Journal of Geophysical Research: Atmospheres

Photochemical ozone loss in the Arctic as determined by MSX/UVISI stellar occultation observations during the 1999/2000 winter



[1] The combined SAGE III Ozone Loss and Validation Experiment and Third European Stratospheric Experiment on Ozone 2000 (SOLVE/THESEO 2000) campaign during winter 1999/2000 sought, in part, to quantify ozone loss within the Arctic polar vortex using a variety of aircraft-, balloon-, ground-, and space-based instrument platforms. The Midcourse Space Experiment/Ultraviolet and Visible Imagers and Spectrographic Imagers (MSX/UVISI) suite of instruments performed 31 stellar occultation observations from 23 January through 4 March 2000 in and near the Arctic polar vortex. Using a newly developed combined extinctive-refractive algorithm, ozone mixing ratio profiles, along with profiles of total density, pressure, and temperature, were retrieved. Retrieved temperature and ozone profiles are shown to agree well with other measurements. Diabatic trajectory calculations are used to remove the effects of subsidence within the vortex, allowing photochemical ozone loss to be inferred from these occultation measurements. A maximum ozone loss of about 1 ppmv is found at 400–500 K (∼16–21 km) during the 41-day period for which occultation data are available, in agreement with several other ozone loss analyses of the campaign. This corresponds to an average daily loss rate of ∼0.024 ppmv/day. The combined extinctive-refractive stellar occultation technique is demonstrated to accurately measure stratospheric ozone loss during polar winter.

1. Introduction

[2] The increased frequency of cold Northern Hemisphere stratospheric winters over the past decade [Pawson and Naujokat, 1999] leading to significant photochemical ozone depletion [Newman et al., 1997; Solomon, 1999] has heightened interest in modeling and predicting the future of Arctic ozone loss and recovery. To gain a better understanding of the processes controlling stratospheric ozone at mid-to-high northern latitudes, the SAGE III Ozone Loss and Validation Experiment (SOLVE) campaign was conducted simultaneously with the Third European Stratospheric Experiment on Ozone 2000 (THESEO 2000), based in Kiruna, Sweden (68°N, 20°E), from November 1999 through March 2000. The combined campaign coordinated measurements from a wide range of aircraft-, balloon-, ground-, and space-based instruments. A key goal was to assess quantitatively the rate of high-latitude stratospheric ozone loss.

[3] The winter 1999/2000 Arctic polar vortex was particularly stable over a long period of time [Manney and Sabutis, 2000]. Trajectory calculations run over the SOLVE time period show that the vortex was well isolated from midlatitude air. Under these conditions, it is possible to remove the effects of subsidence from ozone measurements within the vortex and determine the photochemical loss of ozone. Ozone loss during SOLVE has also been the subject of several other analyses using remotely sensed and in situ data [e.g., Santee et al., 2000; Sinnhuber et al., 2000; Richard et al., 2001; Hoppel et al., 2002; Rex et al., 2002; Schoeberl et al., 2002; Lait et al., 2002].

1.1. Occultation Techniques

[4] Occultation techniques have been used for many years to characterize the structure and composition of the Earth's atmosphere, as well as those of other planets and their satellites. Occultation provides a self-calibrating method of measuring absorbing species in the atmosphere and monitoring their long-term trends. Traditionally, two distinct approaches have been utilized. Extinctive occultations yield measurements of absorbing/scattering species resulting from wavelength-dependent extinction. For example, the total transmission of sunlight, attenuated as it passes through the atmosphere, is related to the column densities of absorbing species along the light path. A series of measurements made as the Sun rises or sets may be inverted to yield number density profiles of trace gases in the atmosphere as a function of altitude. This technique has been used by a number of space-based solar occultation instruments with great success, including: SAGE II [McCormick et al., 1989], HALOE [Russell et al., 1993], ATMOS [Gunson et al., 1996], and POAM II and III [Lucke et al., 1999]. Another occultation approach, capitalizing on the refraction of light in the atmosphere resulting from vertical density gradients, provides the bulk properties of the atmosphere (i.e., total density, temperature, and pressure). For example, refractive occultation of radio waves from the Global Positioning System has been used to measure temperature profiles in the atmosphere [Ware et al., 1996].

[5] The Sun is a powerful light source and endows solar occultation with an excellent inherent signal-to-noise ratio. With the Sun come certain limitations, however. Measurements are restricted to the terminator, so for space-based solar occultation, this means only two occultations per orbit are possible. Though a weaker source of light, stars provide numerous point sources, ideally suited for occultation measurements in the nighttime atmosphere. Stellar occultation has been used to measure column ozone abundance from the ground [e.g., Roscoe et al., 1994; Fish et al., 1994; Roscoe et al., 1997; McDonald et al., 1999] and ozone profiles from balloon-based platforms [e.g., Renard et al., 1996; Payan et al., 1999]. Yee et al. [2002] have developed a further innovation to this star-based approach by combining the complementary (though historically separate) extinctive and refractive occultation techniques, which simultaneously provides both the bulk properties of the atmosphere and mixing ratio profiles of absorbers, including ozone, in the atmosphere from space.

1.2. Combined Extinctive-Refractive Stellar Occultation

[6] High in the atmosphere, where refraction is not significant and airglow contamination is minimal, atmospheric composition may be retrieved from spectrographic information alone, provided the spectral coverage is sufficient for the species of interest. Airglow emissions become significant in a layer from 80 to 100 km, however, and their contamination must be removed from the stellar signal when the tangent altitude of the line of sight is at or below the layer (light passing through a tangent altitude below the source of the emissions still passes through the emission layer on its path to the observer). An instrument with spectrographic imaging capability is therefore required. Knowledge of airglow emissions above and below the tangent altitude allows for suitable background correction.

[7] Refraction becomes significant in the lower atmosphere, below 35 km, changing the apparent position of the star (and thus the altitude registration) and producing a curved light path that leads to enhanced extinction. In addition, refractive attenuation (photon flux divergence) causes a reduction in the overall signal, and small-scale density fluctuations along the line of sight generate refractive scintillation, which introduces noise and additional uncertainty into the retrieval. Though climatologies may be used to approximate the curved light path, the resulting uncertainties would substantially reduce the efficacy of the technique for addressing science issues. A better approach is to measure refraction simultaneously, allowing for accurate calculation of the refracted path and inference of the atmospheric density profile. A high-angular-resolution, visible-light imager that measures the apparent position of the star can provide absolute refraction angles from which the atmospheric density profile may be retrieved. This retrieved density profile can be used to account for Rayleigh scattering directly, thereby allowing for ozone measurements in the lower atmosphere with greater accuracy. Because the density profile may be used to infer both pressure and temperature, another benefit gained from this unique, combined extinctive-refractive method is that potential temperature and constituent mixing ratio profiles can be calculated directly, thus eliminating the need for approximations based on climatologies or meteorological analyses. The approach may be summarized as follows:

equation image

2. MSX/UVISI Instrument

2.1. Instrument Description

[8] The Midcourse Space Experiment (MSX) was launched on 24 April 1996, into a nearly Sun-synchronous, circular orbit with an inclination of 99.4°, an ascending node of 5:38 pm (which had processed to approximately 7:30 pm during SOLVE), and an altitude of ∼900 km [Mill et al., 1994]. MSX is a 3-axis stabilized satellite with extraordinarily accurate and stable pointing. It can point at a fixed inertial position (e.g., a star) with an absolute accuracy of ∼100 μrad and relative stability of 10 μrad (1-σ) for several minutes.

[9] The Ultraviolet and Visible Imagers and Spectrographic Imagers (UVISI) instrument suite aboard MSX consists of four visible and ultraviolet imagers and five spectrographic imagers (SPIMs) covering a wavelength range of 100–900 nm [Carbary et al., 1994; Paxton et al., 1996]. All nine instruments constituting UVISI are co-aligned to within 50 μrad on a common optical bench. In the past, we often used the narrow field-of-view visible imager to determine the absolute star refraction angles, but during SOLVE, the imaging capabilities of the SPIMs were utilized instead. The SPIMs use a Wadsworth design and have wavelength resolutions of 0.7–2.2 nm and roughly 0.1° × 1° slit fields of view with 0.025° spatial resolution. The slits can be swept by scan mirrors over an additional 1°, for a 1° × 1° total field of regard. The focal plane units consist of intensified photocathodes fiber-optically connected to 2-dimensional charge-coupled devices (CCDs) and have been optimized for the wavelength range of the particular sensor. An automatic gain/gate system, combined with various filters, provides for a total dynamic range on the order of 1013. The SPIMs accumulate data at a rate of 2 or 4 Hz. Though the instrument was not designed for occultation measurements, MSX/UVISI satisfies the general design requirements for extinctive-refractive occultation experiments, described above.

2.2. Occultation Procedure

[10] The general MSX/UVISI stellar occultation procedure has been described in detail by Yee et al. [2002] and is summarized here. During a stellar occultation observation, the UVISI boresight remains fixed and centered on the inertial position of the star, with the SPIM slits oriented vertically with respect to the horizon. The star is acquired when it is well above the attenuating atmosphere, and several hundred spectra are collected before atmospheric extinction begins, thus ensuring an accurate description of the unattentuated light source. As the star sets, its position in the SPIM slits remains unchanged until refraction becomes significant, below about 35 km. Refraction causes the image of the star to drift slowly up and finally out of the top of the slit at a tangent altitude of about 7–8 km, over a period of roughly 30 sec.

2.3. Data Reduction and Analysis

[11] The data reduction and analysis procedure for MSX/UVISI extinctive-refractive stellar occultation observations has been described by Yee et al. [2002]. Following the determination of the apparent star positions and thus refraction angles, the imaging capabilities of UVISI are used for the identification and subtraction of airglow contamination from the stellar spectra. The refraction angles imply an index of refraction profile, from which density, pressure, and temperature profiles are retrieved [Vervack et al., 2002]. The density profile determines the refracted light path and constrains the amount of Rayleigh scattering, yielding stellar transmission spectra from which atmospheric composition (e.g., ozone profile) is retrieved [DeMajistre and Yee, 2002].

3. SOLVE Stellar Occultations

[12] There were 31 MSX/UVISI stellar occultation observations during SOLVE, from 23 January through 4 March 2000. The details of each observation are shown in Table 1. Each occultation provides profiles of pressure, temperature, and ozone mixing ratio throughout the stratosphere and troposphere, down to an altitude of approximately 7 km. Of the 31 available events, six were found to lie outside the Arctic polar vortex (as indicated in Table 1) and have thus been excluded from our analysis. The inside-/outside-vortex distinction was made on the basis of occultation profile potential vorticity (PV) and position relative to stream function maxima on the 480- and 550-K potential temperature surfaces [e.g., Nash et al., 1996]. There was a clear separation between relatively high- (inside-vortex) and low-PV (outside-vortex) profiles. At 500 K, for example, the inside- and outside-vortex modified PV [Lait, 1994] values lay clustered above 28 and below 22 × 10−6 m2 km−1 s−1, respectively.

Table 1. MSX/UVISI Stellar Occultation Profiles During January–March 2000
ProfileDateaTime, UTLatitude, °NLongitude, °EVortexbStar NameVisual MagnitudeSpectral Type
  • a

    Date: MMM DD (day of year).

  • b

    Profile deemed inside (outside) the vortex by virtue of its relatively high (low) potential vorticity and position relative to stream function maxima on the 480- and 550-K potential temperature surfaces.

1Jan 23 (23)0.4571.3−63.0inα Lep2.60F0Ib
2Jan 24 (24)0.5171.2−65.4inα Lep2.60F0Ib
3Jan 24 (24)17.6971.235.8inα Lep2.60F0Ib
4Jan 25 (25)17.7571.233.4inα Lep2.60F0Ib
5Jan 26 (26)17.8071.131.1inα Lep2.60F0Ib
6Jan 27 (27)16.1471.154.6inα Lep2.60F0Ib
7Jan 28 (28)17.9271.026.4inα Lep2.60F0Ib
8Jan 31 (31)16.3770.845.3inα Lep2.60F0Ib
9Feb 01 (32)16.4370.842.9inα Lep2.60F0Ib
10Feb 06 (37)15.0479.336.6inβ Ori0.12B8Iab:
11Feb 07 (38)16.8279.17.2inβ Ori0.12B8Iab:
12Feb 08 (39)16.8679.128.0inκ Ori2.05B0Iab:
13Feb 09 (40)16.9278.924.5inκ Ori2.05B0Iab:
14Feb 10 (41)15.2678.847.0inκ Ori2.05B0Iab:
15Feb 11 (42)15.3178.643.6inκ Ori2.05B0Iab:
16Feb 12 (43)15.3778.440.3inκ Ori2.05B0Iab:
17Feb 14 (45)15.4472.365.9inα CMa−1.47A1V
18Feb 15 (46)17.2272.237.6outα CMa−1.47A1V
19Feb 16 (47)17.2772.135.2outα CMa−1.47A1V
20Feb 17 (48)17.3372.132.7inα CMa−1.47A1V
21Feb 18 (49)17.3972.030.3inα CMa−1.47A1V
22Feb 19 (50)17.4471.927.8inα CMa−1.47A1V
23Feb 20 (51)15.7871.951.3inα CMa−1.47A1V
24Feb 21 (52)15.8471.848.8inα CMa−1.47A1V
25Feb 21 (52)17.5671.823.0outα CMa−1.47A1V
26Feb 23 (54)15.9571.644.0outα CMa−1.47A1V
27Feb 24 (55)0.5471.6−85.5inα CMa−1.47A1V
28Feb 27 (58)16.1361.745.5outδ CMa1.84F8Iab:
29Feb 28 (59)14.4659.067.7outε CMa1.51B2Iab:
30Mar 02 (62)16.3458.742.5inη CMa2.40B5Iab:
31Mar 04 (64)16.4763.949.6inρ Pup2.81F6IIp…

3.1. Retrieved Temperature

[13] A comparison of MSX/UVISI temperature profiles with that from the analysis reported by the United Kingdom Meteorological Office (UKMO) [Swinbank and O'Neill, 1994] is shown in Figure 1. The UKMO temperature data have been interpolated onto a 1-km altitude grid for this comparison. The envelope represented with dotted lines shows the 2-σ standard deviations around the mean temperature difference at each altitude level. Overall, the agreement is excellent, with an average offset of −0.8 (±2.7, 1-σ) K for the altitude range shown. A similar, somewhat larger negative temperature bias between 30 and 35 km relative to UKMO has also been observed in aircraft lidar data during the SOLVE campaign [Burris et al., 2002]. In the ozone loss analysis (see below), we focus on the 400–600-K altitude range (∼16–24 km). Over this range, the temperature difference is −1.4 ± 2.3 K.

Figure 1.

MSX/UVISI-UKMO temperature difference. The MSX/UVISI retrieved temperatures from stellar occultations within the vortex (see Table 1) are compared with UKMO temperature (MSX/UVISI − UKMO), interpolated onto a 1-km altitude grid. The mean difference as a function of altitude is represented with a thick solid line, and the mean ±2-σ standard deviations at each altitude are represented with dotted lines. The average calculated 1-σ uncertainty in the temperature profiles due to the stellar occultation measurements and retrieval over 400–600 K is ±1.8 K (not shown). Average potential temperatures corresponding to the geometric altitude scale are included for reference.

[14] The MSX/UVISI-UKMO differences in potential temperature exhibit the same general characteristics (not shown) as temperature. The average difference is −0.8 (±7.5, 1-σ) K, with a negative bias from 30 to 35 km, similar to that in Figure 1. From 400 to 600 K, the average difference in potential temperature, which provides altitude registration, is −1.0 (±7.1, 1-σ) K, or −0.04 ± 0.29 km at 500 K.

3.2. Retrieved Ozone Mixing Ratio

[15] The results of the MSX/UVISI stellar occultation ozone mixing ratio retrievals for the 25 in-vortex profiles are shown in Figure 2 as a function of potential temperature in the lower stratosphere. The ensemble average descent rates for air parcels within the vortex observed by stellar occultation, averaged over 25-K intervals, are shown to indicate the extent of subsidence. Ozone loss is evident, particularly below 500 K. Due to other commitments of the MSX spacecraft, we were limited to relatively sparse sampling of the vortex, compared to other satellite measurements (e.g., POAM III makes approximately 15 solar occultations per day in each hemisphere [Hoppel et al., 2002]). The time of each profile is indicated at the top and bottom of Figure 2. Though there are only 25 profiles, the general ozone trend is apparent.

Figure 2.

Ozone mixing ratio in the lower stratosphere within the Arctic polar vortex, as measured by MSX/UVISI using the stellar occultation technique, from late January through early March 2000 during SOLVE. Altitude is represented in terms of potential temperature. The times of the MSX/UVISI stellar occultations are marked with diamonds along the upper and lower abscissas and identified by profile number, corresponding to the listing in Table 1. Vertical dotted lines indicate 1 February and 1 March. The dashed lines denote the ensemble average diabatic descent of air parcels within the vortex passing through the tangent points of the MSX/UVISI observations. The descent rates, averaged over ±12.5-K intervals, were calculated by diabatic trajectory analysis.

[16] Several representative MSX/UVISI ozone profiles have been presented by Yee et al. [2002] along with near-coincident ground- and space-based observations for intercomparison, including the 7 February 2000, SOLVE profile (#11 in Table 1). This particular profile was centered close to Ny-Ålesund (79°N, 12°E), where lidar data were available from a similar time period. The stellar occultation ozone number density is within roughly 5% of the lidar profile at the ozone peak, with an expected 1-σ statistical uncertainty in the occultation retrieval of ∼2.5%.

[17] To demonstrate the accuracy of the stellar occultation technique in the context of the SOLVE campaign, the MSX/UVISI-POAM III ozone mixing ratio differences are shown in Figure 3. Three-dimensional ozone fields were reconstructed from POAM III data collected within ±3-day windows around each MSX/UVISI occultation using ozone-PV correlations [Randall et al., 2002] and then interpolated to the occultation geographic location for comparison. While POAM III measures ozone down to the tropopause, ozone fields are not reconstructed below 400 K (∼16 km). The UVISI-POAM ozone difference over the altitude region shown is −0.14 (±0.60, 1-σ) ppmv. Over 400–600 K, the difference is +0.02 ± 0.36 ppmv (+1.5 ± 12.9%).

Figure 3.

MSX/UVISI-POAM III ozone difference. The MSX/UVISI retrieved ozone mixing ratios from stellar occultations within the vortex (see Table 1) are compared with ozone derived from POAM III ozone: potential vorticity reconstructions [Randall et al., 2002] in terms of (a) absolute difference (MSX/UVISI − POAM III) and (b) percent difference ((MSX/UVISI − POAM III)/POAM III × 100%). The mean differences as functions of altitude are represented with thick solid lines, and the means ±2-σ standard deviations at each altitude level are represented with dotted lines. The average calculated 1-σ uncertainty in the ozone mixing ratio profiles due to the stellar occultation measurements and retrieval over 400–600 K is ±0.11 ppmv (not shown). Average potential temperatures corresponding to the geometric altitude scale are included for reference.

[18] Hoppel et al. [2002] found that POAM III ozone measurements during February were not significantly affected by latitudinal sampling (and thus solar exposure) biases. Despite the relative paucity of MSX/UVISI data, which makes it impossible to derive a true vortex average ozone morphology from UVISI SOLVE stellar occultations alone, the excellent average UVISI-POAM agreement suggests an absence of substantial solar exposure bias in the air masses remotely sensed by stellar occultation.

4. Photochemical Ozone Loss

[19] One goal of SOLVE was to quantify the rate of high-latitude photochemical ozone loss. During the polar winter, there is large-scale, continuous subsidence within the cold vortex [Schoeberl et al., 1992]. Before ozone retrievals, such as those in Figure 2, can be used to make any statement about chemical ozone loss, the effects of dynamics within the vortex must be accounted for. Ozone mixing ratios increase with altitude within the lower stratosphere, thus wintertime subsidence tends to increase the ozone mixing ratio at any particular point in the atmosphere with time and mask part of the effect of actual chemical ozone destruction. We have used diabatic trajectory calculations to keep track of individual air parcels and relate observed ozone profiles within the vortex after the effective removal of dynamical influence.

4.1. Trajectory Analyses

[20] Diabatic back and forward trajectories were calculated with the GSFC trajectory model [Schoeberl and Newman, 1995] using UKMO winds from every vortex profile listed in Table 1. Air parcels in 5-K increments from 400 to 800 K (∼16–28 km) were advected back to 15 January and forward to 31 March to produce a continuous 2.5-month history for the parcels (2511 total) in the lower stratosphere at the tangent points of each of the stellar occultation events. There was largely uniform descent within the lower stratosphere vortex, with the 500-K (∼21-km) surface sinking ∼40 K from 15 January to 31 March (∼25 K from 23 January to 4 March).

4.1.1. Vortex Average Descent Analysis

[21] Diabatic trajectory calculations were used to compute average descent rates within the vortex. Multiweek trajectory calculations are reliable only in a statistical sense, but on average, however, the trajectories accurately describe vortex subsidence. The 2025 parcel trajectories that were calculated for the 25 in-vortex occultations were divided into 5-K layers on 23 January (initial ozone profile), and average descent rates were computed for each layer. With knowledge of the average descent rates, it was then possible to map the initial ozone profile diabatically onto each subsequent profile. The difference in ozone mixing ratio represents the degree of chemical ozone change since 23 January. Though the initial condition in this case comprises a single profile, it lies within the range of vortex profiles derived from POAM III data, thus providing a reasonable representation of the initial state of the vortex for this analysis. The relative differences in the retrieved ozone are shown in Figure 4. The contour plot, which is constructed similarly to that in Figure 2, reveals the actual, chemical ozone loss within the vortex relative to the 23 January observation computed not along potential temperature surfaces, but along the diabatic descent curves.

Figure 4.

Photochemical ozone loss within the polar vortex relative to the first ozone profile, obtained on 23 January. The construction of the plot is similar to Figure 2. When computing the relative chemical ozone loss, the effects of vortex subsidence were removed by mapping the initial, 23 January occultation onto each subsequent profile using vortex average descent rates and comparing the retrieved ozone mixing ratios.

4.1.2. Individual Trajectory Analysis

[22] MSX/UVISI data were relatively limited during SOLVE and sampled different parts of the vortex over time. The ozone loss rates implied by the occultations (and corresponding trajectory calculations) over several weeks, however, should provide a good indication of the average loss rate over the time period.

[23] Trajectories were used to follow individual air parcels from the occultation profiles forward to 4 March, the date of the final occultation. Diabatic levels were identified by their potential temperatures on 4 March. Pairs of adjacent trajectories were used to interpolate each profile onto diabatic levels intersecting a 5-K grid on 4 March. Each profile therefore provided a single ozone measurement on every diabatic surface. The results on three such surfaces are shown in Figure 5. Each plot represents vertical cross sections parallel (on average) to the dashed diabatic descent curves of Figure 2. The diabatic surfaces intersect the 575-, 500-, and 425-K potential temperature surfaces on 4 March. The slopes of the measurement points are equivalent to the average daily ozone loss rates at each level during the period and are independent of the initial condition (23 January occultation). There is little net change in ozone mixing ratio at 575 K, but there is increasing loss with decreasing altitude to a maximum loss near 425 K.

Figure 5.

Photochemical ozone loss within the polar vortex along three particular diabatic surfaces corresponding to the (a) 575-, (b) 500-, and (c) 425-K potential temperature surfaces on 4 March, using diabatic trajectory analysis. The ozone mixing ratios indicated were those measured by each of the 25 stellar occultations within the vortex during SOLVE. The error bars denote the 1-σ uncertainty of the occultation measurements and retrieval and do not include errors in the trajectory calculations. The average daily loss rates for the period (i.e., the absolute values of the slopes resulting from weighted linear regressions through the observations on each surface), are also shown, along with the uncertainties in the fits.

4.1.3. Ozone Time Series Trajectory Analysis

[24] Another, closely related approach utilizing diabatic trajectories has been described by Schoeberl et al. [2002], where ozone measurements were injected into a free-running diabatic trajectory calculation. Similar to the preceding analysis, the lower stratosphere on 4 March was divided into 5-K potential temperature layers. All of the parcels within each layer having originated from earlier MSX/UVISI observations were identified using diabatic trajectory calculations. A time series was then constructed for each diabatic layer, with the measured ozone mixing ratios plotted versus the elapsed time from the occultation to 4 March. Such an ozone time series is shown in Figure 6 for the 425-K surface (analogous to Figure 5c). The average ozone loss rate is then simply the slope of the linear regression through the observation points. Because more or fewer than one trajectory air parcel from each occultation may be within each 5-K layer on 4 March, not every profile will be represented with a single data point (there may be more or fewer than 25 points on each surface).

Figure 6.

Ozone loss inferred from ozone time series at 425 K on 4 March. Each symbol represents an air parcel within 2.5 K of 425 K on 4 March, the date of the final MSX/UVISI stellar occultation event, which had been sampled by an earlier occultation. The ozone mixing ratios indicated were those measured at the times of the earlier occultations, and the parcel age is simply the elapsed time since those observations on 4 March. The error bars denote the 1-σ uncertainty of the stellar occultation measurements and retrieval and do not include errors in the trajectory calculations. In each layer, it is possible for there to be more or fewer than one trajectory parcel sampled in an earlier occultation. The average daily loss rate is inferred from the slope of the weighted linear fit. This diabatic surface corresponds to that in Figure 5c.

4.2. Ozone Loss Profiles

[25] The ozone loss rates for consecutive layers, which are individually depicted in Figures 5 and 6, produce a more complete picture of lower stratospheric ozone change when combined to produce a profile, as in Figure 7. All three trajectory analysis methods discussed above for deriving ozone loss are represented in the figure. Though they differ at 5-K resolution (compare Figures 5c and 6), the different analyses provide nearly identical results when averaged over 25-K intervals.

Figure 7.

Profiles of the average daily rates of ozone change and total ozone destruction, from 23 January to 4 March 2000. The results from three different but related methods, utilizing average vortex descent rates and two analyses using individual diabatic trajectory calculations, are shown. The data have been averaged over 25-K layers, and the lined region contains the average 1-σ uncertainties of the linear fits of the ozone measurements (and thus the rates of ozone change) on the individual diabatic surfaces within each layer.

[26] The overall rate of ozone loss below 500 K is very nearly 1 ppmv for this time period, which is consistent with other ozone loss analyses [e.g., Sinnhuber et al., 2000; Richard et al., 2001; Hoppel et al., 2002; Rex et al., 2002; Schoeberl et al., 2002; Lait et al., 2002]. The 450-K average daily loss rate, 0.024 ppmv/day, agrees well with the corresponding loss rate from the ozonesonde analysis of Schoeberl et al. [2002], for example, which was ∼0.025 ppmv/day at 465 K in mid-February. Ozone change rates diminish above 500 K and even become slightly positive above 575 K, as has been seen by Hoppel et al. [2002] and Schoeberl et al. [2002], though perhaps only the result of extravortex contamination [Schoeberl et al., 2002].

[27] It is important to note that the available MSX/UVISI occultations span only the central portion of the time periods of ozone loss in the other analyses, which report higher total ozone loss over larger portions of the December 1999 to late March 2000 period. The quasi-conservative mapping method of Lait et al. [2002], when used to derive the ozone loss rate from ozonesonde data over the same time period used in this paper (23 January through 4 March 2000), yields an average loss rate of 0.030 ± 0.008 ppmv/day at 450 K (L. R. Lait, personal communication, 2001).

5. Summary and Conclusions

[28] We have used an innovative, extinctive-refractive stellar occultation technique to measure atmospheric composition within the Arctic polar vortex. This technique allows for the simultaneous retrieval of total density, pressure, temperature, and ozone mixing ratio. Twenty-five profiles within the vortex were derived from MSX/UVISI measurements over a 41-day period during the SOLVE campaign. Table 2 summarizes the excellent agreement between the stellar occultation retrievals of temperature, potential temperature, and ozone mixing ratio with other measurements.

Table 2. Summary of Comparisons of MSX/UVISI Stellar Occultation Retrievals With Other Analyses
Retrieved QuantityDifferenceaComparison
  • a

    Mean difference (MSX/UVISI − comparison data) ±1-σ standard deviation, over 400–600 K potential temperature (∼16–24 km).

  • b

    Ozone mixing ratio from ozone-PV reconstructions.

Temperature−1.4 ± 2.3 KUKMO
Potential temperature−1.0 ± 7.1 KUKMO
Ozone mixing ratio+0.02 ± 0.36 ppmv (+1.5 ± 12.9%)POAM IIIb

[29] Using diabatic trajectory calculations, we have removed the effects of large-scale descent within the vortex and thus inferred the rate of photochemical ozone loss during polar winter. We find a maximum ozone loss of about 1 ppmv at 400–500 K (∼16–21 km) from 23 January through 4 March 2000, corresponding to an average daily loss rate of ∼0.024 ppmv/day, in agreement with other SOLVE determinations of ozone loss [Newman et al., 2002].

[30] Though the MSX/UVISI instrument was not optimized for stellar occultation and spacecraft commitments limited the number of measurements that could be performed during SOLVE, the data set demonstrates the utility of stellar occultation from a space-based platform as a means of accurately measuring bulk atmospheric quantities and stratospheric ozone. A dedicated stellar occultation instrument, such as the Global Ozone Monitoring by Occultation of Stars (GOMOS) [Bertaux et al., 1991], could provide hundreds of profiles at high vertical resolution and near-global coverage each day. The extinctive-refractive technique described here has the further advantage of the ability to probe the atmosphere down into the upper troposphere (a region that is under-sampled by other satellite techniques and is vital to understanding climate change).


[31] We greatly appreciate the POAM III ozone-PV reconstructions and helpful discussions provided by K. W. Hoppel (Naval Research Laboratory) and C. E. Randall (LASP, University of Colorado), as well as comments by H. K. Roscoe (British Antarctic Survey) and another, anonymous referee. This analysis and MSX/UVISI operations were supported by NASA Atmospheric Chemistry Modeling and Analysis Program grants NAG-5-9988, NAG-5-7552, and NAG-1-2270.