NIGEL MARSH AND HENRIK...
16
COSMIC RAYS, CLOUDS, AND CLIMATE. NIGEL MARSH AND HENRIK SVENSMARK Danish Space Research Institute, Juliane Maries Vej 30, DK-2100 Copenhagen Ø, Denmark. Accepted: 10th May 2000 Abstract. A correlation between a global average of low cloud cover and the flux of cosmic rays in- cident in the atmosphere has been observed during the last solar cycle. The ionising potential of Earth bound cosmic rays are modulated by the state of the heliosphere, while clouds play an important role in the Earth’s radiation budget through trapping outgoing- and reflecting incoming radiation. If a physical link between these two features can be established, it would provide a mechanism linking solar activity and Earth’s climate. Recent satellite observations have further revealed a correlation between cosmic ray flux and low cloud top temperature. The temperature of a cloud depends on the radiation properties determined by its’ droplet distribution. Low clouds are warm ( 273K) and therefore consist of liquid water droplets. At typical atmospheric supersaturations ( 1%) a liquid cloud drop will only form in the presence of an aerosol, which acts as a condensation site. The droplet distribution of a cloud will then depend on the number of aerosols activated as cloud condensation nuclei (CCN) and the level of super saturation. Based on observational evidence it is argued that a mechanism to explain the cosmic ray - cloud link might be found through the role of atmospheric ionisation in aerosol production and/or growth. Observations of local aerosol increases in low cloud due to ship exhaust indicate that a small perturbation in atmospheric aerosol can have a major impact on low cloud radiative properties. Thus, a moderate influence on atmospheric aerosol distributions from cosmic ray ionisation would have a strong influence on the Earth’s radiation budget. Historical evidence over the past 1000 years indicates that changes in climate have occurred in accord with variability in cosmic ray intensities. Such changes are in agreement with the sign of cloud radiative forcing associated with cosmic ray variability as estimated from satellite observations. 1. Introduction The sun is a variable star, which emits both electromagnetic radiation and energetic particles known as the solar wind - these are released as a plasma carrying a fin- gerprint of the solar magnetic field throughout inter-planetary space. Effects from the solar wind are felt at distances well beyond Neptune, possibly up to 200 AU from the sun, in a region of space known as the Heliosphere (figure 1). The flux of the inter-planetary magnetic field (IMF) at 1AU is 5nT. Variability in solar activity affects both the radiative output of the sun and the strength of the IMF carried by the solar wind. The IMF shields the heliosphere from galactic cosmic radiation which, consists of energetic particles, mainly protons that are accelerated through stellar processes in our galaxy. Thus, solar variability modulates both the flux of incoming galactic cosmic radiation (GCR) and the amount of solar radiation received by the planets. Space Science Reviews 00: 1–16, 2000. c 2000 Kluwer Academic Publishers. Printed in the Netherlands. SSR_Paper.tex; 24/05/2000; 14:49; p.1
Transcript of NIGEL MARSH AND HENRIK...
COSMIC RAYS, CLOUDS, AND CLIMA TE.
NIGEL MARSH AND HENRIK SVENSMARKDanishSpaceResearch Institute,
JulianeMariesVej 30,DK-2100CopenhagenØ, Denmark.
Accepted:10thMay 2000
Abstract. A correlationbetweenaglobalaverageof low cloudcoverandtheflux of cosmicraysin-cidentin theatmospherehasbeenobservedduringthelastsolarcycle.Theionisingpotentialof Earthboundcosmicraysaremodulatedby thestateof theheliosphere,while cloudsplayanimportantrolein the Earth’s radiationbudgetthroughtrappingoutgoing-and reflectingincoming radiation.If aphysicallink betweenthesetwo featurescanbeestablished,it would provide a mechanismlinkingsolaractivity andEarth’s climate.Recentsatelliteobservationshave further revealeda correlationbetweencosmicray flux and low cloud top temperature.The temperatureof a cloud dependsontheradiationpropertiesdeterminedby its’ dropletdistribution.Low cloudsarewarm(
�273K) and
thereforeconsistof liquid waterdroplets.At typical atmosphericsupersaturations( � 1%) a liquidclouddropwill only form in thepresenceof anaerosol,whichactsasacondensationsite.Thedropletdistribution of a cloudwill thendependon thenumberof aerosolsactivatedascloudcondensationnuclei (CCN) andthe level of supersaturation.Basedon observationalevidenceit is arguedthatamechanismto explain the cosmicray - cloud link might be found throughthe role of atmosphericionisationin aerosolproductionand/orgrowth. Observationsof local aerosolincreasesin low clouddueto shipexhaustindicatethatasmallperturbationin atmosphericaerosolcanhaveamajorimpacton low cloud radiative properties.Thus,a moderateinfluenceon atmosphericaerosoldistributionsfrom cosmicray ionisationwould have a stronginfluenceon theEarth’s radiationbudget.Historicalevidenceover the past1000yearsindicatesthat changesin climatehave occurredin accordwithvariability in cosmicray intensities.Suchchangesarein agreementwith thesignof cloudradiativeforcingassociatedwith cosmicrayvariability asestimatedfrom satelliteobservations.
1. Intr oduction
Thesunis avariablestar, whichemitsbothelectromagneticradiationandenergeticparticlesknown asthesolarwind - thesearereleasedasa plasmacarryinga fin-gerprintof thesolarmagneticfield throughoutinter-planetaryspace.Effectsfromthe solarwind arefelt at distanceswell beyond Neptune,possiblyup to 200 AUfrom the sun,in a region of spaceknown asthe Heliosphere(figure 1). The fluxof the inter-planetarymagneticfield (IMF) at 1AU is � 5nT. Variability in solaractivity affects both the radiative output of the sunand the strengthof the IMFcarriedby the solarwind. The IMF shieldsthe heliospherefrom galacticcosmicradiationwhich,consistsof energeticparticles,mainlyprotonsthatareacceleratedthroughstellarprocessesin our galaxy. Thus,solarvariability modulatesboth theflux of incominggalacticcosmicradiation(GCR)andtheamountof solarradiationreceivedby theplanets.
SpaceScienceReviews 00: 1–16,2000.c�
2000KluwerAcademicPublishers. Printedin theNetherlands.
SSR_Paper.tex; 24/05/2000; 14:49; p.1
2 MARSH AND SVENSMARK
Figure 1. Schematicof Heliosphere
Satellitemeasurementsof total solarradiationover the past20 yearshave re-vealedthat the total radiative output of the sun variesby � 0.1 % (1.4 Wm��� )over a solarcycle (approx.11 years).Currentlythis is believed to betoo small tosignificantly influenceobserved surfacetemperaturesdirectly, althoughthis mayhave beenlarger backin time (Leanet al, 1995).However, theUV region of thesolar radiationspectrumis observed to vary by up to 10 %. UV is absorbedbyozoneandcanleadto variationsin Stratosphericheating.It hasbeensuggestedthatvariability in this heatingdueto changesin solaractivity canhave aninfluenceonthedynamicalpropertiesof thetropospherebelow (GellerandAlpert,1980;Haigh,1996;Shindellet al, 1999).It remainsto beseenhow this affectssurfaceclimateoncetakinginto accountthemany feedbackmechanismsinvolved.
GCRsinteractwith theEarth’satmospherethroughnuclearcollisionsproducingsecondaryparticles(protons,neutronsandmuons)whichcanpenetratedeeperintotheatmosphere;theseundergo furthercollisions,which leadsto acascadeof parti-cles.Thecascadeof particlesreachesamaximumatabout16km.This is wherethenucleonicpartof thesecondaryparticles,i.e., neutronsandprotons,have reachedenergies that are too low to undergo further collisions as a result of increasingatmosphericdensitywith depth.Below 16km, the charge particleintensitydropsoff andit is mainly the muonsandelectrons,productsof very high-energy GCRcollisions,whichcontributeto thechargedparticleintensitiesat theselow altitudes
SSR_Paper.tex; 24/05/2000; 14:49; p.2
COSMICRAYS,CLOUDS,AND CLIMATE. 3
Figure 2. Temporalvariability of GCR measuredat the Earth surface togetherwith the sunspotnumber(adaptedfrom Svensmark1998).This revealsa clearsolarcycle modulationof theCosmicRayflux. Thetop curve is monthlyaveragesof GCRasmeasuredby theCLIMAX neutronmonitor(cut-off rigidity � 3 GeV),Colorado(1953- 1996),valueshave beennormalisedto May 1965.Themiddle curve is the annualmeanvariation in GCR capturedby ionisationchambers(1934-1994),valuesare normalisedto 1965 (adaptedfrom Ahluwalia 1997).The bottom curve is the relativesunspotnumber.
(Lal andPeters,1967;HermanandGoldberg, 1978).Observationsof bothmuonsandneutronsmadeattheEarth’ssurfacecontaininformationrelatedtoatmosphericionisationatdifferentaltitudesandcanrevealvariability betweendifferentenergiesof theGCRspectrum(figure2).
Cosmicraysareresponsiblefor the ionisationin the lower atmospherebelow35km. Figure3 revealsatmosphericprofilesof ion pair productiontaken duringsolarmaximumandsolarminimum.Thedifferencebetweenmaxandmin peaksataround13kmandcontinuesdown to thesurface.
EdwardP. Ney (1959)observed thatatmosphericionisationwas’themeteoro-logical variablesubjectto the largestsolarcycle modulationin denserlayersoftheatmosphere’.How thismight influenceclimateis opento speculation,but Neysuggestedapossiblelink mightbethroughaninfluenceon ’storminess’.It wasnotuntil 1997thatSvensmarkandFriis-Christensen(1997)showedfrom observationsthationisationmight beinfluencingglobalcloudcover.
SSR_Paper.tex; 24/05/2000; 14:49; p.3
4 MARSH AND SVENSMARK
Figure3.Variabilityoverasolarcyclein theverticalprofilesof Ion pairproductionoverThule,whichis situatedatahighmagneticlatitude(adaptedfrom Neher1971).Dueto geomagneticshielding,thelevels of ionisationarereducedat lower latitudes,but essentialfeaturesin the vertical profile aresimilar.
2. Cloudsand CosmicRays
Figure4 is a compositeof Satelliteobservationsof Earth’s total cloudcover fromSvensmark(1998).The cloud datacomprisethe NIMBUS-7 CMATRIX project(Stowe et al, 1988)(triangles),andthe InternationalSatelliteCloud ClimatologyProject(ISCCP)(Rossow andSchiffer, 1991)(Squares).Finally datafrom theDe-fenceSatelliteMeteorologicalProgram(DMSP)SpecialSensorMicrowave/Imager(SSM/I) (Diamonds)(WengandGrody, 1994).For detailsseeSvensmark(1998).In thefigure the cloud datais comparedwith variationin GCR flux andthe 10.7cm radio flux from the Sun. This radio flux follows closely variationsin totalsolar irradiance,soft X–rays,and in ultraviolet radiationover this period.Thereare clearly differencesbetweenthe variation of GCR and the radio flux with alag of almosttwo yearsbefore1987.Crucially, Earth’s cloudcover follows morecloselytheGCRflux asaproxy for atmosphericionisationratherthanthe10.7cmflux. This observation suggeststhat a mechanismlinking cloud cover with solaractivity will involve changesin ionisationpropertiesof theatmosphereratherthanincomingsolarradiation.
Cloudsinfluencevertically integratedradiative propertiesof theatmospherebycooling throughreflectionof incomingshortwave radiation,andheatingthroughtrappingof outgoinglongwave radiation.The net radiative impactof a particularcloud is mainly dependentuponits heightabove thesurfaceandits optical thick-
SSR_Paper.tex; 24/05/2000; 14:49; p.4
COSMICRAYS,CLOUDS,AND CLIMATE. 5
Figure 4. Compositefigureshowing changesin Earth’s total cloudcover for four clouddatasetsob-tainedfrom satelliteobservations.Also plottedarethecosmicraysfluxesfrom Climax (solid curve,normalisedto May 1965),and10.7 cm Solar flux (broken curve, in units of 10-22Wm Hz�� ).TrianglesaretheNimbus-7data,squaresarethe ISCCP-C2andISCCP-D2data,anddiamondsarethe DMSP data.All the displayeddatahave beensmoothedusinga 12-monthrunningmean.TheNimbus-7andtheDMSPdatais total cloudcover for theSouthernHemisphereover oceans,andtheISCCPdatahave beenderivedfrom geostationarysatellitesover oceanswith thetropicsexcluded.
TABLE I
Radiative forcing from ERBEexperiment(adaptedfrom Hartmann,1993).
High Clouds Middle Clouds Low Total �� � �� �� � � �� �� � �� �� � � �Clouds
GlobalCloudFraction% 10.1 8.6 10.7 7.3 26.6 63.3
NetCRFWm 2.4 � 7.0 1.1 � 7.5 � 16.7 � 27.7
nessascanbeseenfrom TableI. High optically thin cloudstendto heatwhile lowclouds,which aregenerallythick, tendto cool (Hartmann,1993).With a currentclimatic estimatefor the net forcing of the global cloud cover as a 27.7 Wm���cooling in the atmosphere,cloudsclearly play an important role in the Earth’sradiationbudget (Hartmann,1993; Ohring and Clapp,1980; Ramanathanet al,1989;Ardanuyetal, 1991).A significantsolarinfluenceonglobalcloudpropertiesis potentiallyimportantfor theEarth’sradiationbudget,but thesignandmagnitudewill bedeterminedby thecloudtypesaffected.
SSR_Paper.tex; 24/05/2000; 14:49; p.5
6 MARSH AND SVENSMARK
Figure 5. Monthly meanvaluesfor global anomaliesof; a) high ( � 440hPa), b) middle (440 -680hPa), andc) low ( � 680hPa) cloudcover (blue).Galacticcosmicray fluxes(GCR)from Climax(red,normalisedto May 1965)areusedasa proxy for solarvariability. The global averageof thetemporalmeanover this periodfor high,middleandlow IR detectedcloudsis 13.5%, 19.9%, and28.0% respectively.
Recentlyit hasbeenpossibleto observe cloud propertiesat differentaltitudesfrom a global datasetInternationalSatelliteCloud Climate Project(ISCCP-D2)(Schiffer and Rossow, 1983; Schiffer and Rossow, 1985; Rossow and Schiffer,1991; Rossow et al, 1996).Theseare derived from Top Of Atmosphere(TOA)radiancesfor the periodJuly 1983 to September1994.Monthly averagesof in-frared(IR) measurementsarepreferreddueto their superiorspatialandtemporalhomogeneityover visual observationsthat canonly be detectedduring daylight.Cloudcover is obtainedfrom analgorithmusingtheTOA IR statisticsto identifythe cloudinesson an equalareagrid (280kmx 280km).Cloud top temperatures(CT) and pressures(CP) are obtainedfrom an ISCCPIR model constrainedbywatervapourandverticaltemperatureprofilesretrievedfrom theTIROSObservedVerticalSounder(TOVS) (Rossow etal, 1996).CT andCParefoundby assumingan opaquecloud, i.e. emissivity ����� , and adjustingthe cloudspressurelevel(effectively cloud height) in the model until the reconstructedoutgoingIR fluxat TOA matchesthat observed. Basedon retrieved CP, cloudsare divided intoLow 680hPa ( ! 3.2km),Middle " 680-440hPa (3.2-6.5km),andHigh ! 440hPa( 6.5km),Figure5.
SSR_Paper.tex; 24/05/2000; 14:49; p.6
COSMICRAYS,CLOUDS,AND CLIMATE. 7
Figure 6. Global correlationmapof GCRwith anomaliesof Low cloud top temperature(CT). Thefractionof Earthpossessinga correlationcoefficient # 0.6 is 29.6% which is significantat # 99.9%.
Clearly, GCRflux over this periodis correlatedwith a $ 2 % absolutechange( $ 7 % relative change)in low cloud cover, while there is no correlationwithmiddle andhigh clouds.Sincelow cloudstendto be optically thick they areef-ficient reflectorsof sunlightandhave a negative impacton the Earth’s radiationbudget(TableI). Basedon observationsmadeunderthe EarthRadiationBudgetExperiment(ERBE) theaveragelow cloudradiative forcing contributesa coolingof % 16.7 W/m2. Thusa $ 2 % absolutechangein low cloud cover over a solarcycle,wouldgivea changein netlow cloudforcingof $ 1.2W/m2.
In additionto cloudcover, theIR modelalsogeneratesanadditionalcloudpa-rameter, cloudtop temperature(CT), at Low, Mid, andHigh altitudes.Thespatialcorrelationmap in Figure6 shows how low cloud top temperaturescovary withGCR flux, particularlyover oceansin regionswherestratocumulusandmarines-tratuscloudsaredominant(for detailssee(MarshandSvensmark,2000).Thelackof correlationat high latitudesis currentlynot understood,but may be relatedtodifferencesin processesof cloudformation.
However, theopaquecloudassumptionexcludesmicrophysicalpropertiesandsoconstrainscloudvariability to appearonly in cloud ’model height’, thusintro-ducinganelementof artificial variability into CT. Observedpropertiesof low levelmaritimecloudssuggeststhat they arenot opaque(Heymsfield,1993).Relaxingtheopaqueassumptionallows cloudvariability to manifestitself throughchangesin cloudopticaldepth,i.e., &('*) .
SSR_Paper.tex; 24/05/2000; 14:49; p.7
8 MARSH AND SVENSMARK
3. Cloudsand Aerosols
Cloudopticaldepthdependsonprocessesaffectingtheclouddropsizedistribution,andcloud thicknesswhich is influencedby atmosphericvertical temperaturepro-files.Sincelow cloudsarewarm( + 273K) they consistof liquid waterdropswhichform on cloud condensationnuclei (CCN), andso the drop sizedistribution de-pendson thedensityof atmosphericaerosolsactivatedasCCN.TheabundanceofCCNis determinedby boththelevel of supersaturationandthenumberof aerosolspresentin the atmosphereable to grow into CCN. Increasesin supersaturation,typically between0.1% andafew percent,activatesincreasinglysmalleraerosols.A solarsignalcould enterlow cloud propertiesthroughinfluencingatmosphericverticaltemperatureprofiles,watervapour, or aerosolto CCNactivationprocesses.
TOVS observations of the vertical profiles of water vapourand temperaturedemonstratelittle correlationwith GCR(figure7). Thissuggeststhatsolarinducedvariability of local thermodynamicpropertiesin theatmospherecannotbeentirelyresponsiblefor the observed changesin low cloud properties.Sincethe verticalatmosphericstructuredependson radiatively driven large-scalecirculation, thismightnotbesurprisinggiventhatvariability in solarirradianceagreespoorlywithcloudproperties(figure4).
4. Aerosolsand CosmicRays
Assumingtypicalatmosphericwatervapoursaturationsof up to a few percent,theabundanceof CCNis determinedthroughpropertiesof thebackgroundaerosolsizedistribution ( , 0.1-1.0mm). Productionof aerosolcanbe dueto many processesinvolving: gas-particleconversion,droplet-particleconversion,i.e.,evaporationofwater dropletscontainingdissolved matter, and bulk particlesfrom the surface,e.g.,smoke, dust,or pollen(PruppacherandKlett, 1997).Observationsof spectrain regionsof low cloudformationindicatethataerosolsareproducedlocally. In thetroposphereit is believed that ionisationcontributesto thegas-particleformationof ultrafine( - 20nm)aerosol,andthesubsequentgrowth into thematuredaerosoldistributions which act as CCN (Turco et al, 1998; Horrak et al, 1998; Yu andTurco,2000).Figure8 shows theionisationdensityprofile of theatmosphereandthemajor ion speciesproduced.Thecomplexity of ionsincreaseswith decreasingaltitude.It is theions,producedwithin thetroposphereby cosmicrays,which arepotentiallyimportantfor aerosolproduction.
Currentlyit is uncertainwhethervariability in atmosphericionisationdueto theGCR flux could have a significanteffect on eitheraerosolproductionor dropletgrowth. However, modelshave revealedthatnucleationthroughion-ion recombi-nation(Turcoetal, 1998)is capableof maintainingabackgroundaerosoldistribu-tion with realisticconcentrationsin thetroposphere,TableII.
SSR_Paper.tex; 24/05/2000; 14:49; p.8
COSMICRAYS,CLOUDS,AND CLIMATE. 9
Figure 7. Global correlation maps of GCR with anomaliesof a) atmosphericwater vapour(1000-680hPa), and b) atmospherictemperatureat 740 hPa. The water vapour and temperaturearegiven by TOVS observationscovering theperiodJuly 1983- September1994(Rossow, 1996).Anomaliesandcorrelationcoefficientsarecalculatedasin figure5 and6 (seebothfigurecaptions).Theshadingcorrespondsto filled contours,wherecorrelationcoefficient r . 0.8 - black,r . 0.6 -grey, andr / -0.6 - hatched,thereareno regionswith r / -0.8. The fractionof Earthpossessingacorrelationcoefficient . 0.6 is a) 11.7% andb) 3.4% which is not significant.
TABLE II
Concentration- cm0 1CCNin a maritimeair massat 1 % supersaturation 2 100
over thePacific. (PruppacherandKlett, 1997)
Backgroundaerosolin a maritimeair over the 2 200
Pacific. (PruppacherandKlett, 1997)
Backgroundaerosoldistributionmaintainedby 2 200-500
ion-ion recombination.(Turcoet al, 1998)
SSR_Paper.tex; 24/05/2000; 14:49; p.9
10 MARSH AND SVENSMARK
Figure 8. VerticalProfileof ionisationdensityin theatmosphere.Includedarethemajorion speciesatvariousaltitudes(adaptedfrom ViggianoandArnold, 1995).
This is justonepossiblemechanism,but it emphasisesthepotentialimportanceof ionisationfor troposphereaerosolproduction.It is not unrealisticto imaginethatasystematicchangein cosmicray ionisationcouldleadto achangein aerosolpopulationsactingasCCN andthusinfluencecloudproperties.
Note that the processof aerosolsacting as a sourceof CCN for low cloudsis quite differentto thesourceof nuclei for higheraltitudeice clouds.Therefore,theinfluenceof ionisationon aerosolprocessesaffectinglow cloudCCN neednotbe importantfor high clouds.A detaileddiscussionof ice nuclei canbe found inPruppacherandKlett 1997.
5. Influenceof the AerosolDistribution on Cloud Radiative Properties
The effect of increasedaerosolon cloud radiative propertiescanbe observed di-rectly from ship tracks.Ship exhaustemits aerosolinto low cloud layers thusincreasingthelocal backgroundaerosol.Consequently, theclouddropletdistribu-tion is modifiedwhichinfluencestheradiativepropertiesof thecloud.Two satelliteimagesat differentwavelengthscanbe seenin figure 9, wherea numberof shiptracksareobservable. In the visible (left-handimage),theseshow up as regularbright white lines indicatinganincreasein cloudalbedo.Theright-handimageis
SSR_Paper.tex; 24/05/2000; 14:49; p.10
COSMICRAYS,CLOUDS,AND CLIMATE. 11
Figure 9. Imagesof Ship CondensationTrails Off the California Coasttaken on 15 July 1998byGOES-9.Left imageis in theVisible (0.7 3 m)andright imageis theIR Fog Product(10.7 3 m - 3.93 m).
Figure10.Visible0.7 3 m (left) andNearIR 2.2 3 m (right) aircraftobservationsof shiptracksin lowcloud(adaptedfrom King etal, 1993).
thedifferencebetweenNIR andIR wherethesametracksshow upasdarkregions.TheNIR (3.2 4 m) is dominatedby absorptionpropertiesof thecloudwhile theIR(10.7 4 m) is theresultof thecloudemissivity. SubtractingtheNIR imagefrom theIR enhancesregionsof increasedabsorptionandemissionsuchasthedark tracksdueto shipexhaust.
In-situ aircraftmeasurementsmadethroughshiptrackcontaminatedcloudgivedetailto theradiativechanges(King etal,1993).Figure10showsradiationintensi-tiesfrom anaircraftpaththroughthemiddleof suchcloudlookingbothin thenadir(down) andzenith(up). In the visible, the greatestinfluenceis from backscatterbelow the aircraft,while the NIR revealsstrongabsorptionboth from above andbelow. In fact, theabsorptionis muchstrongerfrom above indicatingthecloud isabsorbinga greaterflux of incomingradiation.This suggeststhat an increaseinaerosolgivesa decreasein dropletsizeandanincreasein cloudliquid water, thusenhancingabsorptionof incomingradiationfrom above thecloud.Consequently,the cloud will heat in theseregions, which explains the darkenedtracks in theright-handsatelliteimageof figure9.
SSR_Paper.tex; 24/05/2000; 14:49; p.11
12 MARSH AND SVENSMARK
Figure 11. Reconstructedlow cloudcover (solid) from galacticcosmicray fluxesmeasuredwith aneutroncounterat Huancayo(cut-off rigidity 12.91GeV). Also shown is the 12 monthsmoothedglobal low cloudcover (dashed)from figure5c. Theright axis is anestimatedof changesin CloudRadiative Forcing(CRF)basedon theERBEresultsdisplayedin Table1. SeeTable3 for estimatesof long termtrendsin CRFfrom this reconstruction.
From the ship tracks,it is apparentthat increasesin aerosolleadsto both anincreasein albedoandan increasein absorptionandthereforecloud temperature.If an increasein GCR can lead to a similar increasein aerosol,then this wouldexplain the positive correlation’s betweenGCR and low cloud top temperatures(figure7) over a largefractionof theglobe.
6. Estimatesof PastCloud Radiative Forcing
Theradiative impactfrom cosmicray influenceon cloudscanbeextendedbackintimefrom areconstructionassumingalinearresponsefunction(Svensmark,1998).It wasarguedabove, thatcosmicray ionisationin thelower troposphereaffectingaerosoldistributionscouldexplainthesolarvariability observedin low clouds.Thehigh-energy secondaryparticlespenetratethe atmosphereto depthsof low cloudformation,thustheHuancayoneutroncounterwith cut-off rigidity at12.91GeVisusedin thereconstruction.Fromthefollowing relationship:
576 8 9 :<; = > ?@8 = 6 ACB 9 DE8 B 9 ?@AF8 9 G�6 >IHFJLKNMPO75RQLS(1)
a historyof low cloudcover is found,whereJ
andS
areobtainedfrom a leastsquaresfit over theperiodof availableISCCPlow clouddata(1983- 1994).Thereconstructedlow cloudcover is shown in figure11.
Levels of cosmicraysreceived at Earthare influencedby the strengthof thesolarmagneticfield. Thus,relative changesin thesolarmagneticfield canbeused
SSR_Paper.tex; 24/05/2000; 14:49; p.12
COSMICRAYS,CLOUDS,AND CLIMATE. 13
TABLE III
Estimatesof radiative heatingfrom thereconstructionof low cloudsshown in figure11.
Estimated SolarMax-Min 30yeartrend Centurytrend
Relative Change(%) in: 1983-1987 1964-1995 1901-1995
CosmicRay( T 12.91GeV) 3.5 U 3.5 U 11.2
SolarSourceMagneticFlux 41 131
GlobalLow CloudFraction(ISCCP) 6.1 U 3.5 U 8.6
NetChange(WmV W ) in
ERBELow CRF( U 16.7Wm V W ) U 1.0 0.5 1.4
to estimaterelative changesin cloud forcing. Lockwoodet al (1999)have recon-structeda historyof solarsourcemagneticflux for thepastcenturywhich agreeswell with theavailablesatelliteobservationsfor theperiod1964- 1995.Thisallowsanestimatefor cloudradiative forcing to bemadefor theperiodbeforecosmicrayobservationswereavailable.Thesolarmagneticfield is observedto havemorethandoubledover thepastcenturywhile at thesametimecosmicradiationmeasuredatHuancayohasdecreasedby 3.5%(TableIII). Usingthereconstructionof low cloudcover in figure11andtheERBEestimatesfor cloudradiative forcing(TableI) onecaninfera X 1.4WmY@Z warmingattributedto low cloudcoverchangesoverthepastcentury(1901-1995).This is potentiallyimportantwhenconsideringthatover thesameperiodthe estimatedheatingfrom increasedCOZ emissionsis X 1.5WmY@Zandchangesin solar irradiancereceived at Earthare0.4 WmY�Z (Lockwood andStamper, 1999).
A comparisonbetweencosmicrays and climate can be extendedyet furtherbackin time,asa historyof Solaractivity is well known for thepast10000years.This is from observingchangesin the productionof radioisotopesin the atmo-sphereby galacticcosmicrays.Variationsin [ \ Cproductionfor thelastmillenniumtogetherwith majorclimateepochscanbeseenin figure12.
Although we do not have a goodestimateof the absoluteglobal temperaturerecordcovering this period,someof themajorchangesin [ \ C appearto coincidewith significantclimate shifts. Solar activity was particularly high from 1000 -1300AD, known asthemedieval warmperiod,whenwine wasmadefrom grapesgrown in England,and the Vikings colonisedGreenland.This was followed bya cold periodknown as the Little Ice Age, which lasteduntil the middle of the19thcentury. During thisperiod,theVikings left GreenlandandtheRiverThamesregularly frozein London.TheMaunderMinimum (1645-1715)wasaparticularlycoldspellwhentherewasacompleteabsenceof sunspotsindicatingvery low solaractivity. Thisevidencesuggeststhatin thepastpeopleexperiencedclimatechangesthathave occurredin concurrencewith extremesin solarvariability.
SSR_Paper.tex; 24/05/2000; 14:49; p.13
14 MARSH AND SVENSMARK
Figure 12.Changesin ] ^ C levelsover thepast1000yearsrelative to 1950. ] ^ C is producedthroughnuclearinteractionsof cosmicray particlesin the atmosphere.Cosmicraysaremodulatedby thesolarwind; thus,variability in ] ^ C productionis mainly a resultof changesin solaractivity. Highsolaractivity leadsto strongershieldingfrom thesolarwind andthusa reductionin theproductionrate.Notetheinvertedaxis for ] ^ C anomalies.Thesharpdeviation of ] ^ C duringthelastcenturyisdueto theburningof ] ^ C depletedfossil fuels(theSuesseffect).
7. Conclusions
Basedon the ISCCP-D2IR cloud datathereis a clearcorrelationbetweensolaractivity andpropertiesof low cloudsin contrastto middle andhigh clouds.Thecorrelationis seenin two independentparameters;a)Low cloudcover, andb) Lowcloudtoptemperature.Observationsof atmosphericparametersfrom TOVS donotsupporta solar-cloud link throughtroposphericdynamicsinfluencedby UV ab-sorptionin thestratosphere.However, sincelow cloudsarewarmandcompriseofwaterdropletsnucleatedby aerosol,a link maybefoundthroughaGCRinfluenceontheatmosphericaerosoldistribution.Shiptrackssuggestthatlow cloudradiativepropertiesaresensitive to the backgroundaerosoldistribution suchthat radiativeforcing dueto increasedaerosolleadsto a warmingof thecloud.If anincreaseinGCRcanleadto similar increasesin aerosol,thenthis would explain thepositivecorrelationbetweenGCR and low cloud top temperaturesover a large fractionof theglobe.Historical recordssuggestthatsystematicchangesin globalclimatefollow observedchangesin GCR.This is consistentwith low GCRcorrespondingto fewer low cloudsandthereforea warmerclimate.Crudeestimatesof changesin cloudradiative forcingover thepastcentury, whenthesolarmagneticflux morethandoubled,indicatesthataGCR-cloudmechanismcouldhavecontributed _ 1.4
SSR_Paper.tex; 24/05/2000; 14:49; p.14
COSMICRAYS,CLOUDS,AND CLIMATE. 15
Wm�a to the observed global warming.Theseobservationsprovide compellingevidenceto warrantfurtherstudyof theeffect of GCRonclouds.
References
Ardanuy, P., L.L. Stowe, A. Gruber, andM. Weiss. Shortwave, longwave, andnet cloud-radiativeforcing asdeterminedfrom Nimbus-7observations.J Geophys.Res., 96:18537,1991.
Geller, M.A., andJ.C.Alpert. Planetarywave coupling betweenthe troposphereand the middleatmosphereasa possiblesun–weathermechanism.J. Atmos.Sci., 37:1197,1980.
Haigh,J.D. Theimpactof solarvariability onclimate.Science, 272:981,1996.Hartmann,D.L. Radiative effectsof cloudson Earth’s climate,in Aerosol-Cloud-ClimateInterac-
tions. AcademicPress,1993.Herman,J.R.,andR.A. Goldberg. Sun,Weather, andClimate.NASA SP426,1978.Heymsfield,A.J. MicrophysicalStructuresof Stratiform and Cirrus Clouds,in Aerosol-Cloud-
ClimateInteractionsAcademicPress,1993.Horrak,U., J.Salm,andH. Tammet.Burstsof intermediateionsin atmosphericair. J Geophys.Res.,
103:13909,1998.King, M.D., L.F. Radke,andP.V. HobbsOpticalpropertiesof marinestratocumuluscloudsmodified
by shipsJ. Geophys.Res.98:2729,1993.Lal, D., andB. PetersCosmicray producedradioactivity on theEarth,in Encyclopaediaof Physics
Springer-VerlagBerlin, 1967.Lean,J., J. Beer, andR. Bradley. Reconstructionof solar irradiancesince1610: Implicationsfor
climatechange.Geophys.Res.Lett., 22:3195,1995.Lockwood,M., R. Stamper, andM.N. Wild A doublingof theSun’s coronalmagneticfield during
thepast100years.Nature, 399:437,1999.Lockwood, M. andR. Stamper Long-termdrift of the coronalsourcemagneticflux andthe total
solarirradiance.Geophys.Res.Lett., 26:2461,1999.Marsh, N.D. and H. Svensmark Low cloud propertiesinfluencedby solar activity Annales
Geophysicae, submitted,2000.Neher, H.V. Cosmic-Raysat high latitudesandaltitudescovering four solarmaxima. J. Geophys.
Res., 76(7):1637,1971.Ney, E.R. Cosmicradiationandtheweather.Nature, 183:451,1959.Ohring,G., andP.F. Clapp Theeffect of changesin cloudamounton thenetradiationat thetop of
theatmosphere.J. Atmos.Sci., 37:447,1980.Pruppacher, H.R.,andJ.D.Klett Microphysicsof CloudsandPrecipitation.Kluwer Dordrecht,1997.Ramanathan,V., et al. Resultsfrom theEarth’s radiationbudgetexperimentScience, 243:57,1989.Rossow, W.B., andR.A. Schiffer ISCCPCloudDataProducts.Bull. Am.Met.Soc., 72:2,1991.Rossow, W.B., A.W. Walker, D.E. Beuschel,and M.D. Roiter InternationalSatelliteCloud Cli-
matology Project (ISCCP): Documentationof New Cloud Datasets. World MeteorologicalOrganizationGeneva,1996
Schiffer, R.A., andW.B. Rossow The InternationalSatelliteCloud ClimatologyProject(ISCCP):Thefirst projectof theWorld ClimateResearchProgramme.Bull. Am.Met.Soc., 64:779,1983.
Schiffer, R.A., andW.B. Rossow ISCCPGlobalRadianceDataSet:A New Resourcefor ClimateResearch.Bull. Am.Met.Soc., 66:1498,1985.
Shindell,D., D. Rind,N. Balabhandran,J.Lean,andP. LonerganSolarcyclevariability, ozone,andclimate.Science, 284:305,1999.
Stowe,L.L, C.G.Wellemayer, T.F. Eck,H.Y.M. Yeh,andtheNimbus-7team.J. Clim., 1:445,1988.Svensmark,H., andE. Friis-ChristensenVariationof CosmicRayFlux andGlobalCloudCoverage
– A MissingLink in Solar-ClimateRelationships.J. Atm.Sol.Terr. Phys., 59:1225,1997.
SSR_Paper.tex; 24/05/2000; 14:49; p.15
16 MARSH AND SVENSMARK
Svensmark,H. Influenceof cosmicraysonclimatePhys.Rev. Lett., 81:5027,1998.Turco,R.P., J.-X. Zhao,andF. Yu A new source of troposphericaerosols:Ion-ion recombination.
GeophysicsRes.Lett., 25:635,1998.Viggiano,A.A. andF. Arnold Ion ChemistryandCompositionof theAtmosphere,in Handbookof
AtmosphericElectrodynamics.CRCPress,1995.Weng,F., andN.C. Grody J. Geophys.Res., 99:25535,1994.Yu, F., andR.P. Turco Ultrafineaerosolformationvia ion-mediatednucleationGeophys.Res.Lett.,
27(6):883,2000.
Addressfor Offprints: Nigel MarshandHenrikSvensmark,DanishSpaceResearchInstitute,JulianeMariesVej 30,DK-2100CopenhagenØ, Denmark,[email protected]
SSR_Paper.tex; 24/05/2000; 14:49; p.16
