Adjoint CO source inversion using MOPITT, SCIAMACHY and AIRS

1
CONSISTENCY among
MOPITT, SCIAMACHY, AIRS and TES measurements of
CO using the GEOS-Chem model as a
comparison platform
Motivation towards estimating CO sources
Monika Kopacz, Jenny Fisher, Daniel Jacob,
Jennifer Logan, Lin Zhang, Meghan Purdy
Michael Buchwitz, Iryna Khlystova, John Burrows,
(SCIA Bremen), Annemieke Gloudemans, Jos de Laat
(SCIA SRON), W. Wallace McMillan (AIRS)
Aura Science Team Meeting Columbia, MD, October
30, 2008
2
Satellite instruments providing CO data
MOPITT
SCIAMACHY
AIRS
TES
100
100
Pressure (mb)
300
500
700
1000
0 0.1 0.2 0.3 Averaging kernels
0.2 0.6 1.0 1.4 Averaging kernels
0 0.1 0.3 Averaging kernels
validated data product (5 high bias)
sensitive throughout the column
provides O3 data
extremely dense (daily) coverage
v7.4 (SRON retrieval), v0.6 (Bremen retrieval)
v3 retrieval
v5 retrieval
v2 retrieval
3
Available satellite CO (column) data
May 2004 averages (on 2 x 2.5 resolution)
AIRS
MOPITT
SCIA Bremen
TES (2006)
1018molec/cm2
0 0.88 1.75 2.62
3.50
CO columns expected to be different due to
different vertical sensitivity, but are they
consistent?
4
Chemical Transport Model (CTM) the comparison
platform
satellite 1
satellite 2
satellite 3
SATELLITE DATA
global Chemical Transport Model (CTM)
in situ observations
TRUTH
but very sparse in time and space
5
GEOS-Chem Chemical Transport Model (CTM) the
comparison platform
Chemistry detailed chemical mechanism Meteorology
NASA/Goddard data assimilated
meteorology Resolution horizontal 2 x 2.5,
vertical 1 km, temporal 15 min
Compare with in situ data
Compare with satellite data
200 150 100 50
MOPITT CO columns
model data
MOZAIC
200 150 100 50
GMD
GEOS-Chem MOPITT AK
300 200 100
Vienna
6
Model satellite correlations
4 3 2 1 0
r2 0.65
r2 0.73
May 2004 May 2005 global daytime columns
(averaged on 2x2.5 resolution)
GEOS-Chem CTM
Red line Reduced Major Axis regression
AIRS
MOPITT
4 3 2 1
r2 0.24
r2 0.83
r2 0.29
GEOS-Chem CTM
GEOS-Chem CTM
SCIA SRON
SCIA Bremen
TES
0 1 2 3
4
0 1 2 3
4
0 1 2 3
4
Unit 1018 molec/cm2
TES data start at the end of September 2004
7
Amount of a priori information in
model-satellite correlations
4 3 2 1 0
r2 0.65 slope 0.76
r2 0.73 slope 0.71
GEOS-Chem CTM
Measure of information content degrees of
freedom (DOFs)
MOPITT
AIRS
0.5 1.0 1.5
0.5 1.0 1.5
4 3 2 1
r2 0.83 slope 0.88
r2 0.65 slope 0.74
GEOS-Chem CTM
Note DOFs not available for SCIA reprocessing
with MOPITT a priori does not change SCIA
correlations
TES w/ MOPITT a priori
TES
0.5 1.0 1.5
0.5 1.0 1.5
0 1 2 3
4
0 1 2 3
4
8
Amount of a priori information in
model-satellite correlations
4 3 2 1 0
r2 0.65 slope 0.76
r2 0.73 slope 0.71
GEOS-Chem CTM
Measure of information content degrees of
freedom (DOFs)
MOPITT
AIRS
0.5 1.0 1.5
0.5 1.0 1.5
4 3 2 1
r2 0.85 slope 0.89
r2 0.63 slope 0.70
GEOS-Chem CTM
Note DOFs not available for SCIA reprocessing
with MOPITT a priori does not change SCIA
correlations
TES w/ MOPITT a priori
TES
0.5 1.0 1.5
0.5 1.0 1.5
0 1 2 3
4
0 1 2 3
4
9
Seasonal variability of CO in datasets and model
satellite CO columns averaged over NH
3.0
MOPITT
SCIA Bremen
2.5
AIRS
2.0
1018 molec/cm2

• spring deviation partly due to differences in
  vertical sensitivity

TES
1.5
SCIA SRON
1.0
May04
Jan
July
Nov
Mar May
Sept
2.2
GEOS-Chem CO columns averaged over NH
w/ SCIA SRON AK, SCIA Bremen AK, AIRS AK, MOPITT
AK, TES AK, GEOS-Chem w/ no AK
2.0
1018 molec/cm2
1.8
1.6
May04
Jan
Sept
Mar May
July
Nov
10
Time and space consistency of datasets
(data-model)
difference
model
data-model difference
Consistency Spring CO gtgt non spring CO in all
datasets MOPITT CO AIRS CO Inconsistencies TES
differs from MOPITT or AIRS SCIA
Bremen and SCIA SRON
differ
all data NH spring
11
Implications for further use of data
vs.

• MOPITT and AIRS are very consistent, but differ
  slightly over ocean

• Small inconsistencies in the interhemispheric
  gradient
• SCIA Bremen dataset has valuable info in the
  boundary layer, despite noise quite consistent
  with MOPITT and AIRS
• SCIA SRON is very different from SCIA Bremen and
  does not appear useful for source inversion

• TES (or its Averaging Kernels?) are
  systematically different from MOPITT and AIRS

Data requirements Each (datai-model) difference
provides consistent CO source constraints, not
necessarily data1 data2
12
Acknowledgements NASA funding (graduate
fellowship), MOPITT and TES teams for providing
data
END