Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessing the "ballast" hypothesis for carbon transport in the ocean : global sediment trap data analysis… Izumi, Ryusuke 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-UBC_2010_fall_izumi_ryusuke.PDF [ 2.77MB ]
Metadata
JSON: 24-1.0053295.json
JSON-LD: 24-1.0053295-ld.json
RDF/XML (Pretty): 24-1.0053295-rdf.xml
RDF/JSON: 24-1.0053295-rdf.json
Turtle: 24-1.0053295-turtle.txt
N-Triples: 24-1.0053295-rdf-ntriples.txt
Original Record: 24-1.0053295-source.json
Full Text
24-1.0053295-fulltext.txt
Citation
24-1.0053295.ris

Full Text

ASSESSING THE “BALLAST” HYPOTHESIS FOR CARBON TRANSPORT IN THE OCEAN: GLOBAL SEDIMENT TRAP DATAANALYSIS AND SIMULATION IN AN EARTH SYSTEM MODEL by RYUSUKE IZUMI B.A., Keio University, 1999 B.A., University of Hawaii at Hilo, 2003 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Oceanography) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2010 © Ryusuke Izumi, 2010 Abstract Sediment trap particle flux data analysis and development of a model representation of mineral “ballast” mechanism for carbon transport in the ocean is presented in this study. The validity of several classical POC remineralization models as well as recently hypothesized “ballast” mechanism based POC remineralization models were tested by analyzing data from selected 79 sediment traps at >1500 m from around the world. POC flux modelled with different variations of model representations at each sediment trap site was statistically compared with corresponding measured POC flux in order to evaluate the overall predictability of each model at the global scale. A CaCO3 single-mineral-ballast model could explain up to —79% of the global F variability at depth >1500 m and suggests that CaCO3may potentially be the mineral type that has dominant control on the vertical transport of Fc,c from sea surface to depth in the open ocean. In addition, ai assessment of the impact of reduced CaCO3 production (as a result of ocean surface acidification) on the marine carbon cycle and implications for future atmospheric CO2concentration under the assumption of mineral ballasting of POC is presented. A CaCO3single-mineral-ballast model derived from the data analysis is incorporated into GENIE-i, a computationally efficient carbon-climate Earth System Model of intermediate complexity. Simulation results from a “business as usual” future carbon emissions scenario in GENIE-i suggest that, by year 2300, calcification response of marine calcifying organisms to increased atmospheric CO2 concentrations in a CaCO-ballasting ocean is —63% weaker compared to that in a non-ballasting ocean. With the “ballast” effect in operation, the net effect of climate feedback and calcification feedback is a global decrease in POC export production, except for in some high latitude regions where enhanced POC production due to decreased sea-ice coverage overrides. If the “ballast” hypothesis is true, a CaCO3-ballasting mechanism could completely counter the reduction in atmospheric CO2 concentration by calcification feedback alone in an ocean where no ballasting mechanism is present. 11 Table of Contents Abstract.ii Table of Contents iii List of Tables V List of Figures Vi List of Abbreviations Viii Acknowledgements ix Dedication x I Introduction 1 1.1 POC:PIC “rain ratio” and “carbonate compensation 1 1.2 POC remineralizaion 2 1.3 CaCO dissolution 3 1.4 “Ballast” hypothesis 4 1.5Summary 5 2 Data Analysis 6 2.1 Data 6 2.2 Classical models 7 2.2.1 PP-based model 7 2.2.2 EP-based model 7 2.3 Ballast models 9 2.3.1 Single-ballast model 9 2.3.2 Multiple-ballast model 12 2.4 TE-based analysis 14 2.4.1 TE-based single-ballast model 14 2.4.2 TE-based multiple-ballast model 15 2.5 Summary 15 3 Modeling in GENIE-I 17 3.1 GENIE-i 17 3.1.1 POC flux in the Non-ballasting Ocean 17 3.1.2 CaCO3flux in the Non-ballasting Ocean and the CaCO3-ballasting Ocean 18 3.1.3 POC flux in the CaCO3-ballasting Ocean 19 3.1.4 Model experiment setup and permutations 19 3.i.5NBOvsCBO 20 3.2 Non-ballasting Ocean 21 3.2.1 NBC ‘no feedback’ 21 3.2.2 NBC ‘+climate’ 21 3.2.3 NBC ‘i-calcification’ 21 3.2.4 NBC ‘i-climate i-calcification’ 22 3.3 CaCO-ballasting Ocean 22 3.3.1 CBC +climate’ 22 3.3.2 CBO ‘i-calcification’ 22 3.3.3 CBC ‘i-climate +calcification’ 22 3.4 Multiple-ballasting Ocean 23 3.5 Summary 23 4 Discussion 24 4.1 GENIE-i simulation results 24 4.1.1 NBC ‘i-climate’ 24 4. 1. 2 NBC ‘+calcifiation’ 24 4.1.3 CBC ‘i-climate’ 25 4.1.4 CBO ‘i-calcification’ 25 4.1.5 CBO ‘+climate +calcification’ 25 4.1.6 Multiple-ballasting Ocean 26 4.1.7 Non-ballasting Ocean vs CaCC3-ballasting Ocean 26 4.2 Shallower CaCO3dissolution9 27 111 4.3 Or is it the other way around7 285 Conclusions 29References 55Appendices 59Appendix A: Exchange of CO2 between the atmosphere and ocean 59Appendix B: Carbonate buffer 60Appendix C: Calcium carbonate precipitation/dissolution 61Appendix D: Silicate-carbonate carbon cycle 62 iv List of Tables Table 1. Data 30 Table 2. Correlation matrix 31 Table 3. Linear regression equations, regression coefficients, and R2 against measured data 32 Table 4. GENIE-i simulation results time-series (global) 33 V List of Figures Figure 1. Data 34 Figure la. 79 sediment trap locations showed on the GENIE-i Longitude-Latitude horizontal grid 35 Figure lb. Sea surface PP field on the GENIE-I grid 35 Figure Ic. Sea surface SST field on the GENIE-i grid 35 Figure Id. Sea surface MLD field on the GENIE-i grid 36 Figure ie. Sea surface f-ratio field on the GENIE-i grid 36 Figure if. Sea surface EP field on the GENIE-I grid 36 Figure 2. Data analysis results 37 Figure 2a. F00 measured from 79 sediment traps at different depths> 1500 m 38 Figure 2b. TE measured from 79 sediment traps at different depths> i500 m 38 Figure 2c. Linear regression between F00 measured and F00 modeled with Equation (2a) 38 Figure 2d. Linear regression between b measured and b-modeled with Equation (2c) 38 Figure 2e. Linear regression between F00 measured and F00 modeled with Equation (2e) 38 Figure 2f. F00 and Fm normalized to F 39 Figure 2g. F0, F0 and F1 normalized to Fm 39 Figure 2h. Correlation between F0 and Fm F0, F0, F1 39 Figure 2i. Linear regression between F00 measured and F00 modeled with Equations (3a-3d) 39 Figure 2j. Linear regression between F00 measured and Fy00 modeled with Equations (9a-9d) 39 Figure 2k. Linear regression between Fy00 measured and 00modeled with Equations (12k, 13a, 13b) 40 Figure 21. Correlation between TE and Fm, F0, F0, F1 40 Figure 2m. Linear regression between TE measured and TEmodeled with Equation (16b, 17a, 17b) 40 Figure 2n. Linear regression between TE measured and TEmodeled with Equation (19, 20a, 20b) 40 Figure 3. GENIE-I simulation results time-series (global) 41 Figure 3a. Global carbon emissions inventory from historical record and prescribed future scenario 42 Figure 3b. Global carbon emissions rate from historical record and prescribed future scenario 42 Figure 3c. Global atmospheric pCC2 trajectory obtained from model simulations (NBC) 42 Figure 3d. Global atmospheric pCC2 trajectory obtained from model simulations (CBO) 42 Figure 3e. Difference in atmospheric pCC2with respect to ‘no feedback’ (NBC) 42 Figure 3f. Difference in atmospheric pCC2with respect to ‘no feedback’ (CBC) 42 Figure 3g. Global mean sea surface [H], [CC2], [HCC] and [CC32](NBC) 43 Figure 3h. Global mean sea surface [H], [CC2], [HCC3]and [CC32](CBC) 43 Figure 3i. Global mean sea surface calcite saturation state (NBC) 43 Figure 3j. Global mean sea surface calcite saturation state (CBC) 43 Figure 3k. Global CaCC3 export production (NBC) 43 Figure 31. Global CaCO3export production (CBC) 43 vi Figure 3m. Global POC export production (NBC) 44 Figure 3n. Global POC export production (CBO) 44 Figure 3o. Global mean sea surface [PC4] (NBC) 44 Figure 3p. Global mean sea surface [PC4] (CBO) 44 Figure 4. GENIE-I simulation results time-slices (zonal mean) 45 Figure 4a. Zonal mean (NBC ‘i-climate) 46 Figure 4b. Zonal mean F0 (CBC i-climate’) 46 Figure 4c. Zonal mean [PC4] (NBC ‘i-climate’) 47 Figure 4d. Zonal mean [PC4] (CBC ‘i-climate’) 47 Figure 4e. Zonal mean F0 (CBC ‘i-calcification’) 48 Figure 4f. Zonal mean [PC4] (CBC ‘i-calcification’) 48 Figure 5. GENIE-i simulation results time-slices (surface layer) 49 Figure 5a. Sea surface CaCC3flux at 1765 50 Figure Sb. Sea surface PCCflux at i76S 50 Figure Sc. Sea surface calcite saturation state (NBC ‘i-climate +calcification’) 51 Figure Sd. Sea surface calcite saturation state (CBC ‘+climate i-calcification’) Si Figure Se. Sea surface CaCC3flux (CBC ‘i-climate i-calcification’) 52 Figure Sf. Sea surface PCP flux (CBC ‘i-climate i-calcification’) 52 Figure 5g. Sea surface PCC flux (CBO ‘i-climate i-calcification’) 53 Figure Sh. Sea surface [PC4] (CBO ‘i-climate i-calcification’) 53 Figure Si. Fractional sea-ice extent (NBC ‘i-climate’) 54 vii List of Abbreviations ALK Alkalinity BIOGEM = Biogeochemical Model CBO = CaCO3-Ballasting Ocean DIC = Dissolved Inorganic Carbon DOP = Dissolved Organic Phosphorus EP = Export Production EMBM = Energy-Moisture Balance Model EMIC = Earth System Model Of Intermediate Complexity ESM = Earth System Model GENIE = Grid Enabled Integrated Earth System Model MBO = Multiple-Ballasting Ocean MLD (or z0) = Mixed Layer Depth NBC = Non-ballasting Ocean PlC = Particulate Inorganic Carbon POC = Particulate Organic Carbon POP = Particulate Organic Phosphorus PP = Primary Production SST = Sea Surface Temperature TE = Transfer Efficiency viii Acknowledgements Ryusuke Izumi acknowledges and thanks support from -the Canadian Foundation for Climate and Atmospheric Sciences for research funding. -Dr. Andy Ridgwell for supervising and providing the GENIE-i model. -Dr. Roger Francois for supervising and providing helpful advices on the data analysis. -Dr. Philippe Tortell for supervising and providing overall support throughout the project. ix Dedication /?72 x I Introduction Carbon-bearing “sinking particles” in the open ocean, in their organic and inorganic forms, are one of the major pathways for transportation of carbon from the surface ocean to the deep ocean. One of the highest concerns for those involved in the study of the ocean’s role in the global carbon cycle is in the deciphering of the mechanisms of these carbon-bearing sinking particles as the processes of production, sinking, remineralization of particulate organic carbon (POC) and particulate inorganic carbon (PlC) control the distribution of dissolved inorganic carbon (DIC) and alkalinity (ALK) throughout the ocean reservoir. The distribution of DIC and ALK in turn regulates the strength of the oceanic uptake of carbon dioxide (CQ) gas from the atmosphere. Despite historical efforts by the oceanography community in revealing their nature, the exact controls that determine the composition and magnitude of the fluxes of these sinking particles throughout the ocean water columns as well as their response to future climate change remains poorly understood, which calls for continuous study. Quantitative understandings of the processes involved in the vertical flux of carbon-bearing sinking particles in the ocean are essential for modelling the marine carbon cycle and its role in the regulation of the atmospheric CO2 concentration. The efficiency of the biological pump depends not only on the absolute magnitude of POC and PlC export fluxes that fall down from the euphotic mixed layer of the ocean but also, and more critically, on their vertical remineralization length-scale profiles through the water column. In other words, the transfer efficiency of carbon-bearing particles from the sea surface to the deep ocean. Here, I briefly review the classical paradigms in oceanography on the behaviour of carbon-bearing sinking particles (and discuss the recent shifts in some of those views in later sections). Particular attention is placed here on the PlC component, which plays an important role in determining the distribution of ocean chemical properties and thus the buffering capacity of the ocean against acidification induced by the increasing uptake of atmospheric CO2 gas at the air-sea interface. PlC in the ocean occurs in the form of biogenic calcium carbonate (CaCO3)and the two terms will be used interchangeably in this paper. See Appendix A-D for more detailed descriptions of the chemical reactions and mechanisms referred to in this paper. 1.1 POC:PIC “rain ratio” and “carbonate compensation” In order to maintain a steady state of ALK in the ocean on timescales of thousand of years, weathering inputs of ALK into the ocean must be balanced by equivalent outputs, and any perturbation that drives a change in the CaCO3 burial rate must be compensated for by some kind of a mechanism which adjusts the carbonate ion concentration [CO32]in the ocean to restore balance; a process termed “carbonate compensation” (Broecker and Peng, 1987). One of the most plausible biogeochemical mechanisms for controlling the level of atmcspheric CO2 concentrations over timescales of thousands to hundred of thousands of years, is that of the POC:PIC “rain ratio” (i.e. POC to PlC molar ratio of the sinking particles) (Archer and Maier-Reimer, 1994). This rabo has an influence on the preservation of PlC in the ocean water columns and sediments. While the classical estimate for the export rain ratio of POC:PIC is somewhere around 4 (i.e. 1 mol PlC per 4 mol of POC) (Li et al., 1969), a recent WOCE data analysis of phosphate and alkalinity gradients in the 100-200 m by Sarmiento et al. (2002) 1 has suggested 10 or more. CaCO3dissolution is not only determined by the saturation state of CaCO3 (C)) (see Section 3.1.2 for definition of 0) in the ambient environment but also by the amount of metabolic CO2 (i.e. acidity) released through the respiration of POC. The change in ocean [CC32], in turn, impacts the atmospheric partial pressure of CO2 (pCO2)due to its well established inverse relationship between atmospheric pCO2(see Appendix B: Carbonate Buffer).On the assumption that any changes in the POC:PIC export rain ratio at the surface ocean will be transmitted to the deep sea in a direct proportional manner, any change in the magnitude of POC flux relative to CaCO3flux will affect the fraction of CaCO3 that dissolves. In other words, the strength of respiratory dissolution of CaCO3 can be altered by changing the surface ocean productivity of POC and PlC, relative to each other. Earlier numerical model studies have reported that carbonate compensation only fully operates on a timescale of 5-10k yr(Archeretal., 1997) and that any perturbations to the POC:PIC rain ratio over relatively shorter timescales could affect the strength of oceanic CO uptake by altering the ocean pH before the disturbance can be smoothed out. Recent studies on future projections of surface ocean C) with respect to calcite and aragonite over the next several hundred years strongly suggest that in the course of increasing atmospheric pCO2 and subsequent ocean acidification, calcification rates of marine organisms would significantly decrease due to a pronounce lowering of [C032](Orr et at., 2005; Caldeira and Wickett, 2005). A reduction in the export rate of CaCO3 relative to POC caused by negative calcification responses of calcifying organisms to increasing CO2 invasion (Gattuso et at., 1999; Riebesell et at., 2000; Zondervan et al., 2001) could initiate a faster carbonate compensation process by leaving a larger fraction of ALK in the ocean. Carbonate compensation to this effect is an initial decrease in CaCO3burial that is eventually compensated for by an increase in the [CC32] through enhanced CaCO3 dissolution that leads to CO2 uptake by the ocean (i. e. decrease in atmospheric pCO2). 1.2 POC remineralizaion Despite the very close specific gravity of POC to that of seawater, observations from sediment traps deployed at different depths around the world ocean have confirmed that POC generated in the surface ocean is sinking rapidly (50-200 m d1) to the deep sea. This has been explained mainly as results of aggregation of small individual particles and fecal pellets into larger particles (>0.5 mm) (Honjo, 1996), often referred to as “marine snow”. Some more recent studies postulate either a “ballast” effect or “protection” against biodegradation by other sinking particles of mineral phases (i.e. CaCO2, biogenic opal (SIC2), lithogenic dust) (Armstrong et al., 2002; Klass and Archer, 2002). The ratio of export production (EP) to net primary production (PP) in a certain area is defined as the f-ratio. The production of large rapidly settling cells such as diatoms in regions of high f ratios are known to make a greater contribution to the biological pump than the production of small suspended particles originating from pico- and nano-plankton in low f-ratio regions (Ganeshram, 2002). While the carbonate-dominated low f-ratio regions are considered to contribute less to the transferring of POC to the deep sea, Francois et at. (2002), proposed that the fraction of POC exported in these regions is more efficiently transferred to the deep sea. 2 Perhaps the most influential formulation of POC remineralizaion in the oceanography community during the past few decades is that published by Martinet al. (1987). Martin et al., by placing an empirical fit to observed data, described the variation of sinking POC flux with depth as a power-law function; F0(Z) EP(zIzo)’ EP(z/1 00)0858, where the depth of the surface mixed layer (MLD or z0) from which POC is exported is set to 100 m and -0.858 is a dimensionless negative constant to account for unexplainable variables. In other words, the POC flux at any depth z can be predicted from the export flux of POC out of the surface mixed layer. While the depth dependent term is necessary for the algorithm to reflect the gradual remineralization of POC, the effect of depths becomes rather subtle in the bathypelagic zone (> 1500 m depth) as most of the POC is remineralized at shallower depths. Many ocean biogeochemistry simulation models to date have this power-law curve of Martinet al. or its variation for the purpose of predicting the depth distribution of POC remineralization. However, it has to be emphasized that the “Martin et al. curve” was based on measurements restricted spatially to the Northeast Pacific and only from floating sediment traps at relatively shallow depths (<2000 m), and also temporally to short-term (6-34 days). Its applicability to the global ocean and entire depth range of the water column, therefore, remains highly questionable. It has to be also noted that the assumptions in measurements of sinking particle fluxes by sediment traps, in a simple 1-dimension mass balance, are that particles settle vertically between traps at different depths, that all traps collect particles with equal efficiency, and that particles trapped in the collection cup are not disturbed. Such caveats may result in an over- or under- trapping due to hydrodynamic biases such as current turbulence, horizontal advection, and the presence of heterotrophic swimmers. While deep-sea sediment traps moored in the bathypelagic zone have been shown to fairly measure the vertical flux of sinking particles, measurements from sediment traps deployed at depths above the mesopelagic zone (< 1500 m) are observed clearly susceptible to under-trapping (Yu et al., 2001; Scholten et al., 2001). 1.3 CaCO dissolution The major producers of biogenic CaCO3 in the open ocean are known to be planktonic calcifying organisms such as coccolithophores and foraminifera, both of which secrete calcite, and pteropods which form shells of aragonite. The production of CaCO3 by calcifying organisms in the surface ocean, by taking up ALK and DIC from the ambient water in a 2:1 ratio through the reaction Ca2 + 2HCO —* CaCO3 + CO2 (aq) + H20, raises the pCO2 of seawater and acts as a potential source of CO2 to the atmosphere. It is important to recognize that, in the real ocean, the production of CaCO and release of CO2 does not occur in a 1:1 ratio. In the modern ocean, the CO2 release to CaCO3 production ratio (‘1’) is found to be approximately 0.6 (Ware et al., 1992). This owes to the buffering capacity of the more abundant bicarbonate (HCO) and carbonate (C032)ions in seawater. The decrease in pH by an increase in CO2 is largely compensated for by the reaction of hydrogen ion (H), the other hydrolysis product when CO2 reacts with seawater to form carbonic acid (H2C03),with C032.This ratio is predicted to increase as pCO2 increases (Frankignoulle et al., 1994), that is, the buffering capacity decreases as both H and C032 are consumed in the course of ocean acidification. An emerging concern on timescales that are directly relevant to our society is changes in the partitioning of 3 ALK and DIC in the ocean owing to changes in CaCO3 production and dissolution due to the adverse response of planktonic calcifying organisms to increasing pCO2 in the surface ocean. The depth range over which seawater is saturated with respect to CaCO3,is called the saturation horizon. At depths below the saturation horizon, CaCO3will spontaneously dissolve if exposed to seawater for a sufficient period of time. CaCQ3 dissolution below the saturation horizon is often expressed at a rate proportional to (2 - i)’, where i is the dissolution rate power (i.e. measure of the strength of the non-linear response of calcification to changes in ambient [C032])and is very difficult to constrain (Zhong and Mucci, 1993). Historically, it has been strongly held that pelagic CaCO3 is thermodynamically stable and exhibits a conservative nature in the surface ocean and down to several kilometres depth and that significant dissolution of biogenic CaCQ3 only occur at great depths below the calcite lysocline; the depth at which shell dissolution starts to have a detectable impact on the preservation of CaCO3 (Broecker and Peng, 1982). 1.4 “Ballast” hypothesis Although often not very obvious, the distinction between PIC:POC export rain ratio (i.e. at the base of the mixed layer) and its sediment rain ratio (i.e. at the sea floor) is important. Measurements of POC fluxes at different locations and depths around the world have discovered that, although absolute POC fluxes vary by more than an order of magnitude, in the deep sea (>1000 m) they appear relatively constant when normalized to total mass flux, making up —5% of dry weight fluxes (Armstrong et al., 2002). Another remarkable feature found from global sediment trap data set is the constancy of the POC:PIC rain ratio found in the deep sea. While the absolute fluxes of POC and PlC vary widely between different sites, the deep sea rain ratio appears quite constant falling in a narrow range between 0.5 -1.0. In search for a plausible explanation, researchers have investigated the importance of relatively denser sinking particles such as CaCO3,biogenic opal and lithogenic dusts (collectively termed “ballast minerals”) on the control of POC sinking flux (Armstrong et al., 2002; Klaas and Archer, 2002; Francois et al. 2002). These minerals have been identified as major constituents of the total sinking particle flux, often comprising more than half the mass of the flux leaving the surface (Honjo, 1996). On a basis that the density of POC is so close to that of seawater that individual POC by itself would not sink efficiently, these studies propose that POC flux may be in direct proportion to the flux of ballast minerals, where these minerals either enhance the settling rate of POC by increasing particle density or by protecting it from biodegradation, and that the POC:PIC rain ratio may be fairly fixed in the bathypelagic deep sea. If valid, the ballast model challenges the classical rain ratio hypothesis since it implies that any surface ocean perturbation that decreases CaCO3export will also drive a reduction in POC flux through the water column (i.e. POC flux is coupled to PlC flux), and that the POC:PIC rain ratio in the deep ocean above the sediment surface is rather insensitive to changes in the export rain ratio (Ridgwell, 2003). Remineralization curves of POC derived from classical empirical fitting to data such as in Martinet al. (1987) is fundamentally based on the assumption that all the information necessary for prediction is contained in the POC flux itself and that the fluxes of POC and PlC are independent from each other. From a sediment trap study in the equatorial Pacific and the Arabian Sea, Armstrong et al. (2002) proposed a model where POC flux is subdivided into a physically or chemically ‘protected’ fraction associated with ballast minerals and an 4 ‘unprotected’ fraction that remineralizes exponentially with depth. Their equation states that the protected POC flux decreases asymptotically with depth and is dictated proportionally by the also asymptotic flux of ballast minerals in terms of its “transport ratio”. This transport ratio is a measure of the carrying capacity of POC by ballast minerals (i.e. POC carried per ballast mineral) and has been suggested to vary with ballast composition. Denser CaCO3appears more efficient at transferring POC to the deep sea than less dense opal, and that oflithogenic material varies widely and remains unclear since they rarely dominate the composition of particles in the open ocean (Klaas and Archer, 2002; Francois et al., 2002). Klaas and Archer (2002) extended the analysis of Armstrong et al. to a global sediment trap data set and concluded that up to 83% of the POC flux to the deep sea is associated with the ballasting by CaCO3.However, one puzzle is that density cannot be the sole controlling factor since Si02 is also denser than POC, and yet does not appear to show a strong ballasting capacity (Klaas and Archer, 2002; Francois et al., 2002). Francois et al. (2002) attributes this enigma to differences in the “transfer efficiency” (defined as POC flux normalized to export production; TE = FPOC/EP),illustrating that carbonate-dominated regions with low f-ratios and seasonality may have a higher efficiency in transporting POC compared to opal-dominated regions with high f-ratios and seasonality because systematic differences in the degree of biodegradability and “packaging” factor of sinking POC particles are likely. The ballast hypothesis explains the relative constancy of the rain ratio in the deep sea, which can be described as ‘buffered’ against any POC:PIC rain ratio perturbations in the upper ocean (Ridgwell, 2003). Numerical modelling has shown that the ballast effect can be capable of counteracting the negative calcification response of calcifying organism to increasing pCO2, and may instead result in a net positive feedback (Barker et al., 2003; Heinze, 2004). 1.5 Summary Fluxes of ballast minerals may determine POC fluxes at depth. Thus, a mechanism-based model of POC flux must simultaneously predict fluxes of both POC and PIG as well as other potentially ballast minerals. Since thefactors controlling their remineralization are still poorly understood, development of pure ecological or mechanistic models is in its infant stage. Meanwhile, marine carbon cycle models struggle for finding alternative ways to incorporate remineralization algorithms of carbon-bearing sinking particles in the open ocean, often resorting to ‘tuning’ any unknown parameters (e.g. Martin et al., 1987). Clearly, further progress in a better understanding of the interactions between POC and PlC fluxes and the development of a truly global depth-integral algorithm on the remineralization lengths scales of both fluxes requires more accurate and globallydistributed particle flux measurements, particularly in the upper ocean. The updated global annually averaged sediment trap flux data available at the U. S. Joint Global Ocean Flux Study data directory, with its richer data set, makes it worth reassessing the global applicability of the different algorithms reviewed here. The main objective of this study is to assess the “ballast” hypothesis for the transport of POC through the water column and how an alternative POC-PIC coupled model would compare to the conventional uncoupled model predictions of marine carbon cycle and global climate change. 5 2 Data Analysis Empirical models of POC remineralization through the water column formulated in preceding studies were derived from observations made in limited ocean regions and water depths (Suess, 1980; Martinet al., 1987; Armstrong et al., 2002)]. Statistical analyses of sinking particle fluxes have used different data sets where different criteria were used for data selection (Lutzet al., 2002; Francois et al., 2002; Klaas and Archer, 2002). Discrepancies in space and time from which conclusions of these earlier studies were derived make their direct comparison difficult, which leaves ambiguity in their validity. In order to truly evaluate the applicability of these models to the global ocean, we tested the validity of several classical and ballast POC remineralization models under a single context with our globally distributed 79 sediment trap data. We evaluated the different models in their original forms, and as an elaboration, different variations of the models were also tested. In order to evaluate the overall predictability of each model for the global ocean, POC flux modelled at each selected sediment trap site was statistically compared with corresponding measured POC flux. 2.1 Data A larger number of sediment traps have been deployed in the world ocean over the past few decades. We took the online database of global sediment trap flux data compiled in a U.S. JGOFS Synthesis and Modeling Project (http://usjgofs.whoi.edu/mzweb/smppi/honjo.html). The original database consists of 193 sites from around the global ocean and contains annual integrated data from 467 sediment traps. However, the rich database could not be fully utilized in this study. Since sediment traps deployed at depths shallower than 1200 m, above the mesopelagic zone, are known to be subject to poor trapping efficiency (Yu et al., 2001; Scholten et al., 2001), we have chosen a 1500 m depth cut-off and excluded all traps above 1500 m from our study. In order to minimize seasonal bias, only traps with sample collection duration of more than one full year (>365 days) in high latitudes (>30° N and S) and 10 months (>305 days) in low latitudes (<30° N and S) were selected. Since this study focuses on the flux of sinking particles in the open ocean, traps deployed in coastal seas as well as in coordinates where the relatively coarse resolution of GENIE-I (our Earth System Model) recognizes as ‘land’ were excluded. Finally, traps which met the above criteria but had insufficient measurement data for proper analysis were excluded. Accordingly, we obtained a total of 79 sediment traps distributed over 14 biogeochemical provinces as defined by the Longhurst Biogeochemical Pra’ince Map (Longhurst et al., 1995) around the global ocean for analysis [Table 1]. The complied data used in this study is shown in Table 1. POC flux measured at the selected 79 sediment traps are shown in Figure 2a. Transfer efficiency (TE) of the POC flux is shown in Figure 2b. Flux of particulate organic matter was calculated by Fpom = I .87F0 (Anderson, 1995). CaCO3 flux (F0) was calculated by F0 = 8.33F,0(based on the ratio CaCO3I = 100.0872/12.011 = 8.33). Flux of opal was derived based on its conversion factor between biogenic silica; F0 =2.l4Fsb0(Mortlock and Froelich, 1989). The combined mineral fraction (Fm) of the total sinking particle flux (Fr) and the lithogenic (F1) component of the total mineral flux were calculated simply by difference; Fm = F - Fpom, F1 = F - (F0 + F0 + Fpom). A net primary production (PP) field 6 estimated from the satellite-derived productivity model of Behrenfeld and Falkowski (1997), a sea surface temperature (SST) field from NODC Levitus WOA 98 dataset and a mixed layer depth (MLD or z0) field based on change in potential density (Monterey and Levitus, 1997) was adopted for use in this study. These PP, SST and MLD fields from the source data were re-gridded to a resolution of 36 by 36 equal area grids to match the grids of GENIE-i [Fig. lb-id]. In a seminal paper, Eppley and Peterson (1979) defined the ratio of export production (EP) to PP as f-ratio. f-ratio for each grid were calculated using the algorithm of Laws et al. (2000), where a two- dimensional interpolation computes f—ratio as a function of PP and SST [Fig. le]. Export production for each grid was then calculated from its definition EP = fPP [Fig. If]. 2.2 Classical models The correlation matrix between TE, PP, EP, f-ratio and SST are shown in [Table 2]. A medium positive correlation is found between F0 and PP (R = 0.500) whereas correlation between F0 and EP is poor (R 0.139). These statistical numbers suggest that neither PP nor EP alone contains all the information needed to predict F0 in the water column. 2.2.1 PP-based model We first evaluated the global applicability of the PP-based F0 prediction by Suess (1980), which has been a commonly applied description ofF00(Lutz et al., 2002). F0(z) = PP/(0.0238z + 0.212) (1) Equation (1) was applied to the 79 sediment traps selected in this study. A linear regression analysis between F0 modelled using equation (1) and F0 measured resulted in a coefficient of determination (R2) of 0.445, meaning only —45% of the F0 variability in the world ocean could be explained by this PP-based F0 prediction model. 2.2.2 EP-based model Another widely adopted description of F0 is that formed by Martin et al. (1987). In their original equation, which was derived from observation in the upper 2000 m in the northeast Pacific only, the mixed layer depth (z) is fixed to 100 m and the dimensionless parameterb for unknown factors is tuned to -0.858 in order to obtain a best fit to measured F0C() = EP(zIzo)” = EP(z/i00)°858 (2) 7 By simply applying Equation (2) as it is to our global data set resulted in a poor regression between modelled and measured F00 (R2 0.150). In reality, upon applying equation (2) to the global ocean, it has to be taken into consideration that z0 in fact vary among different ocean regions [Table 1; Fig. id]. Thus, in order to evaluate the applicability of Equation (2) to the global ocean better, we allowed it to take region specific values of in which the 79 sediment traps are located. F00 at each sediment trap was predicted using equations with z0 values specific to each grid. F0(z) = EP(ZIZOgrid)08 (2a) With Equation (2a), regression between modeled and measured dropped to an extremely low R2 value of 0.00350 [Fig. 2c]. These results strongly show that the applicability of Martn et al.’s (1987) original equation to the global ocean is very poor. Figure 2c shows two distinctive clusters of low latitude regions with less negative b and high latitude regions with more negative b. It also demonstrates that the original parameter exponent b (-0.858) systematically overestimates F00 in the low latitude regions and underestimates in the high latitude regions. Overestimation or underestimation of F00 by this equation at a certain sediment trap could be because either b is too high or low, or because Law et al’s (2002) algorithm systematically overestimate/underestimate EP in that region. By accepting Laws et al.’s (2000) EP estimate and taking the assumption that b correlates with the flux of biogenic minerals (i.e. ballast hypothesis; CaCO3 in particular) and other variables such as SST or f ratio (Francois et al., 2002), we calculated ‘b-modeled’ specific to each sediment trap. By re-arranging equation(2), b for the measured at each sediment trap could be calculated. b = ln(F,,jEP)IIn(z/z0)(2b) Multi-linear-regressions including F0 and SST or f-ratio as controlling factors give: e” = kF0 + kSST = 4.780610 F0 +9.53841 Q3. SST +2.119110 (2c) eb = k0F + k,ratiofratio = 7.250910 F0 6.6667.10i f-ratio +5.21471Q (2d) When compared against calculated eb from measured data, modeled eb derived from Equation (2c) and (2d) obtained R2 of 0.755 and 0.574, respectively. Subsequently, eb calculated from Equation (2c) were used to derive modeled values of b. R2 obtained from regression between b-modeled and b measured obtained 0.762 [Fig. 2d]. A more negative exponent b drives shallower F00 remineralization (i.e. lower TE) of carbon to the deep sea, and a more positive exponent b drives POC remineralization at deeper depth (i.e. higher TE). We applied the region specific z0 and the modeled b values to Equation (2): 8 F0(z) EP.(ZIZo)bmd (2e) Equation (2e), with region specific z0 and b, dramatically improved the predictability of F00 up to a much more reasonable R2 of 0.581 [Fig. 2e]. These results confirm that exponent b must vary between different ocean regions rather than globally constant, and also support the speculation that F0 seen in the ocean may be dependent on F0 and seasonality under some yet unknown systematic relationship. 2.3 Ballast models POC flux at depths >1500 m around the world ocean differ considerably between different geochemical regions and within each region, ranging between 0.1 and 5.96 (g rn2 yj [Fig. 2a]. Despite that Fpc measured throughout the world ocean differ by a factor of —60 when POC flux is normalized to total flux, the range of FpOC/Ft becomes strikingly narrow (0.0125-0.205), with a global mean of 0.0529 [Fig. 2f]. This consistency over the world ocean where F00JF becomes virtually constant at depths >1500 m along with the fact that most of F is composed of Fm (mean Fm/Ft = 0.901) and F0 being the major fraction (mean Fe/Fm = 0.545) [Fig. 2gj, makes it plausible to speculate a deep-water asymptotic behaviour of F tightly coupled with F, particularly F0, in the ocean. In order to evaluate the postulated association between POC flux and potentially ‘ballast’ mineral flux in our global data set, we first performed a correlation analysis between Fm and F00. We also repeated the same analysis for F0, F0 and F1to find the strength of correlation between the flux of each mineral type fraction and F0 [Fig. 2h]. Correlation between F and mineral flux (Fm) resulted in a strong positive correlation (R = 0.863). Correlation between F00 and CaCO3 flux (F0) alone (R = 0.860) did not differ from that of Fm. The lack of considerable correlation between opal flux (F0) and F,0, (R = 0.387) suggests that despite the commonly observed high F0 to F0 ratio in some highly productive ocean regions, F0 does not play an important role in the control of F. This could be simply due to the comparatively less dense property of biogenic opal (2.1 g crrn3) than biogenic carbonate (2.71 g cmj, or could be attributed to factors other than particle density, such as lower transfer efficiency (TE) of POC in opal-dominated regions due to higher biodegradability of POC in such low SST high f—ratio regions (Francois et al., 2002). The relatively lower correlation between F00 and the flux of lithogenic minerals (F), R = 0.654, compared to that between F and F0 is most likely because although the density of F1 (e.g. 2.65 g cm3 for quartz) is relatively high, F1 is known to be quantitatively small in bathypelagic depth of the open sea. The potential ballasting effect of F1, if any, may be restricted to coastal oceans and would require evaluations outside the scope of this study. 2.3.1 Single-ballast model From regression analysis, we obtained simple equations for testing the ability of sinking mineral particles in predicting Fpcm, expressed as: 9 F=kbFb (3) Fb is the flux of ballast mineral(s). The coefficient kballast could be interpreted as the ‘carrying capacity’ of each mineral type as ballast. Regression between modeled F0 and measured F0obtained R2 of 0.744 for F0 predicted with all mineral types grouped together as one fraction Fm (Table 3. Eq. 3a). F0 modeled by each mineral fraction F, F0 and F1 alone (Table 3. Eq. 3b-3d) compares with measured F0 by R2 values of 0.739, 0.150 and 0.428, respectively [Fig. 2i]. While R2 obtained between Fm and F were both strong and practically did not differ between each other, R2 obtained between F0 and F were distinctively low. F appears to be the mineral flux which may potentially have dominant control on F, while the influences of F0 and F on F are likely to be negligible, at least at the global scale. Each model was tested with an additional depths dependent term which reflects the possibility of gradual remineralization of ballasted POC flux with depth. Here, we arbitrarily chose a simple inverse function to describe this effect. F = kt,Fb + kz1 (4) Regression with a depth term in the equations resulted in slightly improved R2 values (Table 3. Eq.4a-4d). The additional depth dependent term accounted for up to —4% (mineral ballast) and —2% (CaCQ3ballast) of the measured F00 variability, which suggests that at least a fraction of POC flux still undergoes some gradual remineralization at depth >1500 m while most of the POC in the deep sea is protected by ballast minerals. Armstrong et al. (2002) derived a more mechanism-based single-ballast model from their regional study in the equatorial Pacific and the Arabian Sea. Based on a mechanistic assumption, F is partitioned into two distinctive pools, conceptually written as: F0 (z) = F0QA (z) + F0 E (z) (5) where QA denotes POC flux “quantitatively associated” with ballast minerals and E denotes “excess” POC flux. FpocaA is the ballasted (or protected) POC flux subject to remineralization only when the coupled mineral particles dissolve. F0E is the free POC flux which is not coupled with any mineral flux, therefore quickly remineralizes in the upper water column. It is assumed that both types of F0 are associated with the same aggregates of sinking particles (e.g. flocks, fecal pellets) which provides the increased density needed for both pools of carbon to sink. As described in Armstrong et al. (2002), to incorporate into model experiments, Equation (5) could be expressed as: F — + r . [-(z-zO)/öE]poc (z) — poc () I poc (zO) — poc (.) = Fp0() + [EP — FpOc()].e[_2o)EJ (6) where F() is the asymptotic POC flux quantitatively associated with ballast minerals. [EP — calculates the excess non-coupled free POC flux, where E remineralization length scale. (zO( E EP. 10 When either direct measurements or model predictions of the ballast mineral flux at the mixed layer depth (F)) and asymptotic ballast mineral flux (Fb()) are available, FO() can be evaluated by: FO() = p Fb() = p Fbc[Fb(/Fb(Zo)] (7) where p asymptotic transport ratio (FpOC()/Fb()) and [Fb()/Fb(zo)] proportion of ballast mineral dissolution. The constant of proportionality p describes the proportional coupling between F00(.) and Fb(.). The asymptotic POC flux Fpc is calculated point-by-point by the surface ballast mineral flux Fb , rather than by the surface POC flux Fp0C(Zo). Therefore, for a global scale study, unless direct measurements are available for the entire area of interest (which is often not the case), Equations (6) and (7) must be evaluated with a simulation model that could output Fb > and Fb >. When only Fb () are known from limited numbers of sediment traps (assuming Fb ()E Fb(z>1500m)), as in the case of our data set here, we further assume Fb() Fb() as demonstrated in Francois et al. (2002). With these assumptions, Fb()/Fb() yields 1 and Equation (7) reduces to: Fp00() = P Fb(z>1500m) (8) Armstrong et al. (2002) had suggested parameter values P 0.05 and E 500 m for testing the implications of ballast mineral association in global scale model simulations. Assuming a single-ballast of Fb as the sum of all mineral types (i.e. Fb Fm), Equation (6) at depths >1500m becomes: F0 0.05Fm + [EP - 0.05.Fm].er ZO)(5OO (9) Regression between F0 predicted with Equation (9) and measured at the 79 sediment traps resulted in a value of 0197, which is considerably high and suggests that mineral fluxes may indeed play an important role in the control of F00. In our analysis we calculated p at the 79 sediment traps and obtained a global mean p value of 0.0605. F0 0.0605Fm + [EP - 0.0605Fm]e °°° (9a) R2 obtained from using Equation (9a) was 0.798 [Fig. 2j], which does not differ from that obtained by equation(9). This suggests that the model is not sensitive to a 0.01 change in the global asymptotic transport ratio.WhileArmstrong et al’s (2002) original equation simply considered Fb Fm, we elaborated the idea by testing thepossibility where one particular mineral type may be dominantly responsible for the proposed ballasting effect; Fb F, Fb F0 and Fb F. The global mean asymptotic transport ratio p between POC and each mineral type, assuming ballasting effect by a single mineral type, was calculated from measured data; p Fpc,c(oo)IFc(), Pp F0(c0)IF(o0). Subsequently, the following equations were formulated. 11 F00 0.130F + [EP - 0.130.Fc].ezoofh (9b) F00 0.295F + [EP - (9c) F00 = 0.531F + [EP - 0.531.F].eIzO)/ooJ (9d) R2between F00 measured and F0 modelled obtained with equations (9b), (9c) and (9d) was 0.787, 0.173 and 0.451, respectively [Fig. 2j]. F00 modelled with F0 alone as ballast mineral matched fairly tightly with measured F00as good as that modelled with Fm while predictions made with F0 or F1 alone were poor. These results again suggest that F0 may potentially be the mineral type that has dominant control on Fpc. As >90% of F00 is remineralized at shallow waters >1500 m [Fig. 2b; mean FPOCIEP 0.0634], we tested the possibility that ‘excess” free POC flux F00E at depth >1500 m may be in fact negligible. With this assumption, Equations (9a-9d) reduces to: Fy00 = 0.0605Fm (lOa) F00 = 0.130F (lOb) F00 = 0.295F (lOc) F00 = 0.531F (lOd) R2obtained from Equations (lOa), (lob), (lOc) and (lOd) was 0.744, 0.739, 0.150, and 0.428, respectively. The variability of measured F0 explained by each equation was only up to —5% lower compared to equations with the F0E term. This suggests that, at most, only a very small fraction of FpOCE survive to depth >1500m and experience gradual remineralization in the bathypelagic depths of the ocean. We find that Armstrong et al.’s[2002] mechanism-based single-ballast model could explain up to —80% of the global F00 variability at depth >1500 m. For the equations assuming Fm and F0 as ballast mineral (Equations 1 Oa and lob), it is noteworthy that the asymptotic transport ratio p calculated from measured data and the statistically derived carrying capacity coefficient kb in the single-ballast models (Equations 3a and 3b) are strikingly proximate (p = 0.0605, O.130,kb =6.0966.10.2, 1.125110’). On the other hand, p and kb did not match well for equations assuming F0 and F1 asballast. Armstrong et al.’s [2002] ballast model appears to be valid only when F0 is the dominant mineral flux thatis coupled with F00. 2.3.2 Multiple-ballast model Next, we evaluated Klaas and Archer’s (2002) multiple-ballast model. Conceptually following Armstrong et 12 al.’s [2002] formation: F=kFb+FE (11) To start from a simple form, F0E is first assumed negligible at depths >1500 m so that the ballasted POC fluxes are assumed to sink conservatively. Fb is partitioned into multiple ballast fractions F, F0, and F instead of collectively Fm. F0 = kF + k0F + kF (12) The POC carrying capacity k for each mineral fraction can then be statistically determined by multiple-linear- regression. We obtained coefficients of 9.0426.102, 9.709810 and 8.0843.10.2 for k, k0 and k, respectively (Table 3. Eq. 12). Regression between Fpm measured and modelled with Equation (12) resulted in a remarkably high R2 of 0.867, which suggests that a multiple-ballast algorithm is able to explain —87% of the global variability in F0 at depth >1500 m [Fig. 2k]. We also highlight that the high coefficient for F compared to that of F0 confirms the importance of CaCO3mineral as ballast mineral for POC transportation to the deep water. The high coefficient for F1 could potentially be important if observed lithogenic fluxes were higher. In order to explore the importance of other potential variables in predicting POC flux, additional SST or f-ratio dependent terms were experimentally added. kF + k0F + k1F + kvariable (13) R2 obtained from equations adding SST or f-ratio dependent terms were almost equivalent compared to that obtained from Equation (12) [Table 3. Eq. 13a, Eq. 13b; Fig. 2k]. Although adding the variable SST orf-ratio into the regression does not improve the overall predictability of it is worth noting the signs of their coefficients (positive for SST, negative for f-ratio), which agrees with the correlation matrix in [Table 2]. F0 seems to increase with SST and decrease with f-ratio. Since the combination of high SST and low f-ratio are generally found in carbonate-dominated regions with low seasonality whereas the combination of low SST and high f-ratio are found in opal-dominated regions with high seasonality (Francois et al., 2002) [Fig. ic, Fig. le], this trend implies a relationship between the remineralization profile of F and seasonality in different regions. A depth dependent term can be added to Equation (12) in order to account for the small remineralization of the ballasted POC flux sinking through the water column at depth >1500m. F00 = kF + k0F + k1F + kz1 (14) Adding a depth dependent term to the multiple-ballast model improves the Fpc variability explained by only —2% (Table 3. Eq. 14), which again suggests that the ballasted F0 flux at depths >1500 m may be sinking in an almost conservative manner by either ‘fast sinking’ or ‘protection’ effect, while a fraction may still experiences 13 some decay through their journey to the deep ocean as the associated minerals themselves experience dissolution. Subsequently, as in the preceding analyses, we also tested if the ‘excess’ free POC fraction Fpcc E may still be a considerable component of the flux at depth >1500 m. As we could assumed that FE represents most of the EP sinking down from the MLD that appears to almost entirely remineralized above the mesopelagic depth (< 1500 m) [Fig. 2b], this flux could be considered proportional to surface estimates of EP that remineralizes with depth. F00 k0F + k0F + k1F + kz-1 + kepEP = k0F + k0F + kF + kz’ +k0EP (15) F00 modeled from Equation (15) did not differ from that obtained from Equation (14) where FE is considered negligible (Table 3). This suggests there still remains a possibility that ‘excess’ free fraction of POC flux may actually be totally remineralized within the upper 1500 m of the ocean before reaching the bathypelagic waters. 2.4 TE-based analysis An alternative approach to analyzing the relationship between POC flux and mineral flux is to evaluate themin terms of ‘transfer efficiency’. By normalizing the POC flux measured at depth to estimates of export production, the unknown dependence between Fpcm and EP is removed to look specifically at the factors that affect the efficiency of the transfer of carbon from the surface to the deep sea. The range of variability of the normalized data (FPOCIEP) is much smaller than that of EP alone [Table 1]. 2.4.1 TE-based single-ballast model Correlation coefficients between TE and the flux of each mineral type were calculated for our global sediment trap data assuming a single-ballast model, giving R values of 0.551, 0.612, 0.120 and 0.497 for Fm, F0, F0 and F1, respectively [Fig. 21]. F0 showed a reasonable correlation between TE, followed by Fm. F0 and F1 showed no considerable correlation between TE. Assuming single-ballast, TE could be modeled by the following equation. TE=kbFb+kZi1 (16) Regression between TE measured and TE modeled with Equations (16a-16d) showed a highest predictability of transfer efficiency by F0 alone, accounting for —47% of the global TE variability at depth >1500 m [Table 3. Eq.16b; Fig. 2m]. This result shows that although the coefficient of determination (l) is medium, F0 indeed appears to have a dominant control over other minerals on the efficiency of POC transport to the deep ocean interior. Asin the preceding analyses, additional terms of SST and f-ratio were added to Equation (16)to explore the importance of these variables on the control of POC flux. 14 TE = kb Fb + kz-1 + kvariable (17) TE modeled with an additional term of SST or f-ratio improved added —10% to the ability in explaining the measured variability of TE [Table 3. Eq.17a, Eq. 17b; Fig. 2m]..Aain, the positive and negative signs of the coefficients of SST and f-ratio, respectively, should be highlighted. F0at each trap location could be predicted by re-arranging Equation (17b). F0 = EP[3.00721OF+ 8.5165.10*l.z - 1.7268.1Otfratio + 2.444210j (18) F0 modeled using Equation (18) obtained a R2 = 0.568 against F0 measured. 2.4.2 TE-based multiple-ballast model Multiple-linear-regression was carried out to obtain the contribution of each mineral type to transfer efficiency of POC flux. TE = + k0F + krFi + kz1 (19) Equation (19) obtained R2 = 0.550 when compared against TE measured [Table 3. Eq. 19; Fig. 2n]. Again, additional linear terms of SST or f-ratio were added to the regression to explore the importance of these variables on the control of POC flux. TE = + kF0 + ktFi + kz1 + kvahable (20) An additional term of SST or f-ratio further improved the predictability of TE at depth >1500 m, with R2 of 0.646 and 0.688, respectively [Table 3. Eq. 20a, Eq. 20b; Fig 2n]. The same trend is repeatedly seen, where the coefficient for SST and f-ratio is positive and negative, respectively. These results again suggest the importance of these variables when discussing variations between regions of different seasonality. Again, F at each trap location was predicted by re-arranging Equation (20b). Fpoc = EP[1.1976103F+ 2.9703103F+ 2.1761103F+ 6.2472i01z - 3.7008101f-ratio + 6.3772109 (21) Regression between F0 modeled with Equation (21) and F0 measured was considerable at R2 0.633. 2.5 Summary We found that the applicability of classical EP-based model (i.e. Martin etal., 1987) to the global ocean is very poor. Predictability of F0 improves dramatically by applying a region specific mixed layer depth (zo) and a 15 parameter (b) dependent on F and seasonality. A CaCO3 single-mineral-ballast model could explain up to —79% of the global F0 variability at depth >1500 m. Furthermore, a multiple—mineral-ballast model could explain up to —89% of the global variability in F0 at depth >1500 m. The lower regression coefficients seen from the TE approach is due to the removal of the strong influence of large variations in ER The disadvantage of the TE approach in developing algorithms to predict POC flux is that transfer efficiencies are often low, typically less than 10% [Fig. 2bj, so that small errors on the TE results in large variations in These results suggest that CaCO3may potentially be the mineral type that has dominant control on the vertical transport of F from sea surface to depth in the open ocean. 16 3 Modeling in GENIE-I As part of the GENIE project (www.genie.ac.uk), a new coupled carbon climate model called ‘GENIE-I’ was developed. At its core is a 3D frictional geostrophic ocean circulation model, a 2D atmosphere EMBM (Energy Moisture Balance Model) and a dynamic-thermodynamic sea-ice model . To this a single nutrient (phosphate) based representation of the marine carbon cycle ‘BIOGEM’ (BlOGEochemical Model) is coupled. With the resulting EMIC (Earth system Model of Intermediate Complexity), its computational efficiency enables >2000 yrs integration per hour CPU, which makes it suited for conducting multiple model simulations of short to long timescale variability. See Ridgwell et al. (2007a; 2007b) for a more detailed description of the GENIE-i model. 3.1 GENIE-I In this study, we defined and configured two modes of ocean; a Non-ballasting Ocean (NBC hereafter), where F and F are not coupled but independent of each other; and aCaCO3-ballasting Ocean (CBO hereafter), where the two fluxes are coupled and Fpc is dependent on F. 3.1.1 POC flux in the Non-ballasting Ocean In the NBC, F is determined by the default equations adopted in GENIE-i (Ridgwell et al., 2007a). POC export production is dependent on the available surface nutrient concentrations of phosphate (PQ). First, the biologic uptake of PC4 (F) is calculated as: F = uopoPO4I(PC4 + K04)(1 A)III0 (22) where u0 p04 is the maximum uptake rate of phosphate under the assumption of no limitation phytoplankton growth, and K04 is the half-saturation constant of a Michaelis-menten type kinetic limitation of nutrient uptake. Both u0 PO (1.91 pmol kg-i) and K04 (0.21 pmol kg-i) are calibrated in the model for their appropriate values are unknown in the simplified ecosystem function in our model. Modifier I on productivity is applied to represent the effect of sub-optimal ambient light levels, where the strength of local insolation I is normalized to the solar constant 1 to give a linear limitation term. Another modifier A is applied to represent the effect of fractional sea ice coverage of each grid cell (Edwards and Marsh, 2005). The changes in P04 and dissolved organic phosphorus (DOP) concentrations in the surface ocean layer are governed by equations below. aPo4iat = - F + 2. DOP (23) 8DOP/öt = vF - 2DOP (24) A proportion of PC4 uptake (v) by the biota is partitioned into DOP where the relatively labile organic phosphorus is remineralized with a time constant of 1/2.. The values of v (0.66) and X (0.5 yr1) are taken from the 17 assumptions by Najjar and Orr (1999). The export flux of particulate organic phosphorus (in units of mol PC4 rn-2 yr1) is determined directly from P04 uptake. Fpop(zo) 1O-.zO p (1 - v)J’dz (25) where p is the seawater density and z0 is the thickness of the euphotic zone. Export flux of particulate organic phosphorus (F0) is calculated from PC4 uptake. Then F0 is derived simply from the 106:1 molar ratio between dissolved inorganic carbon (DIC) and PC4 upon production of organic matter (Redfield et al., 1963). FPOC(Zo) = lO6Fpop(Zo) (26) Below the surface layer POC, is partitioned into two distinct fractions; a conservative recalcitrant fraction and a labile fraction. Remineralization of labile POC occurs instantaneously with an exponential decay. POC flux at depth z in the water column is expressed as: FOC(Z) =F0C(oy(rC + (1 — rpoc)exp((zo — z)/I0)) (27) POC is partitioned into a recalcitrant fraction (r) and a labile fraction (1 - r0). The recalcitrant fraction (r) and the e-folding remineralization depth of labile fraction (I) are both parameters that are calibrated in the model, with values of 0.055 and 556 m respectively. 3.1.2 CaCO3flux in the Non-ballasting Ocean and the CaCO3-ballasting Ocean In both NBC and CBO, the export flux of CaCO3 is related to the export flux of POC based on a thermodynamical description of carbonate precipitation rate. F(o) = y rocaco3:pocFpoc(zo) (28) ‘y =(O-1) 0>1.0 (29) , =0.0 fl1.0 (30) where r0 CaCO3:POC is a spatially-uniform scalar, and ‘y is thermodynamically-based local modifier of the carbonate production rate. 0 is the ambient surface saturation state with respect to calcite, defined: O = [Ca2][ 0aj/Ksp (31) where [Ca2] and [CC32]are the concentrations of calcium ion and carbonate ion, respectively, and K6 is the 18 solubility constant (Zeebe and WoIf-Gladrow, 2001). r is the thermodynamic calcification rate power. The higher the value of i, the more responsive calcification is to ambient saturation state. r is an unknown parameter constant calibrated to 1 .28 in the model. As in the remineralization of POC through the ocean water columns, the CaCO3 flux below the surface layer is described as: F(Z) = F(y(r + (1 - rc)exp((zo - z)/I)) (32) CaCO3 is partitioned into a recalcitrant fraction (re) and a labile fraction (1 - re). The recalcitrant fraction (re) and the e-folding remineralization depth of labile fraction (ia) are both parameters that are calibrated in the model, with values of 0.489 and 1055 m respectively. 3.1.3 POC flux in the CaCO3-ballasting Ocean In the CBO, POC flux below the surface layer is dependent on the availability of CaCQ3flux and its carrying capacity as ballast mineral, as in Equation (lOb); F= 0.130F. The recalcitrant fraction (r0) is equated to CaCO3-ballasted POC instead of being prescribed as a set parameter, and is now a function of depth rather than a fixed value for the ballasted fraction changes as a function of depth. F00 (z) F00 (ZO)(rC + (1 — rpoc)exp((zo — z)/I00)) = FpOC(Zo)(0.130FC(Z) + (1 - 0.130FC())exp((zO - z)/I0)) (33) Given the observed relationship between POC flux and CaCO3 flux in terms of the mineral serving as ballast controlling the POC flux, we assume that this relationship also holds in all depths of the water column below the surface layer. 3.1.4 Model experiment setup and permutations In both ocean modes, GENIE-I was first spun up for 10,000 years to reach a stable atmosphere/ocean and considered as the ‘pre-industrial’ state at year 1765. From there, lo investigate the range of potential effects of anthropogenic CO emissions on the ocean and its feedback on atmospheric CO2 concentration, GENIE-i is forced with historical pCO records (ice core and observation records) from year 1765 to 2003 (Enting et al., 1994; Keeling and Whorl, 2005) followed by a prescribed future CO2 emissions scenario thereafter from year 2004 to 3000, where the “business as usual” (lS92a) future emissions scenario is applied based on an use up of conservative estimates of fossil fuel resources (coal, oil and gas). In this prescribed scenario, CO2 emissions rate peak at 2100 (20 Gt Cyr1)followed by a linear decline to zero by the end of year 2331 with a total emission of circa 4000 GtC (Lenton, 2000; Lenton et al., 2006) [Fig. 3a, 3b]. Sediment burial of sinking POC and CaCO3 particulates are not considered in this study as these processes control the climate-ocean interaction only on time scales> 1000 years (see Appendix D). In this version of GENIE-i, all of the sinking particles are 19 remineralized within the water column or at the ocean floor. In each ocean modes, model integrations of four permutations with different feedback operating are compared. ‘no feedback’ - Both climate (including ocean circulation) and CaCC3:PO export ratio are fixed to constant. Neither carbonate production nor climatology are responsive to changing chemistry in the ocean (solid line). ‘+climate’ - Only climate is responsive to increasing atmospheric CO2 while CaCC3:PO export is fixed constant. Only the climate responds to the change in atmosphericpCO2while the spatial field of CaCC3export production is fixed constant to pre-industrial (year 1765) state; the rate of calcification in the surface ocean does not respond to CO2 invasion (dotted line). ‘+calcification’ - Climate does not respond and only CaCO3:POC export is responsive to increasing atmospheric CC2. This experiment extracts theCC2-Calcification feedback by forbidding climate to respond to change in atmospheric pCC2while CaCO3export flux varies according to the surface saturation state with respect to calcite (Q) (dashed line). ‘+climate +calcification’ - Both climate and CaCO3:POC export are responsive to increasing atmospheric CO2. This experiment represents a more real ocean and is the overall result of both climate and calcification feedback operating (dot-dash line). 3.1.5 NBC vs CBO It is important to note that the NBC and the CBC are in slightly different equilibrium state at year 1765 for they were spun up with different equations governing the flux of POC. The recalibration of GENIE-I inCaCO3 ballasting mode was outside the scope of this study for we did not have the computing resource and expertise at ths time. As the parameter calibration has been carried out only for the NBC, the initial states (1765) between the NBC and CBO are slightly different. Therefore, the differences we find between the NBC and CBO may be either actually due to ballasting mechanism or due to the differences in the initial states of the two ocean modes. Ideally, we would have equivalently calibrated models of NBC and CBC (and MBC) that are equally good at reproducing the observed biogeochemical cycles. That said, no studies to our knowledge up to date have compared future CO2 trajectories between equivalently calibrated NBC and CBC models. We highlight that even though our two ocean modes had slightly different starting points, where each model were spun up to a equilibrium state (1765) with different mechanisms governing the flux of PCC, still, the initial global CaCC3and POC export productions at pre-industrial equilibrium state (1765) do not differ much between the two ocean modes [Table 4, Fig. 3k-3n]. Furthermore, the spacial variations were also found very small when we compared sea surface CaCC3and PCC export production at 1765 between the two ocean modes, with no considerable difference [Fig. 5a, 5b]. The atmospheric concentrations reached at 2300 between two ocean modes (both with ‘no feedback’,) are practically equal at 1301 ppm and 1300 ppm, respectively [Table 4]. 20 Hence, in this study, we considered it safe and reasonable to say that the two ocean modes are directly comparable. i 3.2 Non-ballasting Ocean The results from NBO model runs are shown to first elucidate the carbonate system feedback on increasing atmospheric CO2 under our more conventional perspective of the ocean; no ballasting mechanism present. We display the experiment results from the four different permutations described above. The trajectory of each major variable is monitored as a time-series and their numbers at major time-slices (1765, 2000, 2100, 2300) are compiled in [Table 4]. We chose these four time-slices as representing pre-industrial state (1765), modern state (2000), time of maximum CO2 emissions rate (2100) and time when atmospheric pCO2 reaches maximum from the prescribed CO2 forcing (2300) [Fig. 3c]. 3.2.1 NBO ‘no feedback’ In the baseline ‘no feedback’ run, neither climate feedback nor calcification feedback are in operation. Both POC export production and CaCO3export production stays constant over time, maintaining their pre-industrial (1765) value throughout the entire period of CO2 forcing. Global atmospheric pCO2 reaches 1301 ppm by 2300, which is more than four times the pre-industrial value [Table 4, Fig. 3c]. The increase in atmosphericpCO2here is purely due to the decrease in seawater buffer capacity as a consequent of change in ocean carbonate chemistry. 3.2.2 NBO ‘i-climate’ Allowing only the climate in GENIE-i to respond to CO2 forcing (‘-i-climate run) adds 116 ppm to the baseline value. This can be attributed to C02-climate positive feedback from restricted vertical transportation of nutrient supply (PC4) to the surface ocean, thus reduced POC export to the ocean interior [Fig. 3m, 3o; Note that the dotted line of ‘+climate’ feedback in PC4 is being overlaid by the dot-dash line of ‘+climate +calcification’]. Global POC export decreases by 1.2 Gt C over the course from 1765-2300. 3.2.3 NBO ‘+calcification’ In the ‘+calcification’ run where climate feedback are absent and only calcification (and CaCO2:POC export ratio) responds to increasing oceanic CO2 uptake,C02-calcification negative feedback (as described in Section 1.1) kick in and depress atmospheric pCO2 by 38 ppm, partially counter-acting the otherwise 116 ppm enhancement by climate feedback alone [Table 4, Fig. 3e]. Compared to the present day (2000) values, oceanic carbon uptake doubles by year 2100 (when the carbon emissions rate reaches its peak) [Table 4] and is followed by a decrease in line with the linear decrease in global CO2 emissions rates [Fig. 3b]. At 2100 the climate feedback suppress global oceanic carbon uptake by 0.8 Gt C yr1, while the calcification feedback counteract by 21 enhancing carbon uptake by 0.4 Gt C yr1. Although not strong enough to fully reverse the direction of CQ invasion into the ocean, the global reduction in calcification cancels out -‘50% of the increase in oceanic uptake by climate response at peak CO2 emission rates (2100). 3.2.4 N BC ‘+climate +calcification’ The overall combined effect of the two counter-acting feedback +climate’ (+116 ppm) and ‘+calcification’ (-38 ppm) is an increase of 77 ppm atmospheric pCO2 by year 2300 (Table 4, Fig. 3e); a net positive feedback on rising atmospheric pCO2. 3.3 CaCO-ballasting ocean The effects of F0 ‘ballasting’ by F are explored in a CBC with the exact same four permutations as in the NBC experiments. While the general behavior of carbonate chemistry in the CBC is the same as in the NBC, there are some noticeable differences seen in the behaviour of F0 and PC4 concentrations between the CBC and the NBC, largely due to the coupled ‘ballasting relationship between F0 and F. 3.3.1 CBC ‘+climate’ With climate feedback only (‘+climate’), PCC flux decreases over time in most part of the global ocean as in the NBC, except the penetration of PCC in the CBC is much deeper in the equatorial water column than in the NBC [Fig. 4b]. 3.3.2 CBC ‘+calcification’ We next look at the calcification feedback only (‘+calcification) run in the CBC. Due to the dependence of Fpc on F in the ballasting model, one may simply expect to see reduced PCC export production asCaCC3export is depressed from the calcification response to ‘ocean acidification’. The picture is a little more complicated than that. As in the NBC, the progressive decrease of sea surface calcite saturation state due to invading acidity leads to a large decrease in global CaCC3 export production over time [Fig. 3k, 31] and thus one might expect PCC export production to decrease in line in the FpOD-FC coupled model. However, the PCC export production time-series shows a global increase over time [Fig. 3n]. 3.3.3 CBO ‘+climate +calcification’ The end results of both climate feedback and calcification feedback operating in the CBC are shown in the global time-series [Fig. 3f] and sea surface maps [Fig. 5d-5h]. At the global scale, we find that while the net effect of ‘+chmate +calcification’ with respect to ‘no feedback’ in the CBC by 2300 is an atmosphericpCC2increase of 96 ppm (7.4%) [Fig. 3f]., whereas the pCC2 increase with respect to ‘no feedback’ is lower in the NBC at +77 22 ppm (5.9%) [Fig. 3e]. Although the increase in global POC export production fuelled by shallower PQ recycling due to calcification feedback (+calcification) counteract against the diminishing effects by climate feedback (+climate), strong climate feedback (e.g. increased stratification) ultimately dominates over the calcification effects. 3.4 Multiple-ballasting Ocean In addition to the CaCO3-ballasting Ocean, we have also experimentally configured a Multiple-ballasting Ocean (MBO hereafter). In the MBO, Fy00 is dependent on F0, F0 and F1 as in Equation (12); F00 = k0F + k0F + krFi. Since there is no physical transport process which corresponds to a negative flux of carbon, the small offset negative constant term (not shown) is removed for implementing into GENIE-i, obtaining regression coefficients of 8.4648.i02, 6.4016i0 and 7.9564.102 for F0, F0 and F1, respectively. The resulting R2 between F00 measured and F00 modelled remained exactly the same as before at 0.867. As in the CBO, the recalcitrant fraction (r00) is equated to multiple-mineral-ballasted POC instead of being prescribed as a set parameter, and is now a function of depth rather than a fixed value for the ballasted fraction changes as function of depth. FOC(Z) = F00(Zo) . (r00 + (1 — r0) . exp((zo — z)/I00)) = F00() ((8.4648i02FC( + 6.40i6i0FO(Z) + 7.9564i02F(Z)) + (1 — (8.464810Fc(z) + 6.40i6i03F(Z)+ 7.956410FI(Z))) . exp((zo - z)/I00)) The exact same four simulation permutations as in the CBO were run. We found very liltIe differences between the end results obtained from the CBO experiments and the MBO experiments, where the global POC and CaCO3 export productions between the two ocean modes did not show a considerable difference [Table 4]. 3.5 Summary With the ‘ballast’ effect in operation, the net effect of climate feedback and calcification feedback is a global decrease in POC export production, except for in some high latitude regions where decreased sea-ice coverage fuels the production of POC. Simulation in GENIE-i showed that calcification response in the CBO is —63% weaker compared to that in the non-ballasting ocean. With both climate and calcification feedback operating, the CBO resulted in a slightly larger net increase of atmospheric pCO2 over time compared to the NBO. The difference in the net increase of atmospheric pCO2 from 1765 to 2300 between the two ocean modes was 12 ppm. 23 4 Discussion 4.1 GENIE-I simulation results 4.1.1 NBO ‘+climate’ A better visualization can be obtained from zonal average cross-sectional plots. The maximum of PC4 concentrations deepen and surface concentrations deplete over time, leading to a net transfer of P04 from the surface ocean to the deeper ocean [Fig. 4c]. The depth of POC flux shallows in line and is most prominent in the equatorial surface waters where strong stratification dampens nutrient supply [Fig. 4a]. It is also interesting to note that an increase in POC flux is seen in the Northern high latitudes just above > 60N [Fig. 4a]. We attribute this to the progress decrease of sea-ice coverage over time in the high latitudes as seen in [Fig. 5i]. A decrease in sea-ice coverage, hence increased sea surface area for biological production to occur, would lead to increased flux of POC and POP. We find that the decrease in the extent of sea-ice coverage is a climate- sensitive factor that dominates over the effect of increased stratification. 4. 1. 2 NBC ‘+calcification’ The magnitude of marine carbonate chemistry change is greatest at around 2300 for this is when atmospheric pCO2 reaches maximum. The increase in carbon uptake by the ocean leads to a notable increase in H ions, and decrease in C0 ions, in the global sea surface [Table 4, Fig. 3g]. The consequence is aO.5 units sea surface pH drop, from 8.1 (1765-2000) to 7.6 by 2300 [Table 4], pushing the surface ocean towards more acidic’, hence commonly referred to as ocean acidification’. Predictions of possible pH shifts by a more complex OGCM (Ocean General Circulation Model) fall within comparable range where a maximum reduction of 0.77 pH units is reported (Caldeira and Wickett, 2003). Perhaps one of the most important trend that has to be highlighted here is, due to the drop in surface concentrations of C032,calcite (and aragonite) saturation state (0) in the surface ocean drops by more than 60% by 2300 compared to the 1765 pre-industrial state [Table 4, Fig. 3i]. Omega progressively approaches closer to 1, which is the thermodynamical threshold for CaCO3 precipitation to occur. A number of studies have suggested that reduced pH and decreased degrees of C) could substantially lower the precipitation rates of calcareous plankton even if C)> 1 (still supersaturated with respect to CaCO3)[Bijma et al., 1999; Riebesell et al., 2000; Zondervan et al., 2001]. In line with the decrease in biogenic carbonate precipitation, in the permutations with calcification response operating, global CaCO3 export production from 1765 to 2300 decreases from 1.3 to 0.3 GtC yr-i; a —77% reduction [Table 4, Fig. 3k]. This number is considerably larger than the —50% reduction of CaCO3 export predicted in a similar experiment (CO2 forcing under no climate feedback) by Heinze (2004) using a more complex general circulation model. Spatial variability should also be highlighted. The evolution of global distributions of 0 and CaCO3 export per surface area from pre-industrial state (1765) through 2300 are shown in Figure Sc. Sea surface saturation state 24 of calcite are not uniform but heterogeneously distributed over different regions of the ocean. Although the drop in 0 is largest in the equatorial regions, the high latitudes are most subject to progression towards 0 < 1 (undersaturated with respect to CaCO3)for degree of calcite saturation in high latitudes are much lower by default. By 2300, 0 in the oceans at high latitudes >600 drops below the threshold value 1, turning the surface ocean into undersaturation with respect to calcite. The same trend is also seen in other 3D model studies (Caldeira and Wickett, 2005). 4.1.3 CBO ‘+climate’ The deeper penetration of POC seen in the equatorial water compared to the NBO may be due to the ‘ballasting’ effect in the CBO. The equatorial waters may be a high POC transfer efficiency region, derived by the ballasting by dominating CaCO3minerals (Francois et al., 2002). A noticeable increase in F0 is seen in the high latitudes just above >60° N as in the NBO [Fig. 4bj. We also found PO concentrations in the CBO behave in the same way as in the NBO (Fig. 4c, 4d]. 4.1.4 CBO ‘i-calcification’ The increase in POC export in the CBO ‘i-calcification’ run is explainable. In the CBO, the penetration of carbon and thus nutrients into the ocean interior depends on the CaCQ3flux. Hence, due to the reduction in CaCO3flux, POC penetrates less deeply in the ocean and nutrients are released back into the water column at shallower depths. Surface ocean nutrient supply to plankton is increased at the surface ocean, fuelling POC export production [Fig. 3p]. However, increased POC export at the surface ocean does not necessarily lead to POC penetration into the deep water columns for remineralization depth shallows. This is better illustrated in zonal sections of F0 and PC4 [Fig. 4e, 4f]. POC penetrates less deeply into the ocean while showing an increase in flux per surface area only in the surface waters. PC4 recycling maximum move to shallower depths and the concentrations at depth <1000 m increase up to 1.5 106 mol kg1.A vertical re-partitioning of nutrients occur due to shallower remineralization of coupled F0 and F in a more acidic and ballasted ocean. 4.1.5 CBO ‘i-climate +calcification’ Spatial variations of F, F0 and P04 concentrations are presented in sea surface maps [Fig. 5e-5h]. Both CaCO3and POC export productions per surface area diminish through almost the entire ocean and experience the largest magnitude of drop in the equatorial regions [Fig. 5f, 5g] where CaCO3fluxes are dominant by default (pre-industrial state). We have also noticed a net increase in POC export production in two distinctive northern high latitude regions >60 N [Fig. 5gj. In these two areas, sea surface P04 concentrations deplete over time and POP export increases [Fig. 5h, 5g]. These regions are identified as opal-dominating regions with relatively high f-ratios [Fig. le]. Calcite saturation state and CaCO3export productions in these high latitude regions were low to begin with [Fig. 5d, Se] at 1765 and hence the effect of reduced calcification, hence 25 weakening of ballasting, due to carbonate chemistry change is minimal. Furthermore, as discussed above, POC production driven from the progressive decrease in sea-ice coverage appears to override the calcification feedback. 4.1.6 Multiple-ballasting Ocean With both climate and calcification feedback in operation, the atmosphericpCO2at 2300 between the two ocean modes differed by only 4 ppm. We thus do not consider opal-ballasting and lithogenic mineral-ballasting of POC flux important, at least at the global scale, and continue to speculate that CaCQ3-ballasting is indeed the strongest candidate as being the major ‘ballast’ mineral for POC flux in the open ocean. 4.1.7 Non-ballasting Ocean vs CaCO3-ballasting Ocean Several previous studies have evaluated the feedback strength of ballast effect on atmospheric C concentration in model simulations with no climate feedback considered (i.e. equivalent to ‘-i-calcification’ in our study). Barker et al. (2003), by incorporating the CaCO3-ballasting equations of Armstrong et al. (2002) into an eight-box ocean-atmospheric model, reported a complete reverse of the negative feedback of reduced calcification on atmospheric CC2 by the positive feedback of ballast effect. This was by assuming a 20% increase in the remineralizations rates of F00 and F0 (i.e. shortening of remineralization depth) as consequence of predicted reduction of CaCO3 saturation state of the surface ocean. Heinze (2004) have also reported that the negative feedback effect on increasing atmospheric pCO2 due to the weakening of CaCO3export could be fully compensated if CaCO3-ballasting mechanism is introduced. It has to be noted that, in the studies, the ballast mechanism was introduced to an model ocean that previously did not have F00 coupled with F0. In other words, the model ocean was neither calibrated nor spun up to an equilibrium state with a ballast mechanism in operation. When we compare our simulation results from GENIE-i NBC and CBO, we see a similar result as the studies mentioned above. In each ocean mode, ‘+climate’ alone increases atmosphericpCO2by 116 ppm (NBC) and 115 ppm (CBO) with no noticeable difference. With CaCO3-ballasting mechanism operating (‘CBO +calcification’), the reduction in atmospheric pCO2 by calcification feedback alone with no ballasting (i.e. NBC ‘+calcification’) are completely countered; 1263 ppm (NBC) vs 1286 ppm (CBC) [Table 4]. We highlight on the difference seen in the strength of calcification feedback between the two ocean modes. In the ‘+calcification’ runs, in the NBC, calcification feedback alone enhance ocean CO2 uptake by 0.2 Gt C and reduces atmospheric pCC2 by 38 ppm (2.9%) by 2300, whereas in the CBC, enhancement of ocean CQ uptake by calcification feedback is largely eliminated, and the decrease of atmosphericpCC2is by only 14 ppm (1.1%), demonstrating much weaker negative feedback on increasing atmosphericpCC [Table 4; Fig. 3e, Fig. 3f]. By comparison, the negative calcification feedback on atmospheric pCC2 is —63% weaker in the CBC than in the NBC. The atmospheric pCO2 trajectory predicted by the model between NBC and CBC with both responses ‘on’ (i.e. both climate and calcification feedback fully operating), reaches 1378 ppm at 2300 in the NBC, and 1396 ppm in the 26 CBO (both ocean modes starts from a pre-industrial atmospheric pCO2 of 278 ppm) [Table 4]. The strength of ballasting’ depends on the change in magnitude of CaCQ3 export production, in other words, the sensitivity of calcifying organisms to ocean acidification. One important factor which was not considered in this study is that the response to change in carbonate chemistry is very different between different calcifying species. The response of calcifying organisms to lowered saturation states differ from species to species and therefore certain species will be affected more seriously than others (Bijma et al., 1999; Riebesell et al., 2000; Zondervan et al., 2001). Thus, simply taking the sensitivity of one single species of calcifying plankton to change in saturation state (the power parameter n) as representative for all marine biogenic calcification is not the best assumption for global scale modelling experiments (Ridgwell et al., 2007). 4.2 Shallower CaCO3dissolution? A reduction in carbonate production and deposition in the shallow water by corals as predicted by Kleypas et al. (1999) implies an increase to our estimates for the strength of theC02-calcification feedback. However, some modeling studies on solution of shallow-water carbonates have concluded that although metastable carbonates could dissolve under increased invasion of atmospheric C02, the shallow-water ocean will not accumulate enough alkalinity to serve as buffer against change in pH and carbonate saturation state (Andersson et al., 2003; Andersson and Mackenzie, 2004). Shallow water calcification is not considered to affect POC export and remineralization and our open ocean model in this study excludes the effect on atmospheric CO2 from a reduction in shallow water calcification. The conventional view of a relatively conservative PlC flux compared to that of POC is often supported by observations that CaCO3 shells collected in sediment traps have almost always found to be intact with no trace of partial corrosion, showing no visual evidence of significant CaCO3dissolution with depth in the water column above the lysocline (Honjo, 1976; Tsunogai and Noriki, 1991). It has also been observed that CaCO2 in the form of aragonite appears to substantially dissolve in the upper water column (Betzer et al., 1984). Dissolution rates measured by sediment traps in the deep water are generally low, ranging from 0.003 to 0.O3pmol kg1 yr1 (Feely et al., 2004). Since sediment traps measure dissolution in the water column only and can not account for sedimentary dissolution, one should also not overlook that considerable CaCO2 dissolution may also occur in the pore waters of sediments and not on the seafloor (Elderfield, 2002). Further more, results from recent studies of ALK distribution and CaCO3 budgets in the ocean started to indicate considerable dissolution of CaCO3 in the water column well above the chemical calcite saturation horizon. The discrepancies found between alkalinity budget derived CaCO3export production in the surface ocean and sediment trap based measure of CaCO3flux to the sediments suggests that as much as up to —60- 80% of the CaCO3produced in the surface ocean may be in fact lost in the upper 1000 m of the water column, presumably owing to biologically mediated dissolution such as in acidic digestion within zooplankton guts and feces or involving microbial oxidation of organic matter (Milliman et al., 1999; Jansen and WoIf-Gladrow, 2001). The existence of such microenvironments may be one critical control on the degree of dissolution that takes place in the water column. The remaining CaCO3 that survives to further depth would experience further 27 dissolution in the deep water column and at the sediment- water interface, thus it may be that only a small fraction than previously thought of CaCO3produced in the euphotic zone is sequestered in the sediments. While such calculation could be an artefact resulting from the different methods of determination between surface and deep fluxes, a more uniform measurement method using floating sediment traps also demonstrates a tight exponential decay curve in which 50-60% of the CaCO3 export flux is lost in the upper 1000 m of the water column (Martin et al., 1993). Accumulation rates of CaCO3from deep sea cores indicate a global accumulation rate of —0.1 Pg C yr1 (Catubig et al., 1998), which implies that —90% of the surface production dissolves before reaching the sediments or at the sediment-water interface. More recently, Feely et al. (2004) estimated the distribution of ALK in the water column by using a water-mass tracer method and concluded that up to 65% (—0.5 Pg C yr1) of total CaCO3export production dissolves back in the water column before reaching the seafloor, and that up to 60% of the dissolution occurs in the upper water column (<2000 m). While it is convenient to explain such dissolution flux to shallower dissolution of the more pressure susceptible aragonite polymorph, it may be difficult to attribute such huge amount of dissolution solely to aragonite minerals since pteropods probably account for only —10% of the total planktonic carbonate production (Fabry, 1990). A compilation of global sediment trap data through 2000 m depth indicate a global average flux of —0.4 Pg C yr1, also suggesting that at least 50% of the CaCO3export is dissolved in the upper water column (lglesias-Rodriguez et al., 2002). Although these individual studies each contain some degree of ambiguity, if not flaw, such corroborating data serves as collective evidence for shallow water dissolution of CaCO3above the thermodynamically predicted lysocline. 4.3 Or is it the other way around? Correlations on its own do not identify cause and effect. The “ballast” and “protection” mechanisms discussed above are still poorly constrained and therefore must be carefully evaluated before accepted as true. It has been claimed that the slow sinking velocity (—0.1 m d1) of small fragments of CaCO3particles, like individual POC particles, is not sufficient to settle significantly except in forms of aggregates or fecal pellets (Steinmetz, 1994), and that large aggregates of POC (e.g. diatoms) which sink fast (50-200 m d1) can, on its way down the water columns, collect large amounts of inorganic minerals including CaCO3,opal and lithogenic dusts that are each too small to sink on their own (Hamm, 2002). On the contrary to the idea that ballast minerals determine POC fluxes, some studies suggest that it is the other way around; it is the sinking POC particles that scavenge inorganic mineral particles to its carrying capacity as they travel through the water column (Passow et al., 1994; Passow, 2004; Passow & De La Rocha, 2006). Aggregation of sinking particles is shown to be a function of extracellular polymers to some degree, which supports the idea that POC may be in fact determining the flux rates of both itself and other inorganic minerals. The apparent importance of biologically mediated CaCOj dissolution also leads to speculate that the flux of ballast minerals are not independent variables but depend on the flux of POC. 28 5 Conclusions We have found that the flux of CaCO3mineral could potentially be the dominant controlling factor on the transport of POC from the surface ocean to the deep ocean. Statistical analyses suggest that up to —79% of the global variability seen in below 1500 m depth could be dependent on F. The differences between the initial states of our two ocean modes are less important than the anthropogenic impact seen from the model simulations. Cur model simulations have shown that, if the “ballast hypothesis” is valid, the larger net increase of atmospheric pCO2 in the CBO compared to the NBC over the prescribed CO2 forcing period could be attributed to the net positive feedback loop between atmospheric pCO2, F and With the presence of ballast mechanism for POC transport in the open ocean, calcification feedback alone in the CBO is —63% weaker than in a conventional NBC. Moreover, the positive feedback of a potential ballasting mechanism on atmospheric pCO2 could overturn the negative feedback of calcification response on atmosphericpCC2.In an acidifying ocean, on top of the negative feedback of marine calcifying organisms on atmosphericpCO,the ballast hypothesis adds additional complication to the direction of the net feedback. Regional applicability of the ballast equations is not investigated in this study. A deeper understanding of the inter-dependence between PCC and PlC at both global and regional scale should be sought and further investigated. 29 Table 1. Data LG6C Proc cc L It d L 5Ld Traptpr MLD(z SOT PP OP (rate F F TO F IF F F/F F/F F/F FJF -________ rn m C 9royr1 OP/PP Jjoryrigmyri FjEP grn4y i gm4yrr North Atlantic Dnitt 49.1 -13.4 3220 201 13.9 292.7 56.3 019 41.3 3.0 0.05 2.0 1.49 35.7 SAT 0.21 0.32 2.18 NorthAtleshoDritt 49.0 -13.9 4000 201 13.9 292.7 56.3 0.19 40.0 2.6 0.04 2.7 0.92 30.4 0.63 0.06 0.31 11.39 NorthAfianscDrifi 47.8 -19.5 3100 201 13.9 292.7 56.3 0.16 22.3 1.9 0.03 IA 1.36 18.7 0.62 0.22 0.16 2.65 North Atanbc Subtropical Gyro 33.0 -220 4150 65 21.2 121.2 10_i 0.16 12.5 CA 0.02 1.1 0.36 11.3 0.76 0.06 0.13 9.77 North Atlantic Subtropical Gyro 31.6 -64.2 1500 Ti 23.2 117.9 19.2 019 hA 09 0.05 0.8 iii 9.8 0.67 0.19 014 3.56 North Atlanbc Subtropical Gyro 31.8 -64.2 3200 71 23.2 117.9 19.2 019 12.8 8.6 0.03 0.9 0.70 11.6 0.64 0.18 0.19 3.63 North Atlantic Subtropical Gyro 29.1 -15.4 3075 59 20.6 196.9 36.5 0_is 16.1 0.8 0.02 1.0 0.8/ is_S o.oi 0_OS 0A4 10.73 North Atlantic Subtropical Gyro 24.6 -22.8 3870 64 22.5 126.3 25.2 0.16 15.1 0.7 0.04 0.9 0.82 13.7 0.54 0.08 0.39 7.18 Pastors Tropical AUanUc 1.8 -iii 3921 22 27.0 146.9 24.1 0.16 34.2 2.0 0.08 2.2 0.90 30.5 0.61 0.16 0.23 3.63 SouthAtlanscTropioal Gyro -20.1 92 1648 52 20.2 192.7 37.6 0.20 25.6 2.9 0.08 1.7 1.70 20.1 0.70 0.16 0.14 4.23 NorthwastArabian Uywalling 17.4 58.8 3141 20 26.3 396.3 64.0 0.21 79A 4.8 0.06 4.7 1.03 ISA 0.50 0.26 0.19 2/5 NorthwestArabian Upwelling 172 59.6 1657 28 27.3 240.3 39.0 0.16 63.0 6.0 0.15 4.7 1.26 71.8 0.55 0.25 0.21 2.22 NorthwootArabias Upwaltiog 17.2 59.6 2871 28 27.3 240.3 39.0 0.16 81.2 4.9 0.12 4.7 1.05 72.1 0.54 0.23 0.23 2.34 NorthwostArabiao Uywalling 16.0 60.0 3020 34 27.7 172.4 28.0 0.16 510 2.6 5.09 3A 0.77 466 0.60 0.25 0.15 2A0 NorthwastArabias Uywellisg 16.0 60.0 3020 34 37.y i72A 28.0 0.16 43.4 2.5 0.09 3.0 0.83 38.7 0.65 0.28 0.07 2.36 NorthwostArabiao Uywalling 16.0 60.0 3020 34 27.7 172.4 26.0 0.16 03.2 3.3 0.12 3.9 0.84 47.0 0.69 0.21 0.10 3.33 NorthwastArabian lipmalling 16.0 60.0 3020 34 27.7 1724 260 0.16 75.6 4.5 0.16 4.3 1.05 67.2 0.53 0.36 0.1/ I AT loden Ocean Monooon Gyros 10.0 68.0 2800 34 27.7 172A 28.0 0.16 41.5 2.6 0.10 2.5 1.11 36.3 0.58 0/5 0.27 3.82 Indian Ooean Monsoon Gyros 15.0 68.0 2800 34 27.7 172.4 28.0 0/6 22A 1A 0.05 1.3 1.05 16.8 0.06 0.17 027 325 Indian Ooaan Monsoon Gyros 15.0 66.0 2800 34 27.7 172A 28.0 0.16 40.3 2.2 0.08 2A 0.92 362 0.55 0.22 0.24 2.54 Indian Ocean Monsoon Gyros 15.0 68.0 2800 34 27.7 172A 28.0 0.16 29.3 1.9 0.07 2.0 0.96 207 0.64 0.22 0.15 2.93 Indon Ocean Monsoon Gyrao 14.0 64.0 2900 34 27.7 i72A 28.0 0.16 34.3 1.9 0.07 2.7 0.70 30.7 0.74 0.13 0.14 5.78 Indian Ocean Monsoon Gyros 14.0 64.0 2900 34 27.7 172.4 28.0 9/6 18.4 1.1 0.04 1A 0.76 16.3 0.73 0.12 0.15 6.33 Indian Ocoon Monsoon Gyros 14.0 64.0 2900 34 27.7 172A 29.0 0.16 41.4 2.6 0.09 2.9 0.90 36.9 0.66 0.14 0.20 4.82 Indian Ocean Monsoon Gyros 10.0 65.0 2363 36 26.2 i32A 21.7 0.16 24A 1A 0.06 1.8 016 21.8 0.70 0.14 0.17 5.05 Indian Ocean Monsoon Gyros 10.0 65.0 3915 36 28.2 i32A 21.7 0.16 21.3 12 0.06 1.0 019 16.0 0.67 Old 0./S d.78 Indian Ooaan Monsoon Gyros 15.2 892 1717 U 28A i27A 22.0 0.18 31.7 2.0 0.09 1.8 1.39 27.9 0.44 0.25 0.3/ 1.76 Indian Ocean Monsoon Gyras 152 892 1717 U 28.4 i27A 22.0 0.18 32.2 2.2 0.10 1.3 1.73 26.1 0.37 028 0.35 133 Indian Ocean Monsoon Gyros 132 8dA 2282 U 28A i27A 22.5 5.18 43.2 2.5 0.11 2.0 129 38.5 0A2 021 0.37 2.00 Indian Ocean Monsoon Gyros 132 844 2282 U 284 127A 22.0 0.18 49.8 29 0.13 1.9 1.53 443 0.36 0.24 040 1.52 Indian Ocean Monsoon Gyros 132 8dA 2282 U 29A i27A 22.5 0.18 63.5 3.1 0.14 1.7 1.82 57.6 0.25 0.24 0.5/ 1.05 Indian Ocean Monsoon Gyroo 4.5 87.3 3006 25 26.6 103.1 16.8 0.16 38A 2.1 012 22 0.94 34.5 0.04 021 0.25 2.08 Indian Ocean Moosoon Gyros 4.0 87.3 3006 25 28.6 103.1 16.8 0.16 31.0 1.9 0.U 1.0 1.02 27.5 0.50 031 0.14 116 Pacific Subarctic 50.0 -190.0 3260 64 0.7 202.3 96.7 CAB 42.0 1.0 0.01 1.4 0.71 40.7 0.28 0.69 0.03 SAl Pauiflo Subarobu 50.0 -105.0 5090 64 5.7 202.3 96.7 0.48 34.6 0.8 0.0/ 0.9 0.65 33.1 0.21 0.73 0.06 0.29 Pacific Subaroflo 44.0 -205.0 2960 60 7.9 231.5 101.7 0.44 57.5 2.2 0.02 1.9 1.22 53.3 0.20 5.63 0.09 0A6 PaoffioSubarotio 44.0 -205.0 4099 60 7.9 231.8 101.7 0.44 43.5 1.6 0.02 1.2 1.30 40.5 0.26 0.61 0.14 042 Paoiflooubarotio 44.0 -205.0 4990 60 7.9 231.9 101.7 0.44 48.6 1.0 0.00 1.1 1.33 45.8 0.20 0.67 0.13 0.30 KuroshioCurnant 40.0 -195.0 2986 72 14.2 204.2 04.5 0.27 35.0 1.3 0.02 1.6 0.81 32.6 SAl 0.54 0.54 016 KuroshioCurront 40.0 -195.0 5025 72 14.2 204.2 54.5 0.27 28.3 1.1 0.02 1.2 0.94 26.3 0.37 0.57 0.06 0.65 Kurouhio Currant 34.2 -216.0 3423 45 20.0 220.7 46.8 0.23 35.5 0.8 0.02 1.9 0A2 34.3 0.46 0.27 0.27 1.74 Kuroshlo Current 34.2 -218.0 5429 45 20.0 2207 49.8 023 26.7 0.6 0.0/ 1.0 0.67 25.5 0.31 0.28 GAO 1.12 Koroshio Current 30.0 -1850 3873 36 238 937 15.3 0.16 17.3 1.0 0.06 I 3 0.73 15.5 0.71 0.10 0.19 7.16 Wast Paofio Warm Pool 120 -2257 4300 37 288 526 9A 018 21 02 0.02 0.2 0.99 2A 0.61 0.20 0.19 3.12 Wast Pacoc Warrrr Pool 5.0 -2212 3130 25 29.1 712 12.0 0.17 4.1 0.3 0.02 0.3 0.79 3.6 0.74 5.19 5.57 3.83 West Pauou Worm Pool 4.1 -2231 4574 25 29.1 71.2 120 0.17 34.6 2.2 0.18 hA 1.56 305 0.39 0A2 0.19 0.92 West PacSo Warm Pool 3.5 -225.0 1590 23 29.3 93.6 13.8 0.17 58.7 3A 025 3.1 ISO 52 3 0.50 0.29 0.22 1.75 Wost Pauifio Worm Pool 3.0 -225.0 3900 23 29.3 83.6 13.8 0.17 56.7 2.9 02/ 2.8 1.02 51.3 0A6 0.29 0.25 1.58 West PauSu Warm Pool 0.0 -184.8 4363 36 28.8 912 15.6 0.17 13.3 0.6 0.04 1.1 0A9 12.3 017 0.19 0.04 4.09 Wastarn Pacific Arohipalago Deep Basin -i 7.8 -2052 2304 35 26.6 62.3 10.5 0.17 3.0 0.2 0.02 0.3 0.77 2.5 0.94 0.07 0.09 1128 Pacific Sobarutic 50.0 -145.0 3900 59 9.1 195A 79.7 041 48.2 2.5 0.03 2.2 1.10 436 0A3 0.48 0.09 0.99 Pacific Suborc6c 50.0 -145.0 3800 59 9.1 195A 79.7 0.41 3SA 0.8 0.01 1.7 GAO 28.9 0.50 043 0.07 1.10 Pacific Saburotic 50.0 -145.0 3800 59 9.1 /95.4 79.7 0.41 44.1 1.1 001 2.7 0.41 42.0 0.54 0.45 0.01 1.19 Puo6oSubarcfic 50.0 -145.0 3800 59 9.1 195.4 79.7 SAl 30.7 0.8 0.01 1.5 057 29.1 SAl 0.57 0.01 013 Paoifiuoubarctic 50.0 -145.0 3800 59 9.1 190A 79.7 0.41 32A 1.1 0.01 1.7 0.65 30.3 0.48 0.47 0.06 1.01 North Paurto Equatorial Counleruurmnt U.S -140.0 1600 25 26.6 93.4 10.2 0.16 15.2 0.9 0.06 1.0 0.85 13.6 0.62 0.34 0.54 1.82 North Pacific Equatorial Coontorcurrant U.S -140.0 3400 25 26.6 93.4 15.2 0.16 13.2 0.6 0.04 0.9 0.68 12.1 060 0.36 034 1.67 North PauBo Equatorial Courtarcurrnrt 9.0 -140.0 2250 23 27.3 049 155 0.16 8.3 0.6 0.04 0.6 0.93 72 0.69 0.26 0.04 2.64 North Pacific Equatorial Countercurrant 9.0 -140.0 2150 23 27.3 94.9 15.5 0.16 5.7 0.6 0.04 0.7 0.93 7.6 0.72 0.23 0.05 3.10 North PauSo Equatorial Countorcurrant 5.0 -139.8 2100 32 27.1 / 07.5 17.9 0.16 27.3 1.6 0.09 2.2 0.76 243 074 0.24 0.02 3.08 North PacificEquatorial Countarcurrant 5.0 -139.5 3800 32 27.1 107.5 17.6 0.16 24.6 IA 0.08 1.9 0.74 21.9 072 0.26 0.02 2.72 North Pacific Equatorial Couotorcurrant 2.0 -140.1 2203 34 26.2 /21.8 16.6 0.10 26.8 1.3 0.06 2.2 0.59 244 074 0.28 -0.02 2.65 North Pacific Equatorial Countercurrent 1.0 -140.5 1895 3d 262 121.8 199 0.16 26.9 1.6 0.08 1.0 0.85 24.0 0.63 0A3 -0.06 1A7 North Pacific Equatorial Couotorcorrent 1.0 -140.0 3495 3d 26.2 /21.8 19 9 0.16 21.8 1.2 0.06 1.5 0.80 19.6 0.63 OAO -0.03 1.58 North Paoifio Equatorial Coustorcurrant 1.0 -140.0 /983 3d 262 /21.8 19.9 0.16 54A 2.5 012 3.6 0.69 49.9 0.60 0.46 -0.06 129 North Pacific Equatortal Countercurrent 1.0 140.0 2905 34 26.2 121.8 10.5 0.16 41.7 1.9 0.10 2.8 0.69 382 0.60 0.46 -0.06 1.3/ Pacoc Equatooal Dioer5ance 0.0 -140.0 2284 31 26.2 /26A 20.6 0.16 35.2 1.7 0.08 2.8 0.60 32.0 013 0.24 0.04 3.10 Pacific Equatonal Dioargnnce 0.0 -140.0 3618 31 26.2 126.4 20.6 0.16 34.8 1.6 0.08 2.8 0.58 31.8 013 0.31 -0.03 2.37 Pao0o Equalorial Divnrganca -2.0 -139.8 3593 31 26.2 /26A 20.6 0.16 31.2 1.3 0.06 2.4 0.55 281 010 0.32 -0.02 2.16 South Pacific Subtropical Gym -5.0 -140.0 2099 45 26.8 104.8 17.4 0.17 22.4 1.0 0.06 1.7 0.58 20.6 0.70 035 0.05 2.75 South PauSo Subtropical Gyra -5.0 -140.0 2209 45 26.8 1048 1TA 0.17 23.6 1.0 0.06 1.6 0.56 21.7 0.68 028 004 2.38 South PaooooubtmpioalGyre -5.0 -140.0 2316 45 26.8 104.8 17A 0.17 23A 1.0 0.06 1.8 0.57 21.5 0.70 027 0.04 2.62 Sooth Pacoc Subtropical Gyra -12.0 -135.0 3584 38 27.6 88.8 14.9 0.17 7.0 0.3 0.02 0.7 0.35 TA 0.84 0.16 0.01 5.35 Antarctic -50.2 5.9 3U0 122 4.6 90.9 21.9 0.24 6A 0.5 0.02 0.3 1.70 0.5 0.46 CAb 0.10 1.02 Antarctic -54.3 -3A 2i9d /18 0.9 59.1 13.8 0.23 285 0.5 0.04 0.2 2A3 27.9 0.06 0.71 0.23 0.09 Anlarofic -54.3 -3.3 2201 /18 0.9 59.1 13.8 0.23 3.1 0.6 0.05 0.1 7.88 1.9 035 028 037 126 Antaroso -54.3 -3.3 2251 119 0.9 59.1 13.8 0.23 7A 0.1 0.01 5.0 6.50 7.1 0.02 019 0.19 0.03 Antarotlo -57.0 -37.0 2000 93 CA 58.6 16.6 0.28 8.0 0.1 0.01 0.1 0.83 y.5 0/3 1.01 -0.14 0.13 AnlaroSo -56.9 -1702 4224 154 2A 97.9 25.0 026 24.6 0.7 0.03 1.4 0.51 23.2 0.51 OAO 0.09 1.29 30 Table 2 Correlatixn matrix F TE PP EP f.,ax SST jxjy gmy - F. 1.000 0.711 0.500 0.139 -0.172 0.325 TE 0.711 3 500 -0 142 -0.389 -0483 0.575 PP 0.500 -0.142 1.000 0.780 0.362 -0.185 OP 0.139 -5.389 0.760 1.000 0.871 -0.567 f.e -0.172 -0483 0.362 0.871 1.000 -0.706 SST 0.325 0.575 -0.185 -- -0.567 -0.766 1.000 31 Table 3. Linear regression equations, regression nostfinients, and against measured dote — — fpg7essian Cseftinient 0 Modal Type Eq# - Enpptinn CaCO3 —— Opal Lithogenin SST f-rabe OP Single-ballast 3a F = 0 096610’ 0 744 3b F= h,F, 1.1201-tO’ 0.739 F,, = h,F, 5.541910’ 0.108 F,,,b,-F, 1.3948.101 0428 4a Fr,, = h,-F,+ b,-r-’ 6.1521 tO’ 2.1079-10.’ 0.787 4b F,,, = k,F,” 6,2’ 1 1180W’ 1.471010+2 0.760 40 F,,, = 6,-F,,” Ic-i’ 64th-b’ 2.4498-10” 0.207 44 F,,, = lçF1” 6,-i’ 1.3795-by’ 1.4391-10” 0.445 16a TE = 6,F,,” 6,z’ I .6691-10’ 1 4534.10+2 0427 160 TO = 6,-F,. h,-z’ 3 2916-10’ t.3t37-1O” 0470 lee TEk-F,+6,-z” 1.1077 10-’ 1.519410” 0.135 104 TE = Ic-F,n h,z’ 42938.10.2 1.2906.10+2 8.340 t7a TE = 6F° ki’° beSOT 2.0943-10’ 1.559910 1.0099.10+2 6566 175 TE = h,F,+ 9,-i’-” 6,,,,,-Lrada 3.007210’ - I 7268-10.’ 8.5165-10” 0.055 MuIt,pln-ballaet 12 F,,, = b,F, + h,-F,+ kF 9.042610’ 9.709510’ 5.0943102 0.597 13a F = SF + 6F+ h,F, + h,SST 8.8753.102 1.170910’ 50447.102 2.847310’ 0.967 135 F,,, = 6,-F, + b,F,+ 0,-F, = lç,,;f’raeo 8.1969-10’ 2.7798.102 7.8132182 -1.0335 0574 14 F,,, = 6,F, + 6,F,+ k,F, + 6,i’ 8.8710102 1.4100-10’ 7.9258-10’ 1.4909-10” 0.696 is F = 6,-F, = 6,-F,. 0,-F, + 6,-i’ = 5n EP 8.9113.10.2 1.1619.10.2 7.9060.10.2 I .1142-10’ I .5624-10” 0.888 19 TO = 6,F, + 6,F,+ 6,-F, + 6,i-’ 2.8372-10° - 5.0911-10’ 2.6612-10’ 1.2079-10.’ 0.550 20a ‘“ 6,F,” 6,F,+ Icr, + 5,i’ 5,,00T 1.5214-10’ 1.0631.10.2 2.3596-10-’ 2.3169-10’ 1.0143.10+2 0.646 205 TO = 6,-F, • 6,-F,. t1-F, + 6,-z”+ heauf-ratm 1.1976-10’ 2.9703-10’ 2.1761-10’ -3.7008-10.’ 6.2472-10” 0.968 32 Table 4. GENIE-I simxlafian resalls lime-series (glabal) 1765 2605 2160 2300 3000 4 399 1937 4140 4193 0 8 20 3 0 278 371 780 1301 981 1417 1080 1263 908 1378 1008 1300 961 1415 1072 1286 949 1396 1050 1392 1044 7.6 7.7 7.5 7.6 7.8 7.7 7.6 77 7.6 7.7 7.5 7.7 7.6 7.7 — —— 7.6 7.7 7.8 7.7 1.71 2.17 I 88 2.29 1.85 239 202 2.54 1.72 2.18 1.89 2.31 1.82 2.31 2.00 247 2.00 248 3.8 04 2.8 0.5 3.8 0.4 3.1 0.5 3.8 0.4 2.9 06 3.6 0.2 6.7 2.9 0.5 8.7 2.9 0.5 1.3 1.3 1.3 1.3 1.2 11 11 1.1 0.6 0.3 0.5 1.1 0.6 0.3 0.5 1.2 1.2 1.2 1.2 1.2 1.2 1.0 1.1 1.1 0.6 0.3 0.5 1.1 06 0.3 05 1.1 — 0.6 0.4 0.5 9.0 9.0 9.0 9.0 5.9 9.5 7.8 9.0 9.0 9.0 90 9.0 8.9 9.5 7.8 8.1 88 88 8.8 88 8.7 8.3 7.5 7.7 88 89 9.2 92 - Ocean Made Experiment cpLbxn Emissians Inxenfx)y (G)) Carban Emisaienx Rate (GI C pr’) Afmxsphekc pCO2(ppm) 980 xx faedbacb +clamate 278 372 778 +calcjfjxafiax 278 370 702 -cclimafe +calcificafixn 278 371 770 C8O xx feedbacb 278 371 760 +xlimafe 278 378 777 +calcifxabxx 278 370 756 +xlimafe -cxalxifxa9xx 278 371 773 M8O +cbmate +xalcificafiax 278 371 773 Sea Sxrface pH NBO nx feedback 8.1 8.1 7.8 xxl,mate 8 1 8.1 7.8 -cxalcihcalixx 8.1 8.1 7.8 -cclimafe +calcificabxx 8.1 8.1 7.8 CBO xx feedbacb 8.1 8 1 7.8 xclimafe 8.1 9.1 7.8 +calcifcabxx 8.1 8 1 7.8 +xlimafe +xalcifxabxx 8.1 8 1 7.8 M8O +climate xcalcifxa5ax 8.1 8.1 7.8 Sea Sxrface0cx N80 xx feedback 5.20 4.45 2.70 +climafe 520 4.51 2.80 xcelcdixafixx 520 449 2.79 +climate +calcificabpn -- 520 4.54 2.93 CBO xx feedback 521 443 2.71 +xlimata 5.21 448 2.86 xcalxipcabxx 5.21 4.49 2.78 xxlimafe +calciflxa5xx 5.21 4.54 2.92 M80 i-climate +calcifxa9xx 5.20 4.53 2.92 Atm-Ocx lCO )Gf C yri) 990 xx feedback 0.0 34 7.3 xclimafe 0.0 32 6.5 4-calxifixafixx 0.0 3.4 7.7 +climafe 4-calcihca9ax 0.0 3.2 6.8 CBO xx feedback 0.0 3.6 7.3 i-climate 0.0 3.5 6.8 -ccalxificefiax 0.0 3.5 7.4 +climata xcalcifca5xx 0.0 3.2 MBO xxlimete xcalxifxesxx 0.0 32 CaCO3Expxo Prxdacbxx (Of C yr’) 980 xx feedback 1.3 1.3 i-climate 1.3 xcalcifcafixx 1.3 - 4-climafe 4-celcoxabxx 1.3 C8O xx feedback 12 xclimafe 12 -ccalxifixatixx 1.2 i-climate 4-calxibcabxx 1.2 — M80 i-climate +xalcificabxx - 14 POC Expxrt Pmdaxhxx (Of C pr) N80 xx feedback 9.0 i-climate 9.0 +calcificafixx 9.0 +xlimate .calc,ficaoax 9.0 C80 xx feedback 8.8 xclimate 8.8 i-cafcificafixx 8.8 i-climafe +cafcifcafixx 8.8 8.7 M80 i-climate +calcifcatxx 9.0 — 8.9 84 8.0 8.3 8.6 8.1 8.5 33 Figure 1. Data Figure la. 79 sediment trap locations showed on the GENIE-i Longitude-Latitude horizontal grid Figure lb. Sea surface PP field on the GENIE-i grid Figure ic. Sea surface SST field on the GENIE-i grid Figure Id. Sea surface MLD field on the GENIE-i grid Figure le. Sea surface f-ratio field on the GENIE-i grid Figure if. Sea surface EP field on the GENIE-i grid 34 a Sediment Trap Locations C -o 4- -I -J Figure 1 a-c Sea Surface Temperature CC) Longitude 40 100 — — — — — — above 300 280 — 300 260 — 280 240 260 220 — 240 200 — 220 180 — 200 160 — 180 140 — 160 120— 140 100 —120 80 — 100 60 — 80 40 — 60 20—40 0 — 20 below 0 bcye — .t,— — — .5— — —2 .5— —2 .5— _ -1 .5.-16 .5—i -1% .5— —10 — .5— —7. — .5— — —4. — .5— -1. elowO • North Atlantic Drift o North Atlantic Subtropical Gyre x Eastern Tropical Atlantic ÷ South Atlantic Tropical Gyre * Northwest Arabian Upwelling Indian Ocean Monsoon Gyres 0 Pacific Subarctic V Kuroshio Current West Pacific Warm Pool <1 Western Pacific Archipelagic Deep Basin > North Pacific Equatonal Counteivurrent * Pacific Equatorial Divergence 0 South Pacific Subtropical Gyre • Antarctic 0) D b 0) -o 4-. 4- Cu -J —260 —200 —140 —80 —20 40 100 Longitude Primary Production (g C m2y1) Longitude —260 —200 —140 —80 —20 35 Figure 1 d-f Longitude —1 elow 0 ov : elow 0 Mixed Layer Depth (m) Longitude f—ratio —6O1 —90 —260 —200 —140 —80 —20 40 100 Export Production (g C m2y1) —260 —200 —140 —80 —20 40 100 Longitude 36 Figure 2. Data analysis results Figure 2a. F00 measured from 79 sediment traps at different depths> 1500 m Figure 2b. TE measured from 79 sediment traps at different depths> 1500 m Figure 2c. Linear regression between F0. measured and F0. modeled with Equation (2a) Figure 2d. Linear regression between b measured and b-modeled with Equation (2c) Figure 2e. Linear regression between F0. measured and F0. modeled with Equation (2e) Figure 2f. and Fm normalized to F, Figure 2g. F0, F0 and F normalized to Fm Figure 2h. Correlation between F0. and Fm, F0, F0, F, Figure 2i. Linear regression between F0. measured and F00 modeled with Equations (3a-3d) Figure 2j. Linear regression between Fpm, measured and F0. modeled with Equations (9a-9d) Figure 2k. Linear regression between F0. measured and F0. modeled with Equations (12k, 13a, 13b) Figure 21. Correlation between TE and Fm, F0, F0, F Figure 2m. Linear regression between TE measured and TE modeled with Equation (16b, 17a, 17b) Figure 2n. Linear regression between TE measured and TE modeled with Equation (19, 20a, 20b) 37 E 0) •0 :3 a) E L Figure 2 a-e —i.e —1.8 —1.6 —1.4 —1.2 —1 —0.8 —0.6 —0.4 —0.2 b modeled Fpoo (g m2y1) 2 3 4 5 6 Transfer Efficiency (Foc/EP) Q05 0.1 0.15 0.2 0.2515000 . A a 2000 ODD 2500 3000 A• * * E3500 :>v 4000 A x A 4500 + 00 * * 0 DO 2000 2500 3000 E E 3500 a, 0 4000 4500 b I ..1 V 3c 0. V * , * * *L * ** Vx 5000 5500 5000 —0.4 —0.6 —0.8 0) (0 a) a) E .0 —1 —1.2 d j,. /vO •0 •:. . —1.4 /. . R2=0.762 —1.6 F modeled = EP•(zzo)° E 0) 0 a) E LL North Atlantic Drifto North Atlantic Subtropical Gyre x Eastern Tropical Atlantic + South Atlantic Tropical Gyre * Northwest Arabian Upwelling Indian Ocean Monsoon Gyres Pacific Subarctic V Kuroshio Current ‘ West Pacific Warm Pool 1 Western Pacific Archipelagic Deep Basin North Pacific Equatorial Countercurrent * Pacific Equatorial Divergence 0 South Pacific Subtropical Gyre • Antarctic 0 1 2 3 4 5 6 7 8 9 10 F modeled = EP.(zIza)°8t 38 2* 3500 a, C x * * x østx ** ++ 2# M- +Hi * *<4* *±++ ++ * *+ ++1 X4 r + + * * ++ 0 10 20 • Fpoc/Ft •t1. + FSFt I>, 2 0) •0 a) 0) to a) 2 ft U- ** *Z x xao< K-” * * >4 K -* K * ** * ** I,., 2 0) I: Figure 2 f-j Flux/Ft ,,0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 2000 2500 3000 Flux/Fm ,.,0 0.1, 0,2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 2000 2500 ‘•x *• * **Kx ** *KKSt *x K * * ‘a + + + ++ + 4+ -r t ++ + 4000 4500 5000 * *3000 2 3500 e C 4000 4500 * 4,-nfl, *X)OO * * *flxxtx K *xx * K K 5000 5500 - 5 4 *. K Fc/Fm I ** * Fo/Fm x Fu’Fm I> 23 0) I h ++ K * + + + + -- ++ +++ +++ + + + + Fm, R = 0.863 ++ 0,R=0.860 F, R = 0.387 x Fi, R = 0.654 30 40 50 S (g m 60 70 3 F modeled North Atlantic Drift o North Atlantic Subtropical Gyre x Eastern Tropical Atlantic + South Atlantic Tropical Gyre * Northwest Arabian Upwelling D Indian Ocean Monsoon Gyres O Pacitic Subarctic V Kuroshio Current West Pacific Warm Pool <1 Western Pacific Archipelagic Deep Basin > North Pacific Equatorial Countercurrent * Pacific Equatorial Divergence O South Pacific Subtropical Gyre • Antarctic 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 F modeled = F..i + [EP — Fp]•e - ZOl/500] F,,l.) = p•Fb(..) 39 -x “0 10 20 30 40 50 60 70 Flux (g m2y1) + Fm,R=0.551 Fc, R0.612 * F,R=0.120 x Fi, R = 0.4970.2 0 0.15 0.1 ci, La I- 0.05 x * + x * + + * + * x + * + + + x * + x + + -I- + XX i-- +:!: + *%% -c++ +* + ++ ++ + fr. _+_i+ + + r*:- + + Fpcc modeled 0 Lu LL 0 La Ca ID E Lu I— TE(Fc,z1),R2 = 0.470 TE(F,z,SST), R2 = 0.566 V TE(Fc,Z,f—ratio), R2 = 0.2 0.25 0.250 0.05 0.1 0.15 TE modeled North Atlantic Drift o North Atlantic Subtropical Gyre x Eastern Tropical Atlantic + South Atlantic Tropical Gyre * Northwest Arabian Upwelling Indian Ocean Monsoon Gyres <) Pacific Subarctic V Kuroshio Current West Pacific Warm Pool <1 Western Pacific Archipelagic Deep Basin North Pacific Equatorial Countercurrent * Pacific Equatorial Divergence r South Pacific Subtropical Gyre • Antarctic Figure 2 k-n 40 Figure 3. GENIE-i simulation results time-series (global) Figure 3a. Global carbon emissions inventory from historical record and prescribed future scenario Figure 3b. Global carbon emissions rate from historical record and prescribed future scenario Figure 3c. Global atmospheric pCO2 trajectory obtained from model simulations (NBC) Figure 3d. Global atmospheric pCC2 trajectory obtained from model simulations (CBO) Figure 3e. Difference in atmospheric pCO2 with respect to ‘no feedback’ (NBC) Figure 3f. Difference in atmospheric pCC2with respect to ‘no feedback’ (CBO) Figure 3g. Global mean sea surface [H], [CC2], [HCC3]and [CC32](NBC) Figure 3h. Global mean sea surface [H], [CC2], [HCC3]and [CC32](CBC) Figure 3i. Global mean sea surface calcite saturation state (NBC) Figure 3j. Global mean sea surface calcite saturation state (CBC) Figure 3k. Global CaCC3export production (NBC) Figure 31. Global CaCO3export production (CBC) Figure 3m. Global POC export production (NBC) Figure 3n. Global PCC export production (CBC) Figure 3o. Global mean sea surface [PC4] (NBC) Figure 3p. Global mean sea surface [PC4] (CBC) 41 4000 3500 r2000 8 w C 0 .0 ‘5 0 2500 1500 1000 500 1800 1900 2000 2100 2200 2300 2400 Year 2500 2600 2700 2800 2900 3000 21 20 19 18 17 16 0 15 14 13 12 11 10 89 Lii 7 6 5 4 3 2 1800 1900 2000 2100 2200 2300 2400 Year 2500 2600 2700 2600 2900 30 NBO 160 150 no feedbacks - . e 140 ÷climate 130 — — i-calcification ..... 120 — i-climate i-calcification 110 100 90 — 80 // 70 1 60 1 . / 50 . :1 40 30 20 10 : —10 - -20 -30 -41) —50 —60 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year Figure 3 a-f 1400 1300 1200 1100 8 1000 900 800 -C a 700 < 600 500 400 300 150 no feedbacks 140 i-climate ...., 130 — — i-calcification 120 — i-climate i-calcification .- — —- — — . — — — - 110 : / — 100 90 : / 80 70 -, 60 -, 50 . -/ 40 30 -, 20 / 10 - —10 - ——-- -20 -30 -40 —50 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year b NBO 1400 1300 1200 1100 E a a 1000 900 800 -C a 700 8 600 500 400 300 CBO C — +climate i-calcification 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year — +climate +calcification 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year CBO > 8 a a 0 > 8 a a 0 42 10_B 1.5 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3( Year NBO Figure3g-I 43 NBO 1 0_2 8 io I a 10 B) 0) • •:• .VVH g ——C02 HCO3 —C03 ------------------ CBO I P _____ h - - H - _.;____._._____._ C02 - V V -—HCO3 V —C03 NBO 1800 1900 2000 2100 2200 2500 2600 2700 2800 2900 30002300 2400 Year CBO - no feedbacks —.-----• V V -i-climate 5 : — — -i-calcification - : — -i-climate i-calcification 4.5 4 - - V 3.5 - 3 - V 2.5 - - 2 V — V I 0 E 0 I io6 I -710 a Co 8 z Cl) a SI(0 1.3 1.2 i)_ 1.1 0 a 0.9 C 0 0.8 2 - 0.7 t 8. 0.6 0.5 0.4 a 8 0 0) a a C,) 1.2 •>. 1.1 0 a 0 C 0.9 C 0 0.8 0.7 0 w 0.6 0.5 0.4 Do 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 Year CBO V V•• : - - - V• V.: : V V V V i-climate ‘ V : — — -i-calcification \ V V — V +climate +calcification - \\ \\ V V V - -k : no feedbacks - V V V V : V +cllmate • V V — — i-calcification V — V i-climate i-calcification \ - - VI ___ — 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year Year 9.3 9.2 9.1 —9 8.9 8.8 o 8.7 8.6 8.5 0 8.4 8.3 0 8.2 8 7.9 7.8 7.7 7.6 7.5 7.4 66 6.7 6.6 6.5 6.4 6.3 6.2 6.1 6 — 5.9 5.8 ‘ 5.7 5.6 5.5 - 5.4 0 a 5.2 5.1 .t 5 c 4.9 a 4.8 4.7 4.6 4.5 4.4 4.3 4.2 4.1 4 3.9 3.8 NBO Figure 3 rn-p 1800 1900 44 NBO no feedbacks rn . . — — +calcification — +climate ÷calcification CBO 9.’ — — no feedbacks 92 — — — — — . . . . +UThite — — — +calcification 9.1 —, — +ciirnate icalcification 58.7 8.6 0 . E83 082 - — — 81 . 8 87.9 7.8 7.7 -. 7.6 .... I 7.5 .1 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year x107 CBO 1800 1900 20002100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year 0 no feedbacks- - +climate — — ÷calcification — ÷climate +calcification N _._-. — 1800 1900 2000 2100 2200 6.7 6.6 6.5 6.4 6.3 6.2 6.1 6 5.9 -r 5.8 .?‘ 5.7 5.6 .g .s -s 5.4 5.3 5.2 5.1 5 4.9 a 4.8 4.7 4.6 4.5 4.4 4.3 4.2 4.1 4 3.9 3.8 no feedbacksp — — . — — i-calcification • — i-climate +calcification - - / / 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year 2300 2400 2500 2600 2700 2800 2900 3000 Year Figure 4. GENIE-i simulation results time-slices (zonal mean) Figure 4a. Zonal mean F0. at 1765, 2000, 2100, 2300 and 2300-1765 difference (NBC ‘+climate’) Figure 4b. Zonal mean F0. at i765, 2000, 2100, 2300 and 2300-i 765 difference (CBO ‘+cliniate’) Figure 4c. Zonal mean [PC4] at 1765, 2000, 2100, 2300 and 2300-1 765 difference (NBC ‘+climate’) Figure 4d. Zonal mean [PC4] at 1765, 2000, 2i00, 2300 and 2300-i 765 difference (CBO ‘+climate’) Figure 4e. Zonal mean F0at 1765, 2000, 2100, 2300 and 2300-i 765 difference (CBO ‘+calcification’) Figure 4f. Zonal mean [PC4] at 1765, 2000, 2100, 2300 and 2300-i 765 difference (CBC ‘+calcification’) 45 - n Co C - I a A id & p th (‘I - n C o C a C . I’ll 8 8 8 C, 8 ± _ 1 1 i d p th — 1 .. e1 ü d d .p th (a a a 0 t a 0 0 0 (0 a a 0 a 0 0) 0 0 (0 a a 0 0) 0 - n C) , , cc C c . C (0 a a 0 C, C 0 0 C a 0 a (0 4 O a th 0 a a C a 0) 0 0) 0 0 a ‘0 CD - 0) C 0 a a 0 C a (0 C 0 0 a a 0 ‘ 0 a 0) C 0 a a ‘0 a 0) 0 AC o C a - - v 8 a 8 8 a 8 , , C o C a CD 8 8 I - — a 4 a a I a a a a 8 a S hi a a S T h t 2 2 [ [ . : l t\ Figure 5. GENIE-i simulation results time-slices (surface layer) Figure 5a. Sea surface CaCO3flux at 1765 (NBC ‘+climate +calcification’), 1765 (CBO ‘+climate -i-calcification) and CBO1765-NB01765 difference Figure 5b. Sea surface POC flux at 1765 (NBC ‘+climate +calcification’), i765 (CBC ‘-i-climate -i-calcification’) and CBO1 765-NBO1 765 difference Figure 5c. Sea surface calcite saturation state at 1765, 2000, 2100, 2300 and 2300-1765 difference (NBC ë-’-climate +calcification) Figure 5d. Sea surface calcite saturation state at 1765, 2000, 2100, 2300 and 2300-1765 difference (CBO ‘+climate +calcification’) Figure 5e. Sea surface CaCO3flux at 1765, 2000, 2100, 2300 and 2300-1765 difference (CBO ‘+climate +calcification) Figure 5f. Sea surface POP flux at 1765, 2000, 2100, 2300 and 2300-1 765 difference (CBO ‘+climate -i-calcification’) Figure 5g. Sea surface POC flux at 1765, 2000, 2100, 2300 and 2300-1 765 difference (CBO +climate +calcification) Figure 5h. Sea surface [PC4] at 1765, 2000, 2100, 2300 and 2300-i 765 difference (CBO ‘-i-climate +calcification’) Figure 5i. Fractional sea-ice extent at 1765, 2000, 2100, 2300 and 2300-i 765 difference (NBC ‘-i-climate’) 49 ‘ 1 1 m 01 ‘ 1 01 8 8 - v 8 Ct 0 8 Ct 8 8 0 + 0 Ct Ct 0 - v 0 Ct C, - Il = 1- fl C, LA - n C o C 1-T I C . B 0’ 0) a 0) 0) 0’ a B 0) 0) B B ‘ 1 C o = - I CD 01,, C o C (3 1 CD 8 8 8 8 8 8 8 8 8 8 CD 8 CD C 8 8 8 + 8 CC 8 0 0 0 8 C, 8 0 C, 8 + C, 0 0 I - , , = a’ - v , 1 C -I m C, e t t B t B a t B C, ’ (B ) fractional xex—joe extent NBC *olinate’ 1765 fractional sex—ice extent NBC *climate 2300 fractional sex—ice extent NBC ‘+climxte 2000 fractional sea—ice extent NBC +climxte’ 2100 fctinna1sna—1cn nxtent fractional sea—ice extent NBC +olinate’ 2300—1765 2300—1765 Figure 51 54 References Archer, D. & Maier-Reimer, E. Effect of deep-sea sedimentary calcite preservation on atmospheric CQ concentration. Nature 367, 260-263 (1994). Archer, D., Kheshgi, H. & Maier-Reimer, E. Multiple timescales for neutralization of fossil fuel CO2. Geophys. Res. Left. 24, 405-408 (1997). Anderson, L. On the hydrogen and oxygen content of marine phytoplankton.Deep-Sea Research 42, 1675-1680 (1995). Andersson, A. J., Mackenzie, F. T. & Ver, L. M. Solution of shallow-water carbonates: An insignificant buffer against rising atmospheric CO2. Geology 31, 513-516 (2003). Andersson A. J. & Mackenzie, F. T. Shallow-water oceans: a source or sink of atmospheric CO2? Front. Ecol. Environ. 2, 348-353 (2004). Armstrong, R. A., Lee, C., Hedges, J. I., Honjo, S. & Wakeham, S. G. A new, mechanistic model for organic carbon fluxes in the ocean based on the quantitative association of POC with ballast minerals. Deep Sea Res. Part II 49, 219-236 (2002). Barker, S., Higgins, J. A. & Elderlield, H. The future of the carbon cycle: review, calcification response, ballast and feedback on atmospheric CO2. Philosophical Trans. Royal SocietyA 36, doi:10.1098/rsta2003.1238 (2003). Behrenfeld, M. J. & Falkowski, P. G. Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnol. Oceanogr. 42, 1-20 (1997). Betzer, P.R., Byrne, R. H., Acker, J. G., Lewis, C. S. & Jolley, R. R. The oceanic carbonate system: A reassessment of biogenic controls. Science 226, 1074-1 077 (1984). Bijma, J., Spero, H. J. & Lea, D. W. Reassessing the foraminiferal stable isotope geochemistry: Impact of the oceanic carbonate system. In Use of Proxies in paleoceanography: Examples from the Atlantic (eds Fisher, G. & Wefer, G.) 489-512 (1339 Springer-Verlag Berlin Heidelberg, 1999). Broecker, W. S. & Peng, T. H. Tracers in the Sea. (Palisades, New York, 1982). Broecker, W. S. & Peng, T.-H. The role of CaCO3 compensation in the glacial to interglacial atmospheric CO2 change. Glob. Biogeochem. Cycles 1, 15-29 (1987). Caldeira, K. & Wickett, M. E. Anthropogenic carbon and ocean pH. Nature 425, 365 (2003). Caldeira, K. & Wickett, M. E. Ocean model predictions of chemistry changes from carbon dioxide emissions to the atmosphere and ocean. J. Geophys. Res. 110, doi:10.1029/2004JC002671 (2005). Catubig, N. R. et al. Global deep-sea burial rate of calcium carbonate during the Last Glacial Maximum. Paleoceanography 13, 298-310 (1998). Elderfield, H. Carbonate mysteries. Science 296, 161 8-1 621 (2002). Eppley, R. W. & Peterson, B. J. Particulate organic matter flux and planktonic new production in the deep ocean. Nature 282, 677-680 (1979). Fabry, V. J. Shell growth rates of pteropod and heteropod mollusks and aragonite production in the open ocean: 55 Implications for marine carbonate system. J. Marine Res. 48, 209-222 (1990). Feely, R.A. et al. Impact of Anthropogenic CO2 on the CaCO3System in the Oceans. Science 305, 362-366 (2004). Francois, R., Honjo, S., Krishfield, R. & Manganini S. Factors controlling the flux of organic carbon to the bathypelagic zone of the ocean. Glob. Biogeochem. Cycles 16, doi: 10.1029/2001 GBOO 1722 (2002). Frankignoulle, M., Canon, C. & Gattuso, J. P. Marine calcification as a source of carbon dioxide: positive feedback of increasing atmospheric CO2. Limnol. Oceanogr. 39, 458-462 (1994). Ganeshram, R. S. Oceanic action at a distance. Nature 419, 123-125 (2002). Gattuso, J-P. et a!. Photosynthesis and calcification at cellular, organismal and community levels in coral reefs: a review on interactions and control by carbonate chemistry. American Zoologist 39, 160-183 (1999). Hamm, C. E. Interactive aggregation and sedimentation of diatoms and clay-sized lithogenic material. Limno!. Oceanogr. 47, 1790-1795 (2002). Heinze, C. Simulating oceanic CaCO3export production in the greenhouse. Geophys. Res. Lett. 13, doi:10.1020/2004G02061 3 (2004). Honjo, S. Coccoliths: production, transportation, and sedimentation. Mar. Micropaleo. 1, 65-79 (1976). Honjo, S. Fluxes of particles to the interior of the open oceans. In Particle Flux in the Ocean, SCOPE 57 (eds Itekkot, V.,Aschauffer, P., Honjo, S. & Depetris, P.) 91-154 (Wiley, New York, 1996). lglesias-Rodriguez et al. Progress made in study of ocean’s calcium carbonate budget. EQS Amer. Geophys. Un. Trans. 83, 365, 374-375 (2002). Jansen, H., & WoIf-Gladrow, D. A. Carbonate dissolution in copepod guts: A numerical model. Mar. Ecol. Prog. Ser., 221, 197— 207 (2001). Klaas, C. & Archer D. Association of sinking organic matter with various types of mineral ballast in the deep sea: Implications for the rain ratio. Glob. Biogeochem. Cycles 16, doi:10.1029/200IGBOO1 765 (2002). Kleypas, J. A. et al. Geochemical Consequences of increased atmospheric carbon dioxide on coral reefs. Science 284, 118-1 20 (1999). Laws, E. A. et a!. Temperature effects on export production in the open ocean. Glob. Biogeochem. Cycles 14, 1231-1246 (2000). Lenton, T. M. Land and ocean carbon cycle feedback effects on global warming in a simple Earth system model. Tellus Ser. B 52, 1159-1188 (2000). Lenton, T. M. et a!. Millennial timescale carbon cycle and climate change in an efficient Earth system model. Clim. Dynam. 26, 687-711, doi: 10.1 007/s00382-006-01 09-9 (2006). Li, Y.-H., Takahashi, T., & Broecker, W. S. Degree of saturation of CaCO3 in the oceans. J. Geophys. Res. 74, 5507 (1969). Longhurst, A., Sathyendranath, S., Platt, T. & Caverhill, C. An estimate of global primary production in the ocean from satellite radiometer data. Jour. Plankton Res. 17, 1245-1277 (1995). Lutz, M., Dunbar, R., & Caldeira, K. Regional variability in the vertical flux of particulate organic carbon in the ocean interior. Glob. Giogeochem. Cycles 16, doi:10.1029/2000GB001383 (2002) Martin, J. H., Knauer, G. A., Karl, D. & Broenknow, W. W. VERTEX: Carbon cycling in the northeast Pacific. 56 Deep Sea Res. Part 134, 267-285 (1987). Martin J. H., Fitzwater, S. E., Gordon, R. M., Hunter, C. N. & Tanner, S. J. Iron, primary production and carbon- nitrogen flux studies during the JGOFS North Atlantic Bloom Experiment. Deep Sea Res. Part 1140, 115-134 (1993). Milliman, J. D. eta!. Biologically mediated dissolution of calcium carbonate above the chemical lysocline? Deep Sea Res. PartI46, 1653-1669 (1999). Monterey, C. & Levitus, S. Seasonal variability of mixed layer depth for the world ocean. NOAA Atlas NESDIS 14, 100 (U.S. Gov. Print. Off., Washington D. C., 1997). Mortlock, R. A. & Froelich, P. N. A simple method for the rapid determination of biogenic opal in pelagic marine sediments, Deep-Sea Research 36, 1415-1426 (1989). Orr, J. C. et a!. Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms. Nature 437, 681-686 (2005). Passow, U., Alldredge, A. L. & Logan, B. E. The role of particulate carbohydrate exudates in the flocculation of diatom blooms. Deep Sea Res. Part 141, 335-357 (1994). Passow, U. Switching perspectives: Do mineral fluxes determine particulate organic carbon fluxes of vice versa? Geochemistry Geophysics Geosystems 5, doi: 10.1 029/2003GC000670 (2004). Passow, U. & De La Rocha, C. Accumulation of mineral ballast on organic aggregates. Glob. Biogeochem. Cycles 20, doi: 10.1 029/2005GB002579 (2006). Ridgwell, A. J. An end to the “rain ratio” reign? Geochemistry Geophysics Geosystems 4, doi:10.1 029/2003GC000512 (2003). Ridgwell, A. J. et a!. Marine geochemical data assimilation in an efficient Earth System Model of global geochemical cycling. Biogeosciences 4, 87-104 (2007). Ridgwell, A. J. et a!. Assessing the potential long-term increase of oceanic fossil fuel CO2 uptake due to CC2- calcification feedback. Biogeosciences 4, 481-492 (2007). Riebesell, U. et al. Reduced calcification of manne plankton in response to increased atmospheric CO2. Nature 407, 364-367 (2000). Sarmiento, J. L. & Bender, M. Carbon biogeochemistry and climate change. Photosynthesis Research 39, 209- 234 (1994). Sarmiento, J. L. et al. A new estimate of the CaCO3 to organic carbon export ratio. Glob. Biogeochem. Cycles 16, doi:10.1029/2002GB001919 (2002). Scholten, J. C. et al. Trapping efficiency of sediment traps from the deep Eastern North Atlantic: The 230Th calibration. Deep Sea Res. Part 1148, 2383-2408 (2001). Sigman D.M. & Boyle E.A. Glacial/interglacial variations in atmospheric carbon dioxide. Nature 407, 859-869 (2000). Steinmetz, J. C. Sedimentation of coccolithophores. In Coccolithophores (eds Winter, A. & Siesser, W.) 179-1 97 (Cambridge Univ. Press, New York, 1994). Suess, E. Particulate organic carbon flux in the oceans-surface productivity and oxygen utilization. Nature 288, 260-263 (1980). 57 Takahashi, T. et a!. Global sea-air CO2 flux based on climatological surface ocean pCO2 and seasonal biological and temperature effects. Deep Sea Research Part l149, 1601-1622 (2002). Takahashi, T. The Fate of Industrial Carbon Dioxide. Science 305, 352-353 (2004). Tsunogai, S. & Noriki, S. Particulate fluxes of carbonate and organic carbon in the ocean: Is the marine biological activity working as a sink of the atmospheric carbon? Tellus Ser. B 43, 256-266(1991). Ware, J. R., Smith, S. V. & Reakakudla, M. L. Coral reefs: sources or sinks of atmospheric CC2? Coral Reefs 11, 127-130(1992). Yu, E.-F. et al. Trapping efficiency of bottom-tethered sediment traps estimated from the intercepted fluxes of 230Th and 231Pa. Deep Sea Res. Part 148, 865-889 (2001). Zhong, S.J. & Mucci, A. Calcite precipitation in seawater using a constant addition technique-a new overall reaction kinetic expression. Geochimica et Cosmochimca Acta 57, 1409-1417 (1993). Zondervan, I., Zeebe, R. E., Rost, B. & Riebesell, U. Decreasing marine biogenic calcification: A negative feedback on rising atmospheric pCC2. Glob. Biogeochem. Cycles 15, 507-516 (2001). 58 Appendices Appendix A: Exchange of CO2 between the atmosphere and ocean Carbon dioxide has a relatively high solubility and chemical reactivity in seawater. Its solubility has been determined experimentally as a function of temperature and salinity (both inverse relationship), where temperature plays a larger role. Colder and less saline water could take up more CO2 than warmer and more saline water. Henry’s law describes the relationship between solubility (S) of gas in liquid and sea water properties. C’ _l* k solubility constant P = overlying pressure of the gas in the atmosphere The exchange of CO2 between the atmosphere and the ocean takes place in the surface layer of the ocean, above the thermocline. Air-sea exchange occurs in the form of gas exchange and is mainly driven by the difference in the pCO2 between the atmosphere and the ocean (A pCO2). For every 1°C temperature increase in seawater, pCO2ocean increases by —4% (Sarmiento and Bender, 1994). pCO2ocean is also influenced by the thermodynamic relationships between different carbon species (i.e. distribution of carbon species within the DIC pool). Under PCO2 atmosphere> pC02ocean conditions, CO2 will diffuse into the ocean. Conversely, fPCO2 atmosphere pCO2ocean, the ocean would outgas CO2 to the atmosphere. The gas transfer rate depends on the magnitude of A pCO2and other factors such as solubility and turbulence. I _‘*C’ * A f’I5 — F\ 0C02 -° P’-”-’ K = piston velocity SCO2 = solubility of CO2 in sea water at a given temperature and salinity A pCO2 pCO2ocean - pCO2atmosphere Since gas exchange across the air-sea is slower than the oceanic processes that affect the partial pressure of CO2 in the surface waters, most surface waters are not in gaseous equilibrium with the atmosphere at one instantaneous time. The rate of gas exchange varies as warming and cooling processes of surface waters changes the saturation state of the surface seawater with respect to CO2. Cold undersaturated and warm supersaturated regions of the ocean are identified as places of carbon ‘sinks’ and ‘sources’, respectively (Takahashi et al., 2002; Takahashi, 2004). The time required for CO2 between the atmosphere and surface ocean to achieve equilibrium is about 6 to 12 months (Sigman and Boyle, 2000) and on this time scale, and on a global spatial scale, the atmosphere and ocean can be considered as in a steady-state with respect to C exchange if there were no anthropogenic disturbances. 59 Appendix B: Carbonate buffer It is important to recognize the rapid speciation of CO2 upon its entering the ocean. Most of the dissolved inorganic carbon (DIC) in the ocean is in the form of bicarbonate ion (HCO3-). When gaseous CO2 is taken up from the atmosphere and dissolve in seawater, a large portion of the CO2 quickly undergoes dissociation by reacting with water and carbonate ion (C032-) to form bicarbonate ions. Acid-base pH equilibrium reactions: CO2 + H20 —* H2C03 H2C03<—* W + HC03 H + C032 HCO3 Overall reaction: CO2 + HO +C032 2HCO3 Adding 1 CO2 unit decreases C032 by 1 unit and increases HC03 by 2. DIC increases by 1. TA (total ALK) CA (carbonate ALK) remains the same. pH decreases (more acidic) since H is also produced (in an instantaneous sense), but doesn’t lower pH very much (well buffered). Such buffering reactions of the carbonate system maintain the pH of seawater to almost constant at —8, where deviations are largely due to kinetic issues that inhibit equilibrium to be achieved. Through this process of carbonate buffering, the carbonate ion concentration is reduced and the bicarbonate ion concentration is increased. The pH equilibrium reaction maintains an inverse relationship between [CO32]and pCO2 and suggests that the CO2 uptake capacity of the ocean is strongly limited by the concentration of C032 in the ocean. The release of CO2 into seawater from biological respiration initiates the same buffering reaction. DIC = [CO2 + [H2C03]+ [HC03]+ [CO32j [HC03]+ [C032] TA CA = 2[C03]+ [HCO3jin milliequivalents of charge present in 1 L of solution (meq L1) Carbonate ion concentration could be approximated from DIC and CA [CO32]= CA- DIC 60 Appendix C: Calcium carbonate precipitationldissolution CaCO3 equilibrium reactions: Ca2 + CO32—* CaCO3 HC03 + H HCO3 + H H2C03 H2C03 CO2(aq) + H20 Overall reaction: Ca2 + 2HCO3 CaCO3+ CO2 (aq) + H20 Precipitation of CaCO3by marine organisms (Ca2 + 2HC03-—* CaCO3 + CO2 (aq) + H20) in the surface ocean decreases the ambient seawater DIG and ALK by 1 and 2 unit(s), respectively. The result is an increase inpCO2 ocean (causes seawater to become more acidic), hence decrease in oceanic CO.2 uptake. It is estimated that, in the modern ocean, for each molecule of CO2 produced by calcification a fraction of about 0.6 is potentially released to the atmosphere, while the rest is taken up by the carbomte buffer (Ware et al., 1992) Here, we find a positive coupling, where the increase in CaCO3production ‘amplifies’ the accumulation of atmospheric CO2. This result may be somewhat counter-intuitive and is often easily confused.. The effect of dissolution (CaCO3 + CO2 (aq) + H20 — Ca2 + 2HCO3jin the deep ocean is the increase in DIC and ALK of the ambient seawater by 1 and 2 unit(s), respectively, thus partially neutralizing the accumulation of metabolic CO2 with depth, creating potential for future CO2 uptake. Here, the weakening of the positive coupling ‘depresses’ the accumulation of atmospheric CO2. An increase in dissolution (or decrease in precipitation) in the surface ocean would create a more alkaline profile water column that weakens the positive coupling. 61 Appendix D: Silicate-carbonate carbon cycle Weathering-metamorphism reactions: CaSiO3 + 2C0 + H20 Ca2 + Si02 + 2HC03 Ca2 + 2HC03 CaCO3 + CO2(aq) + H20 Overall reaction: CaSiO3+ CO2 CaCO3+ Si02 On the time scale of millions of years, CO2 is removed from the atmosphere by chemical weathering on land, deposited in the ocean, subducted, and returned to the atmosphere by volcanic activity. The weathering process depends on the fact that atmospheric CO2 and rain (H20) combine in soils and rock crevices to form carbonic acid (H2C03), a weak acid that slowly attacks the silicate rocks (CaSiO3)chemically. Rivers carry off the dissolved ions to the ocean, then some of the dissolved ions are taken up by benthic and planktonic calcifying organisms that form CaCO3 and silica opals (SiC2). A fraction of the biogenic CaCO3would eventually be buried to the seafloor and be incorporated back to the geological reservoir. CaCO3 sediments are transported downward in the subduction process and are either melted or transformed by high temperature and pressure. Organic matter is decomposed and CaCQ3 reacts with the silica (SIC2)found in the subducted rocks and forms calcium silicate (CaSiO.3). This process termed metamorphism is the reverse of weathenng and is often also referred to as ‘reversed weathering’. These processes return to the atmosphere and complete (close) the tectonic-scale cycle. The reactions that summarize the chemical changes involved in the weathering and metamorphism phases of this long-term carbon cycle are mirror processes. 62

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0053295/manifest

Comment

Related Items