The impact of vibrational Raman scattering of air on DOAS measurements of atmospheric trace gases

Abstract

In remote sensing applications, such as differential optical absorption spectroscopy (DOAS), atmospheric scattering processes need to be considered. After inelastic scattering on N2 and O2 molecules, the scattered photons occur as additional intensity at a different wavelength, effectively leading to "filling-in" of both solar Fraunhofer lines and absorptions of atmospheric constituents, if the inelastic scattering happens after the absorption. Measured spectra in passive DOAS applications are typically corrected for rotational Raman scattering (RRS), also called Ring effect, which represents the main contribution to inelastic scattering. Inelastic scattering can also occur in liquid water, and its influence on DOAS measurements has been observed over clear ocean water. In contrast to that, vibrational Raman scattering (VRS) of N2 and O2 has often been thought to be negligible, but it also contributes. Consequences of VRS are red-shifted Fraunhofer structures in scattered light spectra and filling-in of Fraunhofer lines, additional to RRS. At 393 nm, the spectral shift is 25 and 40 nm for VRS of O2 and N2, respectively. We describe how to calculate VRS correction spectra according to the Ring spectrum. We use the VRS correction spectra in the spectral range of 420-440 nm to determine the relative magnitude of the cross-sections of VRS of O2 and N2 and RRS of air. The effect of VRS is shown for the first time in spectral evaluations of Multi-Axis DOAS data from the SOPRAN M91 campaign and the MAD-CAT MAX-DOAS intercomparison campaign. The measurements yield in agreement with calculated scattering cross-sections that the observed VRS(N2) cross-section at 393 nm amounts to 2.3 ± 0.4 % of the cross-section of RRS at 433 nm under tropospheric conditions. The contribution of VRS(O2) is also found to be in agreement with calculated scattering cross-sections. It is concluded, that this phenomenon has to be included in the spectral evaluation of weak absorbers as it reduces the measurement error significantly and can cause apparent differential optical depth of up to 3 ×10-4. Its influence on the spectral retrieval of IO, glyoxal, water vapour and NO2 in the blue wavelength range is evaluated for M91. For measurements with a large Ring signal a significant and systematic bias of NO2 dSCDs (differential slant column densities) up to (-3.8 ± 0.4) × 1014 molec cm-2 is observed if this effect is not considered. The effect is typically negligible for DOAS fits with an RMS (root mean square) larger than 4 × 10-4.