111 3.30 34.956 250 71 22.5 1.35 20.4 2177 2126 2325 2328 113 3.29 34.954 237 84 25.0 1.43 21.5 2181 2121 2325 2328 114 3.44 34.994 231 90 27.4 1.45 21.7 2184 2117 2330 2330 115 3.65 35.049 246 74 25.3 1.33 19.7 2174 2125 2344 2341 116 3.67 35.053 249 70 21.8 1.30 19.4 2183 2127 2333 2329 117 3.65 35.047 255 65 19.2 1.24 18.8 2172 2120 2334 2331 118 3.54 35.019 259 61 20.3 1.22 18.7 2173 2126 2336 2335 119 3.45 34.995 264 57 17.9 1.20 18.5 2172 2129 2338 2338 120 3.40 34.983 265 56 18.0 1.20 18.5 2168 2127 2321 2322 121 3.34 34.967 269 52 16.3 1.18 18.1 2166 2129 2329 2331 Equivalent to the a 2 = 36 .95° /00 horizon adopted for the TTO measurements. b lC02° = (ECO 2 - 0.75 AOU)(35. 0/S); Alk° = ‘ Alk(35 .0/S) • 54 ALK n ( M eq/kg) Ic0 * ( '‘ mol/k « l AOU (^.mol/kg) AOU (yu.mol/kg) O 2 (/xmol/kg) Fig. 23. Salinity-normalized and AOU-corrected £C02 concentrations versus oxygen content along the 36.95°/oo isopycnal horizon in the Atlantic Ocean (upper diagram); salinity-normalized alkalinity versus oxygen content along this same isopycnal horizon (lower diagram). The data are from the TTO expedition in the northern Atlantic (Williams 1981b). 55 Fig. 24. Dissolved oxygen versus potential temperature for the a 2 = 36.95°/oo isopycnal horizon north of 20°S. (Bainbridge 1981, Williams 1981b). 56 CP \ o E O h 2 o Fig. 25. "NO" (O 2 + 9 NO 3 ) versus potential temperature plots for the GEOSECS stations (Bainbridge 1981) and the TTO stations (Williams 1981b). The density level a 2 * 36.95°/oo corresponds to 04 * 45.73°/oo in the GEOSECS data compilations. 57 obtain the alkalinity change due to CaC 03 dissolution alone. Thus, about (16+4)/2 or 10 pmol/kg must be added to the ZCO 2 difference (over the O 2 range 243 to 283 pmol/kg) to correct for the contribution of CaC 03 dis¬ solution. In this way we get 26 ± 10 ymol/kg for the anthropogenic- induced difference along this isopycnal. It is, of course, important to determine to what extent the ECO 2 and alkalinity trends shown in Fig. 23 are the result of mixing with other water types. The dissolved oxygen/potential temperature plot in Fig. 24 suggests such effects might be important. The TTO data show a trend from high O 2 and low 0 in the northwest to low O 2 and high 0 in the southeast. The warmer temperature for the eastern portion of this isopycnal horizon reflects a small contribution of overflow water from the Mediterranean. By contrast the southern hemisphere values show a trend toward lower O 2 and lower 0 values, which reflects a contribution of waters of Antarctic origin. The plot of "NO" (i.e. , 02 + 9 NO 3 ) against potential tempera¬ ture in Fig. 25 confirms these suspicions; contamination with waters of Antarctic origin occurs in the southern hemisphere and contamination with waters of Mediterranean origin occurs in the eastern Atlantic. Antarctic waters have higher "NO" and Mediterranean waters lower "NO" than do deep waters originating in the northern Atlantic. As Antarctic waters with a density a 2 = 36.95°/oo have nearly the same EC0 2 and alkalinity as waters of this density in the northern Atlan¬ tic, small amounts of contamination from this southern source will not significantly alter the trends seen in Fig. 23. The situation for the eastern Atlantic is potentially more serious. However, it can be shown that the low O 2 values found along the a 2 = 36.95°/oo isopycnal horizon in the region adjacent to Gibraltar are the result of respiration under the high productivity upwelling zones off the African coast rather than of 58 low O 2 In the Mediterranean component. One piece of evidence in support of this conclusion is portrayed in the plot of alkalinity against dis¬ solved silica shown in Fig. 26. Water with high silica contents have high alkal ini ties • The slope of this trend corresponds to one mole of CaC 03 (dissolved) per mole of opal (dissolved). As Mediterranean overflow water has less dissolved silica than even deep waters in the far northern Atlan¬ tic, its admixing should lower rather than raise the dissolved silica con¬ tent. A similar correlation exists between O 2 and Ht+SiOi*. Thus, while the input of Mediterranean water causes the observed potential temperature (e)-salinity (S) shift, its contribution is too small to seriously alter the ZCO 2 , alkalinity, and O 2 concentrations. The changes seen in these properties are the result of higher than average productivity (and hence respiration and dissolution) along the eastern margin of the Atlantic. A direct examination of the composition of Mediterranean outflow water shows that the contribution needed to make appropriate 9-S shifts is not enough to make a significant ECO 2 or alkalinity shift. Based on this analysis we conclude that the trends in alkalinity along this isopycnal horizon are primarily the result of in situ produc¬ tion rather than of mixing with other water types. Thus, it is appropri¬ ate to correct the ZCO 2 results for the alkalinity difference seen between waters of 283 ymol/kg O 2 and 243 ymol/kg O 2 . The corrected trend is shown in Fig. 23. As no waters with O 2 greater than 283 pmol/kg are found along the 02 = 36.95°/oo isopycnal, we can only speculate how much ECO 2 change should be associated with the O 2 decrease from 320 pmol/kg (the saturation value) to 283 ymol/kg. If the trend observed in Fig. 23 continues with the same slope, then the total anthropogenic effect is 59 ALK (^.eq/kg) T H 4 Si0 4 (^mol/kg) CJ> XL 30- 25- o E 3 O 20 CO X <3 ® *® O ® ° °<®g< 00 15 1-0 TTO x GEOSECS N HEM I. ® GEOSECS S HEMI. » i 10 220 240 260 280 0 2 (^mol/kg) Fig. 26. On the left is shown alkalinity versus dissolved silica as meas¬ ured during the TTO program along the a 4 = 36.95°/oo isopycnal horizon (Williams 1981b) and on the right dissolved silica versus dissolved oxygen on the same horizons as measured during both the GEOSECS (Bainbridge 1981) and TTO (Williams 1981b) programs. 60 or 50 ± 20 pmol/kg 26 ± 10 x (320 - 243) (283 - 243) This may be compared with the preanthropogenic to 1980 equilibrium surface water increase of 29 pmol/kg for the fossil-fuel-C02~only scenario and of 73 pmol/kg based on the C deconvolution scenario. Another possibility is that the water left the surface with only 283/320 or 0.88 of the O 2 saturation value. Neglecting for the moment the uncertainties in the ECO 2 -O 2 trend and its extrapolation to O 2 saturation, interpretation of the observed difference is hampered by the following questions: (1) Can water with an O 2 content 243 pmol/kg be considered free of excess C0 2 ? (2) Do waters descending onto this isopycnal horizon carry an equi¬ librium amount of excess anthropogenic CO 2 ? In connection with the first of these questions, it is informative to consider the simple model of North Atlantic ventilation shown in Fig. 27. In this model, surface water at pCC >2 equilibrium with the atmosphere is injected into the deep sea. Once in the deep sea it moves through a series of reservoirs and finally out the far end of the chain (i.e., into the Antarctic). The advective flushing time of each box is taken to be 40 years. In case "a," no exchange between adjacent boxes is included (i.e., k=o). In case "b," a between-box exchange three times the advective flux is employed (i.e., k=3w). In case "c," the five boxes are treated as a single well-mixed reservoir with an overall advective flushing time of 200 years i.e., (<=«>). Figure 28 shows the excess CO 2 content of each box in 1980 for the two atmospheric C0 2 scenarios given in Fig. 1. In none of 61 Fig. 27. Box model for the ventilation of the 36.95°/oo isopycnal horizon. K is the mixing between boxes, W the flow from box to box and V the volume of a box. 62 Fig. 28. Anthropogenic ZCO 2 excesses in the five deep water boxes of the model shown in Fig. 27 for the year 1980. The upper panel shows the results when the model is forced with an atmospheric CO 2 history shown in Fig. 1, based on fossil-fuel-C02 _ only inputs. The lower panel shows the results when the model is force^with an atmospheric CO 2 history shown in Fig. 1, based on the C deconvolution. 63 these models selected here does the first box in the deep sea chain show more than two-thirds of the full anthropogenic effect (i.e., that for the surface box). Also, in no case is the last box free of anthropogenic C0 2 . The distribution of natural radiocarbon suggests in the upper NADW that model "a" comes closest to the real situation in the Atlantic. As listed in Table 8 and shown graphically in Fig. 29, the bomb-corrected A 14 C value in this horizon decreases from about -74°/oo at the northern end of the Atlantic to about -102°/oo in the eastern southern Atlantic (20°S). As shown in Fig. 30, when these corrected A 14 C values are plotted against the dissolved oxygen content of the water, a linear trend is seen in the 283-to 243-pmol/kg dissolved 0 2 range. A 30°/oo decrease in 14 C:C ratio is seen over this range. As pointed out by Broecker (1979), 14 part of this decrease is the result of mixing with low C water of Ant- 14 n arctic origin. Because this water has a A C value about 88°/oo lower than that for water of northern origin (i.e., -155°/oo vs -67°/oo) for each 10% mixed into the northern water, the A 14 C drops by about 9°/oo. Thus, part of the change observed down the Atlantic could be the result of mixing rather than radiodecay. We estimate that mixing accounts for about one-third of the total but admit that there is considerable uncertainty in this estimate. In Fig. 31 we show that case "a" of the 5-box model produces a dif¬ ference of about 20°/oo in the radiocarbon A li4 C value between 50°N and 20°S. Such a model also confines the bomb 14 C largely to the northernmost model box. To get the right magnitude of the bomb- C effect we must assume that the waters descending from the surface had, as of 1972 , only about 12% of the equilibrium excess of atmospheric bomb radiocarbon 64 Table 8. Radioisotope data in the vicinity of the a 2 = 36.95°/oo isopycnal horizon (Stuiver and Ostlund 1980, Ostlund and Brescher 1982) S tn No. Lat. Long, • Depth (km) 0 (°C) (y 0 2 mol/kg) a 14 c (°/oo) 3 h ( T . U .) A C cor a (°/oo) Western basin 11 63.5 °N 35 . 2 ' 9 W 1.82 3.35 277 -49 2.1 -77 23 60.4 °N 18 . 6 * 5 W 1.98 3.46 276 -63 0.7 -72 23 60.4 °N 18 . 6 * ’w 2.19 3.30 275 -53 0.7 -62 3 56.9 °N 42 . 5 * >W 1.64 3.41 276 -60 0.9 -72 5 56.9 "N 42 . 5 * >w 2.05 3.15 277 -60 0.9 -72 3 51.0 °N 43 . 0 * >W 2.05 3.36 278 -51 1.5 -71 27 42.0 °N 42 . 0 * 3 W 2.19 3.15 273 -75 0.4 -80 29 36.0 °N 47 . 0 * 5 W 2.12 3.53 269 -63 0.5 -70 29 36.0 °N 47 . 0 ‘ 5 W 2.65 3.01 269 -76 0.5 -83 120 33.3 °N 56 . 5 ‘ ’w 2.26 3.40 265 -76 0.1 -77 117 30.7 °N 39 . 0 * s w 1.95 3.78 255 -78 0.1 -79 117 30.7 °N 39 . 0 * ’W 2.51 2.90 260 -92 0.1 -93 31 27.0 °N 53 . 5 * 5 W 2.23 3.18 255 -87 0.0 -87 33 21.0 °N 54 . 0 * ’w 2.24 3.15 254 -94 0.0 -94 37 12.0 °N 51 . 0 * 5 W 1.98 3.30 263 -80 — -80 40 3.9 °N 38 . 5 * >W 1.98 3.45 253 -86 — -86 48 4.0 °S 29 . 0 * 9 W 1.80 3.58 251 -89 — -89 48 4.0 °S 29 . 0 * ’w 2.26 3.05 255 -90 — -90 49 7.9 °S 28 . 2 * 5 W 2.09 3.10 252 -97 — -97 54 15.0 ”S 29 . 5 * J w 1.84 3.09 248 -94 — -94 56 21.0 °S 33 . 0 * 5 W 1.87 3.38 253 -86 — -86 56 21.0 °S 33 . 0 * *w 2.07 3.14 256 -83 — -83 Eastern bas in 115 28.0 °N 26 . 0 * J w 2.49 3.01 250 -97 — -97 113 11.0 °N 20 . 5 * 9 W 1.75 3.56 232 -97 - -97 113 11.0 °N 20 . 5 * >W 2.20 3.05 241 -99 - -99 111 2.0 °N 14 . 0 * 5 W 2.07 3.19 250 -92 — -92 107 12.0 °S 2 . 0 * >E 1.76 3.40 220 -101 - -101 107 12.0 °S 2 . 0 * 9 E 2.07 3.02 230 -107 — -107 a The correction deep waters in is based on the slope of 14°/oo per T.U. observed for the northern Atlantic during the GEOSECS program. 65 1 if 0 Fig. 29. A C (circles) and H (squares) values for samples close to the ° 2 ~ 36.95°/oo isopycnal horizon as a function of latitude. Also jshown (by the arrows) are the corrections for the input of bomb C based on the relationship: A 14 C* = 13 3 H, observed dur¬ ing GEOSECS for the northern deep waters of the Atlantic Ocean. 66 cm) h O2 (^.mol/kg) 14 14 Fig* 30. Plot of bomb- C-corrected A C values against dissolved O 2 centrations for samples from close to the 02 = 36.95°/oo iso pycnal horizon. con- 67 oo Fig. 31. 14 C distribution as yielded by the five-box NADW model for the case of advection only (i.e., no between-box mixing). 68 14 (because the atmosphere in the northern hemisphere had an excess A C of about 475°/oo over the prenuclear value, this means that the surface waters descending onto the 36.95°/oo isopycnal in 1972 carried a A 14 C excess of 57°/oo). The fact that the 14 C excess is so low suggests that the CO 2 content of these waters may not have achieved equilibrium with the atmosphere during the densification process. The relationship between pC02 and ZCO 2 for waters with the tempera¬ ture, salinity, and alkalinity of those on the a 2 = 36.95°/oo isopycnal horizon is shown in Fig. 32. Also shown is respiration-and dissolution- corrected ZCO 2 content for waters of O 2 concentration equal to 243 ymol/kg (i.e. , waters with the least anthropogenic contamination). In this figure the 5 pmol/kg correction between the shipboard titrator and Keeling labor¬ atory manometric results has been taken into account (i.e. , the titrator ECO 2 values have been reduced by 5 ymol/kg). This water appears to have 6 had a PCO 2 of about 260 x 10” atm when it left contact with the atmo¬ sphere. The pC02 for the atmosphere at the time of the TTO expedition was about 335 x 10~ 6 atm (the corresponding equilibrium ECO 2 content for this PCO 2 is 2138 pmol/kg). Also included on this diagram are vectors showing the anthropogenic ECO 2 increase (as of 1980) expected from fossil fuel CO 2 alone (29 pmol/kg) and from this combined fossil fuel and forest-soil scenario (73 pmol/kg). In each case the base of the vector lies slightly below the observed value for water with 243 pmol/kg O 2 . This is because even the box #5 (see Fig. 27) waters in the model are, as of 1981, contaminated 13 with anthropogenic CO 2 . Because, in the case of the C-based scenario, the mean age of the anthropogenic CO 2 molecules is considerably larger, the portion of the vector extending beneath the line is greater than that for the fossil-fuel-C02~only case. 69 380 340 300 260 220 >- _j 2 O CJ o o _J LU 3 U_ _J -EQUIL. WITH ATM (1980) £ t 29 — 02*243 WATER JL- CVJ O O o co^o O (C o . LU 3 Li¬ en uj cr: o cn x cn 3 O § Ll. 2 2200 2175 2150 2125 2100 2075 1 SUMMER SURFACE > WATER N. NORWEGIAN I SEA IN 1981 2050 Fig. 32. Comparison of the atmospheric equiliibrium ZCO 2 value (as of 1980) and the respiration- and dissolution-corrected ZCO 2 content of water of O 2 = 243 ymol/kg on the 02 = 36.95°/oo isopycnal horizon. Also shown are estimates of the preanthropogenic EC02 content of water on this isopycnal for the fossil-fuel-C02~only scenario and for the combined fossil fuel and forest-soil sce¬ nario calculated using the chain model discussed in text. These estimates are shown by the bottoms of the arrows. The arrows show the anthropogenic effects (i.e., 1980 atmosphere equilibrium value minus the preanthropogenic value) for these two scenarios. Also shown are the summer ZCO 2 concentrations and the correspond¬ ing pC02 values observed for 3°C water in the northern Norwegian Sea during the TT0 program (1 atm = 1.013 x 10 Pa). 70 As can be seen, the fossil-fuel-CC> 2 - only scenario predicts that the water descending onto the 36.95°/oo isopycnal currently has a pC 02 of just over 300 x 10 -6 atm, while the combined scenario suggests a pC02 of about 380 x 10~ 6 atm. To date, only summer values for the pC02 of cold waters in the northern Norwegian Sea are available. Values of about 200 x 10~k atm were obtained for 3°C water in the northern Norwegian Sea during the TTO expedition. These lie well below even the respiration- and dissolution-corrected values for water with an O 2 content of 243 pmol/kg O 2 on the 36.95°/oo isopycnal. This suggests that strong seasonal cycles in the pC02 of surface water must exist in the northern Atlantic. Until these cycles are understood and the geographic distribution of winter values is known, nothing can be concluded from the above approach regarding the magnitude of the anthropogenic change. Although there are not as yet any published winter surface water pC02 measurements, we can get some idea of what these values might be by look¬ ing at the reconstructed pC 02 values for waters beneath important source regions for NADW (i.e., the Norwegian Sea, the Labrador Sea, and the Den¬ mark Straits). As listed in Table 9 and summarized in Table 10, these values range from 260 x 10 -6 atm for the coldest of these source waters (i.e., those from the bottom of the Norwegian Sea at -1.1°C) to 320 x 10 -6 atm for the warmest of these source waters (i.e., those in the deep Labrador Sea at 3.3°C). As these waters have aged for periods of a few to a few tens of years, these values must be taken as lower limits on the PCO 2 values for the winter waters descending in 1980. The summer to win¬ ter pC02 difference of at least 100 x 10~ 6 atm [-200 x 10 6 for 3°C summer C. surface water and >300 x 10 atm for 3°C winter surface water as recon¬ structed from the data in Table 11] must be the result of the stripping of 71 Depth 0 S 0 2 AOU N0 3 P0 4 Aik Alk° ZC0 2 ZC0 2 ° pC0 2 ° (km) (°C) (°/oo) (ymol/kg)(ymol/kg)(ymol/kg)(ymol/kg) (yeq/kg) (yeq/kg) (yraol/kg)(ymol/kg) (yatra) I O uo 00 r-H CO O CM CM X O CM -3- m m X) X X X X X X X —H r—H o r-H CM CM CM CM CM CM CM CM CM CM CM co co CO CO O' i—H CO CM CO r-^ I-'- r-H X CM X CO O' oo CM O O O o o O r-H •—H O o r-H r-H o o r-H CO CO CO CO CO co co CO CO CO CO CO CO CO co CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM - f- as nO CN st 00 r-H O >3- CN UO CN UO •3- m m m st m UO uo uo St st st UO uo st UO NO st UO uo UO NO *—H rH rH «—1 r-H «H r-H r-H r-H H r—H r-H r-H r-H r-H r-H rH rH rH rH rH rH CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN 00 •H sr 00 on O UO UO 00 f'- as St St st 0 r-H rs UO O CN ON 0 rH rH rH O r-H r-H r-H r-H O 0 0 r-H r-H r-H rH h1 0 Hi O rH rH co CO co co CO co CO co co co CO CO co co co co co CO co co co co CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN CN s SS S2 ,3 3 0 O 0 O 5 0 0 00 m NO nO O co Cs • • • • rH • • M CN 0 UO IH CO CO 0) m m UO UO U CO CO 4-1 0) to z nO O' H-« st z 00 r-H r-H st z UO CN 0 O z 00 nO st > z CN nO co 00 z U0 Sr 0 O O rH r-H O O 0 r-H r-H r-H 0 O O O r-H 0 O O rH 0 0 O O 0 O 0 rH v£> CO CO co co vO CO co co co CO sD in UO UO UO UO as co CO CO rH CO CO CO CO CO co « • CN CN CN CN • CN CN CN CN CN • CN CN CN CN CN • CN CN CN CO • CN CN CN CN • CN 0) 00 Os c- sC 4-1 st sr CO m in uo UO IN nO NO M vy U O CO r-H U0 1''. u 00 TJ as m co rs 00 ON CO UO as as r-H CO UO nO on CO NO |S> V) NO as O rH O vO O' CO Hi O O O O r-H 0 O 0 O O H O 0 O O O r-H O O O Hi 00 O' Os O' Hi 00 h * • • • • • C r—■1 rH *H r-H C r-H r-H r-H r-H r-H c r-H »-H r-H r-H r-H c ^H r-H rH Is C 0 c O 0 C 0 (0 4-1 u cc 4J 4-1 •J CO tn CO CO E CO CO C • 0 p p p on O on 00 ON O 00 00 as a Os 00 00 ON ON 00 O 0 0 O 0 C CN CO CN CN CN co CN CN CN CN CN CN CN CN CN CN CN co CO CO CO CO flj /*-N O co St O 'O O' 00 as UO CN r-H O O CN UO O co 00 CO CO 00 0 00 O' 0 CO CN in rH st r-H 00 f—H co m UO St as CN uo uo 0 CN uo 00 r-s CO Z 00 00 00 on 00 00 00 as On 00 00 00 00 as oc 00 O' 00 00 00 00 00 1 • • • • • • • • • • • • • • • • • • • • • • ro st st St st st st st sr St •«t St sr st St st st sr st st St sr st O CO co CO CO CO co CO CO CO co co co co CO CO CO CO CO CO CO CO CO Z + st \£> 00 nO in CO vO rs. st r- as nO co uo 00 00 uo rH CN st 00 rH sr h1 CN CN m *-H CO CN 0 U0 co CN CO r-H uo CO CN O as CN CN 00 •-J • • • • • • • • • • • • • • • • • • • • • • < CO CO CO CO CO co CO CO co CO CO CO co CO co CO CO rH 0 O O O H O uo 0 0 00 nO vO 0 0 0 0 nO st in O uo CO CN NO O CN NO Hi CN 1"- CO as CN 00 U0 as CO <3- as UO as UO ON uo CN CN CO CO 00 < • • • • © c 0 ^H rH O O *“ 1 r-H CN 0 O O ^H ^H 0 e rH CN CN CN CN *■* 73 CO CO CM o o a CO So M CO E E 3 C/j 00 CM CTn 6 \ CM CM CM CM O Csl rH c-H f—1 •—H f-H f-H o o CM CM CM CM CM c_> CM bt OcM 35 O CJ f-H CM O E a. C bC in CO o CM vO O 35 ■—i f-H CM CM t—i 35 \ co CO CO CO CO f-H CT CM CM CM CM CM <3 CM II D 35 r—I <3 p o o © -t* S I •2 2 . ?? _ ««. |2 IS s C 8 to 8 og to 30 ® 3 « ' TJ S’ 3- rt © © 5 1 ° 3 8 . §Q3« ? 2 3 £ O O fo O CD •Pt CJ1 7 © o © m 3 ©