Diffuser plate spectral structures and their influence on GOME slant columns

As soon as GOME spectra were available for analysis, several groups started to apply their own DOAS algorithms to derive slant columns of O3 and NO2, and later also BrO, OClO, SO2, HCHO, H2O and O4. It soon became obvious, that the residuals obtained from the DOAS fit were not as good as expected from the SNR of the instrument, and that the residuals were not of random nature but systematic and rather stable over longer time periods. It turned out that using an earth-shine instead of the solar irradiance as a background improves the residual significantly, but this introduces the problem of an unknown amount of absorption in the background measurement. Several factors have been identified that contribute to the larger than expected residuals: • The undersampling of the spectra in conjunction with the Doppler shift of the solar irradiance spectrum • Small errors in the Ring spectrum • Small variations in the effective slit function for not uniformly illuminated ground pixels However, even with these effects considered, the quality of the residuals is poorer than expected.