Ozone

Vertical column densities below the aircraft have been retrieved along the flight track from 26 April, 6 May, 10 July and 21 September 1997. Retrieval in the near-UV using 320-340 nm revealed that the air mass factors are quite sensitive to the amount of ozone below the aircraft used when making the calculations. Thus, the air mass factors must be calculated using the correct amount of ozone below the aircraft, which clearly is not known beforehand. This can be handled by iterating the retrieval and AMF calculation until the amount used to calculate the air mass factors agrees with the retrieved amount. Alternatively, the air mass factor sensitivity to ozone is reduced if the DOAS fitting window is shifted towards longer wavelengths as the ozone cross-sections decrease rapidly with increasing wavelength in this region. For the retrievals presented in this section, the 330-360 nm range was used. There are three independent sources of surface albedo which can be used when calculating the air mass factors. The first is derived from the CPFM at 600 nm and is simply the ratio of nadir radiance to downwelling irradiance. At this wavelength, the effects of scattering below the aircraft should be small in the absence of clouds. The second is derived from AVHRR (Advanced Very High Resolution Radiometer) radiances between 550 and 680 nm. The final source is from TOMS radiances at 360 nm with an atmospheric correction to account for scattering. As the CPFM albedo is obtained over the same field of view, it will be the primary source used. Any value of AVHRR or TOMS albedo larger than one or smaller than zero was set to one and zero, respectively.

Comparisons have been made against VCDs below the aircraft derived from total column EarthProbe TOMS (Total Ozone Mapping Spectrometer) measurements and a climatological Arctic profile. The climatological profile was generated using measurements from ozonesondes launched in the Arctic throughout the spring and summer of 1997 (S. Gaines, personal communication, 1998; J. Davies, personal communication, 1998). These profiles have been analysed by a research group at the Applied Physics Laboratory (APL) and scaled to match the TOMS total vertical column value along the flight track (S. Lloyd, personal communication, 1998). It should be noted that the TOMS footprint is approximately 24 km wide at the surface. While no direct comparison can be made, VCDs can also be calculated by vertically integrating the scaled ozonesonde measurements (hereafter referred to APL-TOMS columns) between specified heights. The average VCD between 0 and 20 km over 15 ozonesondes launched between April 20 and May 15 was 198 DU with a minimum of 166 DU and a maximum of 223 DU.

  

  

The first flight examined was from 26 April 1997, shown in Figure 6.14, in which the ER-2 flew to the pole and back. The ACDs, obtained using a wavelength range of 320-350 nm, are shown in panel (a) along with the altitude of the ER-2. The mid-flight dip in ozone corresponds to a dive by the ER-2 down to 15 km. Panel (b) shows the CPFM and APL-TOMS VCDs below the aircraft along with the CPFM surface albedo. For the first and last quarter of the flight, the APL-TOMS VCDs are 10-20 DU larger while, throughout the middle half the CPFM VCDs are consistently 30-50 DU larger. Coinciding with this increase through the middle of the flight is a decrease in the albedo of about 0.2. The dashed vertical line represents a latitude of 90$^{\circ }$N and this increase is almost perfectly centered about it. The increased CPFM VCDs start near 77$^{\circ }$N on the northward leg and end at 78$^{\circ }$N on the southward leg. It is interesting to note that at the bottom of the dive, the two sources of VCDs are nearly identical which would suggest that if this feature is real then the increased ozone measured by the ER-2 resides between 15 and 20 km. There is also a noticeable drop of 25 DU near 62.5 ksec.

The dependence of ACD on latitude and solar zenith angle are shown in Figure 6.14c and 6.14d, respectively. There is a clear increase in ACD with latitude and the jump near 78$^{\circ }$N is quite evident. The overall increase with latitude is consistent with increasing total columns and a larger fraction of the total column below the aircraft. Also, the northern and southern legs (differentiated by different symbols) are quite consistent. The SZA dependence is due to its dependence on latitude.

Of the flights examined, only for this one did the three source of albedo differ significantly. The VCDs were recalculated using the other two sources, as shown in Figure 6.15. The AVHRR albedo (dotted) follows the CPFM albedo fairly closely but is always smaller, especially for the first and last quarter. The TOMS albedo (dashed) is consistently larger and hovers between 0.8-0.9 throughout the flight. The increase in VCD towards the middle of the flight remained regardless of which albedo is used, although the TOMS albedo reduced the CPFM VCDs by about 25 DU which drew it much closer to the APL-TOMS values.

To help assess if the increase centred about the pole is real or the result of an incorrect albedo, the total ozone maps from EarthProbe TOMS and ADvanced Earth Observing Satellite (ADEOS) for 26 April are given in Figure 6.16. They give roughly the same total VCDs although the ADEOS is consistently about 10-15 DU larger. The ER-2 flight track is also indicated with the circles corresponding to one-hour intervals. The feature of interest is the fairly steep increase along the flight track just south of the pole (about one-hour south of the pole) and is more pronounced in the ADEOS map. The increase in the CPFM VCDs begins at 2 hours into the flight and ends at 3.6 hours when the ER-2 begins its dive. This corresponds to this narrow ozone bulge observed in the maps. However, the magnitude of the CPFM increase, about 50 DU (using the CPFM albedo) is about twice that from the maps, about 20-30 DU (or 2-3 contour intervals).

To examine if the mid-flight increase is the result of a changing profile, the total VCD and VCD above the ER-2 are examined. All VCDs (above, below, and total) are shown in Figure 6.17. The VCDs above the ER-2 are determined from the CPFM measurements using the Brewer method as outlined in Section 3.3. They appear fairly constant throughout the flight (the dive notwithstanding) except for a small decrease between 2.7 and 3.6 hours which partially cancels the increase in the VCDs below the ER-2. Although the ER-2 mid-flight ozone increase and the ozone bulge in the maps do not exactly coincide, it seems likely that they are related. One implication could be that the CPFM can better capture the tropospheric component.

     

Figure 6.19: As Figure 6.18 but for the 10 July 1997 flight.
Figure 6.20: As Figure 6.18 but for the 21 September 1997 flight.

Results from the 6 May 1997 flight are given in Figure 6.18. Throughout much of the flight APL-TOMS VCDs are 20 DU larger and as all three sources of albedo generally agree, an incorrect albedo is likely not responsible for this discreprency. Similar to the 26 April flight, there is better agreement at the bottom of the dive which could be an indication that the APL-TOMS climatological profile near 20 km is not correct for these two flights. The best comparison was observed for the 10 July 1997 flight in which the ER-2 flew in a tight circle making measurements in a series of altitude steps between 8 and 20 km, as shown in Figure 6.19. At nearly every altitude step, there is good agreement with the APL-TOMS VCDs. The high frequency fluctuations present in the CPFM VCDs result from the albedo.

Finally, VCDs from the Fairbanks to Hawaii transit flight are presented in Figure 6.20. The APL-TOMS VCDs are not available as the Arctic ozonesonde profiles are not applicable. As this flight was over open ocean, the large values of albedo throughout the first half suggest substantial cloud cover. For the second half the albedo was generally less than 0.05. The ACDs steadily decrease with latitude and sun angle until the mid-flight dive, after which they remain constant. The cloud cover over the first half of the flight likely shields the ozone below. This acts to artificially reduce the ACDs and VCDs by an estimates 20%. This is based on an assuming a cloud top of 10 km and a ratio of 1 to 4 of ozone between 0-10 km and 10-20 km. By including an additional 20% over the first half of the flight, the VCDs would follow the expected general decrease throughout the entire flight as the ER-2 proceeds south. It may be possible to estimate the cloud top through the relative response of O₄ and O₂ absorption. The mid-flight (39$^{\circ }$N) down to 16 km gave a VCD of 70 DU. The post-dive VCDs are roughly constant near 140 DU which is reasonable based on the latitude and season. To properly analyse nadir measurements containing cloud-reflected light, their impact on ACDs needs to be assessed through forward modeling.

 

As a result of the low noise present in the O₃ ACDs, height information can be obtained from the nadir by simply taking the difference between successive ACDs and dividing by the AMF and the change in ER-2 altitude. The expression used is,

$$n_{\rm O_3}(z_i,z_{i+1})=\frac{{\rm ACD}(z_{i+1})-{\rm ACD}(z_i)} {\delta(z_i,z_{i+1})} \cdot \frac{1}{(z_{i+1}-z_i)}$$ (11.32)

where $n_{\rm O_3}$ is the ozone number density between z_i and z_i+1 and $\delta(z_i,z_{i+1})$ is the air mass factor over this range. As pressure is measured, these can be converted into mixing ratio. Comparisons are made with an in-situ dual-beam UV-absorption ozone photometer instrument which is also flown on the ER-2. By comparing the absorption of 254 nm light through two chambers, one containing the air sample with ozone and the other without ozone, ozone can be measured to an accuracy of 3% (Proffitt and McLaughlin, 1983; Proffitt et al., 1989). These ozone measurements are made every second so that for comparison purposes, a 20 s average is used centred at the time of the CPFM measurements. This comparison is shown in Figure 6.21 for portions of the 6 May 1997 flight. The ascent, panel (a), produced the best comparison, with agreement usually to within 10%, although there were a couple of points in which the difference was much larger. The two dives, in panels (b) and (c), did not track the in-situ measurements well at all. Two separate CPFM-derived profiles are observed in panel (b). During the descent portion of this dive, the slope is too small compared with the in-situ while during the ascent, the slope is too large. This behaviour follows the difference in pre- and post-first dive VCDs from Figure 6.18. During the descent, panel (d), there was a systematic underestimation by the CPFM-derived mixing ratios of 10-40%. A possible explanation for this difference could be the result of an offset in time as there is about 100 s between when a nadir spectrum is measured and a horizontal flux spectrum is measured.

 

VCDs of ozone below the ER-2 are now retrieved using the Chappuis bands in the visible using a wavelength range of 480-540 nm. Ozone cross-sections in the visible are about 100 times smaller than those in the UV. To improve the signal to noise, a spatial 1-2-1 filter was convolved with both the nadir radiances and the horizontal flux. That is, the radiance measured at a given time is taken as a weighted average with that immediately preceding and following it. Figure 6.22 shows the results from the 6 May 1997 flight. Despite the averaging, there remains considerable noise in the ACDs. Also, the expected decrease over the first of the two dives is not as pronounced. Comparing the VCDs in the visible with those from the UV, as shown in Figure 6.18, the values in the visible are noticeably larger, about 30-70 DU throughout the flight.

There are a number of possible reasons for this difference. The first is that the ozone signal is too small for ozone VCDs to be retrieved accurately. However, as the ozone trend for both dives was more or less captured combined with the small residuals evident from Figure 6.9, it would seem reasonable to conclude that this is not the major cause of the discreprency. The second possibility relates to the ability of the UV retrieval to see near-surface ozone. The larger optical depth in the near-UV decreases the sensitivity to the lowest levels of the atmosphere. Upon examining the air mass factors calculated at 340 nm, there was reduced sensitivity but only for the lowest two or three kilometres is expected to contain, at most, about 10 DU (for 40 ppbv over 3 km). (The air mass factors at 340 nm for ozone are similar to those for BrO at 350 nm, as shown in Figure 6.13.) Further, this would assume that the VCDs from the visible, which peak at 260 DU, are more representative. This is likely not the cause. Another possibility is related to the air mass factors. It is possible that the air mass factors are being underestimated. However, the optically thin, plane-parallel, geometric enhancement factor, $1+\sec{\theta_o}$, is consistently smaller than the air mass factors calculated for this retrieval (except near the end of the flight when the albedo decreased to about 0.1). As the majority of this flight would be over snow and ice, the albedo is not likely wavelength dependent. Small errors in the albedo, even 0.1, would account for much of the difference between in the near-UV and visible VCDs.