Quantifying Glacial Ocean Carbon Cycling
Samar Khatiwala, Juan Muglia & Andreas Schmittner
Supported by NSF CEOAS OEB Seminar Fri. Jan 12, 2018
Jouzel et al. 2007
Lüthi et al. 2008
Lisiecki & Raymo 2004
Motivation
Various processes have been proposed to explain glacial CO2
drawdown of ~ 90
ppm, e.g.
1. Sea Ice (Stephens and Keeling, 2000, “67 ppm”)
2. Iron (Martin, 1999; Lambert et al. 2015, “10-20 ppm”)
3. Circulation (Sarnthein et al., 2013; Skinner et al., 2017, “65 ppm”*)
4. Temperature (but thought to account only for 15-30 ppm; Wallmann et al.
2016 “16 ppm”; Sigman & Boyle, 2000 “30 ppm”; often ignored e.g. Hain et
al., 2010)
5. Whole Ocean Alkalinity (Boyle 1988; Brovkin et al, 2007, “10-30 ppm”)
Here we want to quantitatively understand these processes better using a proper
decomposition of the ocean’s carbon cycle.
*these and many other recent studies (incl. my own) suggest an increase in the biological pump, often quantified as Corg ~ AOU
Ocean Carbon Basics
Carbon storage in the ocean is dominated by dissolved inorganic carbon (DIC).
DIC is affected by
1. physical processes such as the temperature dependent solubility of CO2
in
seawater (solubility pump), and by
2. biological processes such as remineralization of organic matter (biological
pump).
Air-sea gas exchange is relatively slow (~1 yr) such that the surface DIC is not in
equilibrium with atmospheric pCO2
.
Broecker and Takahashi (1984), Volk and Hoffert (1985)
Theory
Dissolved Inorganic Carbon: DIC = Cpre + Creg
Preformed Carbon: Cpre = Ceq + Cdis
Regenerated Carbon: Creg = Corg + Ccaco3
Equilibrium Carbon: Ceq = DIC(T,S,ALK,pCO2
) = Ceq,phy + Ceq,bio
Disequilibrium Carbon: Cdis = Cdis,phy + Cdis,bio
Biological Carbon: Cbio = Corg + Ccaco3 + Cdis,bio + Ceq,bio
Physical Carbon: Cphy = DIC - Cbio = Ceq,phy + Cdis,phy
Cpre
Creg
DIC=Cpre+Creg
Biology
Circulation
Often used approximations: Cbio = Corg, Corg ~ AOU (we’ll show that those are not warranted and lead to large errors)
Based on previous work: e.g. Toggweiler et al. 2003, Sarmiento and Gruber 2006, Williams and Follow 2011
Methods
● Use data constrained models to quantify carbon components*
● Model of Ocean Biogeochemistry and Isotopes (MOBI) and Transport Matrix
Method (TMM)**
● Use model without biology to calculate Ceq,phy and Cdis,phy
● Use TMM to propagate preformed properties Ceq, Cdis, potALKpre and
PO4
pre into the interior
● Ccaco3 = (potALK - potALKpre)/2; potALK = ALK + NO3
● Corg = RC:P
(PO4
- PO4
pre)
● Cdis,bio = Cdis - Cdis,phy
● Ceq,bio = Cdis - Ceq,phy
*Data constraints for the modern ocean include T, S, NO3, PO4, DFe, ALK, DIC, AOU.
**TMM uses different advection scheme to extract matrices, which leads to small differences when compared to online model
Pre-Industrial
Control (PIC)
Ceq,phy = 36,209 Pg
Potential Solubility Models of Toggweiler et al. (2003), Murnane et al. (1999)
Depth(m)
Ceq,phy = 36,209 PgC
Cdis,phy = -785 PgC
Cphy = 35,424 PgC
PIC
Cdis,phy
Cphy
Solubility Models of Toggweiler et al. (2003), Murnane et al. (1999)
Ceq,phy = 36,209 PgC
Cdis,phy = -785 PgC
Corg = 1070 PgC
PIC
Corg
Cphy
Ceq,phy = 36,209 PgC
Cdis,phy = -785 PgC
Corg = 1070 PgC
Ccaco3 = 208 PgC
PIC
Ccaco3
Cphy
Ceq,phy = 36,209 PgC
Cdis,phy = -785 PgC
Corg = 1075 PgC
Ccaco3 = 208 PgC
Cdis,bio = 1070 PgC
PIC
Cdis,bio
Cphy
Toggweiler et al. (2003b), Murnane et al. (1999)
Ceq,phy = 36,209 PgC
Cdis,phy = -785 PgC
Corg = 1075 PgC
Ccaco3 = 208 PgC
Cdis,bio = 1070 PgC
Ceq,bio = -293 PgC
PIC
Cphy
Ceq,bio
DIC
Ceq,phy = 36,209 PgC
Cdis,phy = -785 PgC
Corg = 1075 PgC
Ccaco3 = 208 PgC
Cdis,bio = 1070 PgC
Ceq,bio = -293 PgC
Ctot = 37,484 PgC
PIC
Cphy,deep - Cphy,sfc =
100 mol m-3
Cbio
Cbio,deep - Cbio,sfc =
163 mol m-3
DIC
PIC Carbon Pumps
Deep - Surface
(mmol m-3
) (%) (%)
Cphy
Ceq,phy 158 60
38Cdis,phy -58 -22
Cbio
Corg 35 13
62
Ccaco3 20 8
Cdis,bio 90 34
Ceq,bio 19 7
Ctot 263 100 100
Deep > 2 km
Surface 50S - 50N
PIC
Ceq
Atlantic Pacific
Surface
mol m-3
Dominated by temperature variations
PIC Cdis
Atlantic Pacific
Surface
mol m-3
PIC Corg
Organic carbon remineralizes at relatively shallow depths and
accumulates over time.
Atlantic Pacific
Corg 1075 PgC
Often used approximation
Corg = AOU*RC:O
1476
PgC
AOU approximation
overestimates Corg in
deep ocean due to
oxygen disequilibrium.
PIC Ccaco3 Ccaco3 remineralizes deeper than Corg.
Atlantic Pacific
PIC Cdis
Disequilibrium component is net result
of two opposing effects:
Cdis,bio = 1070 Pg
Cdis,phy = -785 Pg
Cdis,phy is negative because slow
air-sea gas exchange limits the
ingassing of carbon at high latitudes.
Cdis,bio is positive because slow
air-sea gas exchange limits the
outgassing of biologically sequestered
carbon at high latitudes.
PIC
Cdis,phy
Atlantic Pacific
Surface
mol m-3
Upwelling of cold water in the tropics and
associated downward surface heat flux
causes positive Cdis,phy.
Poleward transport and associated heat loss
causes negative Cdis,phy at high latitudes
and in deep ocean.
PIC
Cdis,bio
Atlantic Pacific
Surface
mol m-3
Last Glacial Maximum (~20,000 yr BP)
Model constrained by δ15
N, δ13
C and radiocarbon (Muglia et al., submitted).
Features:
● prescribed pCO2
= 190 ppm
● colder (ΔSST = ΔT = -2.4°C*)
● increased sea ice cover
● shallower, weaker (~50%) AMOC** (required to reproduce δ13
C and Δ14
C)
● Δage = +600 14
C years***
● 10x enhanced iron solubility in Southern Ocean (required to reproduce δ15
N,
δ13
C)
*consistent with ΔT = -2.6°C from ice core noble gas measurements (Bereiter et al. 2018, Nature)
**Atlantic Meridional Overturning Circulation, forced by decreasing atm. meridional moisture flux in southern hemisphere
***consistent with Skinner et al. (2017) and Sarnthein et al. (2015); Model ideal age is younger!
Transport Matrix Method (TMM) calculates efficiently BGC tracer distributions at
equilibrium (Khatiwala 2007). Includes prognostic atmospheric pCO2
(but no land
carbon and no sediments). Two sets of TMM experiments:
1. PIC->LGM
2. LGM->PIC
In each set 5 individual variables are changed one at a time: temperature,
salinity, circulation, sea ice, iron and one experiment in which all are changed
together. Non-linearities are evaluated as the difference between the sum of the
individual experiments and the all experiment. It turns out that averaging (1-2)/2
the two experiments reduces the non-linearities.
LGM Set-Up
-44
+16
-26
+5
+4
-67
+16
+14
+42
+48
+92
-1
PIC->LGM
Temperature and
iron cause large
and robust
decrease.
Together they
account for 77±10
ppm CO2
drop.
Sea ice and
circulation effects
are small.
-44±1
-32±6
-4±9
+3±13
3±0
-3±20
-77±10
-91±4
PIC->LGM
Cdis quadrupled in
LGM!
Much smaller changes
in Corg and Ccaco3.
Ceq has decreased
mostly because lower
pCO2
.
Corg decreased*
Ceq
Cdis
Corg
Ccaco3
*Corg ~ AOU increases; a Corg decrease is contrary to many previous studies (e.g. Sarnthein et al. 2013, Skinner et al.
2017, Schmittner and Somes 2016, who used Corg ~ AOU, Jaccard & Galbraith 2011, ....)
PIC->LGM
Large increase in Cdis is dominated by Cdis,bio
Temperature
increases Cdis.
Iron increases
Corg and Cdis.
Circulation and
sea ice increase
Cdis but
decrease Corg.
PIC->LGM
PIC->LGM
Temperature increases
Cdis,phy.
Iron increases Cdis,bio.
Circulation increases both.
Sea ice increases Cdis,bio
but decreases Cdis,phy.
Temperature
Global mean ΔT = -2.5 C consistent with Bereiter et al. (2018; ΔT = -2.6 C)
Theory (e.g. Williams and Follows 2011) => ΔCO2
= -25 ppm.
But total T effect is -45 ppm.
Thus, spatial differences in ΔT must lower CO2
by an additional 20 ppm through
Cdis,phy.
How?
PIC->LGM
Reduced meridional
SST gradient at high
latitudes in LGM
decreases surface
heat flux.
This decreases the
(negative) Cdis,phy
thus increases Cdis.
Temperature Increases Cdis,phy
Atlantic Pacific
PIC->LGM
This effect causes temperature to
reduce CO2
by much more than
just the effect of global mean
cooling of ~2.3°C (23 ppm).
White stippling indicates qualitatively not
robust changes (i.e. the same sign in
PIC->LGM and LGM->PIC experiments.
+186 ± 40 PgC
Iron increases Corg and Cdis
Corg
Atlantic Pacific
Cdis
Export production
increased by 0.72 PgC/y.
+146 ± 2 PgC
+356 ± 164 PgC
Circulation Changes LGM AMOC much weaker and
shallower than PIC. Reduced
upwelling in Indian and Pacific.
PIC
LGM
PIC
LGM
Indian and PacificAtlantic
Muglia et al. (submitted)
Circulation Decreases Corg
Atlantic Pacific
Decreased AMOC causes reduced export
production by -and thus lower Corg.
Export production increased by 0.46 PgC/y.-194 ± 8 PgC
Circulation Increases Cdis
Atlantic Pacific
Because of reduced AMOC more of
the ocean is filled with high Cdis
AABW
+340 ± 187 PgC
Sea Ice
Decreases Cdis,phy
Increases Cdis,bio
Cdis
Cdis,phy
Cdis,bio
Atlantic Pacific
+63 ± 136 PgC
+245 ± 13 PgC
-182 ± 186 PgC
Sea Ice Decreases Corg
Because it decreases export production (by 0.26 PgC/y) due to enhanced light
limitation.
-67 ± 46 PgC
heat
flux
carbon
flux
PIC
sea ice
heat
flux
carbon
flux
LGM
cold
warm
Cdis,phy < 0
colder
Cdis,phy increased
cold cold
No Biology
cold
sea ice
Cooling of mid-latitudes
decreases heat flux and
increases Cdis,phy
heat
flux
carbon
flux
LGM
Cdis,phy decreased
sea ice
More sea ice decreases carbon
fluxes (but not heat fluxes) thus
decreases Cdis,phy
cold
cold
Net effect is that Cdis,phy doesn’t change much
warm
Temperature Effect Sea Ice Effect
DIC = … + Cdis,phy DIC = … + Cdis,phy
DIC = … + Cdis,phy
Cdis,phy is negative because
carbon flux is into the ocean,
but too slow to achieve the full
potential (Ceq,phy)
carbon
flux
PIC
sea ice
carbon
flux
LGM
sea ice
Corg = 0
Corg > 0
Cdis,bio > 0 Cdis,bio increased
Biology
DIC = … + Cdis,bio DIC = … + Cdis,bio
Corg = 0
Corg > 0
Cdis,bio is positive because carbon flux
is out of the ocean, but it cannot get rid
of all of the Corg that is upwelled and
converted to Cdis,bio at the surface.
Increased sea ice decreases the outgassing
thus increasing Cdis,bio. Any increase in
upwelling Corg (e.g. through iron fertilization)
would also increase Cdis,bio.
Conclusions
● Data-constrained LGM model with weaker AMOC and increased Fe solubility
in SO decreases pCO2
by 77±10 ppm
● Corg decreased but Cbio increased due to large increase in Cdis,bio
● Temperature has large effect (-44±1 ppm) not only through Ceq but also by
increasing Cdis,phy due to reduced sea-air heat fluxes at mid- to high
latitudes
● Iron is also important ~30 ppm due to increasing both Corg and Cdis,bio
● Compensating effects reduce the overall impact of circulation and sea ice
changes
○ Both increase Cdis but decrease Corg
○ Corg is decreased due to reduced export production (more sea ice enhances light limitation*
and reduced AMOC -> less upwelling)
*not considered by Stephens and Keeling (2000)
Implications
● Important steps towards understanding glacial-interglacial CO2
problem
● The large effect of temperature is surprising and may explain the tight
coupling of Antarctic temperatures with CO2
in the ice core observations
● The large effect of iron is contrary to recent iron model simulations but
required by δ15
N and δ13
C reconstructions
○ Something wrong with iron models
○ Iron solubility was enhanced during LGM
○ Another process causes enhanced nutrient utilization (e.g. dust effects on sinking speed)
Thanks
Creg and EP
Corg and EP
changes are of
same sign except
for temp.
Carbon Pumps (deep - surface)
PIC LGM
(mmol m-3
) (mmol m-3
)
Cphy
Ceq,phy 158
99
135
83Cdis,phy -58 -52
Cbio
Corg 35
164
31
252
Ccaco3 20 16
Cdis,bio 90 162
Ceq,bio 19 44
Ctot 263 335
Mean age:
picdefault: 519.06
lgmdefault/diffq01: 450.73
lgm-pic: -68.33
C14 age:
picdefault: 1308.76
lgmdefault/diffq01: 1831.84
lgm-pic: 523.08
PIC Cdis
PIC->LGM
Increase in Cdis is caused by
Cdis,bio.
Quantifying glacial ocean carbon cycling
Quantifying glacial ocean carbon cycling
Quantifying glacial ocean carbon cycling
Quantifying glacial ocean carbon cycling
Quantifying glacial ocean carbon cycling

Quantifying glacial ocean carbon cycling

  • 1.
    Quantifying Glacial OceanCarbon Cycling Samar Khatiwala, Juan Muglia & Andreas Schmittner Supported by NSF CEOAS OEB Seminar Fri. Jan 12, 2018 Jouzel et al. 2007 Lüthi et al. 2008 Lisiecki & Raymo 2004
  • 2.
    Motivation Various processes havebeen proposed to explain glacial CO2 drawdown of ~ 90 ppm, e.g. 1. Sea Ice (Stephens and Keeling, 2000, “67 ppm”) 2. Iron (Martin, 1999; Lambert et al. 2015, “10-20 ppm”) 3. Circulation (Sarnthein et al., 2013; Skinner et al., 2017, “65 ppm”*) 4. Temperature (but thought to account only for 15-30 ppm; Wallmann et al. 2016 “16 ppm”; Sigman & Boyle, 2000 “30 ppm”; often ignored e.g. Hain et al., 2010) 5. Whole Ocean Alkalinity (Boyle 1988; Brovkin et al, 2007, “10-30 ppm”) Here we want to quantitatively understand these processes better using a proper decomposition of the ocean’s carbon cycle. *these and many other recent studies (incl. my own) suggest an increase in the biological pump, often quantified as Corg ~ AOU
  • 3.
    Ocean Carbon Basics Carbonstorage in the ocean is dominated by dissolved inorganic carbon (DIC). DIC is affected by 1. physical processes such as the temperature dependent solubility of CO2 in seawater (solubility pump), and by 2. biological processes such as remineralization of organic matter (biological pump). Air-sea gas exchange is relatively slow (~1 yr) such that the surface DIC is not in equilibrium with atmospheric pCO2 . Broecker and Takahashi (1984), Volk and Hoffert (1985)
  • 4.
    Theory Dissolved Inorganic Carbon:DIC = Cpre + Creg Preformed Carbon: Cpre = Ceq + Cdis Regenerated Carbon: Creg = Corg + Ccaco3 Equilibrium Carbon: Ceq = DIC(T,S,ALK,pCO2 ) = Ceq,phy + Ceq,bio Disequilibrium Carbon: Cdis = Cdis,phy + Cdis,bio Biological Carbon: Cbio = Corg + Ccaco3 + Cdis,bio + Ceq,bio Physical Carbon: Cphy = DIC - Cbio = Ceq,phy + Cdis,phy Cpre Creg DIC=Cpre+Creg Biology Circulation Often used approximations: Cbio = Corg, Corg ~ AOU (we’ll show that those are not warranted and lead to large errors) Based on previous work: e.g. Toggweiler et al. 2003, Sarmiento and Gruber 2006, Williams and Follow 2011
  • 5.
    Methods ● Use dataconstrained models to quantify carbon components* ● Model of Ocean Biogeochemistry and Isotopes (MOBI) and Transport Matrix Method (TMM)** ● Use model without biology to calculate Ceq,phy and Cdis,phy ● Use TMM to propagate preformed properties Ceq, Cdis, potALKpre and PO4 pre into the interior ● Ccaco3 = (potALK - potALKpre)/2; potALK = ALK + NO3 ● Corg = RC:P (PO4 - PO4 pre) ● Cdis,bio = Cdis - Cdis,phy ● Ceq,bio = Cdis - Ceq,phy *Data constraints for the modern ocean include T, S, NO3, PO4, DFe, ALK, DIC, AOU. **TMM uses different advection scheme to extract matrices, which leads to small differences when compared to online model
  • 6.
    Pre-Industrial Control (PIC) Ceq,phy =36,209 Pg Potential Solubility Models of Toggweiler et al. (2003), Murnane et al. (1999) Depth(m)
  • 7.
    Ceq,phy = 36,209PgC Cdis,phy = -785 PgC Cphy = 35,424 PgC PIC Cdis,phy Cphy Solubility Models of Toggweiler et al. (2003), Murnane et al. (1999)
  • 8.
    Ceq,phy = 36,209PgC Cdis,phy = -785 PgC Corg = 1070 PgC PIC Corg Cphy
  • 9.
    Ceq,phy = 36,209PgC Cdis,phy = -785 PgC Corg = 1070 PgC Ccaco3 = 208 PgC PIC Ccaco3 Cphy
  • 10.
    Ceq,phy = 36,209PgC Cdis,phy = -785 PgC Corg = 1075 PgC Ccaco3 = 208 PgC Cdis,bio = 1070 PgC PIC Cdis,bio Cphy Toggweiler et al. (2003b), Murnane et al. (1999)
  • 11.
    Ceq,phy = 36,209PgC Cdis,phy = -785 PgC Corg = 1075 PgC Ccaco3 = 208 PgC Cdis,bio = 1070 PgC Ceq,bio = -293 PgC PIC Cphy Ceq,bio DIC
  • 12.
    Ceq,phy = 36,209PgC Cdis,phy = -785 PgC Corg = 1075 PgC Ccaco3 = 208 PgC Cdis,bio = 1070 PgC Ceq,bio = -293 PgC Ctot = 37,484 PgC PIC Cphy,deep - Cphy,sfc = 100 mol m-3 Cbio Cbio,deep - Cbio,sfc = 163 mol m-3 DIC
  • 13.
    PIC Carbon Pumps Deep- Surface (mmol m-3 ) (%) (%) Cphy Ceq,phy 158 60 38Cdis,phy -58 -22 Cbio Corg 35 13 62 Ccaco3 20 8 Cdis,bio 90 34 Ceq,bio 19 7 Ctot 263 100 100 Deep > 2 km Surface 50S - 50N
  • 14.
  • 15.
  • 16.
    PIC Corg Organic carbonremineralizes at relatively shallow depths and accumulates over time. Atlantic Pacific
  • 17.
    Corg 1075 PgC Oftenused approximation Corg = AOU*RC:O 1476 PgC AOU approximation overestimates Corg in deep ocean due to oxygen disequilibrium.
  • 18.
    PIC Ccaco3 Ccaco3remineralizes deeper than Corg. Atlantic Pacific
  • 19.
    PIC Cdis Disequilibrium componentis net result of two opposing effects: Cdis,bio = 1070 Pg Cdis,phy = -785 Pg Cdis,phy is negative because slow air-sea gas exchange limits the ingassing of carbon at high latitudes. Cdis,bio is positive because slow air-sea gas exchange limits the outgassing of biologically sequestered carbon at high latitudes.
  • 20.
    PIC Cdis,phy Atlantic Pacific Surface mol m-3 Upwellingof cold water in the tropics and associated downward surface heat flux causes positive Cdis,phy. Poleward transport and associated heat loss causes negative Cdis,phy at high latitudes and in deep ocean.
  • 21.
  • 22.
    Last Glacial Maximum(~20,000 yr BP) Model constrained by δ15 N, δ13 C and radiocarbon (Muglia et al., submitted). Features: ● prescribed pCO2 = 190 ppm ● colder (ΔSST = ΔT = -2.4°C*) ● increased sea ice cover ● shallower, weaker (~50%) AMOC** (required to reproduce δ13 C and Δ14 C) ● Δage = +600 14 C years*** ● 10x enhanced iron solubility in Southern Ocean (required to reproduce δ15 N, δ13 C) *consistent with ΔT = -2.6°C from ice core noble gas measurements (Bereiter et al. 2018, Nature) **Atlantic Meridional Overturning Circulation, forced by decreasing atm. meridional moisture flux in southern hemisphere ***consistent with Skinner et al. (2017) and Sarnthein et al. (2015); Model ideal age is younger!
  • 23.
    Transport Matrix Method(TMM) calculates efficiently BGC tracer distributions at equilibrium (Khatiwala 2007). Includes prognostic atmospheric pCO2 (but no land carbon and no sediments). Two sets of TMM experiments: 1. PIC->LGM 2. LGM->PIC In each set 5 individual variables are changed one at a time: temperature, salinity, circulation, sea ice, iron and one experiment in which all are changed together. Non-linearities are evaluated as the difference between the sum of the individual experiments and the all experiment. It turns out that averaging (1-2)/2 the two experiments reduces the non-linearities. LGM Set-Up -44 +16 -26 +5 +4 -67 +16 +14 +42 +48 +92 -1
  • 24.
    PIC->LGM Temperature and iron causelarge and robust decrease. Together they account for 77±10 ppm CO2 drop. Sea ice and circulation effects are small. -44±1 -32±6 -4±9 +3±13 3±0 -3±20 -77±10 -91±4
  • 25.
    PIC->LGM Cdis quadrupled in LGM! Muchsmaller changes in Corg and Ccaco3. Ceq has decreased mostly because lower pCO2 . Corg decreased* Ceq Cdis Corg Ccaco3 *Corg ~ AOU increases; a Corg decrease is contrary to many previous studies (e.g. Sarnthein et al. 2013, Skinner et al. 2017, Schmittner and Somes 2016, who used Corg ~ AOU, Jaccard & Galbraith 2011, ....)
  • 26.
    PIC->LGM Large increase inCdis is dominated by Cdis,bio
  • 27.
    Temperature increases Cdis. Iron increases Corgand Cdis. Circulation and sea ice increase Cdis but decrease Corg. PIC->LGM
  • 28.
    PIC->LGM Temperature increases Cdis,phy. Iron increasesCdis,bio. Circulation increases both. Sea ice increases Cdis,bio but decreases Cdis,phy.
  • 29.
    Temperature Global mean ΔT= -2.5 C consistent with Bereiter et al. (2018; ΔT = -2.6 C) Theory (e.g. Williams and Follows 2011) => ΔCO2 = -25 ppm. But total T effect is -45 ppm. Thus, spatial differences in ΔT must lower CO2 by an additional 20 ppm through Cdis,phy. How?
  • 30.
    PIC->LGM Reduced meridional SST gradientat high latitudes in LGM decreases surface heat flux. This decreases the (negative) Cdis,phy thus increases Cdis.
  • 31.
    Temperature Increases Cdis,phy AtlanticPacific PIC->LGM This effect causes temperature to reduce CO2 by much more than just the effect of global mean cooling of ~2.3°C (23 ppm). White stippling indicates qualitatively not robust changes (i.e. the same sign in PIC->LGM and LGM->PIC experiments. +186 ± 40 PgC
  • 32.
    Iron increases Corgand Cdis Corg Atlantic Pacific Cdis Export production increased by 0.72 PgC/y. +146 ± 2 PgC +356 ± 164 PgC
  • 33.
    Circulation Changes LGMAMOC much weaker and shallower than PIC. Reduced upwelling in Indian and Pacific. PIC LGM PIC LGM Indian and PacificAtlantic Muglia et al. (submitted)
  • 34.
    Circulation Decreases Corg AtlanticPacific Decreased AMOC causes reduced export production by -and thus lower Corg. Export production increased by 0.46 PgC/y.-194 ± 8 PgC
  • 35.
    Circulation Increases Cdis AtlanticPacific Because of reduced AMOC more of the ocean is filled with high Cdis AABW +340 ± 187 PgC
  • 36.
    Sea Ice Decreases Cdis,phy IncreasesCdis,bio Cdis Cdis,phy Cdis,bio Atlantic Pacific +63 ± 136 PgC +245 ± 13 PgC -182 ± 186 PgC
  • 37.
    Sea Ice DecreasesCorg Because it decreases export production (by 0.26 PgC/y) due to enhanced light limitation. -67 ± 46 PgC
  • 38.
    heat flux carbon flux PIC sea ice heat flux carbon flux LGM cold warm Cdis,phy <0 colder Cdis,phy increased cold cold No Biology cold sea ice Cooling of mid-latitudes decreases heat flux and increases Cdis,phy heat flux carbon flux LGM Cdis,phy decreased sea ice More sea ice decreases carbon fluxes (but not heat fluxes) thus decreases Cdis,phy cold cold Net effect is that Cdis,phy doesn’t change much warm Temperature Effect Sea Ice Effect DIC = … + Cdis,phy DIC = … + Cdis,phy DIC = … + Cdis,phy Cdis,phy is negative because carbon flux is into the ocean, but too slow to achieve the full potential (Ceq,phy)
  • 39.
    carbon flux PIC sea ice carbon flux LGM sea ice Corg= 0 Corg > 0 Cdis,bio > 0 Cdis,bio increased Biology DIC = … + Cdis,bio DIC = … + Cdis,bio Corg = 0 Corg > 0 Cdis,bio is positive because carbon flux is out of the ocean, but it cannot get rid of all of the Corg that is upwelled and converted to Cdis,bio at the surface. Increased sea ice decreases the outgassing thus increasing Cdis,bio. Any increase in upwelling Corg (e.g. through iron fertilization) would also increase Cdis,bio.
  • 40.
    Conclusions ● Data-constrained LGMmodel with weaker AMOC and increased Fe solubility in SO decreases pCO2 by 77±10 ppm ● Corg decreased but Cbio increased due to large increase in Cdis,bio ● Temperature has large effect (-44±1 ppm) not only through Ceq but also by increasing Cdis,phy due to reduced sea-air heat fluxes at mid- to high latitudes ● Iron is also important ~30 ppm due to increasing both Corg and Cdis,bio ● Compensating effects reduce the overall impact of circulation and sea ice changes ○ Both increase Cdis but decrease Corg ○ Corg is decreased due to reduced export production (more sea ice enhances light limitation* and reduced AMOC -> less upwelling) *not considered by Stephens and Keeling (2000)
  • 41.
    Implications ● Important stepstowards understanding glacial-interglacial CO2 problem ● The large effect of temperature is surprising and may explain the tight coupling of Antarctic temperatures with CO2 in the ice core observations ● The large effect of iron is contrary to recent iron model simulations but required by δ15 N and δ13 C reconstructions ○ Something wrong with iron models ○ Iron solubility was enhanced during LGM ○ Another process causes enhanced nutrient utilization (e.g. dust effects on sinking speed)
  • 42.
  • 43.
    Creg and EP Corgand EP changes are of same sign except for temp.
  • 44.
    Carbon Pumps (deep- surface) PIC LGM (mmol m-3 ) (mmol m-3 ) Cphy Ceq,phy 158 99 135 83Cdis,phy -58 -52 Cbio Corg 35 164 31 252 Ccaco3 20 16 Cdis,bio 90 162 Ceq,bio 19 44 Ctot 263 335
  • 46.
    Mean age: picdefault: 519.06 lgmdefault/diffq01:450.73 lgm-pic: -68.33 C14 age: picdefault: 1308.76 lgmdefault/diffq01: 1831.84 lgm-pic: 523.08
  • 48.
  • 49.
    PIC->LGM Increase in Cdisis caused by Cdis,bio.