Probing Saturn's tropospheric cloud with Cassini/VIMS

1.5 MB·

by  Joanna K Barstow

(cid:13)c2016. ThismanuscriptversionismadeavailableundertheCC-BY-NC-ND 4.0
licensehttp://creativecommons.org/licenses/by-nc-nd/4.0/
6
1
0
2

n
a
J

3
1

]
P
E
.
h
p
-
o
r
t
s
a
[

2
v
8
7
9
2
0
.
1
0
6
1
:
v
i
X
r
a
1
Probing Saturn’s tropospheric cloud with Cassini/VIMS
J.K.Barstowa,∗,P.G.J.Irwina,L.N.Fletchera,b,R.S.Gilesa,C.Merleta
aAtmospheric,OceanicandPlanetaryPhysics,ClarendonLaboratory,UniversityofOxford,ParksRoad,
Oxford,UK
bDepartmentofPhysicsandAstronomy,UniversityofLeicester,UniversityRoad,Leicester,UK
Abstract
InitsdecadeofoperationtheCassinimissionhasallowedustolookdeepintoSaturn’s
atmosphereandinvestigatetheprocessesoccurringbelowitsenshroudinghaze.Weuse
VisualandInfraredMappingSpectrometer(VIMS)4.6—5.2µmdatafromearlyinthe
missiontoinvestigatethelocationandpropertiesofSaturn’scloudstructurebetween
0.6and5bars.Weaveragenightsidespectrafrom2006overlatitudecirclesandmodel
the spectral limb darkening using the NEMESIS radiative transfer and retrieval tool.
We present our best-fit deep cloud model for latitudes −40◦ < λ < 50◦, along with
retrievedabundancesforNH ,PH andAsH . WefindanincreaseinNH abundance
3 3 3 3
attheequator,acloudbaseat∼2.3barandnoevidenceforcloudparticleswithstrong
absorptionfeaturesin the 4.6—5.2µm wavelengthrange, allof whichare consistent
with previous work. Non-scattering cloud models assuming a composition of either
NH orNH SH,withascatteringhazeoverlying,fitlimbdarkeningcurvesandspectra
3 4
atalllatitudeswell;theretrievedopticaldepthforthetropospherichazeisdecreasedin
thenorthern(winter)hemisphere,implyingthatthe hazehasa photochemicalorigin.
Ourabilitytotestthishypothesisbyexaminingspectraatdifferentseasonsisrestricted
by the varying geometry of VIMS observations over the life of the mission, and the
appearanceoftheSaturnstormtowardstheendof2010.
∗CorrespondingAuthor
Emailaddress:[email protected](J.K.Barstow)
PreprintsubmittedtoElsevier AcceptedJanuary2016
1. Introduction
It has long been known that clouds are present on the giant planets in our solar
system, but attempts to predict their location and composition using microphysical
models have so far been relatively unsuccessful (e.g. Atreyaetal. 2005, describing
the cloudpatternsfoundon Jupiterby the Galileo spacecraft). Cloudsare intimately
linked with planetary dynamics and chemistry, so understandingtheir formation and
behaviourisakeypartofstudyinganyplanetaryatmosphere.
ThearrivaloftheCassinimissionatSaturnprovidedanunprecedentedopportunity
tostudyitsatmosphere. Inthesubsequentdecade,Saturn’sstratosphericcomposition
hasbeenmonitoredduringthechangingseasons(Fletcheretal., 2010;Sinclairetal.,
2013; Fletcheretal., 2015); a spectacular hexagonalvortex has been observedat the
north pole (Fletcheretal., 2008; Bainesetal., 2009); and the developmentof a dra-
matic,largescalestormhasbeentracedoveraperiodofseveralmonths(Fletcher,L.N.etal.,
2011;Fischeretal.,2011b;Sa´nchez-Lavegaetal.,2011;Fletcheretal.,2012;Hesmanetal.,
2012;Sromovskyetal.,2013;Sayanagietal.,2013;Achterbergetal.,2014).Cassini’s
suiteofinstrumentsincludestheVisualandInfraredMappingSpectrometer(VIMS),
whichprovideswavelengthcoveragebetween0.3and5.1µm ataspectralresolution
of ∼16 nm. Absorption bands due to methane, ammonia, phosphine and other trace
gasesarepresentinthiswavelengthrange;wecanobservethereflectedsunlightsigna-
turefromthedaysideatshorterwavelengths,andonthenightsidethethermalemission
fromtheplanetbeginstoemergeataround4.6µm. Thisbroadwavelengthcoverage
providessensitivityoveralargealtituderange,makingthisinstrumentextremelyuse-
ful for atmospheric sounding; also, due to the typical size of particles (Romanetal.
2013 finds tropospheric haze particles have radii of approximately 2 µm), VIMS is
highlysensitive beyond4.6µm to the spectraleffectofcloudsandhaze in the lower
atmosphere(between1and8bars).
Inthiswork,weuseVIMS4.6—5.2µmthermalemissionspectrafromthenight-
3
side of Saturn to investigate the tropospheric cloud and haze. Stratospheric and tro-
pospherichazepropertiescanbeexploredusingreflectedlightfromthedayside(e.g.
KarkoschkaandTomasko 2005; Sromovskyetal. 2013; Romanetal. 2013) but sun-
lightdoesnotpenetratefarenoughintoSaturn’satmospheretoeasilyprobecloudmuch
beyondthe 1-bar pressure level. On the other hand, thermalemission from the deep
atmosphere is absorbed and scattered by clouds in this altitude region (Bainesetal.,
2006; Choietal., 2009), as discussed by Fletcheretal. (2011a), who presented the
first detailed exploration of thermal emission from Saturn using VIMS. They inves-
tigated the sensitivity of the spectrum to properties of the tropospheric cloud and
haze, as well as determining the latitudinal dependence of PH , NH and AsH gas
3 3 3
abundances, but found considerable solution degeneracy. We build on this previous
work by using spectroscopic limb darkening within latitude circles to provide fur-
therconstraintonthepropertiesofthecloudandhaze. Ground-basedobservationsby
Yanamandra-Fisheretal.(2001)showedstronglatitudinalvariationin5.2µmbright-
ness,attributedtovariationincloudproperties;weaimtogainabroad,globalpicture
ofSaturn’stroposphericaerosolpropertiesasafunctionoflatitude.
Uncertaintyastothecompositionandsize, andthereforescatteringproperties,of
theKroniancloudsisamajorcontributortosolutiondegeneracyintheFletcheretal.
(2011a)study. AtreyaandWong(2005)useanequilibriumcloudmodelforSaturnto
predictthepresenceofNH iceandsolidNH SHcloudsinthetroposphere.TheNH
3 4 3
ice cloud is estimated to form a little above the 2 bar level, with the deeper NH SH
4
cloudformingataround5bars. Belowthislevelwemayalsoexpectwatericeclouds
toform,butitisunlikelythatthesewillpersisttohighenoughaltitudesfortheVIMS
measurementstobesensitivetothem(atthesewavelengths,VIMSismostlysensitive
topressuresbetween1and8bar,Fletcheretal.2011a).
Previous observational work on Saturn’s cloud (e.g. KarkoschkaandTomasko
2005; Fletcheretal. 2011a; Sromovskyetal. 2013; Romanetal. 2013) has indicated
4
thepresenceof bothstratosphericandtropospherichazes, with thetropospherichaze
locatedinthe regiondirectlyabovewherethe NH cloudis predictedtoform. How-
3
ever, infrared observationssensitive to the deeper tropospherehave provided no evi-
dence for the two distinct troposphericcloud decks(NH and NH SH) above the 10
3 4
barlevelpredictedbyAtreyaandWong(2005)(seeSromovskyetal.2013). Instead,a
singleclouddeckbeneaththetropospherichaze,inthe1—5barrange,ispreferred,lo-
catedinbetweenthepredictedbasepressuresforNH andNH SH.Thismayindicate
3 4
that the deep cloud is in fact a mixture of these two components, or is composed of
either NH or NH SH but also contains impurities. Based on the AtreyaandWong
3 4
(2005) model predictions, we consider NH and NH SH compositions for the tro-
3 4
posperic cloud and NH for the tropospheric haze. Adopting compositions from ab
3
initio models in this way reduces the degeneracy of the problem and allows a more
informativeexplorationofothercloudparameterssuchasparticlesize.
2. Dataandreduction
WeusenightsideVIMScubesfromApril2006(latenorthernwinter/southernsum-
mer;Table1)toinvestigatethecloudlimbdarkeningproperties.Wechoosecubesfrom
this yearas Cassini’s fairly equatorialorbitat thattime allows usto investigatefrom
theequatoruptothemid-latitudesofbothhemispheres. Italsofacilitatescomparison
withFletcheretal.(2011a),whousedthesamecubes. Theseareoverlappingobserva-
tionstakeninasinglesessionwhileSaturnrotatedunderneath,suchthatalllongitudes
wereobserved. TheyareshowninFigure1, inwhichitcanbeseenthatthenorthern
latitudesaremuchbrighterthanthesouthat5µm.
Similar coverage was obtained during 2007, and in Section 5.5 we compare the
limbdarkeningforthetwoyears. Weattemptedtoinvestigatefurtherintothemission
toseeifthesetrendsbegantochangeasSaturnmovedtowardsvernalequinoxandinto
northernsummer. However,ourabilityto dothiswas restrictedbythe unavailability
ofsimilardataproducts. Between 2008and2010,wecouldnotlocatea sequenceof
5
Observation Date IntegrationTime(s)
CM1524383985 2006-04-22 480
CM1524388848 2006-04-22 480
CM1524393612 2006-04-22 480
CM1524400806 2006-04-22 480
CM1524403247 2006-04-22 480
CM1524408018 2006-04-22 480
CM1524412815 2006-04-22 480
CM1524417617 2006-04-22 480
Table1:Listof2006datacubesusedinthecurrentresearch.
VIMS images covering a wide simultaneous range of latitudes and emission angles,
which is required for a study of this kind. This was due to the spacecraftmovingto
an inclined orbit. Towards the end of 2010 the large Saturn storm emerged, causing
great disruption to the atmosphere with effects that persisted for several Earth years,
preventinganyfurtherstudy.
Weinvestigatethelatitudinaldependenceofthecloudpropertiesbyexploitingthe
change in emission angle along a latitude circle as viewed by VIMS. To first order,
wedonotexpectsignificantlongitudinalvariabilityatthesepressurelevelsonSaturn
(see Yanamandra-Fisheretal. 2001 for details of 5 µm variability observedfrom the
ground),somuchofthebroadzonalvariationevidentinFigures1and2isduetolimb
darkening. Toaverageoutsmall-scalelongitudinalvariationwetake all8 datacubes
togetherandbinthespectrainlatitudeandemissionangle.Anydatapointswithasolar
incidenceangle of less than 105◦ are rejected, to avoid contaminationfrom reflected
sunlight.Weextractlatitudecirclesfrom40◦Sto50◦N,takingallpixelsatintervalsof
10◦ inlatitudewithaspreadof±3◦. By binningoverabroadlatitudebandwe hope
toaverageoveranysmall-scalevariation,althoughitmeansthatwedonotresolvethe
finedetailsofthelatitudevariationasreportedbyFletcheretal.(2011a).
We alsoaveragespectrainthelongitudinaldirectionevery10◦ inemissionangle,
to ensure that any localfeaturesare smoothedout. This resultsin an emission angle
rangebetween5◦ and45◦ attheequator,and35◦ and85◦ atthehighestlatitudes. The
6
Latitude(◦) Emissionanglerange(◦)
-40 ∼35—85
-30 ∼25—65
-20 ∼15—55
-10 ∼5—55
0 ∼5—45
10 ∼5—55
20 ∼15—55
30 ∼25—65
40 ∼35—85
50 ∼45—85
Table2: Emissionanglerangesforeachlatitudecircle. Therangesrefertothecentralangleofthehighest
orlowest10◦rangeincludedforeachlatitudecircle. Theanglesincreasetowardshigherlatitudesbecause
thesub-spacecraftpointliesclosetotheequatorforallobservations.
averagelimbdarkeningat5.1µmforlatitudecirclesand20◦northandsouthareshown
inFigure3. Thesmallscale variationisapparent,butitisalsoclearthattheaverage
captures the basic limb darkeningtrend well. The variation in emission angle range
isduetotheequatoriallocationofthespacecraftduringtheseobservations,leadingto
generallyhigheremissionanglesfurtherfromtheequator. Atlowandequatoriallati-
tudes,theimagesdonotextendtothelimboftheplanet,truncatingthelimbdarkening
range.AnexampleofthemanipulationofasinglecubeisshowninFigure2.
ThecubesweredownloadedfromtheNASAPDSarchiveandcalibratedusingthe
standardISISpipeline(Fletcheretal.,2011a). ThecubesareprojectedontoaSystem
IIIplanetographiclatitude/longitudegrid. Radiometricerrorsare conservativelyesti-
matedtobe12%oftheaveragefluxbetween4.7and5.1µm;the12%valueisbased
ontheerrorestimatesusedinFletcheretal.(2011a),butweassumeaconstanterrorof
12%oftheaverage4.7—5.1µmfluxacrossallwavelengthsandallemissionangles,
whichfavoursthespectralregionswherethesignalislargest.
3. Modelatmosphereandretrieval
TheSaturnmodelatmosphereisbasedontheworkofFletcheretal.(2011a). Line
datasourcesareasinthispreviouspaperandGilesetal. (2015). AfterFletcheretal.
7
Figure1: MapprojecteddatacubesasusedbyFletcheretal.(2011a),whichwealsouseinthiswork. A
clearhemisphericalasymmetryinthe5µmfluxisapparent,withthenorthernlatitudesappearingtobemuch
brighter. ReprintedfromIcarus,214,Fletcher, L.N.etal.,Saturn’stroposphericcompositionandclouds
fromCassini/VIMS4.6-5.1µmnightsidespectroscopy,510—533,Copyright(2011),withpermissionfrom
Elsevier
8
Figure2:MapprojecteddatacubeCM1524383985,showntoillustrateourdataselectionprocedure.White
contours show emission angles. Hatching indicates region rejected due to sunlight contamination (solar
incidenceanglelessthan105◦).Paleshadedstripesshowthelatitudecirclesused.
Figure3:5.1µmradianceforallpixelswithinasinglelatitudecircleat20◦northandsouth. Whilstpixel-
levelvariations areapparent, theshapeofthelimbdarkening relationiswellrepresented bytheaverage.
Thisencompassesbothlatitudinalandlongitudinalvariation.
9
Figure4:Jacobians(functionalderivatives)fortemperature,PH3,NH3andAsH3foratypicalcloudymodel
atmosphere as used in this work. A compact tropospheric cloud is located at 2.3 bars and an extended
hazebetween0.1and0.6bar. Jacobiansshowsensitivitytochangesintemperature(perK)andPH3,NH3
andAsH3abundances(perlogvolumemixingratio)atdifferentaltitudes. TheVIMSinstrumentismostly
sensitive topressures between 1and8bar. Theeffect ofthecloud can beclearly seen intheincreased
sensitivityabovethe2.3-barlevelinthetemperatureandPH3Jacobians,whereasthehazeistoohighupto
haveasimilareffect.
(2011a),weuselatitudinallyvaryingtemperatureprofilesbasedonretrievalsfromthe
Composite Infrared Spectrometer (CIRS) instrument averaged over the 2004—2008
periodofobservations,extrapolatedtoanadiabatinthedeepatmosphere(Fletcheretal.,
2007,2010). CIRS(FP3/FP4)operatesinthe7—16µmwavelengthrange,makingit
highly sensitive to Saturn’s thermalemission from 600—100 mb and 10—1 mb and
therefore an ideal probe of temperature in the upper troposphere and middle strato-
sphere. The temperatureat higherpressures, closer to the regionsin which VIMS is
sensitive(seeFigure4),isnotprobedbytheCIRSinstrumentandcannotbeindepen-
dentlyconstrainedusingVIMSdata; however,we expectthe temperatureto bemore
stable in the deeper regions of the atmosphere. We do not, therefore, expect small-
scalevariabilitytoaffectourresults,especiallyasweaverageoverspectrafromeight
differentdatacubes.
3.1. RetrievalAlgorithm
WeusetheNEMESISradiativetransferandretrievalalgorithm(Irwinetal.,2008)
tosimultaneouslyretrieveseveralatmosphericpropertiesfromtheVIMSspectra.After
Fletcheretal.(2011a),we onlyvarythemodelparameterstowhichthe4.6—5.1µm
10