Analysis for BrO in zenith-sky spectra: An intercomparison exercise for analysis improvement

[1] The analysis for BrO using the technique of differential optical absorption spectroscopy as applied to spectra of light scattered from the zenith sky has historically presented something of a challenge, leading to uncertainty about the accuracy of measurements. This has largely been due to the large sensitivity of the measurement to many analysis parameters and due to the small size of the absorption features being measured. BrO differential slant columns have been measured by six different groups taking part in an intercomparison exercise at Observatoire de Haute-Provence in France from 23 to 27 June 1996. The data are analyzed in a collaborative attempt to improve the overall analysis for BrO through investigation of a series of sources of errors in the instrumentation, calibration, input to the analysis, and the spectral analysis itself. The study included comprehensive sensitivity tests performed using both actual measurements and synthetic data. The latter proved invaluable for assessing several aspects of the spectral analysis without the limitations of spectral quality and instrument variability. The most significant sources of error are identified as the wavelength calibration of several of the absorption cross sections fitted and of the measured spectra themselves, the wavelength region of the fitting, the temperature dependence of the O-3 absorption cross sections, failure to adequately account for the so-called I-0 effect, inadequate offset correction, and inadequate measurement of the individual instrument slit functions. Recommendations for optimal analysis settings are presented, and comparing the results from the analysis of the campaign data shows BrO differential slant column observations from the various groups to be in agreement to within 4% on average between 87degrees and 90degrees solar zenith angle, with a scatter of 16%.