Comparison of Atmospheric Column Density Measurements over Eureka

Comparison of Atmospheric Column Density Measurements over Eureka, Nunavut by two Fourier Transform Spectrometers Keeyoon Sung, Kaley Walker, Chris Boone, and Peter Bernath University of Waterloo, Canada and Richard L. Mittermeier, Hans Fast Environment Canada International Symposium on Molecular Spectroscopy 61 st Meeting June 19 -23, 2006

Introduction SCISAT-1 • The Atmospheric Chemistry Experiment (ACE) A Canadian satellite mission that makes measurements of the Earth's atmosphere. Also known as SCISAT-1. • The ACE Arctic Validation Campaigns To validate the ACE satellite results, groundbased measurements were made in the Canadian Arctic at Eureka, Nunavut. • Aim of this study – To investigate the sources of the differences seen in the two FTSs making side-by-side measurements of the daily mean column density in the Arctic atmosphere.

Eureka Weather Station Eureka (80ºN, 86ºW) the Canadian Arctic PEARL • Eureka Weather Station • The Polar Environmental Arctic Research Laboratory (PEARL) • One of primary Arctic stations of the Network for Detection of Stratospheric Change (NDSC, now NDACC)

PARIS-IR (Portable Atmospheric Research Interferometric Spectrometer) ACE-FTS The primary instrument on SCISAT-1 Operating spectral region, 750– 4400 cm-1 Max. OPD = 25 cm (RES. = 0. 02 cm-1) PARIS-IR A portable ground-based version of the ACE-FTS PARIS-IR during the 2005 Arctic Campaign The Max. OPD was limited to 10 cm, so the spectral resolution used in this study was 0. 05 cm-1 PARIS-IR

DA 8 -FTS • BOMEM FT-IR DA 8 is stationed at the PEARL research facility in Eureka, and operated on a campaign basis by Environment Canada since 1993. • Spectral resolution Max. OPD is 250 cm. Apodized resolution is 0. 004 cm-1 • A series of filters It records spectra using filters. • Validated in 1999 The DA 8 results were validated by an intercomparison campaign. • Joined the ACE Arctic Validation Campaigns

Data acquisition § Solar absorption atmospheric spectra § During Feb. 19 – Mar. 7, 2005 PARIS-IR DA 8

Intercomparison between PARIS-IR and DA 8 Species and Observing date § For stratospheric species (O 3, HCl, HF, HNO 3) Sensitive to line shape § For one of good observing days: Mar. 05, 2005 Least affected by measurement condition Differences in retrieved daily mean column density and the selected fitting microwindows (typical width of 0. 5 ~ 5 cm-1) DA 8 PARIS-IR Column diff. : (PARIS-DA 8)/DA 8 O 3 2775, 3040, 0782 {2775, 2779, 2781}, 3040, 0782, 0982 1. 4 % (0. 8%, 13%, -8. 7%, n/a) HCl 2925 {2925, 2843, 2963} 10. 8 % HF 4039 {4000, 4039} - 2. 0 % (7. 9 %) HNO 3 0868 {0873, 0868} 9. 2 % Note fitting microwindows with the spectral location given in wavenumbers.

Investigation methods Data sets analyzed spectrum PARIS-IR DA 8(original) Res. Degraded DA 8 (DA 8 -rd) Max. OPD (resolution) 10 cm (0. 05 cm-1) 250 cm (0. 004 cm-1) Integration time 400 sec. 16 min. 10 cm (0. 05 cm-1) Fitting algorithm SFIT 2 SFIT 1/SFIT 2 Line list HITRAN 1992+ HITRAN 2004 and HITRAN 1992+ HITRAN 2004 Other fitting options tested Fit option A priori MWs EAP (ILS) : SFIT 2 s(scale fit) vs. SFIT 2 p(profile fit) : DA 8 set vs. PARIS-IR set : narrow MW vs. wide MW : Retrieval switch on/off for DA 8; Linefit (v. 11) for PARIS-IR

Ozone: residuals from MW 3040 and MW 0782 Line list effect: MW 3040 in the high resolution spectra HITRAN 1992+updates HITRAN 2004 Resolution effect: MW 0782 with EAP retrieval in SFIT 1/2 High res. + narrow MW Low res. + wide MW O 3 HITRAN 2004 line list produced higher columns by ~8% , but poorer fitting residuals. CO 2 The resolution degraded DA 8 produced lower columns by ~ 9%.

Vertical Columns (molecules/cm 2) Ozone: Line list effects on columns (MW 3040) HITRAN 2004 higher columns by ~ 8 % HITRAN 1992+ Time of observation (UT) Test with four DA 8 spectra on the day Ø Different symbols represent different combinations of fitting options. Ø Retrieved columns were observed to increase with HITRAN 2004, regardless of other fitting options/parameters/retrieval methods.

Primary column difference source: Resolution effect on HCl columns (in 1. E 15 mole. /cm 2) Algorithm MW 2925 SFIT 1 Narrow MW multi-MW MOPD SFIT 2 p Wide MW Single MW fit Max. OPD = 250 cm 3 MWs Max. OPD = 10 cm Spectra DA 8 -rd PARIS Mean (diff. ) 3. 24 (ref. ) 3. 29 3. 34 3. 56 3. 59 ( 10. 8 %) (ref. ) 0. 8% New diff. Ø Low resolution DA 8 spectra (DA 8 -rd) produced higher columns by ~ 9 % than the original. Ø Column difference was reduced to < 1 % under the same fitting conditions.

Primary column difference source: a priori vmr effect on HF columns (in 1. E 15#/cm 2) MW 4038 Spectra SFIT 1 and SFIT 2 (scale fit) DA 8 Width(cm-1) column Diff. New. diff DA 8 -rd DA 8 -rd 1992+ HITRAN MOPD(cm) DA 8 SFIT 2 p (profile fit) PARIS 2004 0. 32 (narrow) 0. 90 (wide) 250 250 10 10 2. 54 2. 76 2. 85 2. 48 2. 45 2. 49 Ref. 7. 9 % 8. 7 % 12. 2 % -2. 4% -3. 5% - 2. 0 % Ref. 1. 6% Ø The original difference was – 2. 0 %, which is misleadingly good. Note that these results were obtained when unrealistic subsidence correction was applied to a priori volume mixing ratio profile of HF only. Ø When the daily mean DOSA value of the day was applied, the difference was 8%. ØHowever, regardless of resolution, MW width, linelist, the SFIT 2 profile fit produced consistent results with PARIS-IR, where the difference is now 1. 6 %.

Advantage of using the SFIT 2 profile fit SFIT 1 Fitting (scale fit) residuals of DA 8 SFIT 2 (profile fit) Comparison of the a priori vmr with climatology data sets DA 8 a priori It improves fitting residuals, and it works well even with a poor a priori vmr.

Primary column difference source: Line list on HNO 3 column (in 1. E 16#/cm 2) 2005. 03. 05 DA 8(SFIT 1) HITRAN 1992 PARIS-IR HITRAN 2004 2. 60 2. 84 MW 0868 Difference 9. 2 % New Results with SFIT 2 + HITRAN 2004 2005. 03. 05 DA 8 -rd PARIS-IR Difference Max. OPD 250 cm 10 cm (PARIS-DA 8 -rd) /DA 8 -rd MW 0868 2. 96 (14 %↑) 2. 93 (13 %↑) 2. 84 – 3% Ø The difference was 9. 2 %. Ø With HITRAN 2004 for DA 8, the difference was reduced to 3 %. Ø Note that this regions have been updated in HITRAN 2004.

Summary and conclusion Ø The sources of differences depend on both species and fitting microwindows. species MW Primary difference sources O 3 3040 Line list (HITRAN 2004 vs. HITRAN 1992+) O 3 0782 Spectral resolution and Line list HCl 2925 Spectral resolution HF 4038 A priori vmr profile and retrieval program HNO 3 0869 Line list Ø These findings should be considered in validating ACE measurements by the two ground-based FTS measurements. Ø The spectroscopic lines play a critical role in remote sensing. Sun dog Ø Using a profile fit option (SFIT 2) is recommended. observed at Eureka

Acknowledgements PEARL Inukshuk • Canadian Space Agency • Natural Sciences and Engineering Research Council (NSERC), Canada • Eureka Weather Station • Frank Hase for Linefit (ver. 11) [F. Hase, et al. Appl. Optics, 38: 3417 -3422, 1999]