sensitivity co2 sources and sinks to ocean versus land-dominated observational networks
DESCRIPTION
Prabir K. Patra Acknowledgments: S. Maksyutov, K. Gurney and TransCom-3 modellers TransCom Meeting, Paris; 13-16 June 2005. Sensitivity CO2 sources and sinks to ocean versus land-dominated observational networks. Yet another sensitivity study!. Plan of the talk - PowerPoint PPT PresentationTRANSCRIPT
Prabir K. Patra
Acknowledgments: S. Maksyutov, K. Gurney and TransCom-3 modellers
TransCom Meeting, Paris; 13-16 June 2005
Sensitivity CO2 sources and sinks to ocean versus land-dominated
observational networks.
Yet another sensitivity study!
Plan of the talk
– Why network sensitivity (using IAVs in flux anomalies)
– Experimental setup (based on T3-L1 & L2)– Some results (may be useful for synthesis)– Conclusions
64-Regions Inverse Model(using 15 years of interannually varying NCEP/NCAR winds)
Patra et al., Global Biogeochem. Cycles., revised, 2005a
]/)(/)[([1 21 0
121
12S
M
MDpredicted
N
N CSSCDDT
Inv. Setup Chi222 reg 2.1564 reg 1.1164+IAV 0.99
CS = cs1 + cs2…
Flux anomaly (6-month running averages) and initial conditions
Flux anomaly = TD
I Flux – avg. sea. cyc
Comparison of land flux anomalies
Comparison of ocean flux anomalySource: C
. Lequere
Sensitivity to networks
and inversion
methods(!)
Thanks to:Philippe BousquetChristian Rodenbeck
Validation…
Validation…
Conclusions: IAV in fluxes (and fluxes indirectly) is controlled mainly by network selection
Assumption: Biases in flux estimation are linked mainly to transport model errors
Inverse model framework and present day network (70% real data for the period 1999-2001)
Land Fluxes – Netw
ork and m
odel Dependency
Ocean Fluxes –
Netw
ork Dependency
Signal gradients at optimal stations - tropical
Signal gradients within regions – high/midlats
Global &
hemispheric S
caleFluxes – N
etwork D
ependency
Land and Ocean Fluxes (70% real) – ocean versus all networks
Land Seasonal C
ycle
Ocean S
easonal Cycle
Conclusions1. The IAV is controlled mainly by observational
network selection, less on techniques
2. Biases in fluxes estimation are linked to transport model errors
3. For synthesis of CO2 sources and sinks, we need to revisit the estimations
• Different networks• Separate time period for inversion
4. Finally, any suggestions are welcome
Do not reject the land stations, but be careful …