Transcript Slide 1

VALIDATION OF SCIAMACHY CH4
SCIENTIFIC PRODUCTS USING
GROUND-BASED FTIR MEASUREMENTS
B. Dils, M. De Mazière, C. Vigouroux, C. Frankenberg, M.
Buchwitz, A. Gloudemans, T. Blumenstock, F. Hase, I.
Kramer, E. Mahieu, P. Demoulin, P. Duchatelet, J.
Mellqvist, A. Strandberg, K. Petersen, J. Notholt, R.
Sussmann and T. Borsdorff
Introduction
• Validation using FTIR measurements
commenced in ~2004
• Since then improvements on the SCIAMACHY
algorithms
• Also the FTIR comparison dataset has evolved
•  Hard to inter-compare results from different
validation studies
•  Re-evaluate ‘all’ SCIAMACHY CH4 algorithms
with a ‘standard’ FTIR dataset
Timeline for CH4 validation
2005, Dils et al. ACPD,5: WFMDv0.41 XCH4, IMAPv0.9 XCH4, IMLMv5.5 CH4
Covering 2003 only
CO2 normalized XCH4 for IMAP and WFMD, total columns for IMLM
Channel 6 for IMAP, Channel 8 for WFMD and IMLM
Channel 8 affected by Ice layer build-up and decontamination phases
Very limited datasets (large gaps in annual coverage)
2006, Dils et al. ACP,6: WFMDv0.5 XCH4, IMAPv1.0 XCH4, IMLMv6.3 CH4
Covering 2003 only
WFMD, now also using Channel 6
Solar zenith angle (sza) dependence of WFMD data
Inverse seasonality of southern hemisphere IMAP XCH4
2007, Dils et al. ACVE-3 proceeding, ESA SP-642: WFMDv1.0 XCH4
2003+2004
Overall improvement of data quality
sza issue resolved
2008, Current HYMN validation effort: IMAPv4.9 XCH4, WFMDv1.0/C XCH4
2003+2004+2005
Updated CH4 spectroscopy for IMAPv4.9
WFMD XCH4 CO2 normalised data using carbon tracker data in stead of a
constant value ( algorithm itself remains the same (v1.0))
The contributing FTIR-NDACC network
Spatial coordinates of the ground-based FTIR stations.
Station
Lat N
Lon E
Alt (m)
NY.ALESUND
78.91
11.88
20
KIRUNA
67.84
20.41
419
HARESTUA
60.22
10.75
580
BREMEN
53.11
8.85
27
ZUGSPITZE
47.42
10.98
2964
JUNGFRAUJOCH
46.55
7.98
3580
IZAÑA
28.30
-16.48
2367
UFTIR, http://www.nilu.no/uftir
Currently a harmonized global FTIR dataset is being developed within
the HYMN project! (http://www.knmi.nl/samenw/hymn/)
Validation Issues
• Time of measurement (limited overlap  small dataset)
• Compared the SCIA data with a 3rd order polynomial fit
through the FTIR data or
• Compared Monthly averages.
• Used Spatial collocation grid around location of gb station
• Large grid = Lat ± 2.5° Lon ± 10°
• Small grid = Lat ± 2.5° Lon ± 5°
• Altitude of FTIR station vs ‘altitude’ of SCIA data
• Conversion of total column data to effective mean volume
mixing ratios (with ECMWF model data)
• Assumes constant VMR with altitude!
• extra vmr correction using TM4 model data
• FTIR airmass vs. SCIA airmass (averaged over pixel)
• gets worse with grid! (two grids allows us to assess the
impact)
• Retrieval parameters, averaging kernels etc. (minor impact)
Validation Parameters
• Bias:
• Weighted bias of the SCIAMACHY measurements with
respect to the FTIR polynomial fit
• weighted mean [(SCIA-FTIR)/FTIR]
• the corresponding weighted standard error
= 3*std/sqrt(N)
Weight = 1/ (error of SCIA data point)2
• Scatter:
• Weighted standard deviation around the polynomial FTIR fit,
shifted with the bias, acting as the mean.
• R:
• Correlation coefficient between SCIAMACHY and FTIR
weighted monthly means
•FTIR stations in Europe only, thus limited variability
Evolution of CH4 quality (year 2003 data)
CH4 2003
WFMDv041
WFMDv0.5
WFMDv1.0
WFMDv1.0/C
LG Bias -6.95 ± 0.28
-3.45 ± 0.05
-2.70 ± 0.04
-1.17 ± 0.04
LG scat 8.38
1.75
1.40
1.29
LG R 0.37
0.55
0.65
0.65
LG N 9131
33958
21331
17084
IMAPv0.9
IMAPv1.1
IMAPv4.9
LG Bias 12.6 ± 0.09
-0.87 ± 0.03
-1.015 ± 0.026
LG scat 1.32
1.12
1.04
LG R 0.33
0.58
0.69
LG N 4585
20695
36238
IMLMv5.5
IMLMv6.3
LG Bias -1.88 ± 0.12
-3.00 ± 0.13
LG scat 2.59
3.16
LG R 0.23
0.60
LG N 6248
6433
Evolution of CH4 quality
Evolution of R and scat for all validated versions of SCIAMACHY CH4
algorithms
*IMLM is markedly different from IMAP and WFMD, since it does not
include a dry air normalisation step (using CO2 data) and uses a different
spectral window: → Needs strict cloud filtering → Less datapoints → More
scatter
Now IMLM focuses on CO retrievals, IMAP (Frankenberg) now developed
at SRON
Time series  seasonal variation
• These plots show the weighted monthly mean and
error of the SCIAMACHY data aafo time together with
daily mean FTIR data.
• Months that have less than 10 data points are not
shown.
Evolution of quality (IMAP,WFMD and IMLM):
*Large improvements for all
algorithms
*Better seasonality for IMAPv1.1
than IMAPv4.9??? (fig A)
Current status (IMAPv4.9 and WFMDv1.0/Carbon
Tracker)
• More scatter in WFMD data
• low values for February
IMAP?
Current status
• seasonality is not that
well captured
•Slightly worse for
IMAPv4.9?
Conclusions
• Overall, one can state that all SCIAMACHY algorithms
have evolved significantly over time. Both the
correlation as well as the scatter have improved with
each new development. Correlation coefficients of ~0.7
and scatter values of ~1% have been obtained.
• However several issues still remain. The latest IMAP
product (v4.9) seems to do a worse job at capturing the
seasonality than v1.1
• An in depth validation study, using the latest HYMN
harmonised date (http://www.knmi.nl/samenw/hymn)
(using a ‘quasi-global’ FTIR dataset), is currently
undertaken. This expanded dataset will allow a closer
look at station-to-station biases in the SCIAMACHY data.