CO2 Mixing Ratio Comparisons

• Comparison of model using 3-hourly fluxes shows significant skill insimulating observed variability over range of time scales at altitudes of 30 to396 m. Lack of vertical resolution limits ability to resolve very high values inshallow nighttime boundary layer at 11 m (not shown).

• Monthly average fluxes (not shown) capture little of the sub-seasonalvariability at this continental site in summer.

Flux Mismatch

• Examination of flux time series for 2000 shows model under-predicts CO2surface uptake in spring and overestimates uptake in peak of growing season(Jul-Aug) at LEF but PAR and T differences are not apparently responsible.

• Average flux comparison shows spring (summer) under- (over-) estimate ofvegetation uptake appears to be systematic for this site. Note also that CASAis nearly balanced but observations indicate residual CO2 source toatmosphere. Comparisons vary at other sites, e.g., Harvard Forest relativebehavior is nearly the converse of LEF.

CO2 Uptake

• Comparison of CO2 mixing ratios at LEF shows mismatches correspondingto flux differences seen at left. This suggests that observed LEF local flux isrepresentative of larger region and that model flux response needsadjustment regionally.

CO2 Surface Fluxes

• Monthly global biosphere fluxes at 1x1º for 1998-2004 generated fromCASA using monthly mean GEOS-4 analyzed meteorology (Tsfc,shortwave solar radiation, and precipitation) and monthly GIMMS NDVI.• 3-hourly fluxes produced using 3-h analyzed radiation and temperature inthe method of Olsen and Randerson, JGR, [2004]:

• Fossil Fuel emission fluxes mapped for 1998 at 1 x 1º from CDIACcountry data.• Ocean fluxes based onhttp://www.ldeo.columbia.edu/res/pi/CO2/carbondioxide/air_sea_flux/fluxdata.txt

Abstract and Introduction

Progress in better determining CO2 sources and sinks will almost certainlyrely on utilization of more extensive and intensive CO2 and relatedobservations including those from satellite remote sensing. Use ofadvanced data requires improved modeling and analysis capability. UnderNASA Carbon Cycle Science support we seek to develop and integrateimproved formulations for 1) atmospheric transport, 2) terrestrial uptake andrelease, 3) biomass and 4) fossil fuel burning, and 5) observational dataanalysis including inverse calculations. The transport modeling is based onmeteorological data assimilation analysis from the Goddard Modeling andAssimilation Office. Use of assimilated met data enables model comparisonto CO2 and other observations across a wide range of scales of variability.In this presentation we focus on the short end of the temporal variabilityspectrum: hourly to synoptic to seasonal. Using biospheric CO2 fluxes at 3-hourly temporal resolution from the CASA model, we examine the models’ability to simulate CO2 variability in comparison to observations at differenttimes, locations, and altitudes. The influence of key processrepresentations is inferred. The high degree of fidelity in these simulationsleads us to anticipate incorporation of real-time, highly resolvedobservations into a global multiscale carbon cycle analysis system that willbegin to bridge the gap between top-down and bottom-up flux estimation,which is a primary focus of NACP.

Wintertime CO2

• Model continues to perform well into fall and winter season when respirationand fossil fuel fluxes dominate this site. Similar behavior is seen at ARM andother sites. Note that monthly-resolved fluxes perform just as well as hourlyduring this season.

Frontal Passage

• Model simulates passage of frontal systems, which separate airmasses ofwidely varying CO2 mixing ratio, in close correspondence to CMDLobservations at Texas tower site (WKT). Synoptic wind fields showadvection from regions of varying characteristic CO2 surface flux. Front isseen at all vertical levels on tower. Many such examples are found in thedata from various sites.

Progress in Modeling Global Atmospheric CO2 Fluxes and TransportS. R. Kawa1, G. J. Collatz1, A. S. Denning2, D. J. Erickson3

1NASA Goddard Space Flight Center2Colorado State University Department of Atmospheric Science

3Oak Ridge National Laboratory

Toward Carbon Data Assimilation

The results of these modeling studies will help to prepare for the use of satelliteCO2 and other data in a multi-disciplinary carbon data assimilation system foranalysis and prediction of carbon cycle changes and carbon/climate interactions.The high degree of fidelity in these simulations leads us to anticipate incorporationof real-time, highly resolved observations into a system that will reduce uncertaintyin the terrestrial CO2 sink and lead toward credible, tested predictive models ofclimate and carbon needed for informed policy decisions. Analysis is ongoing. Wewelcome collaboration and comparison with data and other models on the NACPcontinental scales to better connect bottom-up and top-down processcharacterization globally and reduce uncertainty in global carbon source/sinkinferences.

Author Contact Information:S. R. Kawa (kawa@maia.gsfc.nasa.gov),NASA GSFC, Code 613.3,Greenbelt, MD, 20771

Parameterized Chemistry/Transport Model (PCTM)

• Meteorological fields from the Goddard Global Modeling and AssimilationOffice (GMAO), version GEOS-4.

- 3-hour averages from analysis used in off-line transport- Flux-form semi-Lagrangian transport algorithm [Lin and Rood, Mon.Weather Rev., 1996]- Model Grid: 2º x 2.5º x 28 levels to 0.4 mbar, hybrid terrain-followingcoordinate- Parameterized convective and PBL diffusive transport in troposphere- Global fields output hourly, plus interpolation to selected sitelocations- Runs done for year 1998-2004

• Model evaluation using climatological CO2 fluxes in Kawa et al., JGR,[2004] and Bian et al., Tellus, [2006].

BOTTOM-UP CO2 FLUXES ATMOSPHERIC CO2 VARIABILITY

K.5

CO2 SYNOPTIC EVENTS

AcknowledgementsNOAA ESRL Carbon Cycle Greenhouse Gases GroupS. Pawson, Z. Zhu (NASA Goddard Space Flight Center)I. Baker, (Colorado State University)M. Butler (Penn State U)J. T. Randerson (University of California, Irvine)

NASA Carbon Cycle Science

CASA Flux Evaluation

Time series comparison of flux model output and forcing data with flux tower dataat CMDL WLEF site show general good agreement for seasonal to dailyvariations.

TRANSPORT MODEL AND FLUXES

-30 -15 0 15 30-30

-15

0

15

30

Observations

WLEF Mean Daily Air Temperature

r=0.80

0 10 20 30 40 50 600

10

20

30

40

50

60

70

Observations

WLEF Daily PAR

r=0.57

-6 -3 0 3 6-6

-3

0

3

6

Observations

WLEF Daily NEP

r=0.244

1998 1999 2000 2001 2002-5

-4

-3

-2

-1

0

1

2

3

4

5WLEF Daily NEP

Observations CASA

1998 1999 2000 2001 20020

20

40

60

80WLEF Daily PAR

1998 1999 2000 2001 2002-30

-20

-10

0

10

20

30WLEF Daily Temperature

2000.000 2000.083 2000.167 2000.250 2000.333 2000.417 2000.500 2000.584 2000.667 2000.750 2000.834 2000.917 2001.000-5

-4

-3

-2

-1

0

1

2

3

4

5WLEF Daily NEP

Observations CASA

!

RScale3hr =

R

R

month

"

TScale3hr =

Q10

Q10

month

"

GPP3hr = RScale

3hr # (2 # NPPmonth )

RH3hr = TScale

3hr # (2 # NPPmonth ) $ NEPmonth[ ]

NEE3hr = RH

3hr $ NPP3hr

Net E

cosy

stem

Exc

hang

e (g

m-2

mon

-1)

July, 2002

R: 3 hourly surface incident shortwave radiationfrom met analysis

T: 3 hourly 2-m air temperature from met analysis

!

Q10

=1.5T"3010

Daily Net Ecosystem Productivity (NEP)

Photosynthetically Active Radiation (PAR)

Temperature

1998 1999 2000 2001

NEP PAR Temperature

2000.000 2000.083 2000.167 2000.250 2000.333 2000.417 2000.500 2000.584 2000.667 2000.750 2000.834 2000.917 2001.0000

20

40

60

80WLEF Daily PAR

CASA

Observations

Observations= -69.5 gC/m2/yrCASA= -6.4 gC/m2/yr`

2000.000 2000.083 2000.167 2000.250 2000.333 2000.417 2000.500 2000.584 2000.667 2000.750 2000.834 2000.917 2001.000-30

-20

-10

0

10

20

30WLEF Daily Temperature

Temperature

NEP

PAR

J F M A M J J A S O N D

(CASA)

396 m

30 m

30-d high pass filter applied

CMDLCTM CASA 3-h

— CMDL— CTM CASA 3-h