The T3L3 experiment; current status and plans Sander Houweling (SRON/IMAU) Stefan Körner (BGC-Jena) All participants.
Download
Report
Transcript The T3L3 experiment; current status and plans Sander Houweling (SRON/IMAU) Stefan Körner (BGC-Jena) All participants.
The T3L3 experiment;
current status and plans
Sander Houweling (SRON/IMAU)
Stefan Körner (BGC-Jena)
All participants
Aim of the experiment
What other factors, besides transport model
differences,limit the robustness of CO2 inversions?
(…. optimization method, boundary conditions)
How important are these factors?
Experimental outline
Inversion 1: Reproduce a very simple bayesian
set-up
(same methods, same bound. cond.)
Inversion 2: as 1 using own preferred method
(different methods, same bound. cond.)
Inversion 3: as 2 using own data & assumptions
(different methods, different bound. cond.)
Who participated?
JMA-CDTM; Maki; change Globalview - WDCGG
LMDZ; Carouge; 35 regions, 21 additional stations
MATCH; Law; mass-balance inversion
MATCH; Law; 116 region inversion
NIES; Patra; 64 regions
NIRE; Taguchi; Actual meteorology
TM3; Roedenbeck/Houweling; 730 regions +covar
Problems …
Inversion 2 / 3 not purely separated
Sometimes only one of both submitted
Some inversions are perturbations of T3L2
Inversion 1 against reference
South America Tropical
Subm.
Ref.
Inversion 1 against reference
North America Boreal
Ref.
Subm.
Example of Inv1, Inv2, Inv3
Eurasia Temperate
Inv 1:
Example of Inv1, Inv2, Inv3
Eurasia Temperate
Inv 2:
Example of Inv1, Inv2, Inv3
Eurasia Temperate
Inv 3:
Offset corrected & larger scales …
NH. Extra Tropics
Inv 1:
Offset corrected & larger scales …
NH. Extra Tropics
Inv 2:
Offset corrected & larger scales …
NH. Extra Tropics
Inv 3:
Even larger …
All Continents
Inv 1:
Even larger …
All Continents
Inv 2:
Even larger …
All Continents
Inv 3:
Long-term means
Tropics
NH-ex.trop
SH-ex.trop.
Continents Oceans
Globe
Long-term trend
Tropics
NH-ex.trop
SH-ex.trop.
Continents Oceans
Globe
Correlation of AIV
N. Atlantic Temp.
Ind. Trop. Ocean
Europe
Conclusions
‘robust’ AIV looses some robustness in T3L3
- The role of method/boundary conditions not yet clear
- Only the largest scales remain rather robust
Long/term mean en trend seem less affected
by methodology
Inversions remain robust regarding e.g.:
- Correlation El-Nino, Pinatubo
- Continental sink temperate N.H.
- Less then Takahashi uptake in the Southern Ocean
How to proceed?
Further study of the difference between inv 2, inv 3
doesn’t look promising right now…
Archive of state of the art inversions in comparable
format still relevant!