Abstract.Ground based millimeter wave measurements of Arctic stratospheric ozone in the winter 1996/97 are presented. The measurements have been performed at one of the primary Arctic stations of the Network for the Detection of Stratospheric Change (NDSC) in Ny-Ålesund, Spitsbergen ( 78.9°N, 11.9°E). Over the period 11 February to 26 April the measurements show in the lower stratosphere an ozone mixing ratio decrease of 1.3 ppm or 44 %. Correspondingly, stratospheric ozone column densities decreased by more than 50 DU. Taking into account the transport of ozone due to diabatic decent, we estimated chemical ozone loss rates of 22 ppb/day in February decreasing to 15 ppb/day in late April 1997.
Chemical depletion of lower stratospheric ozone in the Arctic has been
observed during the last years in winter and spring [WMO,1994;
European Commission,1997]. Although the extent of the Arctic
ozone depletion is still less than in the Antarctic Ozone Hole, the
processes leading to the ozone depletion are believed to be the
same in both hemispheres. Polar stratospheric clouds (PSC)
form at very low temperatures and enable heterogeneous reactions to activate
chlorine which, in the presence of sunlight, can efficiently destroy ozone.
However, dynamical processes usually lead to a high ozone variability in the
Arctic, that makes it difficult to seperate chemical ozone depletion.
In late winter and spring 1997 again substantial ozone loss has been
observed in the Arctic [Manny et al.,1997;Newman
et al.,1997;Müller et al.,1997.]
The meteorology of the winter 1996/97 stratosphere was unusual
[Coy et al.,1997]. The polar vortex formed unusually late
and was stable until the first days in May. Temperatures stayed
low enough to allow the formation of PSCs until the end of March.
In this paper we derive chemical ozone loss rates from ground based
millimeter-wave ozone measurements. The measurements have been
performed at one of the primary Arctic stations of the Network
for the Detection of Stratospheric Change (NDSC) in Ny-Ålesund,
Spitsbergen (78.9°N, 11.9°E).
These continuous measurements of ozone profiles inside the polar
vortex from February to the end of April allowed the investigation
of the chemical ozone loss evolution during spring 1997.
The Radiometer for Atmospheric Measurements (RAM) is a ground
based millimeter-wave radiometer operated continuously at
Ny-Ålesund, Spitsbergen. An ozone emission line is
measured at 142 GHz over a spectral band-width of 1.6 GHz.
The high spectral resolution allows the retrieval of ozone profiles
between 12 and 55 km from the pressure broaded line shape.
The vertical resolution is about 6-8 km. One to four ozone
profiles per hour are obtained all year round,
largely independent of tropospheric weather conditions.
Temperature profiles required for the ozone profile retrieval
are taken from analyses of the National Center for Environmental
Prediction (NCEP, former NMC). The precision of the measured
ozone volume mixing ratio (VMR) profiles is about 0.1 ppm.
The RAM measures alternately ozone at 142 GHz and chlorine
monoxide (ClO) at 204 GHz. The ClO measurements of late
winter-spring 1997 are presented in a companion paper by
Raffalski et al. [1998]. For a detailed description
of the RAM see Langer et al. [1997].
Fig. 1 shows the RAM
measurements of the ozone VMR at the 475 K level
(approximately 19 km altitude) between mid of February and
begin of May 1997. A continuous decrease from nearly 3 ppm
in February to about 1.6 ppm at the end of April can be seen,
corresponding to an overall reduction of 44 %.
Diabatic decent is expected to increase the lower stratospheric
ozone VMR over this time period. The observed ozone decrease
therefore has to be attributed to chemical ozone depletion.
Potential vorticity (PV), derived from analyses of the European
Centre for Medium Range Weather Forecast (ECMWF), indicate
that the ozone measurements during this period have been
performed well inside the polar vortex, with the exception
of only a few days in mid April, when Ny-Ålesund was
at the vortex edge as seen in the data by a slight increase
of the ozone mixing ratio. The vortex edge
was determined using the PV gradient with respect to equivalent
latitude [Nash et al.,1996].
We defined the inner vortex as the region of relatively low
PV gradients, compared to the high PV gradients at the
vortex edge.
We performed domain filling trajectory calculations
[Sinnhuber et al.,1996] to show that the inner vortex
at 475 K was well isolated from February to the end of April
1997. No intrusion of outer-vortex air into the inner
vortex occurred.
After April 26 the inner vortex moved away from Spitsbergen.
Higher ozone VMR of up to 2.5 ppm have been measured in the
vortex edge region between April 14 and 17 and after April 26.
After May 10 measurements have been made clearly outside the
vortex, with typical mid-latitude ozone mixing ratios of
2 ppm at 475 K.
Stratospheric ozone column densities above 12 km have been
computed from the RAM measurements. They show a constant
decrease between mid of February and the begin of April,
see Fig. 2 (dots).
Fig. 2 also shows UV/visible total ozone measurements
performed at Ny-Ålesund (asterisks, data provided
by F. Wittrock). The RAM column densities above 12 km
are in excellent agreement with the UV/visible total ozone
measurements, when a mean offset of 66 DU
accounting for the remaining ozone column below 12 km is
added. Between the beginning of February and beginning
of April the ozone column densities decreased bymore than
50 DU. The decrease of the ozone column densities inside
the polar vortex was also observed by the two satellite-borne
instruments Total Ozone Mapping Spectrometer (TOMS)
[Newman et al.,1997] and Global Ozone Monitoring
Experiment (GOME) [Bramstedt et al.,1997].
From the RAM ozone measurements at 475 K loss rates have
been determined by fitting a linear trend over periods
of 20 days. A 20 day period was chosen because the
measurements show some variability on time-scales of a
few days, indicating that ozone is not horizontally
homogeneously distributed inside the vortex. Fitting a trend
over 20 days averages out this horizontal variability.
Measurements not made within the inner vortex were excluded
from the trend calculations.
The observed ozone trends are shown in
Fig. 3 (dots with errorbars).
The errorbars indicate two standard deviations of the fit.
To derive chemical ozone loss rates from the observed ozone
trends, we have to take into account how much of the chemical
loss is masked by the diabatic decent of air inside the polar
vortex. The ozone change on a given isentropic surface due to
diabatic decent is given as
is the ozone VMR,
Q the diabatic heating rate,
are pressure and reference pressure respectively,
the ratio of of the dry air gas constant to the specific heat
at constant pressure (2/7), and
is the ozone gradient
with respect to the potential temperature
[Braathen et al.,1994]. The heating rates include solar
heating through the absorption of shortwave radiation and cooling
through longwave emission into space.
The longwave cooling rate calculations have been performed with
a narrow-band model with a resolution of
[Shine,1991]. The computation included ozone,
water vapor and carbon dioxide. Daily profiles of the RAM ozone
measurements at Spitsbergen have been used for the heating rate
calculations. Since no routine measurements of stratospheric water
vapor were available, a typical late winter vortex profile was
assumed with a minimum of 3.5 ppm at 400 K increasing to 6 ppm
above 675 K [Ovarlez and Ovarlez,1994. However, water
vapor has only a small impact on both, longwave and
shortwave stratospheric heating rates. The carbon dioxide VMR
was assumed to be 360 ppm.
Temperature profiles averaged over the inner vortex, have been
taken from the ECMWF analyses. The error through the use of vortex
averaged temperatures for the heating rate calculations,
,
instead of averaging the heating rates
after the computations,
,
is estimated to be less than 0.05 K/day.
The integration of the solar heating rates over the
vortex has been performed by computing
the heating rates for every 10 degrees of latitude. All
calculations have been done for clear sky conditions only.
Rosenfield et al.[1994] found that tropospheric
clouds increaselower stratospheric cooling rates only by
a few percent, while the effect of PSCs on stratospheric
cooling rates is negligible.
The resulting vortex averaged net heating rates at 475 K
decrease from about 0.3 K per day in February to nearly
zero at the end of March. There is slightly less cooling
during February and March 1997 than found by
Rosenfield et al.[1994] for February and March 1992.
Our calculations showed no significant diabatic heating during
the considered period in 1997. This is a result of the
unusually low ozone mixing ratios in April. Since the solar
heating in the lower stratosphere depends almost exclusively
on ozone absorption, higher ozone VMR would have caused diabatic
heating in April.
From the diabatic heating rates and observed vertical ozone
gradients, the diabatic transportrates are computed using
Eq. (1) Fig. 3 (open diamonds).
Significant chemical ozone loss rates are observed throughout
the whole period. The chemical loss rates stayed nearly constant
at about 20 ppb/day throughout February and March and decreased
to 15 ppb/day in the second half of April. They are in good
agreement with results of the 1997 MATCH campaign, which are
available until the beginning of April (P. von der Gathen,
personal communication, see also von der
Gathen et al. [1995]).
The analysis of the millimeter-wave ozone measurements shows substantial chemical ozone loss at the 475 K isentropic level until the begin of April 1997. During this period we also measured high levels of ClO [Raffalski et al.1998]. Ozone loss rates computed from the ClO measurements agree well with the observed ozone loss. The still high ozone loss rates during March and the begin of April are partly due to the high amount of sunlight available inside the vortex at that time. Fig. 4 shows the estimated ozone loss rates per sunlit hour (open diamonds). They have been derived from the loss rates shown in Fig. 3 divided by the average hours of sunlight per day available inside the vortex during the corresponding periods. Between February and March, the ozone loss rates per sunlit hour at 475 K decreased by more than a factor two. For comparison the area of temperatures below 195 K at the 475 K level, where PSCs may exist, is also shown in Fig. 4 . It shows that the reduction of the possible PSC existance area between February and March is correlated to the vortex averaged chemical ozone loss rates per sunlit hours. The ozone loss rates of 15 ppb/day in the second half of April 1997 - three weeks after temperatures were low enough to allow the existance of PSCs - are about a factor of two or three higher than we would expect under this condition from usual gas-phase chemistry without chlorine activation, i.e. due to the NOx- and HOx-cycles alone [Lary1997]. Catalytic chlorine cycles on the other hand would require a vortex averaged chlorine activation of about 0.5 ppm ClO to explain the observed ozone loss in the second half of April, as calculations with the Bremen Atmospheric Photochemical Model indicate [Raffalski et al.,1998]. Under this situation the ClO/BrO- and ClO/O-cycles become dominant compared to the ClO dimer cycle, which means that the ozone loss would be proportional to the ClO concentration. It is presently unclear, what is the cause for the unexpected high ozone loss in late April. The possibility that there may be chlorine activation with ClO mixing ratios as high as 0.5 ppm during this period is disscussed by Raffalski et al. [1998]. The millimeter-wave observations performed at Ny-Ålesund showed that substantial ozone loss occurred again in winter and spring 1997, in agreement with other observations. The late break-up of the polar vortex and record low temperatures during March and early April led to chemical ozone loss that lasted longer than in previous years. However, the overall ozone loss for the winter 1996/97 was lower than that reported for previous years [European Commission,1997; Rex et al.,1997].
We thank the staff at Ny-Ålesund for their support. Keith P. Shine is gratefully acknowledged for providing the radiation model for cooling rate calculations. Meteorological data were provided by the European Centre for Medium-Range Weather Forecast and by the National Center for Environmental Prediction. We thank Folkard Wittrock for the use of the UV/visible data prior to publication. Part of this work was supported by the Commission of the European Community, by the German Bundesministerium für Bildung und Wissenschaft, and the Alfred Wegener Institute for Polar Research, Germany.
Figure 1. Ozone volume mixing ratio at the 475 K isentropic level as measured by the ground based millimeter-wave radiometer at NY-Ålesund in late winter 1997. A continuing decrease of the ozone mixing ratio between mid of February and end of April is evident.
Figure 2. Ozone column densities above 12 km as measured by the ground based millimeter-wave radiometer (dots) and total ozone from UV/visible measurements (asterisks) at NY-Ålesund in late winter 1997.}
Figure 3. observed ozone change at 475 K derived from millimeter wave measurements at NY-Ålesund (dots with errorbars). A linear trend was computed over a period of 20 days. The errorbars represent a 2 sigma error of the fit. Compared to computed ozone change due to vortex averaged diabatic decent of ozone (solid line). Estimates on the chemical ozone loss rates indicated by open diamonds.}
Figure 4. Estimated chemical ozone loss rates per sunlit hour at 475 K (open diamonds). For comparison the possible area of PSC existance at 475~K is displayed, approximated by the area of temperatures within the vortex lower than 195 K.}
