Validation of SCIAMACHY total ozone: ESA/DLR V5(W) and IUP WFDOAS V2(W) M. Weber, S. Dikty, J. P.Burrows, M. Coldewey-Egbers (1), V. E. Fioletov (2), S. M. Frith (3), and D. Loyola (1) Contact: (1)DLR Oberpfaffenhofen (2)Environment Canada (3)NASA GSFC SQWG Meeting, Bremen, Germany, June 2013
The datasets ESA/DLR V5(W) WFDOAS V2m(W) – with V7 L1 m-factor WFDOAS V2(W) – without V7 L1 m-factor
Correlative datasets WOUDC database (brewer/dobson/filter) – monthly mean zonal mean data (Fioletov et al. 2002) – Daily station averages (collocated data) SBUV merged data V8.6 – Monthly mean zonal mean data (Frith et al., 2012)
Bias and drifts of SCIA WFDOAS wrt GOME Drift (%/decade) Bias (% in 2002) w/o m-factors with m-factors
Bias and drifts of SCIA WFDOAS wrt GOME GOME stable over a 16 year period m-factors (Bramstedt et al., 2009 mainly reduces the drifts at low latitudes, little changes above 50° however, the drift and bias pattern looks a bit more complicated (e.g. some seasonal effects) Drift (%/decade) Bias (% in 2002) with m-factors
Zonal mean comparisons with WOUDC ESA/DLR higher than WFDOAS (~1.5%), but both in very good agreement with WOUDC (within ~1-2%, ~3-6 DU) Small (negative) drift evident in ESA/DLR and WFDOAS with m-factor wrt to WOUDC no systematic drifts between ESA and WFDm
Zonal mean comparisons with SBUV V8.6 Very good agreement with SBUV merged for both WFD V2m and ESA V5 (within 2%) at polar latitudes (high SZA) negative biases in ESA/DLR gradient in the bias between SCIA and SBUV from tropics to high latitudes (bias decreases) weak positive drift with time in the tropics
Collocation with ground data Collocation criteria: – 300 km – distance weighted SCIA averages (within collocation radius) Separate comparison with dobsons and brewers – Seasonal cycle in differences to Dobson generally larger than to brewers – constant T in ground retrievals – temperature sensitivity lower in brewers Example: comparison with Brewer at Hohenpeissenberg, Germany (47°N) ESA/DLR WFD(m) WFD
Station-by-station comparison x WFDm: -0.7% WFDm: 0% ESA: +0.5% ESA: +1.0%
Dependence by SZA x WFDm-brewer WFDm-dobson ESA-brewer ESA-dobson Little SZA dependence SZA dependence in Dobson comparison related to seasonal variations (T issues)
Combined ozone and SZA dependency: ESA V5 Low illumination conditions: high ozone and/or high SZA: – Bias to ground increases (straylight issues with both ground and satellite data) Special conditions: ozone hole conditions (very low ozone): – Ground data tend to underestimate by up to 4% (Bernhard et al., 2005)
Combined ozone and SZA dependency: WFDOAS V2 Low illumination conditions: high ozone and high SZA Specual conditions: ozone hole conditions (very low ozone
Summary & Conclusion Very good agreement between SCIAMACHY (ESA & IUP) and WOUDC & SBUV merged (mostly within 1%) Some issues with ESA/DLR at polar latitudes (low bias) Small differences in bias and seasonal patterns (ESA/DLR, WFDOAS) in differences to SBUV and WOUDC are the result of slightly differing settings (different scalings of Bogumil cross- sections, choice of ozone profile climatology, different algorithm approach, and so on) The m-factor approach for L1 V7 successfully removes the drift in SCIAMACHY total ozone data (still some issues in the first year of the data record) WFDOAS V2 with m-factor agrees better than ESA V5, with the new GTO merged dataset (based upon GODFIT, Lerot et al. 2014, Chiou et al., 2013) RECOMMENDATION: GODFIT as the future ESA V6 will be an improvement over SGP 5 (see also Lerot et al. 2014)
APPENDIX
DOAS total ozone retrieval and ozone temperature DOAS satellite retrievals (OMI, GOMEs, SCIAMACHY) – nm (WFDOAS: nm) U Bremen retrieval: Weighting function DOAS (Coldewey-Egbers et al., 2005, Weber et al., 2005, Lee et al., 2008) –scalar temperature shift in the a-priori temperature profile –effective ozone temperature T O3 Both total ozone and temperature depend on ozone cross-section choice Radiation transfer model Coldewey-Egbers et al., 2005 Weighting function DOAS retrieved total ozone retrieved ozone temperature
Ozone and temperature terms in WFDOAS equation –Anti-correlation between ozone and ozone temperature term –Depending on fitting window size and position correlation ranges between r = -0.4 and -0.6 Coldewey-Egbers et al., 2005 GOME
WFDOAS total ozone data sets & cross-section used WFDOAS applied to GOME ( ), SCIAMACHY ( ), and GOME-2 (since 2006) – GOME1/ERS : Burrows et al (GOME FM), shift: nm – SCIAMACHY/ENVISAT: Bogumil et al., 2003 (SCIA FM), scaled 5.3%, shift: nm – GOME2/METOP A: Burrows et al., 1999, convolved, shift: nm agreement to within 1% with WOUDC brewer and dobsons Nevertheless: use of a single cross-section data for all instruments are needed to better understand calibration differences between instruments merged WFDOAS data record (Weber et al. 2011, 2012 )
Satellite vs ground