Estimating regional Estimating regional sources and sinks of sources and sinks of
COCO22 for North for North AmericaAmerica
Using NOAA-CMDL measurements and Using NOAA-CMDL measurements and the TM5 modelthe TM5 model
Wouter Peters, NOAA CMDLTRANSCOM May 13th, 2003
AcknowledgementsAcknowledgements
Maarten KrolPieter Tans
Lori BruhwilerSander HouwelingPeter Bergamaschi
Peter van VelthovenFrank Dentener
Jan Fokke MeirinkJohn Miller
OutlineOutline
My backgroundMy background The TM5 global & regional modelThe TM5 global & regional model The expanded NOAA-CMDL networkThe expanded NOAA-CMDL network Plans for inversionsPlans for inversions
Kalman filterKalman filter 4d-var 4d-var
My backgroundMy background
Master’s in meteorology and physical Master’s in meteorology and physical oceanography from University of oceanography from University of UtrechtUtrecht
Thesis work on LBA-CLAIRE, Thesis work on LBA-CLAIRE, airborne campaign in Suriname, airborne campaign in Suriname, South America.South America.
PhD on tropospheric ozone in the PhD on tropospheric ozone in the tropics with Jos Lelieveld and tropics with Jos Lelieveld and Maarten Krol.Maarten Krol.
TM3 global modelTM3 global model 5x3.75 degrees, 19 layers5x3.75 degrees, 19 layers ECMWF meteorology (1979-1993, and ECMWF meteorology (1979-1993, and
1994-1999)1994-1999) Full chemistry (O3-NOx-CO-CH4, lumped Full chemistry (O3-NOx-CO-CH4, lumped
NMHC’s) + coupled photolysisNMHC’s) + coupled photolysis Extensive budgets of all processesExtensive budgets of all processes Fast parallel version (<2 CPU hrs/yr) Fast parallel version (<2 CPU hrs/yr)
allowing multi-year simulations or allowing multi-year simulations or sensitivity testssensitivity tests
TM5: the next generationTM5: the next generation
New advection algorithm (New advection algorithm (Berkvens, Berkvens, Botchev, Krol, Verwer, 2001Botchev, Krol, Verwer, 2001) allows ) allows online nesting of fine-scaled grids within online nesting of fine-scaled grids within a global domain.a global domain.
TM5 model – coarse gridTM5 model – coarse grid
TM5 model – medium gridTM5 model – medium grid
TM5 model – fine gridTM5 model – fine grid
TM5 model – vertical resolutionTM5 model – vertical resolutionp [hPa] h [km]
0.0 inf1.0 48.23.0 39.37.1 33.316.8 27.639.6 22.160.2 19.187.7 16.9
122.6 14.8165.1 13.0215.0 11.3272.1 9.8336.0 8.4406.1 7.1480.8 5.9558.0 4.8635.3 3.8710.3 2.9780.3 2.1843.3 1.5897.1 1.0940.6 0.62973.0 0.34994.7 0.161007.3 0.051013.3 0.00
values for US standard atmosphere
~ 5 layers PBL
~ 12 layers free troposphere
~ 8 layers stratosphere
25 vertical layers
Hybrid coordinates
TM5TM5
Contributions from IMAU, KNMI, JRC, Contributions from IMAU, KNMI, JRC, and in the near future NOAA.and in the near future NOAA.
Most important improvements: Most important improvements: meteorological input, coding, meteorological input, coding, chemical boundary conditions.chemical boundary conditions.
Special emphasis: ERA40=> re-Special emphasis: ERA40=> re-analysis of ECMWF for 1957-2001.analysis of ECMWF for 1957-2001.
Resolution up to ECMWF operationalResolution up to ECMWF operational
TM5 – latest TM5 – latest developments-developments-
Adjoint version of TM5 under con Adjoint version of TM5 under con structionstruction
4d-var scheme from ECMWF 4d-var scheme from ECMWF implemented and workingimplemented and working
Kalman filter (Baker, Bruhwiler, Kalman filter (Baker, Bruhwiler, Tans) to be implementedTans) to be implemented
222Rn simulations at different resolutionsCourtesy: H. SartoriusIAR-Freiburg
My plansMy plans
Forward modeling of COForward modeling of CO22 + isotopes, + isotopes, SFSF66
Inverse modeling of COInverse modeling of CO2 2 + isotopes+ isotopes Regional scale sources/sinks over the Regional scale sources/sinks over the
USUS Using vertical profile/ tall tower dataUsing vertical profile/ tall tower data Possibly assist in planning and Possibly assist in planning and
optimizing network design/sampling optimizing network design/sampling strategiesstrategies
NOAA CMDL goalsNOAA CMDL goals
??????
Current networkCurrent network
2005 network ?2005 network ?
2010 network ??2010 network ??
Proposed TM5 gridProposed TM5 grid
Proposed TM5 gridProposed TM5 grid
Aspen
Denver
Pueblo
Durango
GrandJunction
Regions for inversionRegions for inversion
Fixed-lag Kalman Fixed-lag Kalman smoothersmoother
‘‘Batch’ inversion technique using Batch’ inversion technique using nn-months -months of information at a time (3 < of information at a time (3 < nn < 9) < 9)
Calculates new fluxes and error estimates Calculates new fluxes and error estimates for for xxJ=0 (where x= fluxes and J = cost J=0 (where x= fluxes and J = cost function)function)
Balances prior flux estimates against new Balances prior flux estimates against new information from observations using prior information from observations using prior flux and model-data mismatchflux and model-data mismatch
Computationally efficient for multi-decadal Computationally efficient for multi-decadal inversionsinversions
Details: David Baker’s talk tomorrowDetails: David Baker’s talk tomorrow
4d-var4d-var
o Uses steepest descend and the Uses steepest descend and the calculated value of calculated value of xxJ to minimize JJ to minimize J
o value of value of xxJ calculated by adjoint J calculated by adjoint modelmodel
o Iterative scheme to converge to Iterative scheme to converge to xxJ J = 0= 0
o Useful for short time periods and Useful for short time periods and many observationsmany observations
Seasonal COSeasonal CO22 Flux Flux EstimatesEstimates
Gt/
yrG
t/yr
Ocean
Land
NCEP (1984-2000)ECMWF (1981-1993)
PRIOR
Annual Average Flux EstimatesAnnual Average Flux Estimates
Ocean
Land
NCEP (1984-2000)ECMWF (1981-1992)
PRIOR
Annual Average Flux Estimates with Annual Average Flux Estimates with Cyclic MeteorologyCyclic Meteorology
Ocean
Land
NCEP (1984-2000)NCEP (1990)
PRIOR
Interface Zoom
Interface:Mixture between‘fine’ and‘coarse’:
=
+
Interface Zoom
XYZV
Write BC
xyzv
vzyx
Update parent
VZYX
vzyx
xyzv