Andrew Schuh1, Thomas Lauvaux2, , Ken Davis2, Marek Uliasz1, Dan Cooley1, Tristram West3, Liza Diaz2, Scott Richardson2, Natasha Miles2, F. Jay Breidt1, Arlyn
Andrews4, Kevin Gurney6, Erandi Lokupitiya1, Linda Heath7, James
Smith7, Scott Denning1 , and Stephen M. Ogle1
Top-down bottom-up comparisons of the Mid-Continental Intensive (MCI) Region
1. Colorado State University, 2. The Pennsylvania State University, 3. Pacific Northwest National Laboratory, 4. NOAA Earth System Research Laboratory, 5. U.S. Forest Service, 6. Arizona State University, 7. U.S. Forest Service
We gratefully acknowledge funding support from the National Aeronautics and Space Administration, Earth Sciences Division, to Colorado State University (agreement #NNX08AK08G).
Main Goal of MCI Synthesis• Compare and reconcile to the extent
possible CO2 fluxes from inventories and atmospheric inversions
C
CO2 CO2
CO2
CO2
CO2
CO2
C
Atmospheric Inversions
Inventories
“Top-down” vs “Bottom-up”
• Accurately captures all C contributions, whether known or unknown
• Integrates and mixes signals, thus generally better used at larger spatial scales then inventory
• Depends on accurate modeling of transport which can be difficult
InventoriesAtmospheric Inversions
• Process based and thus fluxes are “attributable”, good for policy decisions
• Generally tied to valuable commodities and thus tracked well, e.g. crop production, forest inventory, etc.
• Generally sampled at point locations and upscaled and thus possibly not as accurate at larger scales
Main Goal of MCI Synthesis• P. Tans white paper (2002) proposed an area of the country that
might be used as testbed and minimized the potential problems of each method.
• Homogeneous managed landscape, soy, corn, some grasslands, a little forest
• Relatively flat landscape, minimizes possible transport problems due to complex topography
Inventory: CROP NEE
West et al. 2010, Ecol Apps
Inventory: CROP NEE
West et al. 2010, Ecol Apps
HAY
WHEAT
CORN/SOY
SOY/COTTON
Inventory
West et al. 2011 (in prep)
+ FIA (Heath & Smith, US Forest Service)
Inventory
West et al. 2011 (in prep)
+ FIA (Heath & Smith, US Forest Service)
+ Human/Cattle Respiration (West, PNNL)
Inventory
West et al. 2011 (in prep)
+ FIA (Heath & Smith, US Forest Service)
+ Human/Cattle Respiration (West, PNNL)
+ fossil fuel (K. Gurney, ASU)
Inventory
West et al. 2011 (in prep)
+ FIA (Heath & Smith, US Forest Service)
+ Human/Cattle Respiration (West, PNNL)
+ additional contributions (PNNL, CSU, USGS)
+ fossil fuel (K. Gurney, ASU)
Total 2007 NEE (Inventory net fossil)
• Note largest sink driven by crop signal over corn belt• Largest uncertainty is over non-crop lands, presumably forest
driven, on scale of 50% of max sink strength• Note human respiration component over Chicago
MEAN SD
-350gCm-2yr-1
+350gCm-2yr-1 +250gCm-2yr-1
0gCm-2yr-1
Summing over MCI Region
CarbonTracker vs MCI Inventory
-350gCm-2yr-1
100gCm-2yr-1
CarbonTracker vs MCI Inventory
• In general, looks pretty reasonable -350gCm-2yr-1
100gCm-2yr-1
CarbonTracker vs MCI Inventory
MAX CROP SIGNAL
MAX CROP SIGNAL
• In general, looks pretty reasonable• However, max crop signal might be reversed?
-350gCm-2yr-1
100gCm-2yr-1
CarbonTracker vs MCI Inventory
MAX CROP SIGNAL
MAX CROP SIGNAL
• In general, looks pretty reasonable• However, max crop signal might be reversed?• CarbonTracker has little flexibility to adjust sub-ecoregion
scale fluxes, even if fine spatial scale data is available.
-350gCm-2yr-1
100gCm-2yr-1
Regional Inversions?• While some global inversions do reasonably well
(CarbonTracker), can we improve the estimates with regional higher resolution inversions?
• We ran two add’l inversions:– with WRF, regionally at 10KM, w/ prior from offline
SiBCROP fluxes– with RAMS, continentally at 40km, w/ prior from
“coupled” SiBCROP fluxes – both use Marek Uliasz’s LPDM particle model
POSTERA51C-0128. M. Uliasz Regional Modeling Support for Planning Airborne Campaigns to Observe CO2 and Other Trace Gases.
SiB-CROP Prior NEE (TgC/deg2)(June 1 – Dec 31, 2007)
Posterior NEE (TgC/deg2)(June 1 – Dec 31, 2007)
Lauvaux et al. 2011 (in prep)
• Notice the max C drawdown in prior is somewhat similarly placed (NW Iowa/SW MN) to CarbonTracker (CASA).
• The posterior appears to ‘spread’ out the crop signal as well as relocate the max C drawdown location to central/northern Illinois.
• Notice the max C drawdown in prior is somewhat similarly placed (NW Iowa/SW MN) to CarbonTracker (CASA).
• The posterior appears to ‘spread’ out the crop signal as well as relocate the max C drawdown location to central/northern Illinois.
SiB-CROP Prior NEE(June 1 – Dec 31, 2007)
Posterior NEE(June 1 – Dec 31, 2007)
Lauvaux et al. 2011 (in prep)
Yields were better than expected
Yields were worse than expected
Inversion Priors/Posteriors (Jun – Dec, 2007)(GgC /0.5 deg 2 )
Shift in max C drawdown but much stronger sink than inventory
Shift in max C drawdown but sink “appearing” closer to inventory
Inversion Priors/Posteriors (Jun – Dec, 2007)(GgC /0.5 deg 2 )
Magnitude of sink looks reasonable and decently placed but no ability to move source/sink on finer scales
Inversion Priors/Posteriors (Jun – Dec, 2007)(GgC /0.5 deg 2 )
Time series of Inversion Results
• CarbonTracker posterior adheres fairly strongly to CASA prior over MCI• CSU inversion might be biased high (sink) based on uniform inversion adjustment
in sink direction.• All inversions agree on fairly strong drawdown peak not seen in priors
Conclusions• CarbonTracker estimates stronger regional
sink than inventory over MCI but not unreasonable and probably close to accurate at regional scales.
• Mesoscale regional inversions seem to be able to allocate source/sinks better spatially
• Spatial structure of sources/sink seem robust to different driving transport although overall strength of source/sink over region likely varies as a function of uncertainty in vertical transport