A joint CO-CO2 inversion system for studying
the effect of drought on the carbon cycle
NARCISA NECHITA-BANDA, MAARTEN KROL, SOURISH BASU,
THOMAS RÖCKMANN
Motivation
- 122.0 Pg C/yr
115.8 Pg C/yr
2.0 Pg C/yr
- 4.3 Pg C/yr
Photosynthesis
Biosphere flux
Respiration
Biomass burning
Net terrestrial flux
Can we use satellite data to quantify these contributions to the CO2 budget?
Uptake
Correlation between global biomass burning emissions and biosphere flux
­> both climate driven
Emission
Motivation
CO for constraining biomass burning
o well known tracer for biomass
burning
o lifetime of weeks to months
o data available from ground
stations
o satellite data:
o
o
o
o
o
IASI TIR since 2006
MOPPIT TIR/NIR since 2000
TES TIR since 2004
SCHIAMACHI NIR 2002-2012
TROPOMI to be launched in 2017
CO, CO2 emission factors
Biomass burning CO, CO2 emission
factors are driven by combustion
efficiency, and vary with:
o Biome
o Temperature, precipitation, wind
GFED4 emission factors
Biome
g CO / g CO2
Tropical forest
0.056
Temperate forest
0.053
Boreal forest
0.085
Peat
0.123
Gras
0.037
Agricultural waste
0.065
CO+OH → CO2
~1 Pg C from secondary production
CO2
EDGAR, CDIAC, SiBCASA - ‘Total Carbon’
CO2 only fluxes by subtracting CO,
CH4, NMVOC emissions, taking into
account yields
CO
-> CO2 (95%)
CH4
-> CO (90%)
-> CO2
NMVOC
-> CO (~73%)
(10-80%)
-> CO2
GFED4 biomass burning – CO2 only
emissions
Effect on modelled CO2 concentrations
OSSE setup
Assume ‘true emissions’ - From inventories: (GFED4, SiBCASA,
EDGAR)
Run forward (true concentrations)
Synthetic observations
?
Prior emission
TM5-4DVAR
Posterior emission
OSSE setup
Runs on global 6∘x4∘, one year (2009)
Emission resolution – monthly
Run
Observations
Optimized fluxes
COCO2
CO, CO2 surface from COCO2 model run
CO biomass burning (-> CO2 biomass
burning)
CO2 biosphere
CO
CO surface from CO-CO2
model run
CO biomass burning
CO2C
CO2 surface from CO2
model run
C terrestrial (biosphere + biomass
burning)
Comparison to observations
Prior – Obs
Posterior – Obs
Comparison to observations
Posterior emissions
Optimizing emission factors
CO bb emis = DM emis x CO EF
CO2 bb emis = DM emis x CO2 EF
Optimize CO EF per biome (7 parameters)
Assume CO EF + CO2 EF is constant (in terms of g C / kg DM)
=> CO EF, CO2 EF per grid cell (GFED4 biome partitioning)
Optimizing emission factors
Optimize only EFs (true DM, emis)
Optimize EFs, DM, and CO2 bio
Conclusions & outlook
We have a coupled CO-CO2 inverse modelling system, which can constrain
biomass burning and biosphere fluxes separately
The coupled system simulates CO2 production in the atmosphere from
oxidation of other compounds (1 PgC)
Similar performance as the CO and CO2 inversions for individual species
We can optimize emission factors per biome - not feasible with surface
emissions alone
Next steps:
o Quantify uncertainties in posterior fluxes
o Optimize emission factors using satellite data
o Applying the coupled system on biomass burning events using real data
(e.g. Indonesian fires 2015)