Comparisons CH 4 SCIAMACHY / GROUND-BASED FTIR B. Dils, BIRA + HYMN partners Corinne Vigouroux, BIRA, HYMN progress meeting, 7-8 April 2008.
The FTIR stations for HYMN Stations (UFTIR) Ny-Alesund Kiruna Harestua Bremen (CH4) Zugspitze Jungfraujoch Izana Paramaribo St-Denis Collocation grid with SCIAMACHY data: Large grid = Lat ± 2.5° Lon ± 10° Small grid = Lat ± 2.5° Lon ± 5°
Previous work on FTIR / SCIA comparisons Dils et al., ACP, 6, , 2006 European stations (as UFTIR, but different retrieval strategies !) + Egbert, Toronto, Wollongong, Lauder and Arrival Heights 3 Algorithms for SCIA data: 1) IMAP-DOAS (Frankenberg et al.) (only 2003)2) WFM-DOAS (Buchwitz et al.) 3) IMLM (Gloudemans et al.) Scarce data in daily coincidence → Compare the SCIA data with a 3rd order polynomial fit through the FTIR data; therefore daily variability cannot be captured. → Large scatters (except IMAP-D: 1.1%, but worst correlation coefficient: 0.7)
Dils et al., Proceedings of ACVE-3, Dec UFTIR stations only. Homogenize strategy: called this data set “UFTIR data” WFM-DOAS (improved version) Methodology: still polynomial fit (to compare the improvement with respect to previous version WFM-DOAS) → Improvement of the comparisons compared to old data sets WFM-D & UFTIR. Previous work on FTIR / SCIA comparisons
HYMN project Data sets & methodology UFTIR stations + Paramaribo and Reunion, using a common strategy ! → UFTIR or improved ? IMAP-DOAS (new version) 2004 (+ ???) Methodology : for the year 2004 & with new IMAP algorithm, more SCIA data are available → polynomial fit through FTIR data or direct comparisons ?
HYMN project Today, we show : Comparisons new IMAP-D vs WFM-D used lately in Dils et al., ACVE_3: → same UFTIR data set & same methodology (polynomial fit) New IMAP-D, same UFTIR data, but considering direct comparisons of daily mean coincidences Direct comparisons using new HYMN FTIR data set when available.
Improvement in scatter and correlation coefficient scat includes natural variability toward polynomial is close to the FTIR scatter = 0.85% (2004) CH 4 WFM-DIMAP-D BiasWeighted mean [(SCIA-FTIR)/FTIR] ± 3*std/sqrt(N) ± 0.05%-1.23 ± 0.02 % scat Weighted std of SCIA around the polynomial FTIR fit, shifted with the bias 1.40%0.91% RCorrelation coefficient between weighted monthly means NNb coincidences IMAP-DOAS & WFM-DOAS vs FTIR polynomial fit (UFTIR data)
IMAP-D & WFM-D: bias as a function of latitude IMAP-DOAS bias less pronounced and more homogeneous than for WFM-DOAS
Examples of CH 4 monthly time series in 2004
Polynomial fit or direct comparisons ? UFTIR data Polynomial fit UFTIR data Direct comparisons BiasScatterBias (N)Scatter Ny Alesund (10)1.11 Kiruna (37)1.07 Harestua (12)1.17 Bremen (21)1.34 Zugspitze (45)1.42 Jungfraujoch (62)1.11 Izaña (53)0.77 Paramaribo St-Denis Both techniques are in agreement: to be done: check in detail if we see daily effects.
HYMN data set UFTIR Daily means HYMN data setChanges HYMN vs UFTIR Bias (N)ScatterBias (N)Scatter Ny Alesund-0.17 (10) (25)1.10Spectroscopy + new a priori profile Kiruna-0.21 (37)1.07 Harestua-1.16 (12)1.17 Bremen-0.92 (21) (10)1.06Spectroscopy + new a priori profile Zugspitze-1.89 (45) (47)1.30??? Jungfraujoch-0.82 (62) (62)1.29Spectroscopy Izaña-1.39 (53)0.77 Paramaribo+3.14 (20)1.21Spectroscopy + new a priori profile StDenis+1.56 (25)1.46Spectroscopy → Homogenization problem ?
Today, we show : Improvement of new IMAP-DOAS data Methodologies in agreement. Final choice to be decided. It seems that there is some problems of homogeneity with new HYMN FTIR data set: discussion in progress. Questions for FTIR (next meeting May): when should we deliver data for other years than 2004 ? N2O, CO ? ECMWF / NCEP ? (at St-Denis: ~ 1% bias on total columns, ~ 0.2% on tropospheric columns) Conclusions