Simulating oceanic CaCO3 export production in the greenhouse



[1] A model scenario for the change in global marine biogenic CaCO3 export production (CaCO3 = calcium carbonate) due to increasing atmospheric carbon dioxide partial pressure is carried out. Findings from laboratory experiments, which suggest a decrease of biocalcification at higher pCO2, are extrapolated to the world ocean by use of the biogeochemical ocean general circulation model HAMOCC. For an A1B IPCC emission scenario and constant emission rates after year 2100, the simulation predicts a global decrease of biological CaCO3 export production by about 50% in year 2250. The negative feedback due to this drop in CaCO3 export on the atmospheric CO2 concentration is small as compared to the anthropogenic CO2 emissions. This negative feedback will potentially be compensated by a shallower remineralization of organic carbon.

1. Introduction

[2] The ocean can buffer about equation image of any external carbon dioxide inputs into the atmosphere [e.g., Bolin and Eriksson, 1959], given long enough equilibration times over multiple turnover periods of the large scale ocean circulation. Also on the shorter annual-to-centennial time scales, the world ocean is a major sink for anthropogenic CO2 which has been emitted in increasing amounts to the atmosphere since the start of the industrial revolution. At present, about half of the annual anthropogenic CO2 emissions into the atmosphere are taken up by terrestrial systems and the world ocean in approximately equal quantities [Intergovernmental Panel on Climate Change (IPCC), 2001]. The oceanic buffering of excess CO2 in the atmosphere is mainly governed by dissociation of free carbon dioxide (gaseous CO2 and carbonic acid H2CO3) into bicarbonate HCO3 and carbonate CO32− ions as well as by oceanic transport and mixing.

[3] The property of the ocean to buffer excess CO2, however, leads to a large scale acidification of the ocean as a consequence of the uptake of atmospheric excess CO2. In equilibrium, the pH value of seawater shifts towards lower values, the higher the seawater pCO2 is [e.g., Harvey, 1969]. The large scale oceanic pH shift as expected from anthropogenic CO2 emissions during the coming centuries may lead to an acidification of the ocean that has not occurred since the last 300 million years [Caldeira and Wickett, 2003].

[4] A lowering of the seawater pH value may have severe consequences for marine biota. In particular, according to laboratory studies with coccolithophoridae, an acidification of the ocean may lead to a substantial decrease in calcifying primary producers [Riebesell et al., 2000; Zondervan et al., 2001]. A decrease in biogenic CaCO3 production would provide a negative feedback to rising atmospheric CO2 concentrations as long as no other related feedback mechanisms would become important. CO2 is stored in the ocean mainly as bicarbonate HCO3 as well as carbonate CO32− and only to a small fraction as free carbon dioxide (CO2 and carbonic acid H2CO3). Any extraction of CO32− ions - as for the production of biogenic CaCO3 - reduces the ability to dissociate free carbon dioxide into carbonate and bicarbonate. This ability is measured in terms of the alkalinity of the seawater [Harvey, 1969]. For a weaker CaCO3 production in the ocean, respectively, more alkalinity would be retained in the respective ocean water volume at the ocean surface, and a lower seawater pCO2 would result.

[5] Below we use the quantitative feedback between seawater pH (or seawater pCO2) and CaCO3 production as determined by Zondervan et al. [2001] in a global biogeochemical ocean general circulation model. In a future scenario for atmospheric CO2 emissions we extrapolate the suggested effect on planktonic biocalcification to the next 250 years.

2. Model Description

[6] For studying the effect of CO2 buffering on ocean pH and biogenic calcite production we employ here the biogeochemical ocean general circulation model HAMOCC [Maier-Reimer, 1993] in its version HAMOCC4 [Six and Maier-Reimer, 1996; Hofmann et al., 2000]. The model covers the globe with a horizontal resolution of 3.5 × 3.5 degrees and divides the water column into 22 layers with higher resolution in upper waters. The physical fields are taken from a present day climatological mean run with the dynamical LSG model (LSG = Large Scale Geostrophic) [Maier-Reimer et al., 1993]. The biogeochemical HAMOCC4 model includes an NPZD type ecosystem model and includes representations of the marine carbon, nitrogen, phosphorus, oxygen, and silicon cycles. NPZD stands for “nutrient phytoplankton zooplankton detritus”. NPZD models include only two trophic levels which are linked to the large scale biogeochemical cycles through nutrient uptake as basic input and release of dead matter as basic output. The model includes primary and secondary production as well as a representation of the microbial loop. While the physical model field operates on a monthly time step basis, the ecosystem model runs with a 3 day time step. Production of siliceous plankton shell material is preferentially carried out in areas of significant silicic acid supply to the surface ocean. The biogenic CaCO3 export production is parameterized as:

equation image

with PCaCO3 the CaCO3 export production (expressed in units C per unit area ant unit time), PorgC the total organic primary production (in the same units), and PBSi the total biogenic silica production (expressed in units Si per unit time and unit area). In the standard run, the tunable factor A is set to 0.15. The expression on the right hand side of equation (1) cannot become negative, i.e., the BSi production can never exceed twice the amount of the total organic carbon production according to the model formulation for the biogenic silica production applied. A simple diffusive atmosphere is coupled to the ocean model through a gas exchange parameterization. The model further includes an interactive sediment module. Due to the long equilibration time scale between sediment and water column, a slight drift of the modeled tracer distributions in the three model reservoirs ocean, atmosphere, and marine sediment is still going on. Compared to the perturbations in the sensitivity study as describe below, this drift is negligible. The model system itself is, of course, strictly mass conserving.

3. Methodology of the Scenario Experiment

[7] In our scenario we consider as the only forcing the emissions of anthropogenic CO2 into the atmosphere. The physical fields remain unchanged, i.e., we do not consider any change in ocean circulation due to internal variability and anthropogenically induced climate change. We choose the time interval from 1750, the beginning of the industrial revolution, until 2250, i.e., altogether a period of 500 years. As emission scenario we combine the historic emission time series [Marland et al., 2003] for 1751–2000, with the A1B emission scenario as suggested for the 4th IPCC assessment report for the period 2001–2100, and constant emissions thereafter on the level of the year 2100 (Figure 1).

Figure 1.

Scenario for anthropogenic CO2 emissions into the atmosphere for the time 1750–2250 (numbers for 1750–2000 were taken from Marland et al. [2003]).

[8] For the dependency of the biogenic CaCO3 export production on the free carbon dioxide in the water (and hence the CO2 partial pressure and the pH value) we use a parameterization following the laboratory experiments by Zondervan et al. [2001] and modified equation (1) to (see also Figure 2):

equation image

Parameter A was kept at the value 0.15. The difference in concentrations CO2actual and CO2preindustrial was taken individually at each model grid point where CaCO3 export production occurred. The model experiment was restarted from a previous integration at year 1750 and continued until year 2250 using the emission scenario as described above.

Figure 2.

Feedback between rising free carbon dioxide in seawater and biogenic CaCO3 export production applied in the model scenario.

4. Discussion of Results

[9] First of all a time series of global bulk numbers is considered, namely for the simulated global CaCO3 export production and the atmospheric pCO2 (Figure 3). In Figure 4 a corresponding difference time series is given to illustrate the change between an emission scenario using no CaCO3 production feedback and the run with this feedback. These results suggest, that so far a reduction of the pelagic CaCO3 export is barely observable and would just now start to become significant. Given the CO2 emission scenario as adopted and the feedback formulation would be realistic by year 2250 the global biocalcification in oceanic export production would be reduced by 50%. Patterns of the CaCO3 export production for different years are summarized in Figure 5.

Figure 3.

Time series of the global CaCO3 export production (thick line) and the atmospheric pCO2 for the scenario including the feedback and the emissions as given in Figures 3 and 2.

Figure 4.

Difference time series of the global CaCO3 export production (thick line) and the atmospheric pCO2 for the scenario with and without the CaCO3 feedback.

Figure 5.

CaCO3 export production in [gC m−2 yr−1] for the scenario with and without the CaCO3 feedback at different time slices (from above: at A.D. 1750, 2004, 2100, and 2250).

[10] We repeated the experiment by choosing a larger value for the CaCO3 production through doubling factor A in equation (1). The results for the relative changes in CaCO3 production remained similar, though the negative feedback on the atmospheric pCO2 doubled. This negative feedback due to a weakening of the CaCO3 production, however, was small and resulted in a decrease of the atmospheric pCO2 of only 10 ppm (for A = 0.15) or 20 ppm (for A = 0.3) only.

[11] As regions of biogenic CaCO3 production are not homogeneously distributed over the globe, we also investigated whether the CaCO3 feedback would impact the interhemispheric gradient of CO2 in the atmosphere. The effect, however, turned out to be negligibly small.

[12] It should be stressed, that the ocean circulation is kept constant in the present study and global warming itself is not considered. Previous studies investigating the potential effect of a future climate shift including a change in ocean circulation on the carbon cycle resulted in a slight increase of the organic matter production. As in these studies the CaCO3 was coupled to the organic matter production through a fixed rain ratio, a minor positive feedback resulted in these cases [e.g., Plattner et al., 2001].

[13] A weakening of the CaCO3 export could affect the downward transport of organic carbon as well. It has been suggested that biogenic CaCO3 serves as a major ballast for particulate organic carbon (POC), so that POC in areas of high CaCO3 export is efficiently vertically transferred to greater depths before it remineralizes [Armstrong et al., 2002; Klaas and Archer, 2002]. If a decrease in CaCO3 production would lead to a substantial shallowing of the remineralization depth horizon, the weak negative feedback due to the alkalinity effect of CaCO3 formation could be overridden towards a positive feedback. We carried out an additional sensitivity experiment, where the remineralization horizon for POC was reduced corresponding to the weakening of the CaCO3 export production. In the model, the vertical redistribution of organic matter is performed recursively from layer to layer according to:

equation image
equation image

POCkn+1 and POCkn are the POC concentrations in layer k at the new and old time level respectively, fk−1n+1 is the POC flux from above into the respective layer k, Δzk is the layer thickness, and Zp is the scaling depth or “penetration depth” of POC. Examples for the POC flux attenuation in the water column for different values of Zp are given in Figure 6. In the standard run, Zp is weighted with the atmospheric dust deposition onto the ocean surface. The dust deposition rates are taken from Mahowald et al. [1999]. In the sensitivity experiment, Zp was multiplied by a = 0.5 + 0.5 · F, according the weakening of the CaCO3 export production (see equation (2) for the definition of F). This lead to a considerable shallowing of the remineralization horizon for organic carbon under high pCO2 (Figure 6). The inclusion of this ballast effect formulation practically compensated the small reduction in atmospheric CO2 due to the weakening of the CaCO3 export. The values for the atmospheric pCO2 at year 2250 were: 1425 ppm without feedback, 1413 ppm with CaCO3 production feedback, and 1423 ppm with CaCO3 production feedback and ballast feedback.

Figure 6.

Scaling depth for the vertical POC flux attenuation in the model. Top: Preindustrial situation. Center: At year 2250. Bottom: Examples for the POC flux attenuation for different values of Zp (see equation (3)).

5. Summary

[14] A model scenario for extrapolation of laboratory experiment findings about a potential weakening of the CaCO3 production at high atmospheric CO2 partial pressure was carried out. The scenario suggests a drastic reduction of the pelagic biocalcification during primary production to about 50% of the preindustrial value in 250 years from now. The resulting negative feedback on the atmospheric CO2 concentration due to the smaller alkalinity extraction from surface waters during a weaker CaCO3 export is small as compared to the expected anthropogenic CO2 emissions. Taking into account changes in the ballast effect of CaCO3 for vertical transfer of particulate organic carbon compensates for the pCO2 reduction. The longer term effect of these feedbacks and the range of potential associated impacts will have to be assessed by further studies.


[15] I would like to thank Ernst Maier-Reimer for letting me use his model and for his advice. Two constructive reviewers helped to improve the manuscript. This work was supported through grant EVK2-CT-2001-00100 (EU FP5 “ORFOIS”) by the European Commission. This is publication A60 from the Bjerknes Centre for Climate Research.