Seasonal & Interannual Variability in Tropospheric CO:

Global-Scale Analysis of GEOS-CHEM

using MOPITT CO Measurements

​

Prasad Kasibhatla & Quanlin Li

Duke University

Louis Giglio

Science Systems & Applications, Inc.

Jim Collatz

NASA Goddard Space Flight Center

Jim Randerson

University of California Irvine,

Guido van der Werf

University of Amsterdam

​

GEOS-CHEM Users Meeting

April 12, 2007

Questions

• Can seasonal and interannual variations in MOPITT CO provide

constraints on biomass-burning emissions?

​

• Are there persistent biases in GEOS-CHEM simulations of CO

relative to MOPITT measurements?

Simulation Set-up

• 1996-2006 with GEOS-CHEM v7-04-06 driven by GEOS-4 4x5 met fields

(sensitivity runs at 2x2.5 resolution)

​

• Tagged CO with prescribed monthly-mean OH from v5-07-08

(sensitivity runs with OH from GMI-Combo and Isabelle Bey’s simulation)

​

• Sources
• Fossil-fuel: GEIA 1985 CO emissions scaled to 1998 (Logan) and scaled by 18.5% to account for VOC oxidation (Duncan)

(sensitivity runs with EGDAR/Streets/EMEP/NEI/BRAVO inventory)

• Biofuel: Yevich and Logan
• Biogenic VOC oxidation: GEIA isoprene & monoterpene emissions with CO yields of 0.3 and 0.2; CO from methanol = 0.36 CO from isoprene; acetone emissions with CO yield of 0.67 (Duncan)

(sensitivity runs with MEGAN isoprene and monoterpene emissions)

• Methane oxidation: Calculated on-line using prescribed OH
• Biomass burning: GFED2 (with preliminary 2006 product)

​

• Focus here on 2002-2006 – Phase-2 MOPITT & MODIS fire counts

MOPITT-equivalent Model Column CO

MOPITT-equivalent Model De-seasonalized Column CO

MOPITT

MODEL

Model underestimates

tropospheric column CO

in both hemispheres

Model vs MOPITT Column CO

MOPITT

MODEL

Model is relatively successful

in reproducing observed

anomaly patterns, with

some notable exceptions

• we are currently using

anomaly contributions from

basis regions to derive

scaling factors for GFED2

​

Model vs MOPITT De-seasonalized Column CO

NH Model vs MOPITT Column CO

Model vs GMD Surface CO

GMD data from Paul Novelli

30N-50N

10N-30N

MID

MLO

Difference Between MOPITT and Model Column CO in April

MOPITT-Model Column CO (10¹⁸ molecules cm^-2)

MOPITT vs Column CO for Different Model Configurations

in April 2004

Column CO (10¹⁸ molecules cm^-2)

Model results from Susan Strahan

Model results from Isabelle Bey

What is going on?????

SH Model vs MOPITT Column CO

30S-50S

10S-30S

Difference Between MOPITT and Model Column CO in October

MOPITT-Model Column CO (10¹⁸ molecules cm^-2)

MOPITT vs Column October-April CO for

Different Model Configurations in 2004

Oct-Apr Column CO (10¹⁸ molecules cm^-2)

SH Model vs MOPITT Column CO

30S-50S

10S-30S

Model with 2xBB captures

seasonal amplitude of tropospheric

column CO in SH,

but …

Model vs GMD Surface CO

CRZ

POCS30

SYO

SPO

• Model with 2xBB overestimates seasonal amplitude at surface in SH
• Injection of BB emissions into upper free trop appears to improve model

agreement with MOPITT and surface sites in SH

Summary

• Patterns of interannual variability in tropospheric column CO relatively

well simulated

• There are significant biases in modeled CO relative to MOPITT in both

hemispheres (conclusion partly different from Shindell et al., 2006)

​

• NH bias
• Likely not significantly tied to biomass burning emissions
• Difficult to reconcile 🡪 can analysis of aircraft+TES+MOPITT+GMD CO

measurements provide insights?

• Other potential factors to consider 🡪 seasonality and trends of NH anthropogenic emissions, state-dependent MOPITT biases?
• Transport biases???

​

• SH bias
• Also seen with respect to TES (Logan)
• Likely tied to biomass burning emissions
• Difficult to reconcile MOPITT measurements relative to surface

measurements 🡪 can limited aircraft measurements provide a clue?

• Transport biases?

Acknowledgements

• NASA (funding)

• Daniel Jacob, Bob Yantosca, and GEOS-CHEM core team (model)

​

• Jennifer Logan (stimulating discussions)

​

• Susan Strahan and GMI team (GMI results)

​

• Isabelle Bey (model results)

​

• Paul Novelli (GMD data)

​

• Louisa Emmons, Merritt Deeter, and MOPITT team (MOPITT data)

​

• Louis Giglio, Guido van der Werf, Jim Randerson, and Jim Collatz (GFED)