The impact of using different ozone cross sections on ozone profile retrievals from OMI UV measurements

Abstract

We compare three datasets of high-resolution O3 cross sections and evaluate the effects of using these cross sections on O3 profile retrievals from OMI UV (270-330nm) measurements. These O3 cross sections include Brion-Daumont-Malicet (BDM), Bass-Paur (BP) and a new dataset measured by Serdyuchenko et al. (SGWCB), which is made from measurements at more temperatures and in a wider temperature range than BDM and BP, 193-293K. Relative to the BDM dataset, the SGWCB data have systematic biases of -2 to +4% for 260-340nm, and the BP data have smaller biases of 1-2% below 315nm but larger spiky biases of up to ±6% at longer wavelengths. These datasets show distinctly different temperature dependences. Using different cross sections can significantly affect atmospheric retrievals. Using SGWCB data leads to retrieval failure for almost half of the OMI spatial pixels, producing large negative ozone values that cannot be handled by radiative transfer models and using BP data leads to large fitting residuals over 310-330nm. Relative to the BDM retrievals, total ozone retrieved using original SGWCB data (with linear temperature interpolation/extrapolation) typically shows negative biases of 5-10 DU; retrieved tropospheric ozone column generally shows negative biases of 5-10 DU and 5-20 DU for parameterized and original SGWCB data, respectively. Compared to BDM retrievals, ozone profiles retrieved with BP/SGWCB data on average show large altitude-dependent oscillating differences of up to ±20-40% biases below ~20km with almost opposite bias patterns. Validation with ozonesonde observations demonstrates that the BDM retrievals agree well with ozonesondes, to typically within 10%, while both BP and SGWCB retrievals consistently show large altitude-dependent biases of up to ±20-70% below 20km. Therefore, we recommend using the BDM dataset for ozone profile retrievals from UV measurements. Its improved performance is likely due to its better characterization of temperature dependence in the Hartley and Huggins bands. © 2013 .