1. Evaluating the Impact of the Atmospheric “ Chemical Pump ” on CO 2 Inverse Analyses P. Suntharalingam GEOS-CHEM Meeting, April 4-6, 2005 Acknowledgements J. Randerson, N. Krakauer (UCI/CalTech); D. J. Jacob, J. A. Logan, Y. Xiao, R. M. Yantosca (Harvard); A. Fiore (GFDL) Suntharalingam et al. [2005], Global Biogeochemical Cycles, submitted.

2. APPLICATION OF GEOS-CHEM TO EVALUATE CHEMICAL PUMP EFFECT QUESTION : What is impact of accounting for realistic representation of reduced carbon oxidation 1) on modeled CO 2 distributions 2) on inverse flux estimates APPROACH : Use GEOS-CHEM simulations to estimate magnitude of effect

3. ATMOSPHERIC CARBON BUDGET An outstanding question on global CO 2 budget : What is magnitude and distribution of net terrestrial biospheric flux ? ( “ missing sink ” ) The “ Top-down ” approach uses Inverse Analyses of Atmospheric CO 2 ? Net Terrestrial Flux ?

4. CARBON FLUX FRAMEWORK UNDERLYING RECENT ATMOSPHERIC CO 2 INVERSIONS FossilSeasonal Biosphere “ Residual Biosphere ” Land use change, Fires, Regrowth, CO 2 Fertilization Ocean 6 120 Units = Pg C/yr Atmospheric CO 2 90 92 NET LAND UPTAKE ?? ( 0-2 ) All surface fluxes y mod - y obs Concentration residual

5. Model prior distributions for fossil, seasonal biosphere, ocean y mod 76 Surface CO 2 observation stations (GLOBALVIEW-CO2) y obs Estimate “ RESIDUAL ” CO 2 fluxes for 22 regions Inference of Northern Hemispheric Carbon Uptake from Annual Mean Concentration Residuals THE TRANSCOM 3 INVERSE ANALYSES (Gurney et al. 2002) Residuals = y mod – y obs Model simulations (prior fluxes) CO 2 Observations N. Hem. carbon uptake

6. OXIDATION OF REDUCED C SPECIES PROVIDES A TROPOSPHERIC SOURCE OF CO 2 FossilBiomass Burning, Agriculture, Biosphere Ocean ATMOSPHERIC CO 2 CO 0.9-1.3 Pg C/yr Non- CO pathways CH 4 NMHCs Distribution of this CO 2 source can be far downstream of C emission location

7. HOW IS REDUCED CARBON ACCOUNTED FOR IN CURRENT INVERSIONS ? A : Emitted as CO 2 in surface inventories Fossil fuel : CO 2 emissions based on carbon content of fuel and assuming complete oxidation of CO and volatile hydrocarbons. (Marland and Rotty, 1984; Andres et al. 1996) Seasonal biosphere (CASA) : Biospheric C efflux represents respiration (CO 2 ) and emissions of reduced C gases (biogenic hydrocarbons, CH 4,etc) (Randerson et al., 2002; Randerson et al. 1997) Seasonal Biosphere : CASA Fossil Fuel

8. MODELING REDUCED CARBON CONTRIBUTION AT SURFACE PRODUCES BIASED INVERSION ESTIMATES Surface release of CO 2 from reduced C gases Tropospheric CO 2 source from reduced C oxidation CO, CH 4, NMHCs VS. Observation network detects tropospheric CO 2 source from reduced C oxidation y modsurf y mod3D y obs VS.

9. CALCULATION OF CHEMICAL PUMP EFFECT Flux Estimate: x = x a + G (y - K x a ) STEP 1 : Impact on modeled concentrations Adjust y model to account for redistribution of reduced C from surface inventories to oxidation location in troposphere y model y obs Adjustment  y model = y 3D – y SURF ADD effect of CO 2 source from reduced C oxidation SUBTRACT effect of reduced C from surface inventories

10. EVALUATION OF THE CHEMICAL PUMP EFFECT GEOS-CHEM SIMULATIONS (v. 5.07) Standard Simulation CO 2 Source from Reduced C Oxidation = 1.1 Pg C/yr Distribute source according to seasonal 3-D variation of CO 2 production from CO Oxidation Distribute source according to seasonal SURFACE variations of reduced C emissions from Fossil and Biosphere sources CO2 SURF Simulation : y SURF CO2 3D Simulation : y 3D Simulations spun up for 3 years. Results from 4 th year of simulation

11. GEOS-CHEM Model Configuration http://www-as.harvard.edu/chemistry/trop/geos/index.html Global 3-D model of atmospheric chemistry (v. 5.07) 2 o x2.5 o horizontal resolution; 30 vertical levels Assimilated meteorology (GMAO); GEOS-3 (year 2001) CO oxidation distribution from tagged CO simulation using archived monthly OH fields Reduced Carbon Emissions Distributions (spatial and temporal variability) Fossil : Duncan et al. [2005] (annual mean) Biomass Burning : Duncan et al. [2003] (monthly) Biofuels : Yevich and Logan [2003] NMVOCs : Duncan et al. [2005] ; Guenther et al. [1995]; Jacob et al. [2002] CH 4 : A priori distributions from Wang et al. [2004] (monthly)

12. REDUCED CARBON SOURCES BY SECTOR STANDARD SIMULATION : CO 2 Source from Reduced C Oxidation = 1.1 Pg C/yr Sector breakdown based on Duncan et al. [2005] *Methane sources distributed according to a priori fields from Wang et al. [2004] REDUCED CARBON SOURCES Pg C/yr Fossil (CO,CH 4,NMHCs)0.27 Biomass Burning (CO,CH 4,NMHCs)0.26 Biofuels (CO,CH 4 )0.09 Biogenic Hydrocarbons0.16 Other Methane Sources*0.31 TOTAL 1.1

13. SOURCE DISTRIBUTIONS : ANNUAL MEAN Zonal Integral of Emissions Latitude CO2 3D : Column Integral of CO 2 from CO Oxidation CO2 SURF :CO 2 Emissions from Reduced C Sources CO2 3D :Maximum in tropics, diffuse CO2 SURF : Localized, corresponding to regions of high CO, CH 4 and biogenic NMHC emissions CO2 3D CO2 SURF gC/(cm 2 yr) -5050

14. MODELED SURFACE CONCENTRATIONS : Annual Mean CO2 SURF CO2 3D Surface concentrations reflect source distributions: Diffuse with tropical maximum for CO2 3D and localized to regions of high reduced C emissions for CO2 SURF

15. Largest changes in regions in and downstream of high reduced C emissions TAP : - 0.55; ITN : - 0.35; BAL : - 0.35 (ppm) REGIONAL VARIATION OF CHEMICAL PUMP EFFECT  y model = CO2 3D – CO2 SURF

16.  y model : Zonal average at surface CO2 (ppm) CHEMICAL PUMP EFFECT : N/S DIFFERENCES Mean Interhemispheric difference = - 0.21 ppm 0.21 ppm Latitude Impact on TRANSCOM3 Systematic decrease in Northern Hemisphere Residuals 50-50

17. IMPACT ON SURFACE FLUX ESTIMATES Inverse analyses by Nir Krakauer Estimate effect by modifying concentration error vector as : (y – (K x a +  y model )) Then, ‘ adjusted ’ state estimate is: x adj = x a + G(y – (K x a +  y model )) Evaluate with 3 transport models (MATCH, GISS-UCI, LSCE-TM2) Q : What are the changes in estimates of ‘ residual ’ fluxes when we account for chemical pump adjustment  y model Evaluate impact using TransCom annual mean analysis (Gurney et al. 2002)

18. Relative impact of chemical pump adjustment on CO 2 uptake varies across models. 0.220.250.26 Original Uptake -19% -27%-9% 2.50.91.4 % Change REDUCTION IN LAND UPTAKE (Northern extratropics) Systematic Reduction (0.22-0.26 Pg C/year) Pg C/yr

19. SUMMARY Neglecting the 3D representation of the CO 2 source from reduced C oxidation produces biased inverse CO 2 flux estimates. Accounting for a reduced C oxidation source of 1.1 Pg C/yr gives a reduction in the modeled annual mean N-S CO 2 gradient of 0.2 ppm (equivalent to a reduction of 0.2-0.3 Pg C/yr in Northern Hemispheric land uptake in an annual mean inversion.) Regional changes are larger; up to 0.6 ppm concentration adjustment in regions of high reduced C emissions. Impacts on seasonal inverse estimates may be significant and will be examined in future work (N/S  y variation: – 0.32 ppm (January) to – 0.15 ppm (July)).

20. EXTRA SLIDES

21. SOURCE ESTIMATES FROM INVERSE ANALYSIS Minimize cost function: J(x) = (x – x a ) T S a –1 (x - x a ) + (y – K x) T S  –1 (y –K x) Solution: x = x a + G (y - K x a ) where, G = S a K T (K S a K T + S  ) -1 A posteriori errors : S = (K T S  –1 K + S a –1 ) -1 Observed concentrations Modeled concentrations x = state vector (sources) x a = a priori source estimate K = Jacobian matrix (model transport) S a = Error covariance matrix on sources S e = Error covariance matrix on concentration error

22. IMPACT ON SURFACE FLUX ESTIMATES Relative Reduction in N.Hemisphere Land Uptake Varies with Model Reduction in Land Uptake : LSCE-TM2 Reduction in Land Uptake : MATCH

23. CHEMICAL PUMP FLUX ADJUSTMENTS ZONALLY AGGREGATED LAND REGIONS Relative impact of chemical pump adjustment varies across models, though magnitude of zonally aggregated flux adjustment relatively invariant Sum N. extratrop. Land net flux (PgC/yr) - 2.5 - 0.9 -1.4