Simultaneous measurements of OClO, NO 2 and O 3 in the Arctic polar vortex by the GOMOS instrument

We present the ﬁrst nighttime measurements of OClO from a limb-viewing instrument in the Arctic polar vortex. The relationship between OClO, NO 2 and O 3 slant column densities in the Arctic polar vortex are analyzed from the GOMOS measurements. The retrieval process is based on a di ﬀ erential optical absorption spectroscopy 5 (DOAS) method applied on the weighted median GOMOS transmittances. A study about the longitudinal distributions of OClO, NO 2 and O 3 above 65 ◦ north in January 2008 is presented. It shows a strong halogen activation in the lower stratosphere and a strong denoxiﬁcation in the entire stratosphere inside the Arctic polar vortex. Time series of temperatures and OClO, NO 2 and O 3 slant column densities for the winters 10 2002/2003 to 2007/2008 are also presented. They highlight the correlation between temperature, OClO and NO 2 . The GOMOS instrument appears to be a very suitable instrument for the monitoring of OClO, NO 2 and O 3 in the stratosphere during nighttime.


Introduction
Since the discovery of the stratospheric ozone depletion by Farman et al. (1985), several studies have been performed to better understand this recurrent phenomenon (a historical review of this research can be found in Solomon, 1999). Inorganic chlorine species (Cl y ) play an important role in the stratospheric chemical processes that lead to ozone depletion in both Arctic and Antarctic polar regions. The inert reservoir 20 species (like ClONO 2 or HCl) are converted into active chlorine (ClO, Cl 2 O 2 ) by heterogeneous reactions, which occur on the surface of polar stratospheric clouds (PSC) formed during the polar night (Solomon et al., 1986) if stratospheric temperatures are below 198 K (T PSC ) . These active species will strongly contribute to catalytic cycles that destroy ozone. A consequence of the halogen activation is the production of OClO Introduction The study done by Sessler et al. (1995) shows that OClO is a good qualitative indica-5 tor of chlorine activation and a good quantitative indicator of BrO. This study highlights also that OClO is a poor quantitative indicator of the ClO presence. The monitoring of OClO appears to be crucial to better understand the polar stratospheric chemistry by constraining the chemical models. The only significant sink of OClO is its rapid photolysis by the solar radiation in the UV wavelength range. Consequently OClO is almost 10 constant during night. Nitrogen species NO x (NO+NO 2 ) also play an important role by reforming the halogen reservoir species: ClO + NO 2 → ClONO 2 (R4) BrO + NO 2 → BrONO 2 (R5) 15 The reactions R4 and R5 limit the formation of OClO. Nevertheless, in the polar vortex, NO 2 is removed via the formation of HNO 3 . This is the well-known denoxification of the polar vortex.
The first measurements of OClO in the stratosphere were performed in Antarctica by Solomon et al. (1987) from a ground-based station. Since then, other measurements 20 have been performed by using ground-based (Miller et al., 1999), balloon-borne (Canty et al., 2005;Pommereau and Piquard, 1994;Renard et al., 1997;Riviere et al., 2003) or satellite measurements (Krecl et al., 2006;Wagner et al., 2002). Some of these satellite instruments used a nadir geometry to retrieve the total vertical column densities of OClO: it is the case for the instruments Global Ozone Monitoring Experiment (GOME) on ERS-2 (Burrows et al., 1999), SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY (SCIAMACHY) onboard ENVISAT (Bovensmann et al., 1999) and Ozone Monitoring Instrument (OMI) on Aura (Levelt et al., 2006). The instrument Optical Spectrograph and InfraRed Imager System (OSIRIS) onboard the Odin satellite (Llewellyn et al., 2004) uses the limb scattering technique to retrieve ver-5 tical profiles of concentrations of OClO (Krecl et al., 2006). Moreover, SCIAMACHY has a limb-viewing mode and can also retrieve such vertical profiles. However, no results have been published for the moment. Note that the Stratospheric Aerosol and Gas Experiment III (SAGE III) on the Meteor-3M satellite (McCormick et al., 1991) can also perform lunar occultations and limb-scatter measurements in order to retrieve the 10 OClO vertical distributions but, for the moment, no results concerning this has been published. The Global Ozone Monitoring by Occultation of Stars (GOMOS) instrument on ENVISAT (see e.g., Bertaux et al., 1991;Kyrölä et al., 2004) (Fussen et al., 2006). The stellar occultation technique used by GOMOS allows the measurements of OClO during nighttime. GO-MOS is the only satellite instrument able to perform nighttime measurements in the stratosphere. For the moment, the OClO product obtained from the GOMOS spectra has not been validated with data from other instruments. This is why we consider this 20 work as preliminary. We report in this paper the distributions of OClO, NO 2 and O 3 slant column densities (SCD) retrieved from GOMOS measurements during the Arctic winters from 2003 to 2008. After a brief summary of the GOMOS instrument, the retrieval algorithm is described. Then, we present the spatial distributions of OClO, NO 2 and O 3 measured in  Fussen et al., 2006). GOMOS is also equipped with two fast photometers used to correct for star scintillation and to retrieve temperature profiles. GOMOS measures light from several stars that are setting behind the Earth horizon. The transmittance along the line of sight is obtained at each tangent altitude by dividing the stellar spectrum measured through the atmosphere by the reference stellar spec-15 trum measured outside the atmosphere. This method is self-calibrated and offers the advantage of a large number of occultations per day (30 to 50 measurements per orbit compared to only 2 occultations for the solar occultation method). The wide variety of stars used combined with the sun-synchroneous orbit allows a global coverage in about 3 days. Nevertheless, the light intensity of stars is weak, influencing the signal-to-noise 20 ratio of GOMOS measurements that depends on the star used.

OClO retrieval
Due to the difficulties of the detection of OClO in a single GOMOS spectrum, we co-add transmittances (interpolated on a common altitude grid) in latitude bins of 10 degrees with a temporal resolution of one month. The transmittances used are already cor-25 rected for scintillation and dilution effects (Dalaudier et al., 2001). The consistency of Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion each data set is checked by using statistical normality tests after which we calculate for each consistent bin a weighted median spectrum and the associated variances. A differential optical absorption spectroscopy (DOAS, Platt et al., 1979) technique is then applied on this weighted median spectrum in the [355-390 nm] wavelength region. In this spectral window, the main contributions of the total extinction are the 5 molecular scattering, the aerosol extinction and the absorptions by OClO and NO 2 . Even if the ozone absorption in this wavelength range is weak, we take it into account in order to obtain a better accuracy of the retrieval. For the sake of an optimal signal-to-noise ratio, we have decided to use only measurements from the star Sirius (Temperature=11000 K, Magnitude=-1.44) in this study. 10 The key principle of the DOAS method is to separate the total transmittances T (λ) into two components, one varying slowly with the wavelength (T s (λ)) and the other rapidly varying (noted d T (λ) and named experimental differential transmittance). The slowly varying transmittances are calculated using a second order polynomial. It corresponds to the molecular scattering, the aerosol extinction and the slowly varying 15 components of the gaseous absorptions. In the same way, we separate the absorption cross sections (σ i ) of each species into a slowly varying component (calculated also with a second order polynomial) and a high frequency component. The differential cross sections δσ i are then calculated as the differences between the absorption cross sections and the smoothed ones. Thus, the modelled differential transmittance M (λ) 20 can be written as: where N i are the slant column densities (SCD). The N i are then obtained by a nonlinear least-squares minimization of the difference between M(λ) and d T (λ) weighted ACPD 9,2009 OClO, NO 2 and O 3 measurements in the Arctic polar vortex by GOMOS Interactive Discussion by the experimental errors. In the minimization procedure, the wavelength is eventually shifted by ∆λ. The estimated retrieval error is extracted from the jacobian. Note that the cross sections used in the retrieval are the same regardless of the temperature. Unfortunately, for the moment, no direct validation of the OClO SCD obtained can be achieved. In the future, a validation exercise will be done with the OSIRIS, SCIA-5 MACHY, OMI and GOME results. Nonetheless, we can compare the NO 2 SCD obtained from our DOAS procedure and from the GOMOS operational algorithm. In January 2008, the polar vortex was not centered on the geographical north pole. It gradually moved towards Europe. Figure 2 shows a potential vorticity map obtained from the MIMOSA model (Hauchecorne et al., 2002) at 675 K (about 27 km) on 10 January 2008. It highlights that the polar vortex extends from the North pole down to 60 • N latitudes above the regions whose longitude lies between 70 • W and 135 • E. 5 Consequently, a 5 degree latitudinal band in the northern hemisphere was not homogeneous for the highest latitude. In January 2008, only the latitudinal band near 72 • was sounded by GOMOS (Sirius occultations). In Fig. 2, the white circle represents the spatial distribution of these GOMOS Sirius occultations. They are distributed along the entire longitudinal range: some of them were performed inside the polar vortex and 10 the others outside. Hence, we cannot bin together all these measurements. Figure 3 shows that the distribution of the transmittances at 385 nm and at 30 km for the GO-MOS measurements in this latitudinal band are bimodal. Note that for other altitudes and wavelengths, this bimodality of the transmittances is also observed. One expects that high transmittances correspond to measurements inside the polar vortex. Indeed, 15 because of the denoxification of the polar vortex, the NO 2 absorption is weak.
To take into account the specificity of the position of the polar vortex in January 2008 This is in a good agreement with our current knowledge of stratospheric chemistry (Brasseur and Solomon, 2005). According to Sessler et al. (1995), OClO presence is an indicator of halogen activation. The presence of ClO in the lower stratosphere in the polar vortex is confirmed by the instrument Microwave Limb Sounder (MLS) instrument onboard EOS Aura (Earth Observing System, Schoeberl et al., 2006): in Fig. 6, 5 a maximum of ClO volume mixing ratio in the lower stratosphere is located in the Arctic polar vortex (at high equivalent latitude, Santee et al., 2008). Figure 5 shows also the longitudinal and vertical distribution of ozone SCD. It shows a slow decreasing of ozone inside the polar vortex. This is in a good agreement with the halogen activation and with the denoxification observed. One may also notice the low temperatures (lower 10 than 200 K) encountered in the areas where the halogen activation and the denoxification are observed, a condition for the presence of PSCs.

Temporal evolution of OClO, NO 2 and O 3 in Arctic winters
In this section, we study the temporal evolution of the OClO, NO 2 and O 3 SCDs in the Arctic regions during winter from 2002/2003 to 2007/2008. Figure 7 shows the 15 geolocation of the Sirius occultations in the northern hemisphere (latitude greater than 65 • ). All these measurements occurred during polar nights. This study was performed with a temporal resolution of two days. Bins with less than 4 measurements are not considered as statistically significant and are not taken into account. This restriction eliminates only 11 bins among a total of 200 bins. The same method is then applied on 20 each set: the weighted median is calculated and the DOAS retrieval is applied. Thus, we retrieve the OClO SCD, the NO 2 SCD and the O 3 SCD as a function of time in the Arctic winter. Note that, for this study, we do not take into account the position of measurements relative to the polar vortex. In other terms, no longitudinal separation (like in the previous section) has been done because the number of measurements 25 available is not sufficient to perform spatio-temporal binning with a two days resolution. One should keep in mind that this can cause the averaging of inconsistent GOMOS ACPD 9,2009 OClO, NO 2 and O 3 measurements in the Arctic polar vortex by GOMOS  Figure 8 shows the minimum temperature reached in the stratosphere (panels A) and the O 3 (panels B), NO 2 (panels C) and OClO (panels D) SCDs (with the error bars) at the altitude of 19 km as a function of time. This level has been chosen because at this 5 altitude both denoxification and halogen activation are clearly observable. The winter 2006/2007 was not studied due to the lack of data. In the next paragraphs, we will describe the results obtained for each winter.

The Arctic winter 2002/2003
During this winter, stratospheric minimum temperatures retrieved using the GOMOS 10 photometers were below the temperature of formation of PSC from mid November to mid January (panel A). Then, temperatures were higher after mid January, almost always above T PSC . Consequently, the halogen activation must be more important during this cold period: we observed a slow increase of OClO SCDs all along this period (reaching 2.0e15 cm −2 at the end of December) and then, after mid January, 15 OClO SCDs began to decrease slightly. For NO 2 , it is the reverse: we observed a decrease of the NO 2 SCDs during the cold period and then an increase. Concerning ozone, the SCDs are very noisy. This is mainly because ozone is a weak absorber in the wavelength region used in the retrieval.

The Arctic winter 2003/2004
20 This winter is also characterized by two periods: a cold one (from mid November to late December when, the minimum temperatures are below T PSC ) following by a warming. During the cold period, OClO SCDs are increasing, NO 2 and O 3 SCDs are both decreasing. When the temperatures gets higher, NO 2 SCDs are increasing and OClO SCDs are decreasing. However, ozone SCDs remain approximately constant. For this 25 winter, a strong contrast (for OClO and NO 2 SCDs) is observed between the cold and ACPD //www.ozone-sec.ch.cam.ac.uk/scout o3/), the temperatures in the stratosphere were the lowest since 50 years. They remained below T PSC from late November to late February. Consequently, lots of PSC have been observed during the entire winter. The minimum GOMOS temperatures are in agreement with this situation. They are below T PSC from late November to late January. OClO SCDs strongly increased fom 10 late November to mid December, reaching approximately 2.5e15 cm −2 . In the same time, NO 2 SCDs decreased strongly. From mid December, OClO and NO 2 remained roughly constant. Ozone SCDs are weak all along this period. During this exceptionally cold and stable winter, GOMOS has observed strong and long halogen activation and denoxification. 15

The Arctic winter 2005/2006
For this winter, GOMOS has operated measurements only during January. The minimum temperatures retrieved from the GOMOS photometers are not very low. They are below T PSC only for a few days in mid January (during the GOMOS measurements period). Then the miminum temperatures in the stratosphere increased. Conse-20 quently, we observed an increase of the NO 2 SCDs (from 2.5e15 cm −2 in mid January to 6e16 cm −2 in early February). Nevertheless, we cannot observe a diminution of the OClO SCDs as expected. For an unclear reason, the temporal evolution of OClO SCDs is first decreasing and then increasing. Concerning ozone, nothing can be deduced by studying the temporal evolution.

The Arctic winter 2007/2008
The GOMOS measurements for this winter occurred only in January 2008. The minimum temperatures are often lower than T PSC during the whole GOMOS measurement period. Nevertheless, we can distinguish neither denoxification nor halogen activation. This is probably because no longitudinal discrimination is carried out for this temporal 5 study. Indeed, the polar vortex is not centered on the geographical pole and the latitudinal band sounded by the GOMOS measurements is not homogeneous (cf. Sect. 4). This lack of homogeneity of the GOMOS measurements used generates the large error bars observed on the O 3 , NO 2 and OClO SCDs. However, we can observe an anticorrelation between NO 2 and OClO SCDs: in mid January a sudden increase of 10 NO 2 SCD occurred at the same time as a sudden decrease of the OClO SCD. This is followed by a decrease of NO 2 and an increase of OClO SCD.

Conclusions
This preliminary work focuses on the interactions between NO 2 , O 3 and OClO in the Arctic polar vortex. The slant column densities of these species are retrieved during 15 nighttime from Sirius occultations as observed by the GOMOS instrument. First, the longitudinal distributions of these species in January 2008 highlight strong variations of OClO and NO 2 (relative to the longitude) in the lower stratosphere: in the polar vortex, weak NO 2 SCD and strong OClO SCD are observed, and the reverse situation outside the polar vortex. Halogen activation (marked by high OClO SCDs) and denoxification 20 which occur in the polar vortex are clearly detectable. Hence, the GOMOS spectrometer appears to be a suitable instrument to perform a monitoring of such species in the polar vortex. Furthermore, the temporal study confirms that the halogen activation degree and the concomitant denoxification can be effectively monitored by GOMOS. We were able to highlight the correlation between cold temperatures (below T PSC ), weak 25 NO 2 SCDs and strong OClO SCDs inside the Arctic polar vortex for several winters. This paper reports the first study of OClO distribution using limb-viewing satellite measurements performed during night in the Arctic regions. The next step in the use of the GOMOS measurements for the OClO retrieval is to validate this product. For this aim, we need correlative measurements from other satellite instruments (like OSIRIS, SCIAMACHY, GOME or OMI) or balloon-borne in-