Transcript Folie 1 - Uni Bremen

Marine Biogeochemical and
Ecosystem Modeling
Michael Schulz
MARUM -- Center for Marine Environmental Sciences
and
Faculty of Geosciences, University of Bremen
9:15 - 9:45
1. Introduction (Lecture)
- The global carbon cycle, CO2 in seawater
- “Biological pumps”
- Reservoir or box models
2. Modeling Marine Nutrient and Carbon Cycles
(Box-Model Exercise)
- Global oceanic phosphate distribution
- Nutrient – productivity interactions
- Oceanic carbon budget and large-scale ocean
circulation
10:45 - 11:00 break
11:00 – 12:30
2. cont'd
- Circulation-productivity feedback in the global
ocean
3. State-of-the-art Biogeochemical Models
(Lecture)
- 2D and 3D Models
- Included tracers and processes
4. Marine Ecosystem Models (Lecture)
- Why ecosystem models?
- Ecosystem models in paleoceanography
Course Material
www.geo.uni-bremen.de/geomod/
staff/mschulz/lehre/ECOLMAS_Modeling/
This presentation
 Box-model exercises
Basic Literature
Najjar, R. G., Marine biogeochemistry. in Climate system modeling, edited by
Trenberth, K. E., pp. 241-280, Cambridge University Press, Cambridge,
1992.
Rodhe, H., Modeling biogeochemical cycles. in Global biogeochemical
cycles, edited by Butcher, S. S., R. J. Charlson, G. H. Orians and G. V.
Wolfe, pp. 55-72, Academic Press, London, 1992.
Sarmiento, J. L., and N. Gruber, Ocean biogeochemical dynamics, pp. 503,
Princeton University Press, Princeton, 2006.
Walker, J. C. G., Numerical adventures with geochemical cycles, 192 pp.,
Oxford University Press, New York, 1991.
For a climatologist
biogeochemical
cycles usually
translates into
carbon cycle.
Ruddiman (2001)
Carbon-Cycle – Characteristic Timescales
Reservoir Sizes in [Gt C]
Fluxes in [Gt C / yr]
Sundquist (1993, Science)
Average surfaceWater composition
CO2
0.5 %
HCO3- 89.0 %
CO32- 10.5 %
TA
CO2
[HCO3- ] + 2[CO32- ]
[HCO3- ] + [CO32- ]
Thurman & Trujillo (2002)
Biological Productivity in the Ocean
Nutrients:
P, N, (Si, Fe)
Ruddiman (2001)
The Biological Pump
CO2
Atmosphere
CO2
Primary Production
Inorgan. C
Organ.
C
Particle-Flux
Ocean
Remineralisation
Organ. C CO2
Sediments
Fig. courtesy of A. Körtzinger
Photic Zone
Aphotic Zone
Sediments
Biogenic Calcium Carbonate Production
Raises Dissolved CO2 Concentration
pH Reaction:
CO2 + H2O + CO  2HCO
23
3
(1) Biogenic carbonate uptake
(2) More bicarbonate
dissociates
(3) More CO2 is formed
The Calcium Carbonate Pump
CO2
Atmosphere
CO2
Biogenic CaCO3
Formation
3
Lysocline
Ocean
CaCO3 Dissolution
CO32-
Fig. courtesy of A. Körtzinger
Reservoir or Box Models
• Reservoir = an amount of material defined by
certain physical, chemical or biological
characteristics that, under the particular
consideration, can be regarded as homogeneous.
(Examples: CO2 in the atmosphere, Carbon in living organic matter in
the oceanic surface layer)
• Flux = the amount of material transferred from one
reservoir to another per unit time
Single Reservoir Case
Flux In
Reservoir
(mass M)
Flux Out
Basic Math of Box Models
(Rate of change of mass in reservoir) =
(Flux in) – (Flux out) + Sources – Sinks
dM
 Fi  Fo  SMS
dt
Or, for concentration (C [mol/m3]) and water flux (Q
[m3/s]):
dC
V
 Qi Ci  QoC  SMS
dt
Numerical Solution of Box-Model
Equations
dM M M t1  M t0


 Fi  Fo  SMS Solution by
finite-difference
dt
t
t1  t0
method
(approximation!)
Time (in steps of t)
Initial Condition
M t1  M t0  t ( Fi ,t0  Fo ,t0  SMSt0 )
M t2  M t1  t ( Fi ,t1  Fo ,t1  SMSt1 )
M tn1  M tn  t ( Fi ,tn  Fo ,tn  SMStn )
“Euler Method”
Numerical Solution of Box-Model
Equations
dM M M t1  M t0


 Fi  Fo  SMS Solution by
finite-difference
dt
t
t1  t0
method
(approximation!)
Initial Condition
M t1  M t0  t ( Fi ,t0  Fo ,t0  SMSt0 )
M t2  M t1  t ( Fi ,t1  Fo ,t1  SMSt1 )
M tn1  M tn  t ( Fi ,tn  Fo ,tn  SMStn )
Numerical Solution of Box-Model Equations
dM M M t1  M t0


 Fi  Fo  SMS Solution by
finite-difference
dt
t
t1  t0
method
(approximation!)
Time (in steps of t)
Initial Condition
M t1  M t0  t ( Fi ,t0  Fo ,t0  SMSt0 )
M t2  M t1  t ( Fi ,t1  Fo ,t1  SMSt1 )
M tn1  M tn  t ( Fi ,tn  Fo ,tn  SMStn )
“Euler Method”
Euler Method
M
M(tn+1)“Prediction”
Slope =
Fi(tn) - Fo(tn) + SMS(tn)
Error
True Value
M(tn)
t
tn
tn+1
t
Assumption: Slope at time tn remains constant throughout time interval t
Coupled Reservoirs
F12
Reservoir 2
(mass M2)
Reservoir 1
(mass M1)
F21
Principle of mass-conservation requires M1 + M2 = const.
Large-Scale Ocean Circulation
(after Broecker, 1991)
Box-Model of
Oceanic PO4 Distribution
Indo-Pacific
Southern Ocean
Atlantic
Surface
(0-100 m)
20 Sv
AABW_P
(20 Sv)
20
Sv
NADW
(10 Sv)
10 Sv
Deep
(> 100 m)
AABW_A
(4 Sv)
www.geo.uni-bremen.de/geomod/staff/mschulz/lehre/ECOLMAS_Modeling/
bm1_po4_only.gsp
Box-Model Experiment 1
• Vary the water transports and initial PO4
concentration and observe the final PO4
concentration and evolution (time series).
• Q1: How does the final PO4 distribution depend
on these settings?
• Q2: How do these settings affect the time it
takes to reach a steady state? (What
characterizes the steady state?)
Inducing PO4 Gradients –
Biological Productivity
• Assume an average export production of
12 g C/m2/yr
• With a “Redfield ratio” of C:P = 117:1
(molar ratio) and 1 mol C = 12 g C
 Corresponding biological PO4 fixation is
1/117 mol P/m2/yr
Box-Model of Oceanic PO4
Distribution with Productivity
Indo-Pacific
Southern Ocean
Atlantic
Surface
(0-100 m)
AABW_P
(20 Sv)
NADW
(10 Sv)
Deep
(> 100 m)
AABW_A
(4 Sv)
Assumption: Biologically fixed PO4 sinks from the surface layer to the
underlying deep layer, where the organic material is completely remineralized.
www.geo.uni-bremen.de/geomod/staff/mschulz/lehre/ECOLMAS_Modeling/
bm1_po4_fix_prod.gsp
Box-Model Experiment 2
• Q: How does the inclusion of biological productivity affect
the PO4-concentration difference between Atlantic and IndoPacific Oceans in the standard case?
10 m water depth
1750 m water depth
Box-Model Experiment 2
• Vary the water transports (try max. and small values)
and observe how the PO4 distribution changes. Explain
the changes.
• Q: What happens if NADW = 0 Sv? (Keep the
remaining parameters at their default values.) Does this
result make sense in the real world?
• Q: For which initial PO4 concentration do no negative
concentrations result (with NADW = 0 Sv)? Is this a
reasonable increase for Late Pleistocene glacials?
Avoiding Negative PO4 Concentrations –
Nutrient-Dependent Productivity
• Assume that productivity scales with the PO4
availability in the surface layer (variety of
relationships are possible: linear, non-linear with
saturation…)
• PO4 fixation = [PO4]sfc * Volsfc / t [mol/yr],
where t is the residence time of PO4 in the
surface due to biological productivity
• Assume tATL = tIPAC = 5 yr and tSOC = 50 yr (Broecker
and Peng, 1986)
www.geo.uni-bremen.de/geomod/staff/mschulz/lehre/ECOLMAS_Modeling/
bm1_po4_dyn_prod.gsp
Box-Model Experiment 3a
• Run the model for NADW of 0 and 10 Sv
and write down the PO4 concentrations for
the Atlantic boxes for each case.
• Calculate the difference between conc. in
deep and surface box. What do you
observe?
Box-Model Experiment 3a – Atlantic
NADW
PO4 Surface
PO4 Deep
PO4
(Sv)
(mmol/l)
(mmol/l)
(mmol/l)
10
0.24
0.69
0.45
0
0.18
0.88
0.70
Shift of PO4 content from surface
to deep Atlantic as NADW drops
Box-Model Experiment 3b
• Run the model for NADW = {0, 5, 10, 15,
20} Sv and write down the final PO4
fixation in the Atlantic Ocean.
• Sketch NADW vs. PO4 fixation.
• Q:What is the paleoceanographic
implication of this finding?
NADW and Productivity in the
Atlantic Ocean
3.6
PO4 Fixation [1011 mol P /yr]
3.4
3.2
3
2.8
2.6
2.4
2.2
2
0
5
10
NADW Flow [Sv]
15
20
Including the Marine Carbon-Cycle
• Tracers: PO4 ( controls productivity)
DIC (dissolved inorganic carbon)
ALK (alkalinity)
• Aqueous CO2 partial pressure = f(DIC, ALK)
• Redfield ratio of organic matter (C:N:P = 117:16:1)
• Ratio between Corg and CaCO3 production (“rain
ratio”)  assumed to be temperature dependent (a
crude parameterization of ecosystem dynamics)
Rain-Ratio Parameterization
0.16
Rain Ratio = PCaCO3 / PCorg
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0
5
10
15
20
Water Temperature [°C]
25
30
AreaWeighted
Average
Atmospheric
pCO2 ≈ Mean
Oceanic pCO2
www.geo.uni-bremen.de/geomod/staff/mschulz/lehre/ECOLMAS_Modeling/
bm1_c-cycle_fix_prod.gsp
Box-Model Experiment 4
C-Cycle with Fixed Productivity
• Run the model for the default setting. Identify the
sources and sinks with respect to atmospheric
CO2.
• Run the model for NADW of 0 and 10 Sv. Write
down the final global mean pCO2 and the
productivity in the Atlantic Ocean. (Neglect the
negative PO4 conc., identified in the previous exp.)
Box-Model Experiment 4
C-Cycle with Fixed Productivity
NADW
Prod. ATL
Prod. Glob.
Global pCO2
(Sv)
(Pg C/yr)
(Pg C/yr)
(ppm)
10
0.447
5.03
281
0
0.447
5.03
265
16 ppm
Reduction
Box-Model Experiment 5
C-Cycle with Dynamic Productivity
• How will the response of the mean pCO2
change if productivity is no longer constant
but a function of PO4?
www.geo.uni-bremen.de/geomod/staff/mschulz/lehre/ECOLMAS_Modeling/
bm1_c-cycle_dyn_prod.gsp
Box-Model Experiment 5
C-Cycle with Dynamic Productivity
• Run the model for again for NADW of 0
and 10 Sv. Write down the final global
mean pCO2 and the productivity in the
Atlantic Ocean.
• Interpret your results.
Box-Model Experiment 5
C-Cycle with Dynamic Productivity
NADW
Prod. ATL
Prod. Glob.
Global pCO2
(Sv)
(Pg C/yr)
(Pg C/yr)
(ppm)
10
0.447
5.03
281
0
0.350
4.83
275
Only 6 ppm
Reduction
Box-Model Experiment 5
C-Cycle with Dynamic Productivity
NADW = 0  DIC shifted from surface to deep
Atlantic  pCO2 reduced
BUT: PO4 is shifted to deep ocean too  less
nutrients in surface  productivity decreases 
biological pump weakens  pCO2 increases
 Negative Feedback Mechanism
From Box-Models to 2D/3D-Models
Ruddiman (2001)
Structure
of a
Global
Biogeochemical
Model
Ridgwell (2001, Thesis)
Modeling Deep-Sea
Sediments
Ridgwell (2001, Thesis)
Phosphate in the Atlantic Ocean [mmol/l]
2D-Model
(Zonal Mean)
(Schulz and Paul, 2004)
3D-Model
(N-S Section)
(Heinze et al., 1999)
Horizontal Resolution in a 2DBiogeochemical Model
Atlantic Ocean Export Production [gC/(m2 yr)]
90
80
70
60
50
40
30
20
10
0
-80
-60
-40
-20
0
20
Latitude
40
60
80
(Schulz and Paul, 2004)
Horizontal Resolution in a 3DBiogeochemical Model
(Heinze et al., 1999)
A Modeled Sediment Stack in the North Atlantic
Heinze, C. et al., 1999: A global oceanic sediment model for long-term climate
studies. Global Biogeochemical Cycles, 13, 221-250.
Modeled and Observed Modern CaCO3
Content of Deep-Sea Sediments
Model
Observations
 Even the most sophisticated biogeochemical models allow only for a crude
approximation of the real world. Discrepancies are largely due to an inadequate
resolution (e.g. MOR) and a lack of knowledge of the processes being involved.
Heinze et al. (1999)
Marine Ecosystem Models – Why?
• Productivity may depend on more than a single
nutrient (N, P, Si, Fe)
• Export production controlled by ecosystem
dynamics
• Understanding the preferential growth of
different algae groups (e.g. diatoms vs.
coccolithophores)
• Disentangling the seasonal imprint in biological
proxy records
NPZD-Type Ecosystem Model
• 4 Compartments
• Coupled to carbon and
alkalinity
• Nutrients are
transported by ocean
circulation
• Efficient in predicting
seasonal patterns
(after Fasham et al., 1990)
Marine Ecosystem Model Components (Moore et al., 2002)
Marine Ecosystem Model Forcing
Output from global OGCM
Global Foraminifera Model
Fraile et al. (subm.)
Fraile et al. (subm.)
Modeled / Observerd Distribution of N. pachyderma (sin.)
Fraile et al. (subm.)
Brown University Foraminiferal
Database (Prell et al., 1999)
Modeled / Observerd Distribution of N. pachyderma (dex.)
Fraile et al. (subm.)
Brown University Foraminiferal
Database (Prell et al., 1999)
Modeled / Observerd Distribution of G. bulloides
Fraile et al. (subm.)
Brown University Foraminiferal
Database (Prell et al., 1999)
Modeled / Observerd Distribution of G. ruber (white)
Fraile et al. (subm.)
Brown University Foraminiferal
Database (Prell et al., 1999)
Modeled / Observerd Distribution of G. sacculifer
Fraile et al. (subm.)
Brown University Foraminiferal
Database (Prell et al., 1999)
Modeled LGM shift in seasonality of G. bulloides
Fraile et al. (subm.)
Benefits of Paleoecosystem Modeling
• To facilitate model-“data“ comparison
• To obtain a mechanistic understanding of
reconstructed shifts in species
• To assess the potential effect of altered plankton
successions on proxy reconstructions based on
organisms