Modelling Atmospheric CO2 Vertical Profiles TRANSCOM PARIS 2005 Christopher Pickett – Heaps PhD Student (Univ. of Melbourne, AUS) Supervisor:Dr. Peter.

Modelling Atmospheric CO2 Vertical Profiles

TRANSCOM

PARIS 2005

Christopher Pickett – Heaps

PhD Student (Univ. of Melbourne, AUS)

Supervisor: Dr. Peter Rayner

Currently studying at the LSCE

Thank You To…

Individuals• Dr. Peter Rayner (Supervisor)• Dr. Rachel Law (CSIRO)• Dr. Philippe Ciais & others (LSCE)

Organisations• CSIRO Atmospheric Research• LSCE, CEA• University of Melbourne

Aircraft Campaign/Mission Author(s) and/or Data Providers Period

PROFILES Cape Grim Aerial Profile ArchiveCape Grim Aerial Profile Archive GASLAB, CSIROGASLAB, CSIRO 1991 - 20001991 - 2000

Orleans Aerial Profile Archive LSCE, CEA 1997 - Present

CMDL Aerial Profile Archives P. Tans, CMDL/NOAA 1992 - 2002

CAR, FTL, HAA, HFM, LEF, PFA, RTA, SAN

TROPOSPHERIC Bible (A, B, C) Machida et al. 1998, 1999, 2000

JAL (Japan/Aus) Matsueda & Inoue 1993 - 2003

PEM-WEST A dc8 Newell et al. 1996 1991

PEM-WEST B dc8 Hoell et al., 1997 1994

PEM-TROPICS A Hoell et al., 1999 1996

PEM-TROPICS B Raper et al., 2001 1999

TRACE A Andrae et al. 1996 1992

TRACE P Jacob et al., 2003 2001

CARIBIC Brenninkmeijer et al., 1999 1997 - 2001

STRATOSPHERIC ASHOE JPL, NASA 1994

POLARIS JPL, NASA 1997

SOLVE JPL, NASA 2000

STRAT JPL, NASA 1996

DATA PROVIDERS!!!

Why model CO2 vertical profiles?

Inversion validation

Satellite Validation

Potential use of profile data in atmospheric inversions

Inversion Procedure…

Use of data from 77 obs. stations• Not gap filled• (from GLOBALVIEW)

Use of prior estimates

Inversion Procedure…

Monthly fluxes 1989 – 2002 (14 yrs)

High spatial resolution: 136 regions– 92 land regions 44 ocean regions

One yr of response f’ns from 97-98 winds

15 ‘Sampled Inversions’ removing 20 stations from the network

Forward Model Procedure…

CSIRO CCAM model Flux fields from each inversion inserted

into the CCAM model

• Resolution 200km by 18 lvls• Time-step 1hr• Nudged by NCEP data (Inter-annual)• Model operates within a Cubic-Conformal grid• Advection scheme: NON LINEAR!

– Semi-Lagranian horiz. advection with bi-cubic interpolation of fields– ‘Non-negative constraint’ used for trace gas advection (Bermejo &

Staniforth, 1992) and an ‘a posteriori conservation scheme’ is applied– Vertical advection uses a TDV scheme (Thuburn, 1993)

Model results at the surface (Examples)…

Model results at the surface (sampled inversions)…

Solutions to improve model fit…

Iteration procedure

Solve for flux estimate adjustments and adjust original CO2 flux estimates

Re-run forward model with adjusted flux field

Iteration results at the surface (examples)…

Profile fit examples (CGO)

Profile RMS Time Series (CGO) Mean RMS: 0.45ppm

Profile Bias (mean residual) Time Series (CGO)

What is causing errors in the model profiles?

Non-Linearity Issue

Profiles an instantaneous measurement• Fluxes are monthly

High degree of ‘scatter’ in observed profile

Biomass burning events

Transport

Future plans…

Analyse sampled inversion flux estimates

Similar model comparison with LMDz

Use of profiles to estimate carbon flux

Proposed inter-comparison project• Stay tuned!!!