1. Data Assimilation and Carbon Cycle Working Groups
Dylan Jones (U. Toronto)
Kevin Bowman (JPL)
Daven Henze (CU Boulder)
IGC8
1 May 2017

2. What is Data Assimilation?
text
A framework for combining uncertain observations and a “forward” model with uncertain parameters and initial/boundary conditions to produce an “analysis” model that is statistically consistent with both.
Motivation:
Quantitatively attribute differences between predictions and observations to “known” unknowns.
Observations provides limited spatial and temporal coverage => data assimilation as interpolation
Produce more reliable forecasts of the state of the atmosphere using models
Combine data from difference sources, with different uncertainties, to obtain a consistent and optimal model.

3. Composition Data Assimilation with GEOS-Chem
4-Dimensional Variation Data Assimilation (4D-Var)
Full chemistry state and source estimation
CO2, CH4, N2O source (flux) estimation
3D-Var
Full chemistry state estimation
Kalman Filter
CO2 state estimation
Ensemble Kalman Filter (EnKF)
CO2, CH4 flux estimation
Local Ensemble Transform Kalman Filter (LETKF)

4. Examples of Recent Applications
Inverse modeling of OMI NO2 observations
Inverse modeling of in situ surface and aircraft observations of N2O
Wells., et al. (2015), Simulation of atmospheric N2O with GEOS-Chem and its adjoint: evaluation of observational constraints, Geosci. Model Dev., 8, , doi: /gmd
Qu, Z., et al., (2017), Monthly top-down NOx emissions for China (2005–2012): A hybrid inversion method and trend analysis, J. Geophys. Res. Atmos., 122, doi: /2016JD02582.

5. Examples of Recent Applications
Multi-constituent chemical data assimilation and the implementation of an ensemble Kalman filter (EnKF) for chemical data assimilation in GEOS-Chem
Estimated surface CO and NOx emissions in July 2010
Assimilation of TES O3 in GEOS-Chem-EnKF in July 2005
Model simulation
Data Assimilation
500 hPa
750 hPa
Zhang et al., Quantifying emissions of CO and NOx using observations from MOPITT, OMI, TES, and OSIRIS, (in preparation).
Miyazaki et al., Development of a tropospheric chemistry data assimilation system: GEOS-Chem-EnKF, (see Poster A.26 this afternoon).

6. Recent Carbon Cycle Applications
Inverse modeling of regional CO2 and CH4 fluxes
Inverse modeling of land and ocean XCO2 data from GOSAT
Deng, F., et al. (2016), Combining GOSAT XCO2 observations over land and ocean to improve regional CO2 flux estimates, J. Geophys. Res. Atmos. doi: /2015JD
Feng, L., et al. (2017), Consistent regional fluxes of CH4 and CO2 inferred from GOSAT proxy XCH4 :XCO2 retrievals, 2010–2014, Atmos. Chem. Phys., doi: /acp

7. New high-resolution CH4 emission inventory
Maasakkers, et al. (2016), Gridded national inventory of U.S. methane emissions, Environ. Sci. Technol., 50, 13123−13133. 

8. WetCHARTs v1.0: wetland CH4 emissions and uncertainty dataset
3 CH4:CO2 temperature parameter-izations
Monthly 0.5° emission and uncertainty
9 terrestrial carbon cycle models
(MsTMIP & CARDAMOM)
4 wetland extent parameter-izations × = ×
9×3×4× 3 global scale factors = 324 ensemble members
Mean CH4 Emissions
90% confidence range
Bloom et al., 2017, GMD, accepted. Full ensemble ( ) and extended ensemble ( ) emissions will become made available on the ORNL DAAC (doi: /ORNLDAAC/1502) upon publication.
A. Anthony Bloom, ESA LPS, May 2016
IGC8 - WetCHARTs

9. Carbon-GCHP Courtesy Seb Eastham and the GCHP team
GCHP successfully applied to simulate CO2
in C180 (~0.5°) simulation on Pleiades:
Run on 72 cores across 3 nodes (IFORT 15, SGI MPI)
1 simulation month in ~4 hours
Driven by 0.5° MERRA-2 meteorological data*
2015 CO2 fluxes from NASA CMS-Flux (uncorrected)

10. Development Priorities for Adjoint/Assimilation and Carbon Cycle WGs
The insatiable demands of Time and Space:
Spatial resolution of operational atmospheric assimilation models are pushing to 10-15km.
Climate and carbon studies are pushing to decadal scales.
How to adapt to GCHP?
Adjoint/forward model divide: How to keep up?
Advances in data assimilation methodologies:
Multi-constituent chemistry and carbon species
Weak constraint 4D-var
Model reduction strategies
Hybrid schemes
Preparation for the fire-hose of data: TEMPO, GeoCARB and TROPOMI
Carbon Cycle as Earth System Science
Carbon-chemistry
Time-evolving chemical source
Biogenics and carbon
Carbon-climate: future winds?
Carbon coupling: oceans and terrestrial models

11. WG network Transport WG Adjoint/Assimilation WG Carbon WG
High Performance WG