Evolution of ozone and ozone-related species over Kiruna during the SOLVE/THESEO 2000 campaign retrieved from ground-based millimeter-wave and infrared observations

Authors


Abstract

[1] Vertical profiles of stratospheric ozone and ozone-related trace gases were measured in winter 1999/2000 using the ground-based millimeter-wave radiometer MIRA 2 and a Fourier-transform infrared spectrometer (FTIR) located at the Swedish Institute of Space Physics at Kiruna. The MIRA 2 measurements covered all three SOLVE/THESEO 2000 flight phases. An almost complete time series of O3 profiles and complementary profiles of ClO, HNO3, and N2O were achieved. Profiles of O3, HCl, HNO3, and N2O, as well as stratospheric column amounts of NO2, ClONO2, and ClO, were obtained from the continuous ground-based FTIR measurements between January and March 2000. From the measurements of N2O and HF, a diabatic subsidence inside the polar vortex of about 1.2 km between January and March for altitudes of about 20 km was deduced. On 26 and 28 January, an uptake of about 25% of stratospheric HNO3 by polar stratospheric clouds (PSCs) could be measured and between January and March, an significant denitrification of the lower stratosphere was derived from the measurements. Strong chlorine activation was detected by both instruments in January and March, resulting in an ozone loss of more than 1 ppmv in a layer below 23 km. In comparison to recent cold winters, this layer was found to be quite thin. Therefore the total loss measured in column amount of (1.2–1.4) × 1022 molec/m2 was smaller than the losses in previous cold winters.

1. Introduction

[2] Since the discovery of the “ozone hole” over Antarctica by Farman et al. [1985] it has been shown that chlorine plays a key role in ozone destruction in the stratosphere [Molina and Molina, 1987]. It is released from reservoir gases by heterogeneous reactions on the surfaces of polar stratospheric clouds (PSCs) [Crutzen and Arnold, 1986] which are formed at very low temperatures. The same processes cause ozone depletion in the Arctic stratosphere [European Commission, 1997], but usually planetary wave activity disturbs the polar vortex, thus leading to high ozone variability. These effects make it difficult to separate chemical ozone loss from dynamic processes and are the subject of current investigations.

[3] In winter 1999/2000, one of the biggest field campaigns ever in the northern hemisphere took place, the joint European and U.S. SOLVE/THESEO 2000 campaign. Many measurements of stratospheric ozone, aerosols and ozone-related species were performed by various ground-based, airborne and balloonborne instruments. The winter 1999/2000 was one of the coldest in the last ten years and temperature in the lower stratosphere dropped below the PSC type I threshold for long periods from December 1999 until early March 2000 [Manney and Sabutis, 2000].

[4] Ground-based millimeter-wave and infrared observations were performed at the Swedish Institute of Space Physics at Kiruna (67.8 N, 20.4 E, 425 m ASL), Northern Sweden. This site is located in the lee of the Scandinavian mountains ridge and is therefore well suited for ground-based measurements in winter due to the relatively small amount of tropospheric water vapor [Berg, 2000]. In addition, this location is a good compromise of being sufficiently far in the north to investigate Arctic ozone loss and sufficiently southern to have a short polar night for measuring in absorption geometry towards the Sun with infrared spectroscopy.

[5] During winter Kiruna is situated alternately inside the polar vortex, at the vortex edge region or outside the vortex. Of particular interest are lee wave induced polar stratospheric clouds (PSCs) which can be observed regularly over this location. These can cause additional chlorine activation by heterogeneous reactions on the surface of the PSC particles leading to enhanced ozone loss [Carslaw et al., 1998].

[6] The aim of the measurements was the nearly continuous monitoring of several stratospheric constituents which play an important role in ozone chemistry from early winter when the setup of the polar vortex takes place to spring when the vortex breaks up. The present paper focuses on chlorine activation and deactivation in the lower stratosphere in the course of the winter, which plays a key role for chemical ozone depletion, and the resulting ozone loss in the vertical profiles and in total stratospheric column amounts. To eliminate dynamic effects, the inert tracers N2O and HF are used to study the diabatic subsidence inside the polar vortex. Furthermore, the uptake of HNO3 by lee wave induced PSCs is estimated from the measurements.

2. The Millimeter-Wave Radiometer

2.1. Experimental Setup

[7] The millimeter-wave observations were performed with the ground-based Millimeter-wave Radiometer MIRA 2. It was constructed at the Forschungszentrum Karlsruhe and measures O3, ClO, HNO3, and N2O in the frequency range from 268 to 280 GHz in emission geometry. MIRA 2 can measure in every direction, but only northward measurements were used for the present paper. A Schottky diode mixer converts the signal to an intermediate frequency of 1.5–2.7 GHz. The receiver noise temperature is about 800 K. Spectral analysis is performed by an acousto-optical spectrometer with a spectral resolution of about 1.2 MHz. For balanced calibration an internal adjustable reference load is used. A detailed description of the system is given in Berg et al. [1998]. The measured spectra are integrated until either the noise in the resulting spectrum is at least ten times smaller than the intensity of the signature of the desired trace gas or until the noise is significantly smaller than baseline artifacts. Integration times depend on the tropospheric conditions and range typically from 0.5–1 hours for ozone up to 3–4 hours for ClO.

2.2. Data Analysis

[8] The radiative transfer model uses spectroscopic data taken from the HITRAN 96 database [Rothman et al., 1998], except for ClO. For this species spectroscopic data from the JPL catalogue [Pickett et al., 1998] and the JPL measurement made in 1992 for the pressure broadening coefficient is used. For the temperature dependence of the pressure broadening parameter, the measurements of Oh and Cohen [1994] are applied. The daily profiles of pressure and temperature from the National Centers for Environmental Prediction (NCEP) [Kanamitsu, 1989] are taken for the forward calculations. Retrieval is carried out using a modified Optimal Estimation Method [Rodgers, 1970, 1976, 1990], including the simultaneous retrieval of several constituents within the same inversion process [Kuntz et al., 1999], and the fit of sinusoidal undulations caused by standing waves within the inversion [Kuntz et al., 1997].

[9] With these methods volume mixing ratio (vmr) profiles in the altitude range from about 17–55 km can be retrieved. Using the full width at half maximum (FWHM) of the averaging kernels as a criterion, a vertical resolution of the ozone profiles of at best 7 km can be achieved with an uncertainty of at least 1 ppmv caused by errors due to thermal noise, standing waves, and systematic errors [Kopp, 2000]. For ClO the vertical resolution in the retrieved profiles is at best 10 km and the uncertainty 0.5 ppbv at least. The profiles of HNO3 and N2O have a vertical resolution of at best 12 km and an uncertainty of at least 1 ppbv and 50 ppbv, respectively. As an example, the left-hand side of Figure 1 shows the vertical resolution of the MIRA profiles of 31 January.

Figure 1.

Vertical resolution of the profiles measured on 31 January 2000 by MIRA (left) and the FTIR (right) using the FWHM of the averaging kernels as a criterion. Note the different altitudes at which the instruments are sensitive.

3. The Infrared Spectrometer

3.1. Experimental Setup

[10] Since March 1996, infrared solar absorption spectra have been routinely recorded at Kiruna within the framework of the NDSC (Network for the Detection of Stratospheric Change). A Bruker 120 HR spectrometer with a maximum optical path difference of 360 cm is used, yielding a spectral resolution of up to 0.002 cm−1. Two detectors, MCT (Mercury-Cadmium-Telluride) and InSb (Indium-Antimonide), cover the spectral range of 700–5000 cm−1. The NDSC filter set (F. Murcray, personal communication, 1996) is used to increase the signal to noise ratio. Spectra are coadded for up to 15 min during noon and 5 min during sunrise and sunset in order to limit the variation of the solar zenith angle to 0.2°. The signal to noise ratio in the line-free continuum amounts to several hundreds. This results in a signal to noise ratio of several hundreds for the signatures of the constituents discussed in this paper, except for ClO. This species has by far weaker signatures which are in the order of the noise level. An NDSC side by side intercomparison was performed successfully in March 1998 [Meier et al., 1999] and the found differences were less than 3% for all compared species.

3.2. Data Analysis

[11] The FTIR spectra are analyzed with the inversion program PROFFIT (PROFile FIT) [Hase, 2000] using the forward model KOPRA (Karlsruhe Optimized Precise Radiative transfer Algorithm) [Stiller et al., 1998; Stiller, 2000]. The synthetic spectra are calculated using daily pressure and temperature data of the National Centers for Environmental Prediction (NCEP) [Kanamitsu, 1989]. Spectroscopic data are taken from the HITRAN 96 database [Rothman et al., 1998]. The inversion code PROFFIT allows the retrieval of profiles from the absorption line shape using the Optimal Estimation Method of Rodgers [Rodgers, 1970, 1976, 1990], the Phillips-Tikhonov approach [Phillips, 1962; Tikhonov, 1963], or scaling of a priori profiles in user-defined altitude intervals. For some of the species the inversion is performed on a logarithmic scale of the volume mixing ratio in order to avoid negative values. The strength of the regularization constraint is adjusted by reducing the impact of the regularization condition until the fit quality is no longer increased significantly. For profile retrieval an accurate knowledge of the instrumental line shape (ILS) is necessary. The ILS is derived from regular cell measurements using the LINEFIT software [Hase et al., 1999].

[12] Using the FWHM of the averaging kernels as a criterion, four independent layers in the ozone profiles can be retrieved at altitudes ranging from 5–35 km with an accuracy of about 15% in the partial columns in each layer, limited by thermal noise and systematic errors [Hase, 2000]. In the HNO3 profiles, a vertical resolution of three independent layers in the altitude range between 10 and 30 km can be achieved. Measurement accuracy in the total column is about 20%. N2O profiles can be retrieved from 0–30 km with a resolution of three independent layers. The error in the total column is about 13%. ClO column amounts are retrieved with an accuracy of 17% due to systematic errors and 1018 molec/m2 as a result of thermal noise. HCl profiles can be retrieved between 10–30 km with 3–4 independent layers and an error in the total column of 12%. As an example, the right-hand side of Figure 1 shows the vertical resolution of the FTIR profiles of 31 January.

4. Results

[13] The MIRA measurements lasted from late November 1999 to early April 2000 with an interruption from mid-December 1999 to mid-January 2000. During the presence of NASA's DC-8 in Kiruna and under favorable weather conditions, the instrument was tuned alternately for the detection of all species measurable by MIRA. Otherwise, the radiometer measured ozone exclusively. The FTIR measurements started at the end of January 2000 when the Sun rose sufficiently above the local horizon because direct sunlight is needed. To avoid too many breaks in the time series shown in this paper, missing data of single days were interpolated.

[14] Figure 2 shows the potential vorticity (PV) and temperature on the 475 K isentropic level over Kiruna for the entire period covered by the measurements. From January to early February (day 18–37) and from the end of February to mid-March 2000, Kiruna was almost continuously well inside the polar vortex (PV > 42 × 10−6 K m2/kg s) and the temperature was sometimes below the PSC type I threshold (T ≤ 193K). During the rest of the time, Kiruna was outside or at the edge of the vortex.

Figure 2.

Temperature and potential vorticity over Kiruna on the 475 K level in winter 1999/2000 (courtesy of the European Centre for Medium-Range Weather Forecasts).

4.1. Nitrous Oxide

[15] Figure 3 shows the retrieved N2O profiles of MIRA from December 1999 to March 2000 as well as the FTIR measurements from January to March 2000. In December during the vortex setup, MIRA data show no significant correlation with the potential vorticity at 475 K (Figure 2). In January and early February, both data sets show a good correlation with the potential vorticity at 475 K. Especially on 15, 25, and 26 January when Kiruna was outside the polar vortex or at the vortex edge, an enhanced N2O level was observed. It is obvious that MIRA measures more N2O at higher altitudes than the FTIR and shows more variability. This might be an artifact due to the limited sensitivity and altitude resolution of MIRA for N2O (see Figure 1) leading to smoother profiles which range to unrealistically high altitudes.

Figure 3.

Time series of nitrous oxide profiles over Kiruna in winter 1999/2000 as measured by MIRA (top) and the FTIR (bottom).

[16] In February when Kiruna was well outside the polar vortex (days around day 41, 46, and 53), the N2O profiles show larger volume mixing ratios above an altitude of 20 km. During days within that period when Kiruna was at the vortex edge region, e.g., the days around day 44 and day 49, the FTIR results exhibit smaller N2O values above 20 km due to the diabatic subsidence inside the vortex or mixing between intravortex and extravortex air. In March when Kiruna was again well inside the polar vortex (day 57–72), the measurements show N2O at somewhat lower altitudes than in January due to further subsidence inside the vortex.

[17] To estimate the diabatic subsidence inside the polar vortex from the second half of January to the first half of March, levels of constant volume mixing ratio of N2O are used as an inert tracer. In addition, the FTIR measurements of the inert tracer HF are taken for a consistency check. The results are summarized in Table 1, together with 1σ standard deviations. Average values for the diabatic subsidence of 1.0 ± 1.7 km for MIRA and of 1.2 ± 0.3 km for the FTIR are derived. The results agree within the limits of the standard deviations which are very large in the MIRA measurements due to the limited sensitivity of this instrument for N2O. These results are also consistent with the value of 1.6 km between late January and mid-March for the 100 ppbv N2O level as derived from measurements of the ASUR instrument aboard the NASA research aircraft DC-8 [Bremer et al., 2002].

Table 1. Diabatic Subsidence Inside the Polar Vortex From Late January to Early March 2000
 MIRA, kmFTIR, km
Using N2O as Inert Tracer
120 ppbv-isoline1.2 ± 2.51.3 ± 0.4
160 ppbv-isoline0.9 ± 2.41.4 ± 0.5
 
Using HF as Inert Tracer
0.6 ppbv-isoline1.0 ± 0.6
0.9 ppbv-isoline1.4 ± 0.7
1.3 ppbv-isoline1.1 ± 0.9
Mean values1.0 ± 1.71.2 ± 0.3

[18] The measurement results suggest that a value of 1.2 km can be assumed for the diabatic subsidence at altitudes of about 20 km from the second half of January to the first half of March. This value will be used in section 4.4 to calculate the ozone loss.

4.2. Nitric Acid

[19] Time series of HNO3 of the measurements in winter 1999/2000 are shown in Figure 4. Measurements with MIRA were performed in December 1999 and from mid-January until mid-March with some interruptions in February. FTIR measurements were accomplished from the end of January to mid-March with some gaps due to unfavorable weather conditions.

Figure 4.

Time series of nitric acid profiles over Kiruna in winter 1999/2000 as measured by MIRA (top) and the FTIR (bottom).

[20] On 5–16 December temperature was too high for PSCs (Figure 2) and lidar measurements carried out by the Bonn University at ESRANGE near Kiruna did not exhibit any PSC activity for this period [Blum et al., 2000]. At the same time, MIRA measured over 10 ppbv HNO3 at an altitude of about 23 km with small variations only. The MIRA measurements suggest that the HNO3 distribution is quite homogeneous in the polar stratosphere in early winter in the absence of widespread PSCs. During this period, the ASUR instrument measured maximum volume mixing ratios of 8 ppbv up to above 11 ppbv on equivalent latitudes ranging from 60° to 85° [Kleinböhl et al., 2002]. The maximum mixing ratios were found at about 23 km. These results are in good agreement with our findings.

[21] When comparing the time series of MIRA from January to March with the FTIR time series, the better vertical resolution of the latter instrument for HNO3 is obvious. The HNO3 layer appears thinner than that of the MIRA measurements. Therefore volume mixing ratios are higher, which is especially exident on day 31. Both, MIRA and the FTIR reveal reduced maximum volume mixing ratios of HNO3 on several days before 30 January. On these days, the temperature according to the ECMWF analyses was low enough for the formation of PSCs (Figure 2) and the lidar observations at ESRANGE found PSCs of types I and II during this time span [Blum et al., 2000]. On 26 January, the FTIR line of sight crossed a PSC layer and Höpfner et al. [2001] concluded from the spectral signature of the layer that the PSC particles on this day were ice particles. However, this was a low-resolution measurement aiming at the PSC properties. The trace gas profiles and columns discussed in this paper are deduced from a high-resolution measurement taken earlier on the same day.

[22] The fraction of HNO3 included in the PSC particles can be estimated from the measurements of MIRA and the FTIR as follows. Assuming that the horizontal HNO3 distribution was quite homogeneous in the winter polar stratosphere during the absence of PSCs (see the HNO3 results of December in Figure 4), we can compare the stratospheric column amounts of 26 and 28 January with that of 31January. The lidar measurements in the night from 30 to 31 January showed weak PSC but no lee wave activity [Blum et al., 2000; U. Blum, personal communication, 2001], but the measurements of both, MIRA and the FTIR, show no indications for HNO3 uptake by PSCs. The reason for this seeming discrepancy is probably the temporal offset of the measurements since the MIRA and FTIR measurements took place in the afternoon of 31 January. The ECMWF analyses show a strong increase in the temperature from 30 to 31 January, supporting the assumption that temperature was too high for PSCs during the MIRA and FTIR measurements. Therefore we use these measurements as a reference for a day inside the polar vortex and without PSC activity.

[23] Since Kiruna was at the edge of the vortex on 26 January, the effect of the different dynamic situation must be eliminated by dividing the HNO3 columns of 26 and 31 January by the columns of the inert stratospheric tracer HF measured by the FTIR on these days. For example, this procedure is discussed by Chipperfield et al. [1997] and Mellqvist et al. [2002]. After having calculated the percent HNO3 loss of 26 January from these corrected values, the total loss can be calculated from the stratospheric column amount of HNO3 on 31 January. The total HNO3 uptake of 28 January was estimated by direct comparison with the stratospheric column amount of 31 January since the dynamic situation on these days was nearly the same.

[24] For MIRA values between (4–5) × 1019 molec/m2 = 24–29% were obtained. The FTIR values range between (5–6) × 1019 molec/m2 = 20–24% of HNO3 which was included in PSC particles (Table 2). These values are in the order of the typical uptake of HNO3 by PSCs (19–31%) as found by Wegner et al. [1998] in the cold Arctic winters of 1991/1992 and 1992/1993 using a ground-based Michelson Interferometer for Passive Atmospheric Sounding-Laboratory Model (MIPAS-LM) at Kiruna. However, they do not reach the highest uptake of 50% measured on 10 February 1993. From comparisons with model calculations they concluded that the observed HNO3 losses could only be explained by the formation of nitric acid trihydrate (NAT) PSCs. However, it is not possible to distinguish between the PSC types with the measurements presented in the present paper, the more so as the observed PSCs were lee wave induced and possibly inhomogeneous.

Table 2. Results for HNO3 of MIRA and the FTIR
 MIRAFTIR
Amount of HNO3Included in PSCs (molec/m2)
26 January4 × 10195 × 1019
28 January5 × 10196 × 1019
 
Mean Stratospheric Column Amounts (molec/m2)
December1.8 ± 0.2 × 1020
31 January1.7 × 10202.5 × 1020
March0.9 ± 0.1 × 10201.8 ± 0.1 × 1020
 
Total Loss From Late January to Early March (molec/m2)
Total loss8 ± 1 × 10197 ± 1 × 1019
 
Denitrification From Late January to Early March (molec/m2)
Denitrification6 ± 2 × 10195 ± 2 × 1019

[25] It is obvious from Figure 4 that in March both instruments measured smaller HNO3 maximum mixing ratios than on days in December and January when temperatures were too high for PSCs. This is partly a result of the photolysis of HNO3 and the release of NO2. In such a cold stratospheric winter, however, part of the observed loss might be due to denitrification as observed in recent cold Arctic winters [Sugita et al., 1998; Rex et al., 1999; Kondo et al., 2000]. If the mean HNO3 distribution is again assumed to be quite homogeneous in the vortex in early winter, the denitrification can be estimated as follows.

[26] Mean values of the stratospheric column amounts above 10 km of HNO3 were calculated for December and March on days when Kiruna was well inside the polar vortex and temperatures were too high for PSCs (see Figure 2). These values were then compared with the column amount above 10 km of 31 January when Kiruna was well inside the polar vortex and no PSC incidence could be found. From December to 31 January, MIRA results show no significant HNO3 decrease, but from 31 January to March, MIRA and FTIR measurements reveal a significant HNO3 loss. To estimate the amount of HNO3 loss caused by photolysis, the FTIR measurements of NO2 and ClONO2 are taken into account. The sum of the stratospheric column amounts of these two constituents increases by (2 ± 1) × 1019 molec/m2 from January to March. Assuming that this increase was caused by photolysis of HNO3, we can correct the calculated HNO3 loss for this amount. The HNO3 loss caused by denitrification then is (6 ± 2) × 1019 molec/m2 = 35% for MIRA and (5 ± 2) × 1019 molec/m2 = 20% for the FTIR. These results are summarized in Table 2.

[27] These calculations do not consider the increase in the stratospheric column amounts of HNO3 due to air masses originating from lower latitudes and entering the top of the polar vortex. The MIRA measurements can be used to estimate this effect since this instrument is sensitive up to 55 km. Firstly, the diabatic subsidence inside the polar vortex is interpolated linearly, starting from 0 km for an altitude of 10 km and using the value of 1.2 km found in section 4.1 for an altitude of 20 km. Extrapolating this increase in the diabatic subsidence results in a value of about 3 km in the altitude range from 38 km to 35 km. The partial column amounts of HNO3 above 35 km derived from the MIRA measurements inside the polar vortex in March show an increase of (5 ± 1) × 1017 molec/m2 compared to the partial column amounts above 38 km of January/February. It can be concluded that this dynamic effect is negligible.

[28] One reason of the discrepancies in denitrification between MIRA and the FTIR might be the different altitude range in which the respective instrument is sensitive. The MIRA profiles below an altitude of about 16 km are determined by the used a priori profile which was a smooth profile with low total column. The FTIR is sensitive in lower altitudes than MIRA, which would explain the larger column amounts of HNO3. Furthermore, if HNO3 is transported into lower altitudes by sedimentation of PSC particles during winter and then released due to evaporation, it might become undetectable by MIRA, whereas the FTIR would still be able to detect it. This would explain the somewhat smaller total HNO3 loss measured by the FTIR in comparison to MIRA. This interpretation is supported by the findings of Kleinböhl et al. [2002]. They found a NOy deficit between 1.1 ± 1 ppbv and 5.2 ± 2.6 ppbv in the altitude range from 16 km to 20.5 km. This means that significant renitrification took place below 16 km, at altitudes, where the FTIR is still sensitive in contrast to MIRA.

[29] The role of denitrification in Arctic stratospheric ozone loss was shown by Gao et al. [2001] using NASA ER-2 aircraft and balloonborne MkIV FTIR measurements in the 1999/2000 Arctic vortex. They found an increase in chemical ozone destruction from 43 ppbv/day to 63 ppbv/day when comparing 43% denitrified with 73% denitrified air parcels between 18–21 km. Since the values of denitrification derived from the MIRA and FTIR measurements are column amounts, the percent denitrification at certain altitudes probably is much larger than the 35% and 20%, respectively, found by both instruments. Therefore the findings of Gao et al. [2001] may be applicable for the air masses over Kiruna during the ground-based microwave and infrared measurements. The resulting ozone loss found by MIRA and the FTIR will be discussed in section 4.4.

4.3. Chlorine Activation

[30] In Figure 5 the FTIR results for HCl are shown from mid-January to mid-March. For altitudes above 25 km, the measurements show larger amounts of HCl during periods of Kiruna being well inside the polar vortex (January to early February, March) than at times when Kiruna was outside the vortex (February). This is the result of the diabatic subsidence inside the polar vortex. The unrealistically large volume mixing ratios of HCl around 30 km may be caused by uncertainties in the theoretical description of the HCl line shape by a Voigt function as stated by Varghese and Hanson [1984].

Figure 5.

Time series of hydrochloric acid profiles over Kiruna from January to March 2000 as measured by the FTIR.

[31] For the altitude range below 25 km, the measurements exhibit increased values of HCl in February and March when Kiruna was outside or at the edge of the polar vortex. However, the low HCl values in the lower stratosphere on days in January, early February, and the first half of March when Kiruna was well inside the polar vortex, are a first evidence of strong chlorine activation inside the cold polar vortex in winter 1999/2000.

[32] Figure 6 shows the ClO profiles retrieved from the MIRA measurements in November/December 1999 and January 2000. Only profiles around noon are shown. Hence the diurnal variation cannot be seen. In late 1999, no chlorine activation can be found in the profiles, whereas the profiles taken in January 2000 show a ClO layer in the lower stratosphere which is typical of chlorine activation. During this period, Kiruna was mostly well inside the polar vortex and temperatures on the 475 K isentropic level were below the PSC type I threshold (see Figure 2). No lower ClO maximum can be found on 25 and 26 January and also the FTIR results for HCl (Figure 5) show increased HCl values below 20 km, indicating weaker chlorine activation of the air masses observed on these days. The reason is that on 25 and 26 January, Kiruna was at the vortex edge (see Figure 2). However, the temperature on the 475 K isentropic level was low enough for PSCs on 25 January and Voigt et al. [2000] and Höpfner et al. [2001] measured lee wave induced PSCs on 26 January.

Figure 6.

Time series of chlorine monoxide profiles over Kiruna in winter 1999/2000 as measured by MIRA.

[33] For further investigation ECMWF, ten days back trajectories were used to obtain information about the history of the observed air masses. In order to give an overview of the chlorine activation in the course of winter and spring 1999/2000, Figure 7 shows stratospheric column amounts of ClO derived from the measurements. For comparison, the number of hours of possible contact with PSCs within the last ten days before observation is shown in the same figure. On day 40, 74, 76, and 77, the FTIR results are slightly negative, indicating that there was no chlorine activation in the lower stratosphere. Therefore the FTIR values for these days in Figure 7 were set to zero.

Figure 7.

Stratospheric ClO column amounts of MIRA and the FTIR (left) in comparison to the number of hours within the last ten days before observation time when contact of the observed air masses with PSCs was possible (right; courtesy of the European Centre for Medium-Range Weather Forecasts).

[34] On 25 and 26 January, possible contact with PSCs within ten days before observation time lasted for a view hours only in contrast to the rest of the days during this period in January. This would explain why both instruments did not detect chlorine activation on these days. The few hours of prior possible contact with PSCs for air masses observed on 25 and 26 January are an evidence of the PSCs observed on these days (see section 4.2) being a local feature due to lee waves.

[35] In February, when Kiruna was outside or at the edge of the vortex, both the FTIR and MIRA found no chlorine activation. In March, Kiruna was again inside the polar vortex and MIRA and the FTIR show an excellent agreement, except for days 70 and 71 when MIRA yields more chlorine activation. The reason of this discrepancy is not yet clear, and the ten days back trajectories reveal that contact of the observed air masses with PSCs had been possible for some hours. Consequently, there is no clear indication as to which column amount of ClO is plausible. One explanation could be the different directions of observation (MIRA measures in northern, the FTIR in southern direction). Hence these discrepancies might reflect mesoscale features. Between day 70 and 75, both instruments show decreasing column amounts indicating chlorine deactivation due to the release of NO2 from photolysis of HNO3, which is in good agreement with the results for HNO3 and ClONO2 of the FTIR (section 4.2). After day 75, Kiruna was outside of the polar vortex (Figure 2). As expected, no chlorine activation could be found.

[36] These results for chlorine activation are typical of such a cold winter like 1999/2000 and are in good agreement with the slant column amounts of OClO (which is often used as an indicator of chlorine activation) observed by the GOME experiment. Similar to the cold Arctic winters of 1995/1996 and 1996/1997, OClO stayed enhanced for several months in winter 1999/2000, in contrast to the relatively warm winters 1997/1998 and 1998/1999 which were characterized by sporadically elevated OClO amounts [Wagner et al., 2001]. The increase in OClO, indicating chlorine activation in the Arctic, is variable and anti-correlated to the seasonal variation of the stratospheric temperature [Wagner et al., 2001].

4.4. Ozone

[37] Figure 8 shows the retrieved ozone profiles of both instruments. One profile per day was taken for these time series.

Figure 8.

Time series of ozone profiles over Kiruna in winter 1999/2000 as measured by MIRA (top) and the FTIR (bottom).

[38] In November/December 1999, the results of MIRA show small maximum ozone mixing ratios of 4–5 ppmv at 35–40 km, which are typical of early winter. In January and early February 2000, maximum mixing ratios of ozone amount to about 6 ppmv and are now found below 35 km in the MIRA results. The FTIR profiles exhibit smaller maximum volume mixing ratios of about 4 ppmv only and the profiles are smoother than the profiles of MIRA. This is the result of the very low vertical resolution and the very limited sensitivity of ground-based FTIR ozone measurements in the middle stratosphere (see Figure 1).

[39] In the middle of February, Kiruna was outside the polar vortex and maximum mixing ratios increase in both measurements. The reason is the ozone-rich air from lower latitudes prevailing over Kiruna at that time. In the first half of March, Kiruna was well inside the polar vortex and again all volume mixing ratio profiles show reduced maximum mixing ratios of ozone. For the rest of the month, Kiruna was outside the polar vortex, resulting in enhanced maximum ozone mixing ratios in the MIRA measurements.

[40] To investigate the chemical ozone loss in the lower stratosphere, the ozone mixing ratios obtained on the 475 K isentropic level in January/February when Kiruna was inside the vortex were compared with those of the first half of March when Kiruna was again well inside the vortex. The measurements of MIRA in 1999 were not regarded in this study because the potential vorticity on the 475 K level over Kiruna was below 42 × 10−6 Km2/kgs for most of the time during this period (see Figure 1). But measurements using the ASUR instrument show that by far the largest ozone loss took place between late January and early March 2000 [Bremer et al., 2002], thus our measurements are a good estimate for total ozone loss in this winter.

[41] Figure 9 shows O3 mixing ratios on the 475 K isentropic level of the observations from day 22–34 of year 2000 in comparison to day 64–72 of the same year. In the first period the results of both measurements agree well, with mean values of 2.9 ppmv for MIRA and the FTIR. In the second period MIRA measures smaller values than the FTIR, with a mean of 1.7 ppmv corresponding to an ozone loss of 1.2 ppmv (41%) on the 475 K isentropic level. The FTIR measurements provide mean values of 2.1 ppmv and a resulting ozone loss of 0.8 ppmv (28%) (Table 3). The reason of the lower ozone loss found by the FTIR is probably the poorer altitude resolution of this instrument for ozone (see Figure 1): If ozone depletion took place in a quite thin layer, this decrease would be smoothed over a wider altitude range by FTIR measurements compared to MIRA. This results in a seeming lower ozone loss. The percentage ozone losses of both instruments are comparable to the values found by Braathen et al. [1998] for the cold Arctic winters of 1992/1993, 1994/1995, and 1996/1997 (33%, 39%, and 41%, respectively) using ozone sonde data, but do not reach the highest value of 50% that was found in the very cold winter 1995/1996.

Figure 9.

Ozone concentration on the 475 K isentropic level over Kiruna of MIRA and the FTIR in early and late winter 1999/2000.

Table 3. Ozone Loss of MIRA and the FTIR
 MIRAFTIR
Volume Mixing Ratio at 475 K Level, ppmv
Day 22–342.9 ± 0.32.9 ± 0.1
Day 64–721.7 ± 0.52.1 ± 0.2
Ozone loss1.2 ± 0.60.8 ± 0.2
 
Volume Mixing Ratio Calculated Under Consideration of Diabatic Subsidence, ppmv
Day 64–721.5 ± 0.31.8 ± 0.2
Ozone loss1.4 ± 0.41.1 ± 0.2
 
Partial Column Amounts Above 16 km From MIRA and Above 10 km From the FTIR, molec/m2
Ozone loss1.2 ± 0.5 × 10221.4 ± 0.4 × 1022

[42] Up to this point, the diabatic subsidence inside the polar vortex has not yet been taken into account. The value of this subsidence at an altitude of about 20 km was estimated in section 4.1 as 1.2 km from January to March. To consider this effect, the means of the days 64–72 were calculated on isentropic surfaces that were below those of the days 22–34 by an average of of 1.2 km. This yields an ozone loss of 1.4 ppmv for MIRA, corresponding to a percentage loss of 48%. The FTIR measurements provide an ozone loss of 1.1 ppmv, corresponding to a percentage loss of 38% (Table 3). These values are a good estimate for the chemical ozone loss, if dynamic processes like mixing of mid-latitude air into the vortex are negligible. Salawitch et al. [2002] show that dynamic processes could not have made a significant contribution to the observed change in ozone versus N2O vmr relation as observed by the OMS balloonborne observations.

[43] The left-hand side of Figure 10 shows the measured ozone loss in the lower stratosphere in volume mixing ratios calculated from the mean profiles of January/February and March 2000. The diabatic subsidence is considered by lifting the mean profiles of March in a linearly increasing manner from the tropopause so that the descent found in section 4.1 is evened out. It can be seen that most of the ozone depletion in volume mixing ratios took place below an altitude of 25 km, which is in good agreement with other measurements [Salawitch et al., 2002; Swartz et al., 2002]. As mentioned above, the differences between the measurements are a consequence of the different vertical resolution of the instruments, resulting in lower maximum ozone loss in the FTIR profile caused by stronger smoothing. The differences at 30 km might be the result of the very limited sensitivity of the FTIR measurements at higher altitudes due to the stronger Doppler broadening of the infrared spectral lines in comparison to the microwave region.

Figure 10.

Cumulative ozone loss from January/February to March 2000, as derived from the measurements.

[44] The values for ozone depletion show a reasonable agreement with other measurements of different instruments which participated in the SOLVE/THESEO 2000 campaign. These vary between 0.7 ppmv [Klein et al., 2002] and 2.2 ppmv [Müller et al., 2002]. However, most measurements provide values of ozone loss compatible with our results [e.g., Bremer et al., 2002; Hoppel et al., 2002; Rex et al., 2002; Richard et al., 2001; Salawitch et al., 2002; Sinnhuber et al., 2000; Swartz et al., 2002; A. D. Robinson et al., Ozone loss derived from balloonborne tracer measurements and the SLIMCAT CTM, submitted to Journal of Geophysical Research, 2002].

[45] For further investigation, mean values of column amounts were calculated and compared. Column amounts were determined for the FTIR above 10 km and for MIRA from above 16 km due to the limited sensitivity of this instrument below this altitude. This results in an ozone loss of 1.0 × 1022 molec/m2 = 37 DU for MIRA and 1.2 × 1022 molec/m2 = 44 DU for the FTIR.

[46] The change in ozone column amounts due to air masses originating from lower latitudes and entering the top of the polar vortex can be estimated from the MIRA measurements in a manner similar to that described in section 4.2 due to the sensitivity of this instrument up to an altitude of about 55 km. The partial column amounts of ozone above 35 km derived from the MIRA measurements inside the polar vortex in March show an increase of (2.0 ± 0.4) × 1021 molec/m2 = 7.4 ± 1.6 DU in comparison to the partial column amounts above 38 km of January/February. We can consider this increase in the total columns and correct the values of the ozone loss. This results in an ozone loss in the lower stratosphere of 1.2 × 1022 molec/m2 = 44 DU for MIRA and 1.4 × 1022 molec/m2 = 51 DU for the FTIR. The value measured by the FTIR is a very good estimate for the total ozone loss due to the high sensitivity of this instrument for the whole lower and lowermost stratosphere. Since the column amounts of MIRA and the FTIR agree quite well, we conclude that most of the ozone loss took place above an altitude of 16 km.

[47] The ozone loss in column amounts measured by the FTIR compares very well with the value of 53 ± 11 DU found by Rex et al. [2002] using the Match technique and the value of 51 ± 11 DU found by Hoppel et al. [2002] using profiles of ozone measured by POAM III. The somewhat smaller value of MIRA is probably due to some ozone loss that took place below an altitude of 16 km. Salawitch et al. [2002] found an ozone loss of 61 ± 14 DU for the core of the vortex using the JPL O3 photometer, the NOAA LACE gas chromatograph, and the JPL MkIV FTIR. They suggest that this higher loss in comparison to the Match and POAM III estimates might represent the variability of chemical loss between the core of the vortex and the vortex as a whole. Our lower values support this explanation since they are mean values of several days and different positions inside the vortex, and therefore represent estimates for the vortex as a whole like the Match and POAM III estimates.

[48] It is remarkable that the total losses in ozone column amounts as measured in the cold Arctic winter 1999/2000 are quite low in comparison to the previous cold Arctic winters of 1993/1994 to 1996/1997. Cumulative total ozone reductions of 90 DU (1993/1994) up to 160 DU (1994/1995) were measured by five ground-based SAOZ spectrometers in these winters [Goutail et al., 1999; Goutail and Pommerau, 1998; Goutail et al., 1998]. The quite low ozone loss in winter 1999/2000 is also astonishing considering the fact that there were large losses observed at the 475 K isentropic surface. For the earlier cold winters, Goutail et al. [1999] demonstrated that the altitudes where significant ozone depletion took place (more than 1 × 1012 molec/cm3) ranged from below 15 km up to 30 km. The right side of Figure 10 shows the ozone loss in molec/cm3 of the MIRA and FTIR measurements in the lower stratosphere calculated from the mean profiles of January/February and March 2000. It can be seen that the altitudes where significant ozone depletion took place range only up to 23 km. We conclude that in comparison to previous cold winters, the chemical ozone depletion in winter 1999/2000 took place in a thinner layer which results in lower cumulative total ozone reduction.

5. Summary

[49] We have presented the results of ground-based microwave and infrared measurements of ozone and ozone-related species in winter 1999/2000 at Kiruna, Sweden. From the measurements, time series of volume mixing ratio profiles of ozone and ozone-related species were retrieved. In addition, column amounts were calculated and compared, as well.

[50] From measurements of the inert tracers N2O and HF, the diabatic subsidence inside the polar vortex between January and March was estimated to 1.2 km around an altitude of 20 km. On 26 and 28 January, both instruments measured a total uptake of HNO3 by PSCs in the order of 5 × 1019 molec/m2 = 25%. In March significant denitrification of the lower stratosphere of about 5 × 1019 molec/m2 was found.

[51] The HCl and ClO measurements of both instruments show strong chlorine activation of the lower stratosphere. For ClO the microwave and FTIR results show a good agreement and a correlation to the number of hours of possible PSC contact of the observed air masses. Both instruments agree well in detecting chlorine deactivation in March.

[52] The ozone measurements of both instruments show a good agreement in the variations due to dynamic processes in the middle stratosphere. On isentropic levels at about 20 km, an ozone loss between 1.2 ppmv and 1.4 ppmv was found. For the ozone loss in column amounts, values of 1.2 × 1022 molec/m2 = 44 DU above 16 km for MIRA and 1.4 × 1022 molec/m2 = 51 DU above 10 km for the FTIR could be achieved. In comparison to the former cold winters of 1993/1994 to 1996/1997, these total losses are quite low. The reason is that in the winter of 1999/2000 significant ozone depletion took place only in a rather thin layer below 23 km.

Acknowledgments

[53] We would like to thank DG-XII of the European Community for funding the THESEO 2000 campaign under contract EVK2-CT-1999-00047. We also thank the European Centre for Medium-Range Weather Forecasts (ECMWF) for providing the data of dynamics and temperature in the lower stratosphere in winter 1999/2000. Furthermore, we are grateful to the Goddard Space Flight Center for providing the temperature and pressure profiles of the National Centers for Environmental Prediction used for inversion via the automailer system. This work has also benefited considerably from the support by the staff of the Swedish Institute of Space Physics at Kiruna.

Ancillary