Assimilation of ozone data at the GMAO

1
Assimilation of ozone data at the GMAO

• I. Stajner, K. Wargan, M. Mueller, L.-P. Chang,
• H. Hayashi, A. Tangborn, and S. Pawson
• Global Modeling and Assimilation Office (GMAO),
  NASA/Goddard
• NRL ozone assimilation workshop
  June 10, 2005
  

2
Ozone assimilation
Model
Data
Nadir and Limb SBUV MIPAS TOMS POAM OMI MLS R
etrieved Retrieved Radiances, averaging kernels
CTM or GCM Temporal/spatial
resolution Representation of transport,
physics Chemistry parameterization gas
phase stratospheric, tropospheric,
hetrogeneous
PSAS or GSI
Assimilated ozone
Statistical analysis
3
Ozone assimilation
Model
Data
Nadir and Limb SBUV MIPAS TOMS POAM OMI MLS R
etrieved Retrieved Radiances, averaging kernels
CTM or GCM Temporal/spatial
resolution Representation of transport,
physics Chemistry parameterization gas
phase stratospheric, tropospheric,
hetrogeneous
Kris
PSAS or GSI
Assimilated ozone
Statistical analysis
4
Ozone assimilation
Model
Data
Nadir and Limb SBUV MIPAS TOMS POAM OMI MLS R
etrieved Retrieved Radiances, averaging kernels
CTM or GCM Temporal/spatial
resolution Representation of transport,
physics Chemistry parameterization gas
phase stratospheric, tropospheric,
hetrogeneous
Martin
PSAS or GSI
Assimilated ozone
Statistical analysis
5
Ozone assimilation
Model
Data
Nadir and Limb SBUV MIPAS TOMS POAM OMI MLS R
etrieved Retrieved Radiances, averaging kernels
CTM or GCM Temporal/spatial
resolution Representation of transport,
physics Chemistry parameterization gas
phase stratospheric, tropospheric,
hetrogeneous
Andy
PSAS or GSI
Assimilated ozone
Statistical analysis
6
Ozone assimilation
Model
Data
Nadir and Limb SBUV MIPAS TOMS POAM OMI MLS R
etrieved Retrieved Radiances, averaging kernels
CTM or GCM Temporal/spatial
resolution Representation of transport,
physics Chemistry parameterization gas
phase stratospheric, tropospheric,
hetrogeneous
Hiroo
PSAS or GSI
Assimilated ozone
Statistical analysis
7
Ozone assimilation
Model
Data
Nadir and Limb SBUV MIPAS TOMS POAM OMI MLS R
etrieved Retrieved Radiances, averaging kernels
CTM or GCM Temporal/spatial
resolution Representation of transport,
physics Chemistry parameterization gas
phase stratospheric, tropospheric,
hetrogeneous
L.P.
PSAS or GSI
Assimilated ozone
Statistical analysis
8
Operational ozone system

• Total ozone columns and stratospheric profiles
  from SBUV/2 instrument are assimilated in
  near-real time into a parameterized chemistry and
  transport model (CTM) driven by GMAO assimilated
  winds.
• Assimilation provides a sensitive tool for
  monitoring of the quality of ozone data
• Time series of observed-minus-forecast (O-F)
  residuals.
• Comparisons of assimilated ozone with independent
  data
• (Stajner et al, JGR 2004, doi10.1029/2003JD004118
  ).

9
Assimilation of limb viewing data

• Representation of lower stratospheric ozone
  improved by assimilation of
• POAM III over Antarctica (Stajner and Wargan GRL
  2004, doi10.1029/2004GL020846)
• MIPAS globally (Wargan et al. QJRMS, in press)
• These limb and occultation instruments provide
  high quality data in regions where SBUV/2 data
  are of lower quality or missing.
• Aura OMI and MLS ozone data were assimilated for
  January 2005.

10
OMI total column O-F residuals
OMI data January 1-27 MLS data January
2, 3, 9-13, 17, 19, 20, 23
OMIMLS(46-0.15 hPa) OMIMLS(100-0.15 hPa) MLS
data OMI zoom mode data SBUV assimilation
Global RMS (DU)
This is a measure of the disagreement between
incoming OMI data and short-term model forecasts.
OMI residuals are typically 1DU smaller than
SBUV residuals.
11
OMI total column O-F residuals
OMIMLS(46-0.15 hPa) OMIMLS(100-0.15 hPa) MLS
data OMI zoom mode data SBUV assimilation
Global RMS (DU)
Zoom
OMI residuals are typically 1DU smaller than
SBUV residuals.
OMI data from zoom mode increase O-F residuals.
12
OMI total column O-F residuals

• Assimilation of MLS data decreases OMI O-F
  residuals

OMIMLS(46-0.15 hPa) OMIMLS(100-0.15 hPa) MLS
data OMI zoom mode data SBUV assimilation
MLS
MLS
MLS
MLS
Global RMS (DU)
Zoom
OMI residuals are typically 1DU smaller than
SBUV residuals.
OMI data from zoom mode increase O-F residuals.
13
OMI total column O-F residuals

• Assimilation of MLS data decreases OMI O-F
  residuals
• Absence of MLS data increases OMI O-F residuals
• OMI and MLS seem consistent.
• Use of MLS data in lower stratosphere decreases
  OMI O-Fs further.

OMIMLS(46-0.15 hPa) OMIMLS(100-0.15 hPa) MLS
data OMI zoom mode data SBUV assimilation
MLS
MLS
MLS
MLS
Global RMS (DU)
Zoom
OMI residuals are typically 1DU smaller than
SBUV residuals.
OMI data from zoom mode increase O-F residuals.
14
Comparison with SAGE III
Mean profiles from _ SAGE III _ Aura
assimilation _ SBUV/2 assimilation
Sunset near 67N
Sunrise near 37S
MLS data are available for January 9-13. The
comparisons are at SAGE III measurement locations
on January 11-13.
3

• Assimilation of Aura MLS and OMI data reproduces
  the mean profile shape in the lower stratosphere
  and its variability with latitude better than the
  assimilation of SBUV/2 data.

15
Low ozone air
MLSOMI assimilation at 100 hPa on January 9, 2005

• Intrusion of low ozone air at 100 hPa from the
  Tropics is advected eastward and wrapped around
  higher ozone over Canada.
• Circles mark locations where SAGE III ozone was
  less than 10 mPa on the previous day at sunset.

January 11, 2005
January 12, 2005
SAGE III Aura assim.

• This SAGE III profile and the collocated Aura
  assimilated profile agree well.

January 13, 2005
0
180W
4 5 6 7 8 9 10 11 12 13 14
15 mPa
16
Comparisons with sondes
52N 5E
52N 20E
46N 7E
64S 56W
__ Sondes __ MLSOMI __ MLSSBUV
total column
10

• Either assimilation with MLS data captures well
  different profile shapes and variability in the
  middle and lower stratosphere.
• Assimilation of OMI data produces excessively low
  ozone in the troposphere. This may be due to the
  way we are using OMI data, or to data themselves.

100
Pressure (hPa)
Jan. 11
Jan. 12
Jan. 12
Jan. 9
52N 5E
52N 20E
64S 56W
46N 7E
10
100
Jan. 22
Jan. 19
Jan. 19
Jan. 20
Ozone (mPa)
17
Tropical tropospheric ozone
324E 345E 36E 112E 189E
270E
January mean for SHADOZ sondes (1998-2000) MLSO
MI MLSSBUV
Atlantic
Pacific

• Both assimilations reproduce climatology at
  SHADOZ stations. MLSOMI seems better at Natal
  (324E) and Ascension (345E).
• Wave one in tropospheric ozone is captured, with
  higher ozone values over the Atlantic than over
  the Pacific ocean.
• Tropospheric columns from Aura assimilation agree
  within 5 DU with sondes at Nairobi (36E) and
  Samoa (189E).

18
Conclusions

• Aura OMI and MLS ozone data were assimilated.
• Preliminary evaluation shows
• OMI total ozone column O-F residuals decrease
  when MLS data are available indicates that data
  are consistent.
• Aura assimilation agrees better with independent
  SAGE III and ozone sonde profiles than the
  SBUV/2 assimilation, as expected.
• Spatio-temporal variability in the lower
  stratospheric ozone seems well represented in the
  Aura assimilation.
• Wave one in tropical tropospheric ozone is
  represented. Elsewhere, tropospheric ozone
  becomes too low reasons will be investigated.
• Further evaluations and applications will follow.

19
Plans

• Aura OMI and MLS ozone data were assimilated into
  offline ozone system (CTM-based).
• In progress
• Assimilation of MLS and OMI into GEOS-4 using
  replay mode (GCM-based) with parameterizations
  for stratospheric, tropospheric and heterogeneous
  chemistry.
• Planned
• Assimilation of Aura MLS and OMI data into
  GEOS-5.
• Investigate the use/impacts of TES and HIRDLS
  data.
• Investigate impacts of assimilation of Aura ozone
  data on forecast skill and AIRS assimilation
  within GEOS-5.

20
Assimilation of SBUV radiances

• Version 8 of SBUV retrievals provides a fast
  forward model for SBUV radiances using
  linearization around the retrieved ozone profile
• NOAA 16 SBUV/2 radiances were assimilated in a
  GCM based system

21
Total ozone (TOZ)
Retrieved-minus-forecast total column

• Retrieved TOZ and TOZ from radiance assimilation
  typically agree within 4 DU, except at high solar
  zenith angles.
• Fewer SBUV radiances are used in the Tropics than
  at high solar zenith angles.

Zonal mean
DU
22
Ozone at Payerne
January to March 2003
Sonde Radiance assim. V6 retrieval assim.
Tropospheric ozone column (DU)
60
60
0
0
Jan 1, 2003
April 1

• Profile shape and the variability improve over
  the assimilation of Version 6 retrieved data
• Tropospheric column variability is captured

23
Tropospheric ozone columns

• Most estimated tropospheric columns from SBUV/2
  radiance assimilation agree with ozone sondes
  within 10 DU.

24
SBUV/2 radiances conclusions and outlook

• SBUV/2 radiances were assimilated using
  linearized forward model from Version 8 SBUV/2
  retrievals.
• In progress
• Evaluation, comparisons with independent data.
• Perturbation runs (e.g. different error
  statistics)
• Planned
• Multiyear assimilation of SBUV/2 radiances
• Combining SBUV/2 radiances with limb or
  occultation data