10, PAGES 1941-1944, MAY 15, 2001. Spectroscopic Measurements of ... Institut fiir Umweltphysik, University of Heidelberg, Heidelberg, Germany. Abstract.
GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO. 10, PAGES 1941-1944, MAY 15, 2001
Spectroscopic Measurements of Tropospheric Iodine Oxide at Neumayer Station, Antarctica U. Friefi, T. Wagner, I. Pundt, K. Pfeilsticker, and U. Platt Institut fiir Umweltphysik,University of Heidelberg,Heidelberg,Germany
Abstract. First measurements of iodineoxide (IO) in ing of HOI by maritime aerosolparticles could stimulate the Antarctic troposphere are reported. Since March the release of reactive halogen compoundsinto the at1999, a newly developed dual channel spectrograph mosphere [e.g. Vogtet al., 1999].In absence of sunlight, has been continuously performing Differential Opti- IOx is rapidly converted into the night-time reservoir cal AbsorptionSpectroscopy (DOAS) measurements of speciesIONOe, HOI, IeOe, OIO, and HI. zenith scatteredsunlight at Neumayer-Station, AntarcHere we report on first measurementsof iodine oxide tica (70ø39• S, 8ø15• W). The spectralsignatureof IO in the Antarctic troposphere,using DOAS observations was clearly detected by observing five vibrational ab- of zenith scattered sunlight. sorption bands located in the wavelength region between 415 and 461 nm. The observed diurnal
variation
of IO is characterized by a rapid decreasein the dif-
ferentialslant columndensity(DSCD) with increasing solarzenithangle(SZA) duringtwilight. This observa-
Instrument
and data analysis
A DOAS spectrographhas been installed at the Ger-
tion points to a fast conversionof reactive iodine into its nighttime reservoirspecies.It alsostronglyindicates that the detected IO is located in the troposphere.The
man Antarctic researchstation Neumayer (70039' S, 8ø15' W) in March 1999. Sincethen it hasbeencontin-
decrease of the IO DSCD of up to 1.10 TMmolec/cm 2
tion is located on the shelf ice at the coast of the Antarc-
uously performing measurements. The Neumayer Sta-
between 80ø and 95ø SZA is unexpectedly large. Under the assumption that IO is located in the marine
tic continent, at a distance of approximately 7 km to the
boundarylayer (MBL) (below2 kin), IO mixingratios
The instrument collectszenith scattered sunlight using a small telescope The light is fed to the spectrograph using two depolarizing fibre bundles. The spectrograph unit developedat our institute has an optical set-up similar to the balloon-borne instrument already
may reach up to •10 ppt. The seasonalvariation shows higher IO amounts during summer than during winter. This finding is possibly causedby the smaller distance to the open sea, where the iodocarbons are emitted, and by the more efficient photodissociationof the organic iodine precursors.
ocean.
describedin detail by Ferlemannet al. [2000].In brief, the instrument consists of two separate spectrograph
units for the UV (320 nm to 420 nm, 0.5 nm full width at half maximum(FWHM) resolution)andvisible(400 nm to 650 nm, 1 nm FWHM) wavelengthregions.The
Introduction
Recent measurementsreported maximum tropospheric IO mixing ratios of up to 6-7 ppt in the northern mid-
latitudes [Alickeet al., 1999; Allan et al., 2000]. Observations of tropospheric IO in the Arctic have also
incoming light is dispersedwith two holographic grat-
ings(1200grooves/mmand 510 grooves/mm)and it is detected by two 1024 channel photodiode arrays. The detectors are thermostated to -35ø C using water precooled Peltier
cascades in order to minimize
the detector
been reportedrecently [Wittrocket al., 2000]. Inor-
dark current and signal noise. The whole optical setganic iodine in the MBL is believed to originate from up is located inside a vacuum sealedaluminum housing the photolytic destruction of organic iodine compounds xvhich is filled with dry argon. No additional optical
like methyliodide(CH3I), and diiodomethane (CH212), components(e.g., mirrors or quartz windowsin front producedby biological activity in the ocean and subse- of the detectors),which could causeintensitylossesor quentlyemittedto the atmosphere[Schallet al., 1994]. disturbing interference effectsare used. Previous
measurements
have
shown
that
these
com-
pounds may reach mixing ratios of several ppt in the
AntarcticMBL [ReifenhSuser •4Heumann,1992].Their
The spectral retrieval is performed in the wavelength
regionbetween415 and 461.5 nm (175 pixel), encom-
passingfive vibrational absorptionbandsof IO. Absorption crosssectionsof IO, 03, NO2, 04, H20 and OC10 as well as a Fraunhofer reference spectrum recorded daily at 80ø SZA are simultaneouslyfitted to the atmocalsI, IO and OIO. I and IO (=IO/) directlyinitiate the spheric spectra recorded at different SZA as described destructionof ozonein catalyticreactioncycles[Jenkin below. The IO absorption spectrum was recorded reet al., 1985],involvinga secondhalogenatom (Br, C1or cently with a resolution of 0.09 nm at our laboratory
rapid photolyric destruction- with lifetimes ranging between minutes and days- may then initiate inorganic iodine chemistry,in particular involving the iodine radi-
I) or the hydroxylradical(OH) [Chameides andDavis, [HSnninger,1999]. It is scaledto the absolutevalue 1980;Solomonet al., 1994]. Additionally,the scaveng- of the crosssectionfrom Cox et al. [1999].A synthetic Ring spectrumaccountsfor changesof the optical depth Copyright 2001by theAmericanGeophysical Union. Papernumber2000GL012784. 0094-8276/01/2000GL012784505.00
of the Fraunhofer
lines due to rotational
Raman
scat-
tering (Ring effect). The temperaturedependenceof the NO2 absorption is accountedfor by using two NO2 reference spectra, recorded at 223 K and 260 K. All 1941
1942
FRIEg ET AL: TROPOSPHERIC
IODINE
0 Ring
0.1.10-3 to 0.3.10-3 (Figurel, panelD), clearlysmaller thantypicalIO absorption features(about1. l0 -3, see
o
-2 .
,
.
,
.
,
IN ANTARCTICA
An example for a detected IO absorption is shown in Figure 1. It is obtained from a twilight spectrum recorded at SZA - 92ø. Our spectroscopictechnique resultsin root mean square(RMS) residualsof about
4
-4
OXIDE
' NO•- '
' I Figure l, panel C). The retrievalerror at SZA - 92ø varies between 10% and 20%. The uncertainty in the
5
=.. ,e-2 j
o
absolute value of the IO cross section leads to an ad-
-5
ditional systematicerror. Cox et al. do not report on
io
0.5
=_ . . ,,
-2
the error of the IO cross section.
We assume this er-
ror to be in the order of 20%, in agreement with the
0.0
measurements of Hb'lscherand Zellner [1999]. 0.0
-0.2
RMS= 1.4'10 '• i
415
ß
i
420
ß
i
425
,
i
ß
430
i
ß
435
i
440
ß
i
445
ß
i
450
ß
i
455
ß
i
Results
and
discussion
460
Figure 2 showsthe observedIO differential absorp-
Z [nm]
tion structure as a function of SZA during twilight. IO Figure 1. Example for the IO spectral retrieval, show- can be clearly detected for SZAs of up to 95ø. Since
ing the fitted Ring spectrum(panelA) and the absorp- the IO SCD decreases for SZA • 86ø (seebelow),negtion structuresof Ozone(panelB), NO2 (panelC), and ative absorptionsignaturesrelative to the Fraunhofer IO (panel D); dotted lines: retrievedspectralsigna- referenceat SZA -- 80ø are derived during twilight. tures; solid lines: fitted cross sections. Both the atmospheric spectrum and the Fraunhofer reference are recordedon October 30, 1999 at 92ø and 80ø SZA, respectively. The absorption structures of the weak absorbers 04, H20 and OC10 are not shown. cross sections are convoluted
to the instrument
function
as determined by recording the shape of the mercury emissionline at 435.8 nm. A fourth degreepolynomial is included in the fit in order to remove the broad-band
features causedby the atmosphericRayleigh and Mie scattering. The spectral noise is reducedby applying a singletriangular smoothingboth to the logarithm of the spectra and to the crosssectionsprior to the spectral retrieval, a procedurethat only slightly affectsthe retrievedoptical density (• 1%). While the reference spectra and the crosssectionsare fixed in wavelength, the atmosphericspectraare allowedto shift and squeeze in order to compensatepossiblechangesin the wave-
Figure 3 shows an example of the diurnal variation of the inferred
IO differential
slant column den-
sity (DSCD), i.e. the differencebetweenSCD(SZA) and SCD(80ø),on October27, 1999. Throughoutdaytime, the IO DSCDs remainalmostconstant,while they rapidly decreasefor SZA • 86ø. The largest negative IO DSCDs are observed at 95ø SZA when they drop
below-9.5.1013 rnolec/cm 2. For almosteachday in autumn and springof 1999,the diurnal variationof the IO DSCD is similar to the observationon Oct. 27, with
almost no changeduring daytime and a suddendrop during twilight. The data show a larger scatter for cloudydaysand snowdrift conditionsthan for clear sky days,indicatingthat near surfacemultiple Mie scattering modifiesthe radiative transferby changingthe light pathes and accordinglythe observedIO SCD. The amount of IO in the referencespectrum is determined by assumingthat IO completely reacts into its nighttime reservoirspeciesat SZA - 95ø, an assumption that is justified by photochemicalmodel studiesof
lengthcalibration[$tutz and ?latt, 1996]. 2
0
85 ø
0.0-
8
-2 oL•Pm m
-0.5
__. -•
-1.0
A
o %•
2
.5
ß •' -:2.0
•
•
-8
-o -:2.5
, : ", :
-.•
•- -3.0
o
•
-6
-•0 14
. :; ..-.__..'
•
-4.0 Retneved spectral ragnature
-4.5
Fitted IOcross section
SZA
Z [nm]
Figure 2. Detected IO absorption structure during twilight as a function of SZA, observedon October 30, 1999. The Fraunhoferreferencespectrumis recordedon the same day at SZA - 80ø. An offset is added to the individual absorptionstructures and the corresponding SZA is denoted at the right.
•
2
.•o,oc•.•,c., .,..,•.•,.
• I•
c.... species ,o.o
o
rese•oir
C 60
65
70
75
80
85
90
95
SZA [deg]
95 ø
4"'5 '42•0 '42•5 '4•0 '4•5 '440 '4z[5 '4•0 '4•5 '4•0
6
Reference
•o.4
• 0.0
-3.5
O•.2•
a
Figure
3.
Diurnal variation of the IO DSCD on
September5 (panelA) and October27 (panelB), 1999. The position of the Fraunhofer reference spectrum is marked
with
an arrow.
Left-hand
axis:
inferred
IO
DSCD; Right-hand axis: absolute IO SCD under the assumptionthat IO is totally removed from the atmosphere at SZA = 95ø. The error bars denote the retrieval error. Panel B: RMS residual of the spectral retrieval
on October
27.
FRIEg ET AL: TROPOSPHERIC IODINE OXIDE IN ANTARCTICA
reactiveiodinein the stratosphere[Pundtet al., 1998] andthe troposphere [ Vogtet al., 1999;$tutz et al., 1999; McFigganset al., 2000]. Accordingly, definingthe SZA = 95 ø observation
as the zero of the ordinate
Stratospheric
4-
1943
profile assumed
•=;-•.- .•_ .-.....?...
indicates
an absolute IO SCDof about9.5.1013molec/cm 2 for SZAs < 80ø. These values are given in the right-hand
ß
A I==1 i
ß
i
.
i
.
Tropospheric
scaleof Figure 3. Sincesome(but likely not much) IO could remain in the atmosphereeven at 95ø SZA, they
i
ß
i
ß
i
ß
,..,.__. i
ß
profile assumed
(2km box profile) .-..
should be regarded only as a lower limit. Information
about the location
of the observed IO in
the atmosphereis gainedusingthe AMFTRAN multiple
O
i.
scatteringMonte Carlo radiativetransfermodel [Marquardet al., 2000].Two setsof airmassfactors(AMFs) are calculated for • - 435nm, assumingthat IO is either completely located in the stratosphere or in the troposphere. For the calculation of stratospheric airmassfactors, a linearly increasingIO mixing ratio from the tropopause to an altitude of 14 km and a constant mixing ratio above was assumed. TroposphericAMFs are calculated by assuminga constant mixing ratio pro-
file from the ground level up to varying altitudes (1 to 10 km). The actual meteorological conditions(temperature,pressureand ozoneprofile) are taken from a balloonsoundingon Sept. 27, 1999 (G. K6nig-Langlo, AWI Bremerhaven,peps. communication,1999). An albedo of 0.85 was assumed and SAGE II data was used
•½o '
ds
'
'½o '
'½5 '
,•o
'
8'5 '
•o
SZA [deg]
Figure 5. Diurnal variation of the IO VCD on October 27, 1999, under the assumptionthat all IO is removedfrom the atmosphereat 95ø SZA, using the
stratospheric (PanelA) andtropospheric (PanelB) airmassfactorsshownin Figure 4. A layer height of 2 km was assumedto calculatethe troposphericVCD. posphericprofilesare smaller and showa much weaker dependenceon SZA with a maximum around SZA 85ø. The influence of the aerosol extinction profiles on the troposphericAMF was investigatedusing different
aerosolscenarios: (1) no aerosolin the troposphere, (2)
to determine the stratospheric aerosol extinction pro- the abovedescribedstandardscenario,(3) like (2) but file, while the tropospheric aerosolwas modelled using with an of 0.1 km -1 in the lowermost 2 km. Those a linearly increasingextinction coefficientfrom 12 km different aerosolscenarioslead to changesin the tropo-
altitudeto a valueof 5.10-3 km- 1 at ground.Anyinflu- sphericAMF of about 15% for SZA • 80ø. In a next step, IO VCDs are calculated. As shown to the photochemistry (chemicalenhancement) are not in the upper panel of Figure 5, assumingthat IO is loencesof changesin the shapeof the trace gasprofile due
included in the radiative transfer model, although pho- cated in the stratosphere results in a strong increase tochemistry may significantlychangethe stratospheric of the IO VCD towards noon, which contradicts the AMFs duringtwilight (i.e. for SZA > 85ø) [Wittrock expectedphotochemicalsteadystate of the reactiveio-
et al., 2000]. Sincea photochemical steadystate and dine species(I, IO). In contrast,for troposphericIO (lower panel of Figure 5) the expectedand observed
thus small changesof the IO concentrationsat daytime comparedto the rapid variationsduringtwilight can be expected,the discussionof stratosphericAMFs and the
diurnal
variation
of the IO VCD
are in much better
agreement. The IO VCD remainsalmost constantfor resultingvertical columndensities(VCDs) is restricted SZAs • 82ø, which reflects the photochemicalsteady to SZAs < 82 ø. The inferred airmass factors are shown state during day, and decreases rapidly during twilight, in Figure 4. While the stratosphericAMFs increase in agreementwith the modelstudies.Therefore,our obstrongly with increasingSZA, the AMFs for the tro- servationsstronglysuggestthat the bulk of the detected IO is locatedin the troposphere.Further assumingthat IO is completelylocated in the in the lowermost1300 m (which is the altitude of the inversionlayer as inferred 30 -. ,,Airmass factor forstratospheric profile from the ozonesoundingon October27, 1999), the ob25 served IO VCDs would correspondto mixing ratios of • 20 •-6.5 ppt. IO mixing ratios of that amount are also reß- 15 ,
portedfromthe MBL ?t MaceHead,Ireland(3.5- 6.5
• •0
•
5
•
5
ß
Airmass fac'tor fortropospheric p•ofile
'
..._....•:.•............
•__•---: ---:....
15 4
•
3
Layer height:
i
60
ß
-, .. -,.. B • ' ' • -'
-- -2 km - - -5 km ----8 km -- ' - 10 km
Co•cer.• ß
'
i
65
ß
i
70
ß
i
75 SZA
i
80
i
85
ß
i
90
Figure 4. Airmass factors calculated for • = 435nm using the AMFTRAN Monte Carlo radiative transfer model. Panel A: Stratospheric airmass factors for the scenariodescribed in the text. Panel B: Tropospheric airmass factors for a constant IO mixing ratio from
groundup to differentlayer heights(1, 2, 5, 8 and 10 km) as indicatedin the legend.
ppt) [Alickeet al., 1999]and severalotherlocationsin the mid-latitudes[Allan et al., 2000]. The seasonal variation
of the IO DSCD
for SZA
-
92ø - where the best signal to noiseratio for the spectra retrieval is obtained- is shown in Figure 6, indicating that troposphericIO is more abundant during summer than during winter. The following processes may be responsiblefor the observedseasonalvariation:
(1) the strongersolar illumination during summerallows a more efficient photodissociationof the organic iodine precursormoleculesand thus a larger formation
rate of inorganiciodine compounds,(2) due to the retreating sea ice, the distance from the observationsite to the open sea- from which the precursorsare likely to be emitted - is smaller in summer than in winter,
(3) morebiologicalactivity is expectedto occurduring summer,(4) the abundanceof the reactantsneededfor
1944
FRIEB ET AL: TROPOSPHERIC
IODINE
IN ANTARCTICA
Allan, B.J., G. McFiggans, and J.M.C Plane, Observations of iodine monoxide in the remote marine boundary layer,
o -1
o EE
OXIDE
J. Geophys. Res., 105, 14363-14369, 2000. Chameides, W.L., and D.D. Davis, Iodine: Its possiblerole in tropospheric photochemistry, J. Geophys.Res., 85,
-4
7383-7398, 1980. 0 -7: • -8,' co -g'
Cox, R.A., W.J. Bloss,and R.L. Jones,OIO and the atomosphericcycleof iodine, Geophys.Res. Lett., 26, 1857-1860,
• -10'
Ferlemann, F., et al., A new DOAS-instrument for stratosphericballoon-bornetrace gasstudies,Appl. Optics, 39,
1999
1.3
1.4
1.9
1.10
1.11
2377-2386, 2000.
HSnninger, G., Referenzspektrenreaktiver Halogenverbindungenffir DOAS-Messungen (in German),DiplomatheFigure 6. Seasonal variationof the IO DSCD during sis, University of Heidelberg, 1999. 1999 at SZA = 92ø, with respectto daily Fraunhofer Jenkin, M.E., R.A. Cox and D.E. Candeland, Photochemreferencespectrarecordedat SZA - 80ø. ical aspectsof troposphericiodine behaviour, J. A tmos. Date
1999
Chem., 2, 359-375, 1985.
the formationof the night time reservoirsmay have an Marquard, L., T. Wagner, and U. Platt, Improved air mass influenceon the speedof the IO reductionat twilight, factor conceptsfor scattered radiation differential optical and (5) the time elapsedbetween80 and92 degrees SZA absorptionspectroscopyof atmosphericspecies,J. Geois changingwith season,and might influencethe differphys. Res., 105, 1315-1327, 2000. encein IO concentrations if it is not in photostationary McFiggans,G., J.M.C. Plane, B.J. Allan, and L.J. Carpensteady state at low sun.
ter, A modeling study of iodine chemistry in the marine boundary layer, J. Geophys.Res., 105, 14371-14385,2000. HSlscher,D. and R. Zellner, Absorption Cross Section of
Summary and Conclusions
the IO Radical and Kinetics of the IO -b IO and IO -b NO
The absorption structure of iodine oxide has been clearly detected in the zenith sky spectra recorded at Neumayer Station during 1999. Although DOAS observations of zenith scattered sunlight provide no direct information
on the altitude
distribution
of the ob-
servedtrace gases,argumentsbasedon radiative transfer model calculationsand photochemistryindicate that the observed diurnal
variation
of the IO DSCD
is caused
by tropospheric rather than stratospheric IO. Assuming the observed IO to be completely located in the boundary layer leads to mixing ratios of roughly 5-10 ppt during summer, similar concentrationsas typically found in the mid-latitude MBL. However, more direct measurementslike long-path or off-axis DOAS would be necessaryfor a more precisequantificationof the IO amount and a detailed study of the iodine chemistry in the Antarctic
MBL.
Reactions, Proceedingsof the fifth European symposium on stratospheric ozone, EU-report 19340, 264-267, 2000 Pundt, I., J.-P. Pommereau, C. Phillips, and E. Latelin, Upper limit of iodine oxide in the lower stratosphere,J. Arm. Chem., 30, 173-185, 1998. ReifenhSuser,W., and K.G. Heumann, Determinations of methyl iodide in the Antarctic atmosphereand the south polar sea, Arm. Env., 26A, 2905-2912, 1992. Schall,C., F. Laturnus, and K.G. Heumann, Biogenicvolatile organoiodineand organobrominecompoundsreleasedfrom polar macroalgae,Chemosphere,28, 1315-1324, 1994. Solomon,S., R. R. Garcia, A. R. Ravishankara,On the role of iodine in ozone depletion, J. Geophys.Res., 99, 2049120499, 1994.
Stutz, J., and U. Platt, Numerical analysisand estimation of the statistical error of differential optical absorption spectroscopymeasurementswith least-squaresmethods, Appl. Optics, 35, 6041-6053, 1996.
In contrastto the findingsof Wittrocket al. [1999],no Stutz, J., K. Hebestreit, B. Alicke and U. Platt, Chemevidencefor stratosphericIO is found, mainly because istry of Halogen Oxides in the Troposphere:Comparison the expected small stratospheric signature of IO may of Model Calculations with Recent Field Data, J. Arm. be masked by the strong tropospheric signal observed Chem., 3d, 65-68, 1999. at Neumayer. Vogt, R., R. Sander, R. von Glasow, and P.J. Crutzen, Io-
Model studies[Stutzet al., 1999]showthat the IO mixingratiosobservedat mid-latitudes(6 ppt) can destroy • 0.45 ppb ozoneper hour. Therefore IO is probably the dominating sink of BL ozone in Antarctica. Acknowledgments.
We appreciatethe logisticalsup-
port of the Alfred WegenerInstitute, Bremerhaven.We wish to thank the 1999 over- wintering crew of the NeumayerStation, especiallyAndrea Wille, for operatingand maintaining the instrument. The spectral retrieval was per-
dine chemistry and its role in halogenactivation and ozone lossin the marine boundary layer: a model study, J. Arm. Chem., 32, 375-395, 1999. Wittrock, F., R. Mfiller, A. Richter, H. Bovensmann, and
J.P. Burrows, Measurementsof iodine monoxide (IO) above Spitsbergen, Geophys.Res. Lett., 27, 1471-1474, 2000.
formedusing the WINDOAS analysissoftware,developed U. Friefi, T. Wagner, I. Pundt, K. Pfeilsticker, by CarolineFayt and Michelvan Roozendael(BelgianInstitute for SpaceAeronomy,Brussels).The supportof the and U.Platt, Institut fiir Umweltphysik, University of projectby the EU (contractENV4-CT-97-0521)isgratefully Heidelberg,INF 229, 69120 Heidelberg,Germany. (e-mail: acknowledged. udø'friess@iup'uni-heidelberg'de) References Alicke, B., K. Hebestreit, J. Stutz, and U. Platt, Iodine oxide in the marine boundary layer, Nature, 397, 572-573, 1999.
(ReceivedDecember18, 2000; revisedFebruary26, 2001; acceptedMarch 3, 2001.)
