GEOPHYSICAL RESEARCH LETTERS, VOL.23,NO. 13, PAGES 1637-1640, JUNE 15,1996 Validation of ozoneprofile retrievals from infrared ground-based solar spectra N. S. Pougatchev Christopher Newport University, Newport News, Virginia B. J. Connor and N. B. Jones National Institute of Waterand Atmospheric Research, Lauder, New Zealand C. P. Rinsland Atmospheric Sciences Division, NASA Langley Research Center, Hampton, Virginia Abstract. A direct validationof ozoneprofile retrievals from IR ground-based high resolution solar spectra has been performed. The spectra were recorded at0.0035 to 0.01 cm -1 resolution on 11 days betweenFebruary and October 1994 at the Network for the Detection of Stratospheric Change station in Lauder, New Zealand (45.04øS, 169.68øE, 0.37 km altitude) and analyzed for the 0 3 profile with a line-by-line forward model and semi- empirical application of the optimalestimation inversion technique of Rodgers. The line parameters on the 1992 HITRAN compilation were assumed in the analysis. Retrieved ozone columnsin three atmospheric layers (0.37-12 km, 12-20 km, and 20-100 km) have been compared with correlative ozonesondes, microwave and Dobson spectrometer data from the same site. IR total and0 3 col- umns above 20 km agree with Dobson and microwavemeasure- ments to betterthan 2%. For the layers0.37-12 km and 12-20 km thedifferences between IR andintegrated ozonesondes are-7+5% and2+4%, respectively. Special exercises have been performed to show the applicability of the retrieval method to spectra with a res- olution-0.02 cm -1 and better. Introduction Different processes of both natural and anthropogenic origin control the ozone (03) content in different layers of the Earth's atmosphere, therefore, the study of the 03 vertical distribution is one of the most important problems in atmospheric physics and chemistry [WMO Report No 37, 1995].Various passive andactive techniques are now in usefor measuring ozone profiles. A compar- ative analysis of the most widely usedand reliable of thesealgo- rithmswas reported by Rodgers et al. [1988]. The most recent studies confirm the observation of significant negative trend, --6% per decade, in bothupper stratosphere (35-50 kin), andin the lower stratosphere (15-20 km) [Miller et al., 1995]. For the lower stratosphere the cause of ozone decline is still under investi- gation [Miller et al., 1995], andfurthermore, in some cases, results provided by different methods show significantdisagreement, e.g. according to ozonesonde datathe long term trend of the ozone from SAGE I/II data ranges from-5% (at 20 km) to -20% (at 16-17 km) per decade [WMO Report No 37, 1995].This exam- ple is an indication of the need for independent techniques for measurement of theozone profileand its temporal trend in thetro- posphere and lower stratosphere. As shown byPougatchev et al. [1995], ground-based high reso- lutioninfrared solarspectroscopy can be successfully used for ozone vertical profiling. A semi-empirical version of theoptimal estimation technique by Rodgers [1976] was used to retrieve the ozoneprofile from the surface to 30 km with a vertical resolution of 5 km to 10 km. The method hasbeenapplied to spectra recorded at National Solar Observatory on Kitt Peak [Pougatchev et al., 1995].Comparison of theretrievals with TOMS totalozone measurements and ozonesondesounding from Palestine, TX (1000 miles east of Kitt Peak) shows reasonable agreement, but no direct validation was possible because of theabsence of correlative ozoneprofile measurements. In the present work we report comparisons of ozone profiles retrieved from ground-based infrared spectra usingthe method described by Pougatchev et al. [ 1995]with correlative ozonesonde, microwave, and Dobson measurements. Thespectra were recorded atthe Network for theDetection of Stratospheric Change (NDSC) [Kurylo, 1991]primary station in Lauder, New Zealand (45.04øS, 169.68øE, 0.37 km a.s.1.). Measurements The spectral interval 1002.567-1003.203 cm -1 was used for retrieving the 0 3 profile (identical to the interval used in [Pougatchev et al., 1995]). The spectra were recorded between February and October 1994 (11 days, from 1 to 6 spectra per day, totalling 21 spectra) witha Bruker 120 M Fourier transform spec- trometer operating with an unapodized resolution between 0.0035 and 0.01 cm -1 (defined as 0.9 divided by the maximum optical path difference). Spectra wererecorded between 8 h and18 h local time,solar zenith angles were in the range from32 øto 83 ø , and the signal-to-noise ratio was not worse than 200. content in the Northern Hemisphere mid-latitudes 15-20 km A Dobson spectrophotometer was used to obtain total column layer is about-6% perdecade and the same trend inferred measurements [Dobson, 1957]. Precision andaccuracy of the tech- nique are estimated as 0.3% and3.0%, respectively [Margitan etal., 1995]. Thepresent results have been derived assuming new Copyright 1996by theAmerican Geophysical Union. standard, effective ozone absorption coefficients [Komhyr et al., 1993] based on the laboratory measurements of Bassand Paur Paper number 96GL01501 [1985]. 0094-8534/96/96GL-01501505.00 Ozone profiles for the comparison were based on same day 1637