1. Paul Palmer, University of Leeds
Space-based measurements of Carbon Compounds
Paul Palmer, University of Leeds

2. Magnitude of VOC emissions to C budget?61 60
5.5
1.6
+ ballpark flux estimates for fast exchange processes (109 tonnes C)
What are the relative roles of the oceans and land ecosystems in absorbing CO2?
Is there a northern hemisphere land sink?
What are the relative roles of North America and Eurasia
What controls carbon sinks? How will sinks respond to climate change?
Magnitude of VOC emissions to C budget?
IPCC

3. Carbon-containing compounds from space: platforms, instruments, species
ERS-2
Terra
Envisat
Aqua
Aura
???
Launch
1995
1999
2002
2004
2008
Sensor
GOME
MOPITT
SCIAMACHY
AIRS
TES
OMI
OCO
GOSAT
CO2 X CO
HCHO
CH4

4. Horizontal spatial scales
Developing a self-consistent view of the carbon cycle
O(1 km)
O(10s km)
O(10-100s km)
Top-Down
Horizontal spatial scales
Models
Bottom-Up

5. Global Ozone Monitoring Experiment (GOME) & the Ozone Monitoring Instrument (OMI)
Launched in 2004
GOME (European), OMI (Finnish/USA) are nadir SBUV instruments
Ground pixel (nadir): 320 x 40 km2 (GOME), 13 x 24 km2 (OMI)
10.30 desc (GOME), asc (OMI) cross-equator time
GOME: 3 viewing angles  global coverage within 3 days
OMI: 60 across-track pixels  daily global coverage
O3, NO2, BrO, OClO, SO2, HCHO, H2O, cloud properties

6. GOME HCHO columns July 2001 Biogenic emissions Biomass burning
[1016 molec cm-2] 1 2
0.5
1.5
2.5
* Columns fitted: nm * Fitting uncertainty < continental signals

7. Tropospheric O3 is an important climate forcing agentNO
HO2 OH
NO2 O3 hv
HC+OH  HCHO + products
NOx, HC, CO
Level of Scientific Understanding
Natural VOC emissions (50% isoprene) ~ CH4 emissions.
IPCC, 2001

8. May
Jun
Jul
Aug
Sep
1996
1997
1998
1999
2000
2001
GOME HCHO column [1016 molec cm-2] 1 2
0.5
1.5
2.5

9. Relating HCHO Columns to VOC Emissions
hours OH
h, OH
VOC
kHCHO
EVOC =
(kVOCYVOCHCHO)
HCHO
___________
Local linear relationship between HCHO and E
VOC source
Distance downwind
WHCHO
Isoprene
a-pinene
propane
100 km
EVOC: HCHO from GEOS-CHEM and MCM models

10. Isoprene emissions July 1996
GEIA
EPA BEIS2
7.1 Tg C
2.6 Tg C
MEGAN
3.6 Tg C
[1012 atom C cm-2 s-1]

11. Modeling biospheric VOC emissions
PAR – direct and diffuse (GMAO)
Temperature:
Instantaneous (G95)
10-day avg (Petron ‘01)
Fixed base emission factors (Guenther 2004)
Canopy model (Guenther 1995)
Altitude
Isoprene emissions do not generally begin until the forest canopy has reached a mature state, 2 weeks after leaves emerge and photosynthesis had begun.
Emission
April
Sep
LAI
MODIS/AVHRR LAI
Emissions

12. Seasonal Variation of Y2001 Isoprene Emissions
MEGAN
GOME
MEGAN
GOME
May
Jun
Aug
Sep
Jul
3.5 7
1012 atom C cm-2s-1
Good accord for seasonal variation, regional distribution of emissions (differences in hot spot locations – implications for O3 prod/loss).
Other biogenic VOCs play a small role in GOME interpretation

13. GOME Isoprene Emissions: 1996-2001
May
Jun
Jul
Aug
Sep
1996
1997
1998
1999
2000
2001
[1012 molecules cm-2s-1] 5 10
Relatively inactive

14. Surface temperature explains 80% of GOME-observed variation in HCHO
NCEP Surface Temperature [K]
GOME HCHO Slant Column [1016 molec cm-2]
Curves fitted to GOME data
Modeled curves
Time to revise model parameterizations of isoprene emissions?

15. Higher spatial resolution with OMI
13x24 km2 (OMI) vs 40x320 km2 (GOME) at nadir
Biogenic emissions – isoprene, monoterpenes, mbo
10^12 molec emissions flux (mol/cm2/s)
Correlation( r )
0.63
0.59
0.44
0.46
0.27
North America, July 2005
Africa
24 Sept – 19 Oct 2004
MODIS Fire Counts

16. The Orbiting Carbon Observatory (OCO)
“First global space-based measurements of CO2 with the precision and spatial resolution needed to quantify carbon sources and sinks”
Launch in 2008
Spectroscopic observations of CO2 and O2 to estimate the column integrated CO2 dry air mole fraction,
XCO2 = x (column CO2) / (column O2)
Precisions of 1 ppm on regional scales
Global coverage in 16 days
JPL-based instrument: PI D. Crisp; Deputy PI: C. Miller (Crisp et al, 2004)

17. Monitoring CO2 from Space
High resolution (λ/Δλ=17.5k-21k) spectra of reflected sunlight in near IR:
CO m and 2.06 m bands
O m A-band
Retrieval of column average CO2 dry air mole fraction, XCO2 with BL sensitivity
(Also need Ps, albedo, T, water vapour, clouds, and aerosols, provided by OCO)
~0.3% (1ppm) precision
Nadir Mode
Target
Mode
Glint Mode
Why high spectral resolution? Lines must be resolved from the continuum to minimize systematic errors
CO m band is well suited for retrieving CO2 column abundances
CO m band CO2, cloud, aerosol, water vapor, temperature
O2 A-band constrains clouds, aerosols, and surface pressure
OCO samples at high spatial resolution Nadir mode: 1 km x 1.5 km footprint Isolates cloud-free scenes
Thousand of samples on regional scales
Glint Mode: High SNR over oceans
Target modes: Calibration

18. Assessing OCO Performance with OSSEs
CO2 fluxes
XCO2 map
Retrieval
End-to-end retrievals of XCO2 from individual simulated nadir soundings at SZAs of 35o and 75o. The simulations include sub-visual cirrus clouds (0.02c 0.05), light to moderate aerosol loadings (0.05a 0.15), over ocean and land surfaces.
INSET: Distribution of XCO2 errors (ppm) for each case
XCO2 along OCO orbits
CO2, O2 spectral radiances
Kuang et al, 2002
(250 Gb/day)

19. Coordinated calibration/validation activities in the A-train
PARASOL – polarization data
AIRS – T, P, H2O, CO2, CH4
MODIS – cloud/aerosols, albedo
CloudSat – cloud climatology
CALIPSO – vertical profiles of cloud & aerosol; cirrus particle size
**OCO – O2 A-Band Spectra**
OCO – XCO2, P(surface), T, H2O, cloud, aerosol
TES – T, P, H2O, O3, CH4, CO
MLS – O3, H2O, CO
HIRDLS – T, O3, H2O, CO2, CH4
OMI – O3, aerosol climatology, HCHO
Also exploit physically based correlations between measurements of different carbon compounds, e.g., CO and CO2

20. Coupled inversions exploit correlations
REGIONAL CO2:CO CORRELATIONS PROVIDE UNIQUE INFORMATION ON SOURCE REGION AND TYPE
Offshore China
Over Japan
A priori bottom-up
CO2
CO2 CO CO
Slope (> 840 mb) = 51
R2 = 0.76
Slope (> 840 mb) = 22
R2 = 0.45
Top-down information from TRACE-P
Suntharalingam et al, 2004
Coupled inversions exploit correlations

21. Use CO-CO2 correlations to “disentangle” biospheric and combustion CO2 source
State vector (Emissions x)
Consistent CO and CO2 Emissions
Jacobian describes CTM
y = Kxa + 
Forward model
(GEOS-CHEM)
Observation vector y
Inverse model
x = Fluxes of CO and CO2 from Asia (Tg C/yr)
y = TRACE-P CO and CO2 concentration data
x = xa + (KTSy-1K + Sa-1)-1 KTSy-1(y – Kxa) ^

22. Quantification of CO:CO2 Correlations
Colocated Emissions
Atm. Dyn. Processes
Emissions (g gas yr-1)
Emission Factor (g gas/g fuel)
cold air
warm air
E = A F
Coal-burning cook stoves in Xian, China
Activity Rate (g fuel yr-1)
E.g., frontal system
Use A, F, A and F for CO and CO2 from Streets et al, 2003 in a Monte Carlo approach. Correlations <0.1.
Correlations implemented in different ways: Sa and Sy
[CO]:[CO2] correlations from TRACE-P are typically Sa Sy

23. CO:CO2 correlations improve CO2 inversions
China Anth. CO
China Anth. CO2
China Bio. CO2
Tg /Mar 2001
Tg yr -1
Tg yr -1
Sa CO2:CO Correlation
Tg /Mar 2001
Tg yr -1
Tg yr -1
Sy CO2:CO Correlation
A posteriori Source Estimate A posteriori Source Uncertainty
Palmer et al, 2005

24. Closing Remarks Data poor to data rich overnight!
Cal/Val of satellite and in situ data still necessary.
How do we use satellite data to learn about the carbon cycle?
What can we learn?