Contribution to the validation of SCIAMACHY scientific data products for CO, CH4, CO2 and N2O total column amounts using ground-based FTIR network data. M. De Mazi

1
Contribution to the validation of SCIAMACHY
scientific data products for CO, CH4, CO2 and
N2O total column amounts using ground-based FTIR
network data.M. De Mazière, B. Dils, M.
Buchwitz, R. De Beek, C. Frankenberg, A.
Gloudemans, H. Schrijver, M. van den Broek et
al.Belgian Institute for Space Aeronomy,
Ringlaan 3, B-1180 Brussels, Belgium,
martine_at_oma.be, bartd_at_oma.be
2
Coworkers at BIRA-IASB
Contributing PIs
J. Notholt, T. Warneke Institute of
Environmental Physics, University of Bremen,
Germany T. Blumenstock, S. Mikuteit Forschungszent
rum Karlsruhe, IMK, Germany E. Mahieu, P.
Demoulin, P. Duchatelet Institut d'Astrophysique
et de Géophysique, University of Liège,
Belgium J. Mellqvist, A. Strandberg Chalmers
University of Technology, Sweden R. Sussmann, W.
Stremme Forschungszentrum Karlsruhe, IFU,
Germany H. Fast, R. L. Mittermeier
Meteorological Service of Canada (MSC) T.
Kerzenmacher, K. Strong, J.Taylor, A.Wiacek
University of Toronto, Canada S. Wood, D. Smale
 National Institute for Water and Air Research
(NIWA), New-Zealand D. Griffith, N.
Jones University of Wollongong, Australia C.
Rinsland NASA Langley Research Center, USA
J. Granville P. Gérard T. Jacobs J.C. Lambert C.
Vigouroux
3
Correlative dataset
Station Lat N Lon E Alt(m)
NY.ALESUND 78,91 11,88 20
KIRUNA 67,8 20,4 419
HARESTUA 60,22 10,75 580
ZUGSPITZE 47,4 11,1 2964
JUNGFRAUJOCH 46,55 7,98 3580
EGBERT 44,23 -79,78 251
TORONTO 43,7 -79,4 174
IZANA 28 -16 2367
WOLLONGONG -34,45 150,88 30
LAUDER -45,05 169,68 370
ARRIVAL.HEIGHTS -77,85 166,78 190
4
SCIAMACHY retrieval methods
WFM-DOAS Weighting Function Modified
DOAS Channel 8 Two windows CH4 N2O ?
2265.0 2280.0 nm CO ? 2359.0 2370.0 nm CO2
? 1558.0 1594.0 nm (channel 6) IMLM
Iterative Maximum likelihood Method Channel 8
2354.00 2370.45 nm CH4 and CO from the same
window IMAP Interactive Maximum A
Posteriori-DOAS CH4 from channel 6 CO from
channel 8 2324.2 2334.9 nm
5
Comparison remarks

• (FT)IR is the only GB technique to provide
  correlative total column data but number of GB
  measurements is limited - cf. need clear sky
• Comparisons between total column data at high
  altitude stations
• ?
• adopted approach
• ? to use altitude-normalised data which is a
  better compromise for CH4 and N2O than for CO
• still in mountainous regions, pixels do not
  represent uniform elevation
• Comparisons at stations situated near the coast
  (e.g. Wollongong)
• ?
• - cf colocated pixels may be over sea
• ? Verify whether SCIA data are over land

6
Maximizing Data overlap

• Maximizing data overlap
• ? polynomial fit through GB data
• But no extrapolation!
• Good representation of seasonal variability
• (average std 1-2, except CO, 10)
• Loss of information on certain possible short
  term events

7
Selection criteria

• Data processed for two grids around the GB
  stations
• Large grid Lat 2.5
• Lon 10
• Small grid Lat 2.5
• Lon 5
• Sciamachy data cover the jan ? okt/nov time
  period
• Except CH4-IMAP aug-nov

8
Data overlap
Large data loss when restricted to point-to-point
comparison
Global time series for WFMDOAS and GB CO and CH4
(no additional filtering, fixed offset/station,
GB data)
9
Additional Selection criteria

• WFM-DOAS
• Cloud-free, Over land (altitude gt 0), Solar
  Zenith Angle lt 85 deg, Error (fitting) lt10 for
  CH4 and CO2 , lt 60 for CO and N2O
• IMLM-SRON
• Cloud-free, Albedo gt0.01, Error (instrumental)
  lt 2E18 for CH4 (7) and lt1.5E18 for CO (70)
• IMAP
• No further selection needed for CH4,
• variance of the fit residual (without weighting)
  lt 0.017, with weighting between 10 and 0.1, error
  lt 7E17 and 30 for CO
• Note Several selection criteria have already
  been applied to the starting dataset

10
Timeseries plots
IMLM-SRON
11
Timeseries plots
WFM-DOAS
12
Timeseries plots
WFM-DOAS
13
Timeseries plots
IMAP
14
CO
Retrieval algorithm Average Bias over all stations and time (small grid) Average Bias over all stations and time (large grid)
WFM-DOAS (24.56 33.22) (21.76 30.05)
IMLM-SRON (-3.51 82.51) (-1.56 83.60)
IMAP (-21.86 27.58) (-19.14 26.36)
Std GB-FTIR 10.56
Bias mean (SCIA-polyfitFTIR)/polyfitFTIR
IMLM-SRON data is preliminary dataset! ? further
improvement of std likely
15
CH4
Retrieval algorithm Average Bias over all stations and time (small grid) Average Bias over all stations and time (large grid)
WFM-DOAS v4.1 (-1.70 7.93) (-1.75 7.34)
IMLM-SRON (-2.59 15.83) (-3.68 18.39)
IMAP (-4.23 3.54) (-5.22 4.04)
Std GB-FTIR 2.09
IMLM-SRON data is preliminary dataset! ? further
improvement of std likely
16
Latitude dependence?
17
CO2 N2O
Retrieval algorithm Average Bias over all stations and time (small grid) Average Bias over all stations and time (large grid)
WFM-DOAS CO2 (-11.06 7.42) (-11.39 6.64)
Std GB-FTIR CO2 1.06
WFM-DOAS N2O (-1.10 26.83) (-0.34 26.60)
Std GB-FTIR N2O 1.75
18
Remarks

• No information on the time dependence of the
  bias!
• Of major importance in determining the accuracy
  of the retrieval model, is assessing whether or
  not the SCIA data follows the GB data over
  time.
• No data averaging was performed
• Outliner detection and removal (criteria?)
• All algorithms, IMLM (correction for dark current
  over orbits) and IMAP (implementation of the
  cloud retrieval utility for CO), undergo
  continuous improvements.

19
Future work

• In depth analysis of the time/lat dependence of
  the bias
• Comparing daily/ weekly/ monthly averages of SCIA
  and FTIR data
• Further investigation of various selection
  criteria
• Comparisons between Ground Based data and
  Theoretical Models