Abstract. We have performed measurements of stratospheric chlorine monoxide (ClO) in Ny-Ålesund, Spitsbergen (78.9°N/11.9°E), using a ground-based mm-wave radiometer. In this paper we describe the observed degree of chlorine activation inside the polar vortex in the winter 1996/97. We obtained daily averaged ClO profiles on 20 days covering the period from mid-February until the beginning of April. The volume mixing ratio of the lower ClO maximum at an altitude of approximataly 21 km of altitude reached a maximum of 1.6 ppbv. These measurements support the strong chemically induced ozone depletion which was found until April 5 by simultaneous Ozone measurements reported in a companion paper [Sinnhuber et al., 1998], where also an unusual ozone loss in late April is reported, which corresponds to the observed ClO abundance on April 18-22. On several days we observed diurnal ClO cycles which were compared with model calculations.
Anthropogenic fluorchlorocarbons (CFC’s) are the major source of chlorine in the stratosphere. The only known natural source of stratospheric chlorine is methyl chloride (CH3Cl) mainly originating from the world’s oceans and burning of vegetation. According to WMO presently more than 80 % of stratospheric chlorine is of anthropogenic origin [World Meteorological Organization, 1991]. Approximately 90 % of the chlorine is stored in chemically inactive reservoir species such as hydrochloric acid (HCl) and chlorine nitrate ClONO2). Under normal stratospheric conditions the active form ClO, shows a single maximum for its volume mixing ratio (VMR) around 40 km with a relatively small diurnal variation. During the polar winter the temperature inside the polar vortex at around 21km of altitude can fall below the formation temperature (approximately 195 K) for polar stratospheric clouds (PSC) enabling heterogeneous chemical processes [e.g. Hanson and Mauersberger, 1988; World Meteorological Organization, 1991] to occur on the surface of liquid and solid particles. This process releases large amounts of chlorine from reservoir species. However massive destruction of ozone will not start until sunlight reappears, because at night ClO is stored mainly in the dimer form (ClO)2. After sunrise several mechanisms can destroy ozone very effectively. Using a simplified version of the photo-chemical box model BRAPHO (Bremen Atmospheric Photochemical Model) [Trentmann et al., 1997] we have calculated the diurnal cycle of ClO at 21 km for March 13 and 20 considering the following chemical reactions.
All rate coefficients and cross sections were taken from DeMore et al., [1997]. For the calculation of the photolysis frequency of the ClO-Dimer (reaction 2) we used March mean temperatures and pressure profiles from the National Center for Environmental Prediction (NCEP, former NMC) and a mean ozone profile determined from simultaneous mm-wave ozone measurements for March. In Figure 1 calculated ClO profiles for Arctic spring with strong activation of chlorine in the lower stratosphere are presented for day- and night time.
Measurement technique and data processing
The Radiometer for Atmospheric Measurements (RAM) is a ground-based
passive microwave receiver for the detection of thermal emission from
molecular rotational transitions in the frequency range 100-300 GHz.
The vertical VMR profiles of a particular species can be determined
from the information contained in the pressure broadened line shape.
For the analysis spectroscopic parameters are needed for the molecule
under investigation from appropriate compilations [ Pickett et al.,
1992].
The system includes two microwave receivers for the detection of ClO at
204 GHz and O3 at 142 GHz. Both sensors are operated in a time
sharing mode using the same acousto optical spectrometer for the realtime
spectral analysis of the measured emission lines [ Langer et al.,
1996]. For ClO observation we use the reference beam method
[Parrish et al., 1988], which minimizes instrumental effects
such as nonlinearities, and the influence of a rapidly varying
troposphere. When retrieving daily mean ClO amounts we improve the
spectrum quality by subtracting nighttime from daytime spectra, taking
advantage of the large diurnal variation.
The total bandwidth of 1 GHz and residual baseline ripples restrict the
lower altitude limit to approximately 17 km. An integration time of
approximately 100 min is needed to obtain a good signal to noise ratio
for day and for night measurements. Due to tropospheric attenuation
measurements are restricted to clear and cold weather conditions.
At 21 km altitude a small amount of ClO remains at nighttime,
due to thermal dissociation of the dimer. Thus, by subtracting
nighttime measurements from daytime measurements we obtain a spectrum
that slightly underestimates the daytime ClO signal. Since the ClO
remaining at night strongly depends on the stratospheric temperature
we do not account for this effect but rather include it in the error
budget of our measurements.
A typical day minus night spectrum with an integration time of
ca. 100 min (day) and ca. 160 min (night) is shown
in Figure 2. A reduction of
standing waves was achieved by subtracting
two sinusoidal waves which have been fitted to the day minus night
spectra. The two sine waves were restricted to short periods in order
to avoid a detrimental influence on the overall lineshape. The intensity
has been corrected for tropospheric opacity and converted to zenith
direction (elevation angle 90°). The solid line is the
corresponding radiative transfer calculation using the retrieved
ClO profile as input.
For the retrieval the model profile as shown in
Figure 1
is adjusted by allowing only 4 free parameters to vary, namely the
peak mixing ratios and the altitudes for the two maxima in the upper
and lower stratosphere. The retrieved profile is obtained by a best
least square fit of the radiative transfer calculation to the day minus
night spectrum.
In contrast to earlier winters when PSCs were observed until February only, the long lasting extremely low temperatures inside the vortex allowed PSCs to occur until the end of March (according to ECMWF data). An analysis of the meteorology of the polar vortex was presented by Coy et al. [1997] and Sinnhuber et al. [1998], showing that Ny Ålesund was located well inside the polar vortex for almost the entire period. Only for a few days around April 12 (day 103) the vortex edge moved over Ny-Ålesund. During the period from mid February to the beginning of April ClO measurements were possible on 20 days covering the whole period with at least one measurement per week. During the remaining time tropospheric opacity was too high for successful measurements. To improve the signal to noise ratio we have integrated all daytime spectra and all nighttime spectra available for a given day and subtracted the two spectra from one another to retrieve averaged daytime profiles. Spectra have been assigned to nighttime if the data were taken at least two hours before sunrise or two hours after sunset. Corresponding daytime spectra were taken at least two hours after sunrise and 3 hours before sunset. The peak VMR values for the lower stratospheric ClO maximum are presented in Figure 3. High values correspond well with times of large PV as expected for disturbed chemistry. A maximum peak VMR of 1.6 ppbv was reached on March 20. For these mean daily values we have calculated the residual of the measured spectra and the best fit. We find an accuracy of 0.2-0.7 ppbv VMR depending on integration time available. The measurement of February 21 (day 52) of 1.3 ppbv agrees with measurements performed by the Microwave Limb Sounder (MLS) observing the highest ClO value in February of about 1.5 ppbv (vortex averaged) on the same day [ Santee et al., 1997]. The ClO mixing ratio decreased after the middle of March and was down to almost zero by the beginning of April. After a gap of about 10 days due to bad weather we were again able to take data around April 20 (day 110). Due to the end of polar night no nighttime measurements are possible in late April. Therefore we have used nighttime measurements of April 5 to retrieve mixing ratios. These nighttime measurements were taken at solar zenith angles ca.94°. Though, by subtracting these nighttime spectra we more likely underestimate the mean daytime ClO VMR in late April. However, the late April measurements again show enhanced ClO mixing ratios, although the last possible PSC formation inside the entire vortex (according to ECMWF temperature data) occured three weeks prior to these measurements. Simultaneous DOAS measurements showing enhanced levels of stratospheric OClO (F. Wittrock et al., paper in preparation) support the microwave measurements, since there is no other source for stratospheric OClO than ClO [ Sessler et al., 1995]. Late April ClO enhancement is also supported by ongoing ozone depletion retrieved from simultaneous ozone measurements [Sinnhuber et al., 1998]. However, when MLS measurements were resumed after April 10, they showed vortex averaged ClO VMRs of < 0.1 ,ppbv. From our retrieved profiles we calculated column densities of up to 1.5E+15molecules/cm² using temperature and pressure data from meteorological radiosondes launched in Ny-Ålesund. Since we have little information on the upper stratospheric ClO peak we only considered the lower maximum of the VMR profiles. The error due to neglecting the ClO above 35 km however is less than 4 %. Arctic ClO columns of<= 2E+15 molecules/cm² have also been observed by the Airborne-Submillimeter-SIS-Radiometer (ASUR) (J. Urban et al., paper in preparation) for the winter 1996/97 which are in good agreement with the ClO columns derived from our measurements. Also measurements performed in Antarctica under disturbed stratospheric chemistry [de Zafra et al., 1989] are compatible to the mm-wave data presented in this paper considering that the lower stratospheric Antarctic ClO maximum is usually found at lower altitudes than in the Arctic. It should be noted, that ClO column densities retrieved from FTIR observations at Eureka [ Donavan et al, 1997] are larger by a factor of 3-4. However such large colum densities appear to not to be compatible with the overall chlorine loading of the stratosphere [World Meteorological Organization, 1991]. On March 13 and 20 the excellent observing conditions allowed to retrieve ClO data for day and night separately with a temporal resolution of approximately one hour. Figure 4 shows the data in comparison with calculations using the BRAPHO photochemical model. Despite the rather large error bars the typical day night variation is well reproduced with the data, giving credibility in both, the measurements and the model. In order to reproduce the ClO measurements we have chosen 2.2 ppbv and 2.25 ppbv of total ClOx ( ClO + (ClO)2) for March 13 and 20 respectively to initialize the model. According to World Meteorological Organization [1991] the upper limit of ClOx is considered 3.5 ppbv which means that not all the chlorine was converted from the reservoir gases into the active form. To estimate the error for this shortterm integration we have used measurements of February 8 (day 39), well before the lower altitude ClO maximum is expected to develop. From this analysis we conclude the accuracy to be better than 1 ppbv. The sharp increase in ClO is due to the very fast photolysis of the dimer after sunrise. The decrease during sunset is slower because the formation of the dimer is of second order in [ClO]. The temporal discrepancy in the onset of the ClO formation between model and measurements on March 13 was found to be due to observing air masses with different histories. Backtrajectory calculations show that air masses passing Ny-Ålesund in the early morning, unlike the preceding and following air masses, experienced a warming six days prior to the measurements followed by temperatures >200 K inside this air mass until March 13. Finally we also used the BRAPHO model to calculate the ozone loss due to chlorine chemistry according to the mm wave measurements. The chemically induced ozone loss was calculated to be of the order of 27 ppbv/day for March. This is a lower limit of the complete ozone loss, since other chemical processes are involved as well. From simultaneously performed ozone measurements [Sinnhuber et al., 1998] have obtained an ozone loss rate of about 22 ppbv/day at the 475 K isentropic level for an averaged twenty days period during March. Taking into account that the latter is a vortex averaged loss rate this is in good agreement with the modeled loss rates for Ny-Ålesund.
We have found high values of stratospheric chlorine monoxide inside the polar vortex during the winter and spring 1997. The maximum ClO VMR measured was 1.6±0.4 ppbv. With respect to the very stable polar vortex we assume that strong chlorine activation occured from mid February until the beginning of April. Slightly enhanced ClO VMR at the end of April support the unusually late chemically induced ozone loss that was observed by simultaneous ozone measurements in late April [Sinnhuber et al., 1998]. From our measurements we derived diurnal ClO cycles for March. Initializing a photochemical box model with the ClO measurements, we obtained a total ClOx of about 2.2 ppbv for March 13 and 20. According to this result we conclude that not all stratospheric chlorine was activated. From the model calculations we obtained ozone loss rates of about 27 ppbv for March at 21 km. These loss rates are in good agreement with simultaneous millimeter wave ozone measurements leading to a vortex averaged ozone loss rate of about 22 ppbv/day at the 475 K isentropic during March.
This work has been supported by the Alfred-Wegener-Institute for Polar and Marine Research, the German Ozone Research Program of the ministry for education and science (Project no. 01 LO 9528/8) and the Environment and Climate Program of the European Community (Project no. ENV 4-CT95-0136). We also thank the ECMWF and NCEP for providing the PV- and temperature data.
Figure 1. Vertical profile of daytime ClO (solid line) for a situation of severely disturbed stratospheric conditions and a corresponding night time profile (dashed line).}
Figure 2. A typical day minus night spectrum with an integration time of ca.100 min (day) and ca.160 min (night). The intensity is given for zenith direction and has been corrected for tropospheric attenuation. The solid line is a calculated spectrum using the retrieved ClO profile as input.}
Figure 3. The ClO peak VMRs for the lower maximum near 21 km altitude. The shaded area shows the period without sunlight in the stratosphere at 21 km altitude (corresponding to solar zenith angle <94°.}
Figure 4. Diurnal ClO variation obtained from the mm-wave data (full circles with error bars). The integration time is approximately 20 min for each measurement. The solid line represents the diurnal variation according to the model calculations for each day.}
