Trace-Gas Retrievals

The method of Differential Optical Absorption Spectroscopy has been applied to CPFM spectra to extract apparent column densities (ACDs) of ozone, NO₂, and BrO. DOAS was performed on both the limb and nadir fields using the horizontal flux used as the reference. Overall, differential absorption cross-sections were successfully fit to the differential optical depths and meaningful ACDs were extracted. Four spectral regions were examined in detail to retrieve the three trace-gases: 320-340 nm, 345-360 nm, 405-435 nm, and 480-415 nm. Larger spectral windows (>40 nm) generally produced poorer-quality fits which might be related to problems associated with the Ring spectrum used. Ozone in the UV (320-340 nm) was quite successfully retrieved with very low noise as was evident from the stability of ACDs between successive spectra. Using 345-360 nm proved optimum in the recovery of BrO. The BrO signal was very low compared to ozone, on the order of 10^-4-10^-3, yet despite this, the spectral fits in this region were quite good. The most problematic species to retrieve was NO₂. Using 405-435 nm gave poor results and increasing or shifting this range did not appreciably alter the quality of the fits. Ozone in the visible (480-515 nm) produced good spectral fits, even for ACDs as small as 500 DU. A significant signal from water vapour was found to be present, especially in the nadir, and so it had to be accounted for in the fit. Surprisingly, NO₂ was fit reasonable well in this region. Overall, it appears absorption signals as low as the 0.03-0.05% of the total signal can be extracted, although this depended upon the spectral region used. This is also considerably better than the anticipated 0.1%

The retrieved ACDs were then converted into vertical column densities (VCDs) using model calculated air mass factors. The nadir ACDs gave VCDs below the aircraft and these were calculated along several flight tracks. Comparison of ozone VCDs derived in the UV with those derived using TOMS-measured total ozone column combined with a climatological profile were in general agreement. Ozone VCDs derived from the visible were found to be larger for reasons which are not clear at this time, although it could be related to albedo. Similarly, VCDs of NO₂ were found to be too large by a factor of roughly two. Overall, the use of the nadir geometry was found to be quite useful as it isolates the column below the aircraft. More work is required to diagnose some of the inconsistencies.

ACDs from the limb were also used; however, due to the more complicated geometry, direct conversion to VCDs was not possible. An iterative retrieval algorithm was used which solves for VCDs between set altitude levels using an air mass factor weighting matrix. While ten-step limb scans are made, due to the information overlap in many of the steps, only three pieces of profile information could be derived below the aircraft. These layers were chosen to be 0-12 km, 12-16 km, and 16- $z_{\rm ER2}$ km (where $z_{\rm ER2}$ is the altitude of the ER-2). The atmosphere above the ER-2 was taken as the fourth layer. Poor sensitivity to the lowest layers necessitated constraining the sum of these three layers to that of the nadir-derived VCD below the aircraft after each iteration of the retrieval. Ozone results generally agreed with ozonesonde profiles and in-situ measurements made during the same period, although during the 26 April 1997 flight two scans near 90$^{\circ }$N gave smaller-than-expected values in the layer immediately below it. This could be due to the ER-2 encountering processed or ozone-depleted air. NO₂ and BrO results were variable with some scans producing reasonable values and other values which seemed unrealistic.