1. NESTED GLOBAL INVERSION WITH A FOCUS ON NORTH AMERICA: COMPARISON WITH 1994-2003 BOTTOM-UP RESULTS IN CANADA Jing M. Chen, University of Toronto Main Contributors: Feng Deng, Weiimin Ju, Misa Ishizawa, Gang Mo, & Ken Yuan, University of Toronto Kaz Higuchi, Douglas Chan, Doug Worthy, & Lin Huang, Environment Canada Shamil Maksyutov, National Institute of Environmental Studies, Japan TRANSCOM Annual Meeting, Purdue, 23-27 April 2007

2. Upscaling Methodologies Used in Fluxnet Canada/ Canadian Carbon Program 1. Tall tower CO 2 concentration data Site to Landscape 2. Remote sensing and ecosystem modeling (bottom-up) Site to Region 3. Atmospheric inverse modeling (top-down) Region and Globe

3. Bottom-up Modeling

4. Comparison of InTEC-modeled and measured NEP at various sites Data sources: Andy Black, Harry McCaughey, Paul Jarvis, Alan Barr, Brian Amiro, Hank Margolis source sink

5. Datasets Used in Bottom-up Modeling Using InTEC  Land cover 1995  LAI 1994  NPP 1994  Monthly Climate 1901-2003  Forest age map (inventory, large fire polygons, remote sensing)  DEM (hydrological effect on carbon)  Drainage class  Global CO 2 time series  Nitrogen deposition (interpolated from 29 stations)  Soil texture and total carbon  Forest biomass

6. Chen et al. 2003, Tellus

7. Spatiotemporal Carbon Dynamics in Canada’s Forests and Wetlands NBP

8. Chen et al., 2003, Tellus Ju et al., 2006, Tellus.

9. Top-down Modeling

10. Nested Global Inversion System 30 small regions in North America, 20 large regions for the rest of the globe (Transcom 3), and 88 CO 2 stations (GlobalView) Deng et al., 2006, Tellus

11. Models and Data Models –NIES (National Institute of Environmental Studies of Japan), a transport model –BEPS, an ecosystem model, driven by NCEP data, linked with NIES –InTEC NBP results (1994-2003) for partial bottom-up constraint Key Datasets –Globalview baseline station CO 2 data + tall towers (2004 and 2005 versions) –Ocean carbon balance (Takahashi et al., 1997) –Global fields of fossil fuel emission in 1990 + national emissions in 2002

12. Nested Global Inversion Results (1994-2003, 30 Regions in North America and 50 Regions for the Globe) 2005 version of GlobalView data USA: -0.81 ± 0.21 PgC/y (sink) Canada: -0.30 ± 0.18 PgC/y (sink) Red: source Green: sink

13. Comparison of Results Using Two Versions of GlobalView Data 2004 version of GlovalView data: USA: -0.58 ± 0.15 PgC/y (sink) Canada: -0.14 ± 0.14 PgC/y (sink) 2005 version of GlobalView data USA: -0.81 ± 0.21 PgC/y (sink) Canada: -0.30 ± 0.18 PgC/y (sink) Upper Bound Lower Bound

14. Temporal Comparison Between Top-down and Bottom-up Results for Canada’s Forests and Wetlands sink source

15. Spatial Comparison between Top-down and Bottom-up Results for Canada’s Forests and Wetlands (region by region, 10 year average) sinksource Issues: Large uncertainty in top-down modeling Old forest carbon sink C & N coupling Non-forest sinks Lower bound Upper bound

16. Effect of Non-diagonal Co-variance Based on Meteorological Conditions

17. Summary  Both bottom-up and top-down modeling results suggest that Canada’s forests and wetlands were carbon sinks on average in the period from 1994 to 2003.  Top-down results show much larger sinks than bottom-up results. Atmospheric CO 2 data pull the surface flux strongly toward the sink direction, indicating that bottom-up sinks values are underestimated.  There are encouraging similarities in the temporal and spatial variation patterns between top-down and bottom-up results, indicating that this mutual constraining methodology is worth pursuing further under the Canadian Carbon Program. Acknowledgement: Bottom-up modeling is mostly supported by FCRN, and top-down modeling is so far supported by two CFCAS individual grants

18. The issue of old forest sinks Productive forest Non-productive forest

19. Spatial Comparison between Top-down and Bottom-up Results for Canada’s Forests and Wetlands (region by region, 10 year average) After adjusting for old forests Lower bound Upper bound

20. Too Much Carbon and Nitrogen Coupling? FACE: nutrient is less limiting at higher CO 2 (W. Schlesinger) Carbon and nitrogen are completely decoupled in LPJ (C. Prentice) C & N coupling can be relaxed by allowing the soil C/N ratio to increase with CO 2 (evidence?)