Biogeochemical modelling Corinne Le Qur University of East Anglia and the British Antarctic Survey

1
Biogeochemical modellingCorinne Le
QuéréUniversity of East Anglia and the British
Antarctic Survey
2
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

3
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

4

• project the future
• test hypothesis (e.g. CLAW)
• quantify feedbacks
• formalize your ideas

5
SOLAS Science
6
(No Transcript)
7
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

8
SOLAS Science
9
Parameterisation of chemical processes are
0-Dimensional

• known processes
• measured species
• derived rates

10
Typical chemical processes in the atmosphere

• ozone
• NOx
• hydrocarbon (Volatile Organic Carbon)
• OH-
• aerosols
• CO, CH4

11
NOx AND VOC processes (D. Jacobs)
2 km
hn (420 nm)
hn (340 nm)
BOUNDARY LAYER
NO2
NO
HCHO
OH
CO
hours
O3, RO2
hours
VOC
1 day
HNO3
Emission
Emission
Deposition
VOLATILE ORGANIC COMPOUNDS (VOC)
NITROGEN OXIDES (NOx)
12
Tropospheric ozone processes (D. Jacobs)
O2
hn
O3
STRATOSPHERE
8-18 km
TROPOSPHERE
hn
NO2
NO
O3
hn, H2O
H2O2
OH
HO2
Deposition
CO, VOC
13
!
! ! The decay
for CH4 is calculated by ! OH CH4 -gt
CH3 H2O ! k 2.45E-12 exp(-1775/T)
! ! This is from JPL '97. JPL '00 does
not revise '97 value. (jsw)
!
DO L 1, MAXVAL( LPAUSE
) DO J 1, JJPAR DO I 1, IIPAR
! Only consider tropospheric boxes
IF ( L lt LPAUSE(I,J) ) THEN
!jsw Is it all right that I'm using
! 24-hr avg temperature to calc. rate
coeff.? KRATE 2.45d-12 EXP(
-1775d0 / Tavg(I,J,L) ) ! Conversion
from kg/box --gt molec/cm3 ! kg
CH4/box box/cm3 XNUMOL_CH4 molec CH4/kg
CH4 STT2GCH4 1d0 / AIRVOL(I,J,L) /
1d6 XNUMOL_CH4 ! CH4 in
molec/cm3 GCH4 STT(I,J,L,1)
STT2GCH4 ! Sum loss in TCH4(3)
(molecules/box) TCH4(I,J,L,3)
TCH4(I,J,L,3) ( GCH4
BOXVL(I,J,L) KRATE BOH(I,J,L) DT )
! Calculate new CH4 value CH4CH4(1-kOH
delt) GCH4 GCH4 ( 1d0 - KRATE
BOH(I,J,L) DT ) ! Convert back
from molec/cm3 --gt kg/box
STT(I,J,L,1) GCH4 / STT2GCH4 ENDIF
ENDDO ENDDO ENDDO
example of model code from GEOS-CHEM
14
Typical chemical processes in the ocean

• C cycle (CO2, CO32-,HCO3-,CaCO3,H2CO3)
• pH
• Si cycle (SiO2 to Si(OH)4-)
• Fe cycle (Fe3 to Fe2)
• photochemistry (degration of Organic C by light)

15
The Fe cycle in the oceans
growth
dissolved, colloidal
P, B
Fe3
L
h?
Fe(III)L
h?
pFe
organic or inorganic
Fe2
dissolved Fe
sedimentation
hµ photoreduction
Coagulation Dissociation
16
carbon cycle
numbers in PgC/yr
atmosphere
CO2
90
chemical reactions
ocean
17
! Set volumetric solubility constants for co2 in
mol/latm (Weiss, 1974) ! ------------------------
--------------------------------------------------
---------- ! c00 -58.0931 c01
90.5069 c02 22.2940 c03
0.027766 c04 -0.025888 c05
0.0050578 ! ! ln(k0) of solubility of co2 (eq.
12, Weiss, 1974) ! -------------------------------
-------------------------- ! cek0
c00c01/qttc02zqttsal(c03c04qttc05qtt2)
ak0 exp(cek0) smicr ! ! this is
Wanninkhof (1992) equation 8 (with chemical
enhancement), in cm/h ! --------------------------
----------------------------------------------- !
kgwanin(ji,jj) (0.3wsws
2.5(0.5246ttc(0.016256ttc0.00049946))) ! !
convert from cm/h to m/s and apply ice cover !
-------------------------------------------- !
kgwanin(ji,jj) kgwanin(ji,jj)
/100./3600. (1-freeze(ji,jj)) ! Set Schmit
constants ! --------------------------------------
------------------------------------
schmico2 2073.1-125.62ttc3.6276ttc2-0.04312
6ttc3 ! ! compute gas exchange kg in
mol/m2/yr/uatm ! ---------------------------------
-----------------------------------------
gasex kgwanin (660/schmico2)0.5 kg
gasex ak0 1.e3 (3600.24.365.25)
example of model code for CO2 gas exchange
formulation
18
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

19
SOLAS Science
20
Typical biological processes in the ocean

• phytoplankton growth
• zooplankton grazing
• bacterial remineralisation
• particulate dynamics

21
Parameterisation of biological processes are
1-Dimensional

• poorly known processes
• some measured rates
• vertical transport of particles

22
carbon cycle
numbers in PgC/yr
atmosphere
CO2
90
chemical reactions
45
34
ocean
23
biological activity
atmosphere
surface
100 m
mixed layer depth
24
biological activity
atmosphere
100 m
real surface
25
what they do
pico-autotrophs
N2-fixers
phyto-plankton
calcifiers
DMS-producers
mixed
silicifiers
26
what they do
pico-autotrophs
N2-fixers
these bloom
phyto-plankton
calcifiers
DMS-producers
mixed
silicifiers
27
what they do
pico-autotrophs
N2-fixers
these form shells
phyto-plankton
calcifiers
DMS-producers
mixed
silicifiers
28
what they do
pico-autotrophs
N2-fixers
these respond to pH
phyto-plankton
calcifiers
DMS-producers
mixed
silicifiers
29
what they do
pico-autotrophs
N2-fixers
these float
phyto-plankton
calcifiers
DMS-producers
mixed
silicifiers
30
what they need
pico-autotrophs
N2-fixers
phyto-plankton
calcifiers
DMS-producers
mixed
silicifiers
31
(No Transcript)
32
what they do
33
what they do
pico-heterotrophs
bacteria
these control blooms
proto
zoo-plankton
meso
macro
34
what they do
pico-heterotrophs
bacteria
these produce big feacal pellets
proto
zoo-plankton
meso
macro
35
what they need
pico-heterotrophs
bacteria
proto
zoo-plankton
meso
macro
36
time scale
37
phytoplankton growth
38
phytoplankton growth
zooplankton growth
Buitenhuis et al., 2006
39
Modelling strategy

• diagnostic models (Najjar et al., 1992 OCMIP2
  1998-200)

• biogeochemical models (Maier-Reimer et al.,
  1990-1993)

• ecosystem models (Fasham et al., 1993)

Nutrient Phytoplankton Zooplankton Detritus (NPZD)
Dynamic Green Ocean Models (DGOM)
40
! ! Evolution of Mesozooplankton !
------------------------ !
trn(ji,jj,jk,jpmes) trn(ji,jj,jk,jpmes)
mesoge(ji,jj,jk)gramet(ji,jj,jk)
-tortz2(ji,jj,jk)-respz2(ji,jj,jk) !
! Evolution of DOC ! ---------------- !
trn(ji,jj,jk,jpdoc) trn(ji,jj,jk,jpdoc)
rn_sigpocorem(ji,j
j,jk)-olimi(ji,jj,jk)
grarem(ji,jj,jk)(1.-rn_sigmic)grarem2(ji,jj,jk)
(1.-rn_sigmes)-xaggdoc(ji,jj,jk
)-xaggdoc2(ji,jj,jk)
depdoc(ji,jj,jk) ! ! Evolution of POC !
--------------------------------------------------
---------------- ! trn(ji,jj,jk,jpgoc)
trn(ji,jj,jk,jpgoc)
grapoc2(ji,jj,jk)resphy(ji,jj,jk,jpdia,
1)xagg(ji,jj,jk)
tortz2(ji,jj,jk)-orem2(ji,jj,jk)-grazgoc(ji,jj,jk
) xaggdoc2(ji,jj,jk)

(sinking2(ji,jj,jk)-sinking2(ji,jj,jk1))/e3t_0(j
k) ! ! Evolution of dissolved IRON !
--------------------------------------------------
---------------- ! trn(ji,jj,jk,jpfer)
trn(ji,jj,jk,jpfer)-
xbactfer(ji,jj,jk)ferat3(
respz2(ji,jj,jk)respz(ji,jj,jk)
)grafer(ji,jj,jk)
grafer2(ji,jj,jk)ofer(ji,jj,jk)
(1.-rn_siggoc)ofer2(ji,jj,jk)

-xscave(ji,jj,jk)irondep(ji,jj,jk)
depfer(ji,jj,jk)-xaggdfe(ji,jj,j
k) !
example of model code from PlankTOM ecosystem
model
41
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

42
SOLAS Science
43
Typical physical processes in the atmosphere and
ocean

• advection
• diffusion
• mixing
• convection

44
Parameterisation of physical processes are
3-Dimensional

• well known processes with physical equations
• difficult to represent because of size of grid
• sub-grid scale parameterisations developed and
  tuned to give reasonable physical transport

45
convection and horizontal advection
46
vertical advection
47
Eddies and mixing
48
! Horizontal advective fluxes !
----------------------------- !
!
DO jk 1, jpkm1
! Horizontal slab !
!
DO jj 1, jpjm1
DO ji 1, fs_jpim1 ! vector opt.
! upstream indicator zcofi
MAX( zind(ji1,jj,jk), zind(ji,jj,jk) )
zcofj MAX( zind(ji,jj1,jk),
zind(ji,jj,jk) ) ! volume fluxes
1/2 zfui 0.5 e2u(ji,jj) pun(ji,jj,jk)
zfvj 0.5 e1v(ji,jj)
pvn(ji,jj,jk) ! centered scheme
zcenut zfui ( tn(ji,jj,jk) tn(ji1,jj ,jk)
) zcenvt zfvj ( tn(ji,jj,jk)
tn(ji ,jj1,jk) ) zcenus zfui
( sn(ji,jj,jk) sn(ji1,jj ,jk) )
zcenvs zfvj ( sn(ji,jj,jk) sn(ji
,jj1,jk) ) END DO END DO
! Tracer flux divergence at t-point added
to the general trend !
--------------------------------------------------
------------ DO jj 2, jpjm1
DO ji fs_2, fs_jpim1 ! vector opt. zbtr
btr2(ji,jj) ! horizontal advective trends
zta - zbtr ( zwx(ji,jj,jk) -
zwx(ji-1,jj ,jk)
zwy(ji,jj,jk) - zwy(ji ,jj-1,jk) )
zsa - zbtr ( zww(ji,jj,jk) -
zww(ji-1,jj ,jk)
zwz(ji,jj,jk) - zwz(ji ,jj-1,jk) )
! add it to the general tracer trends
ta(ji,jj,jk) ta(ji,jj,jk) zta
sa(ji,jj,jk) sa(ji,jj,jk) zsa
END DO END DO !

example of model code from NEMO ocean physical
model
49
carbon cycle
numbers in PgC/yr
atmosphere
CO2
90
chemical reactions
45
34
ocean
50
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

51

• Example benchmark for marine carbon cycle model
• CO2 sink in 1990 between 1.8-2.6 PgC/y
• export of carbon between 9-12 PgC/y
• primary production between 40-70 PgC/y
• CO2 variability in equatorial Pacific between
  0.6-1.0 PgC
• mezo-zooplankton grazing ltlt micro-zooplankton
  grazing
• all phytoplankton biomass gt 0.02 PgC
• no phytoplankton biomass dominate globally

52
1.e-5

• Example benchmark for marine carbon cycle model
• CO2 sink in 1990 between 1.8-2.6 PgC/y
• export of carbon between 9-12 PgC/y
• primary production between 40-70 PgC/y
• CO2 variability in equatorial Pacific between
  0.6-1.0 PgC
• mezo-zooplankton grazing ltlt micro-zooplankton
  grazing
• all phytoplankton biomass gt 0.02 PgC
• no phytoplankton biomass dominate globally

53
Carbon-cycle model intercomparison Project (OCMIP)
visual evaluation of model results
54
Carbon-cycle model intercomparison Project (OCMIP)
formal evaluation of model results using a Taylor
diagram
55
Cost functions
N Number of Observations D Observational
Data sD Standard deviation Data CF lt 1 very
good,12 good, 25 reasonable,gt5 poor
OSPAR Commission (1998).CF lt 1 very good, 12
good, 23 reasonable, gt3 poor Radach and
Moll (2006).
examples ERSEM Courtesy of I.Allen
56
Model efficiency
D Observational Data D_bar Mean of Data M
Model Results
57
Model Bias
M Model Results D Observational Data
58
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

59
carbon cycle
numbers in PgC/yr
atmosphere
CO2
90
chemical reactions
45
34
ocean
60
energy
water
winds
observed warming trend 1979-2005
Smith and Reynolds 2005 and IPCC 2007
61
biological activity
chemical reactions

ocean
62
sea-air CO2 flux anomaly
63
OPA model

• PISCES-T ecosystem model
• 2 phyto, 2 zoo., 2 sinking particles
• limitation by Fe, P, and Si
• initialise with observations in 1948
• (Buitenhuis et al., GBC 2006)

• OPA General Circulation model
• 0.5-1.5ox2o resolution
• 31 vertical levels
• calculated vertical mixing
• NCEP daily forcing

64
OPA model

• PISCES-T ecosystem model
• 2 phyto, 2 zoo., 2 sinking particles
• limitation by Fe, P, and Si
• initialise with observations in 1948
• (Buitenhuis et al., GBC 2006)

• OPA General Circulation model
• 0.5-1.5ox2o resolution
• 31 vertical levels
• calculated vertical mixing
• NCEP daily forcing for year 1967

65
Change in Southern Ocean CO2 sink in model
real forcing
66
Change in Southern Ocean CO2 sink in model
real forcing
changes in winds
1967 forcing
67
Outline of lecture

• Introduction
• Chemical processes
• Biological processes
• Physical processes
• Model evaluation and benchmarking
• One example (ocean CO2 sink)
• The modellers psychology

68
The modellers psychology
truth
time
69
truth
illusion (everybody is happy)
time
70
truth
illusion (everybody is happy)
time
71
truth
illusion (everybody is happy)
chaos (everybody is happy)
time
72
truth
relief (need a new job)
illusion (everybody is happy)
chaos (everybody is happy)
time
73
climate models
truth
relief (need a new job)
illusion (everybody is happy)
chaos (everybody is happy)
time
74
Putting it all together

• do your best, but simplify to answer your
  question
• use benchmarking to
• i) validate, and
• ii) follow improvements in your model
• EVERYTHING must make sense

75
(No Transcript)