VOLCAN IC ERUPTIONS AN D CLIMATE

Alan Robock Departmentof EnvironmentalSciences RutgersUniversity New Brunswick,New Jersey

Abstract. Volcaniceruptions are an importantnatural an enhancedpole-to-equator temperature gradient, es- cause of climate change on many timescales.A new pecially in winter. In the Northern Hemisphere winter capability to predict the climatic responseto a large this enhancedgradient produces a strongerpolar vortex, tropical eruption for the succeeding2 years will prove and this strongerjet stream producesa characteristic valuable to society.In addition, to detect and attribute stationarywave pattern of troposphericcirculation, re- anthropogenicinfluences on climate,including effects of sultingin winter warming of Northern Hemispherecon- greenhousegases, aerosols, and ozone-depletingchem- tinents.This indirect advectiveeffect on temperatureis icals, it is crucial to quantify the natural fluctuationsso strongerthan the radiative coolingeffect that dominates as to separatethem from anthropogenicfluctuations in at lower latitudes and in the summer. The volcanic the climate record. Studyingthe responsesof climate to aerosolsalso serveas surfacesfor heterogeneouschem- volcanic eruptions also helps us to better understand ical reactions that destroy stratosphericozone, which important radiative and dynamical processesthat re- lowers ultraviolet absorptionand reducesthe radiative spondin the climate systemto both natural and anthro- heating in the lower stratosphere,but the net effect is pogenicforcings. Furthermore, modeling the effects of still heating.Because this chemicaleffect dependson the volcanic eruptionshelps us to improve climate models presenceof anthropogenicchlorine, it has only become that are needed to studyanthropogenic effects. Large important in recent decades.For a few days after an volcanic eruptions inject sulfur gasesinto the strato- eruptionthe amplitudeof the diurnal cycleof surfaceair sphere,which convertto sulfate aerosolswith an e-fold- temperature is reduced under the cloud. On a much ing residencetime of about 1 year. Large ash particles longer timescale,volcanic effects played a large role in fall out muchquicker. The radiativeand chemicaleffects interdecadalclimate changeof the Little Ice Age. There of this aerosol cloud produce responsesin the climate is no perfect index of past volcanism,but more ice cores system.By scatteringsome solar radiation back to space, fromGreenland and Antarctica will improve the record. the aerosolscool the surface,but by absorbingboth solar There is no evidencethat volcaniceruptions produce E1 and terrestrial radiation, the aerosol layer heats the Nifio events, but the climatic effects of E1 Nifio and stratosphere.For a tropical eruption this heating is volcaniceruptions must be separatedto understandthe larger in the tropicsthan in the high latitudes,producing climatic responseto each.

1. INTRODUCTION the main text.) Mitchell [1961] was the first to conducta superposedepoch analysis,averaging the effectsof sev- Volcanism has long been implicated as a possible eral eruptionsto isolate the volcanic effect from other causeof weather and climatevariations. Even 2000 years presumably random fluctuations. He only looked at ago,Plutarch and others[Forsyth, 1988] pointed out that 5-year averageperiods, however, and did not have a very the eruption of Mount Etna in 44 B.C. dimmed the Sun long temperaturerecord. Severalprevious reviews of the and suggestedthat the resultingcooling caused crops to effects of volcanoeson climate include Lamb [1970], shrivel and produced famine in Rome and Egypt. No Toon and Pollack [1980], Toon [1982],Ellsaesser [1983], other publicationson this subjectappeared until Ben- Asaturovet al. [1986],Kondratyev [1988], Robock [1989, jamin Franklin suggestedthat the Lakagigareruption in 1991],and Kondratyevand Galindo [1997].Past theoret- Iceland in 1783 might have been responsiblefor the ical studies of the radiative effects include Pollack et al. abnormallycold summerof 1783 in Europe and the cold [1976], Harshvardhan[1979], Hansenet al. [1992], and winter of 1783-1784 [Franklin,1784]. Humphreys [1913, Stenchikovet al. [1998].The work of H. H. Lamb, in fact, 1940]associated cooling events after largevolcanic erup- was extremely influential in the modern study of the tionswith the radiative effectsof the stratosphericaero- impact of volcanic eruptions on climate [Kelly et al., solsbut did not have a sufficientlylong or horizontally 1998]. Sincethese reviews, a deeper and more complex extensivetemperature databaseto quantify the effects. understandingof the impactsof volcaniceruptions on (Termsin italicare definedin the glossary,which follows weather and climate has resulted, driven by the many

Copyright2000 by the AmericanGeophysical Union. Reviewsof Geophysics,38, 2 / May'2000 pages 191-219 8755-1209/00/1998 RG000054 $15.00 Papernumber 1998RG000054 ß 191 ß 192 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

TABLE 1. Major Volcanic Eruptions of the Past 250 Years

Year of VolcaFlo Eruption VEI D VI/Emax IVI

Grimsvotn[Lakagigar], Iceland 1783 4 2300 0.19 Tambora, Sumbawa, Indonesia 1815 7 3000 0.50 Cosiguina,Nicaragua 1835 5 4000 0.11 Askja, Iceland 1875 5 1000 0.01' Krakatau, Indonesia 1883 6 1000 0.12 Okataina [Tarawera], North Island, New Zealand 1886 5 800 0.04 Santa Maria, Guatemala 1902 6 600 0.05 Ksudach, Kamchatka, Russia 1907 5 500 0.02 Novarupta [Katmai], Alaska, United States 1912 6 500 0.15 Agung, Bali, Indonesia 1963 4 800 0.06 Mount St. Helens, Washington,United States 1980 5 500 0.00 E1 Chich6n, Chiapas,Mexico 1982 5 800 0.06 Mount Pinatubo,Luzon, Philippines 1991 6 1000 ... The officialnames of the volcanoesand the volcanicexplosivity index (VEI) [Newhalland Self, 1982]are from Simkinand Siebert[1994]. The dustveil index(DVI/Emax) comes from Lamb [1970,1977, 1983], updated by Robockand Free [1995].The ice corevolcanic index (IVI) is the averageof Northern and SouthernHemisphere values and is representedas opticaldepth at X - 0.55 pom[from Robockand Free, 1995, 1996]. *SouthernHemisphere signal only; probablynot Askja.

studiesof the impactof the 1991 Pinatuboeruption and London [Symons,1888; Simkin and Fiske, 1983]. This continuinganalyses of the 1982 E1 Chich6n eruption in wasprobably the loudestexplosion of historictimes, and Mexico. the book includescolor figuresof the resultingpressure This paper reviewsthese new results,including the wave'sfour circuitsof the globe as measuredby micro- indirect effect on atmosphericcirculation that produces barographs.The 1963 Agung eruption produced the winter warmingof the Northern Hemisphere(NH) con- largeststratospheric dust veil in more than 50 yearsand tinentsand the new impactson ozonedue to the strato- inspired many modern scientific studies. While the sphericpresence of anthropogenicchlorine. A better Mount St. Helens eruptionof 1980was very explosive,it understandingof the impactsof volcaniceruptions has did not inject much sulfurinto the stratosphere.There- importantapplications in a numberof areas.Attribution fore it had very small global effects[Robock, 1981a]. Its of the warming of the past century to anthropogenic troposphericeffects lasted only a few days[Robock and greenhousegases requires assessment of other causesof Mass, 1982;Mass and Robock, 1982], but it occurredin climate changeduring the past severalhundred years, the United States and so received much attention. includingvolcanic eruptions and solar variations.After Quantificationof the size of theseeruptions is difficult, the next major eruption,new knowledgeof the indirect as different measures reveal different information. For effectson atmosphericcirculation will allow better sea- example,one could examinethe total massejected, the sonalforecasts, especially for the NH in the winter. The explosiveness,or the sulfur input to the stratosphere. impactsof volcaniceruptions serve as analogs,although The limitationsof data for each of thesepotential mea- imperfectones, for the effectsof other massiveaerosol sures,and a descriptionof indicesthat have been pro- loadings of the atmosphere, including meteorite or duced, are discussedlater. comet impactsor nuclearwinter. Volcanic eruptionscan inject into the stratosphere The largesteruptions of the past250 years(Table 1) tens of teragramsof chemicallyand microphysicallyac- have each drawn attention to the atmosphericand po- tive gasesand solid aerosolparticles, which affect the tential climatic effects becauseof their large effects in Earth's radiative balance and climate, and disturb the the English-speakingworld. (Simkin et al. [1981] and stratosphericchemical equilibrium. The volcaniccloud Simkinand Siebert[1994] provide a comprehensivelist of forms in several weeks by SO2 conversionto sulfate all known volcanoesand their eruptions.) The 1783 aerosol and its subsequentmicrophysical transforma- eruptionin Iceland producedlarge effectsall that sum- tions [Pintoet al., 1989;Zhao et al., 1995].The resulting mer in Europe [Franklin,1784; Grattan et al., 1998].The cloudof sulfateaerosol particles, with an e-foldingdecay 1815 Tambora eruption producedthe "year without a time of approximately1 year [e.g.,Barnes and Hoffman, summer"in 1816 [Stommeland Stommel,1983; Stothers, 1997], has important impacts on both shortwaveand 1984; Robock, 1984a, 1994; Harington, 1992] and in- longwave radiation. The resulting disturbanceto the spiredthe book Frankenstein[Shelley, 1818]. The most Earth's radiation balance affects surfacetemperatures extensivestudy of the impactsof a singlevolcanic erup- through direct radiative effectsas well as through indi- tion was carried out by the Royal Society,examining the rect effectson the atmosphericcirculation. In cold re- 1883 Krakatau eruption, in a beautifullyproduced vol- gions of the stratospherethese aerosol particles also ume includingwatercolors of the volcanicsunsets near serve as surfacesfor heterogeneouschemical reactions 38, 2 / REVIEWSOF GEOPHYSICS Robock:VOLCANIC ERUPTIONS AND CLIMATE ß 193 that liberate chlorine to destroyozone in the sameway H20 , N2, and CO2 being the most abundant.Over the that water and nitric acid aerosolsin polar stratospheric lifetime of the Earth these gaseshave been the main cloudsproduce the seasonalAntarctic ozone hole. sourceof the planet's atmosphereand ocean, after the In this paper I first briefly summarizevolcanic inputs primitive atmospherewas lost to space.The water has to the atmosphereand reviewour new understandingof condensedinto the oceans,the CO2 has been changed the radiativeforcing of the climate systemproduced by by plants into 02, with some of the C turned into fossil volcanic aerosols.Next, I briefly review the results of fuels. Of course,we eat the plants and the animalsthat new analysesof ice cores,since they give information eat the plants,we drink the water, and we breathe the about the record of past volcanism,and comparethese oxygen,so eachof us is made of volcanicemissions. The new recordsto past analyses.The effectsof eruptionson atmosphereis now mainly composedof N 2 (78%) and the local diurnal cycle are reviewed. Summer cooling 02 (21%), both of which had sourcesin volcanicemis- and winter warmingfrom large explosiveeruptions are sions. then explained.The impactsof volcanic eruptionson Of these abundant gases,both H20 and CO2 are decadal- and century-scaleclimate changes,and their importantgreenhouse gases, but their atmosphericcon- contributionsto the Little Ice Age and their relative centrationsare so large (evenfor CO2 at only about370 contributionto the warmingof the past century,are next ppm but growing)that individualeruptions have a neg- discussed.Then, I show that the simultaneous occur- ligible effect on their concentrationsand do not directly rence of the 1982 E1 Nifio and the E1 Chich6n eruption impact the greenhouseeffect. Rather, the most impor- wasjust a coincidenceand that it was not evidenceof a tant climatic effect of explosivevolcanic eruptions is cause and effect relationship.Finally, the impacts of through their emissionof sulfur speciesto the strato- volcanic eruptions on stratosphericozone are briefly sphere,mainly in the form of SO2 [Pollacket al., 1976; reviewed. Newhall and Self, 1982; Rampino and Self, 1984] but possiblysometimes as H2S [Luhret al., 1984;Ahn, 1997]. These sulfur speciesreact with OH and H20 to form 2. VOLCANIC INPUTS TO THE ATMOSPHERE H2SO4 on a timescaleof weeks,and the resultingH2SO 4 aerosolsproduce the dominant radiative effect from Volcanic eruptionsinject several different types of volcanic eruptions.Bluth et al. [1992], from satellite particlesand gasesinto the atmosphere(Plate 1). In the measurements, estimated that the 1982 E1 Chich6n past,it wasonly possible to estimatethese volatile inputs eruptioninjected 7 Mt of SO2 into the atmosphere,and basedon measurementsfrom active,but not explosive, the 1991 Pinatubo eruption injected 20 Mt. eruptionsand remote sensingof the resultingaerosol Once injectedinto the stratosphere,the large aerosol clouds from lidar, radiometers, and satellites. The ser- particlesand smallones being formed by the sulfurgases endipitousdiscovery of the ability of the Total Ozone are rapidly advected around the globe. Observations Mapping Spectrometer(TOMS) instrumentto monitor after the 1883 Krakatau eruption showedthat the aero- SO2 [e.g.,Bluth et al., 1992],however, has given us a new sol cloud circled the globe in 2 weeks [Symons,1888]. tool to directlymeasure stratospheric injection of gases Both the 1982 E1 Chich6n cloud [Robockand Matson, from eruptions. 1983] and the 1991 Pinatubocloud [Bluth et al., 1992] The major componentof volcaniceruptions is mag- circled the globe in 3 weeks. Although E1 Chich6n matic material,which emergesas solid,lithic material or (17øN)and Pinatubo(15øN) are separatedby only 2ø of solidifiesinto largeparticles, which are referredto as ash latitude, their clouds,after only one circuit of the globe, or tephra.These particles fall out of the atmospherevery endedup separatedby 15ø of latitude,with the Pinatubo rapidly, on timescalesof minutesto a few weeksin the cloudstraddling the equator[Stowe et al., 1992] and the troposphere.Small amounts can last for a few monthsin E1 Chich6n cloud extending approximatelyfrom the the stratospherebut have very small climatic impacts. equatorto 30øN[Strong, 1984]. Subsequent dispersion of Symons[1888], after the 1883 Krakatau eruption, and a stratosphericvolcanic cloud dependsheavily on the Robock and Mass [1982], after the 1980 Mount St. particular distributionof winds at the time of eruption, Helenseruption, showed that this temporarylarge at- althoughhigh-latitude eruption clouds are seldomtrans- mosphericloading reduced the amplitudeof the diurnal portedbeyond the midlatitudesof the samehemisphere. cycle of surface air temperature in the region of the For trying to reconstructthe effectsof older eruptions, troposphericcloud. These effects, however, disappear as this factor addsa further complication,as the latitude of soon as the particles settle to the ground. When an the volcano is not sufficient information. eruption columnstill laden with these hot particlesde- The normal residual stratosphericmeridional circu- scendsdown the slopesof a volcano,this pyroclastic flow lation lifts the aerosolsin the tropics,transports them can be deadlyto thoseunlucky enough to be at the base polewardin the midlatitudes,and bringsthem back into of the volcano.The destructionof Pompeii and Hercu- the troposphereat higherlatitudes on a timescaleof 1-2 laneum after the 79 A.D. Vesuviuseruption is the most years [Trepteand Hitchman, 1992; Trepteet al., 1993; famousexample. Holton et al., 1995]. Volcanic eruptions typically also emit gases,with Quiescent continuousvolcanic emissions,including 194 ß Robock:VOLCANIC ERUPTIONSAND CLIMATE 38, 2 / REVIEWSOF GEOPHYSICS

0.94

0.92

0.9

0.88

0.86

0.84

0.82

0.8

0.78 1960 1965 19•0 1975 1980 1985 1990 1995

Figure 1. Broadbandspectrally integrated atmospheric transmission factor, measured with thepyrheliometer shownin Plate2. Duttonet al. [1985]and Dutton [1992] describe the detailsof the calculations,which eliminate instrumentcalibration and solar constantvariation dependence, and showmainly the effectsof aerosols. Effectsof the 1963Agung, 1982 E1 Chich6n,and 1991Pinatubo eruptions can clearlybe seen.Years on abscissaindicate January of that year. Data courtesyof E. Dutton.

fumarolesand small episodiceruptions, add sulfatesto radiation by scattering.Some of the light is backscat- the troposphere, but their lifetimes there are much tered, reflectingsunlight back to space,increasing the shorterthan thoseof stratosphericaerosols. Therefore net planetaryalbedo and reducingthe amount of solar they are not important for climate change,but they energy that reachesthe Earth's surface.This backscat- couldbe if there is a suddenchange or a long-termtrend tering is the dominant radiative effect at the surfaceand in them develops.Global sulfur emissionof volcanoesto resultsin a net coolingthere. Much of the solarradiation the troposphereis about 14% of the total natural and is forward scattered,resulting in enhanceddownward anthropogenicemission [Graf et al., 1997] but has a diffuseradiation that somewhatcompensates for a large much larger relative contribution to radiative effects. reductionin the direct solarbeam. The longestcontin- Many volcanic emissionsare from the sides of moun- uous record of the effectsof volcaniceruptions on at- tains, abovethe atmosphericboundary layer, and thus mospherictransmission of radiation is the apparent they havelonger lifetimes than anthropogenicaerosols. transmis.sion record [Duttonet al., 1985;Dutton, 1992] Radiativeforcing (measured at the surface)from such from the Mauna Loa Observatory(Plate 2) shownin emissionsis estimated to be about -0.2 W m -2 for the Figure 1. The effectsof the 1963 Agung, 1982 E1 Chi- globeand -0.3 W m-2 forthe NH, onlya littleless than ch6n, and 1991Pinatubo eruptions can be clearlyseen. anthropogeniceffects. Although the Pinatubo eruption producedthe largest stratosphericinput of the three, the center of the E1 Chich6ncloud went directlyover Hawaii, while only the 3. RADIATIVE FORCING side of the Pinatubocloud was observed.The Agung cloudwas mostlyin the SouthernHemisphere, so only Plate 1 indicatesthe major radiativeprocesses result- the edgewas seenin Hawaii. Figure 2 showsseparate ing from the stratosphericaerosol cloud from a major direct and diffuse radiation measurements,also from volcanic eruption. The most obvious and well-known Mauna Loa, whichshow not onlythe strongreduction of effect is on solar radiation. Since the sulfate aerosol direct radiation by the 1982 E1 Chich6n and 1991 Pina- particlesare about the samesize as visiblelight, with a tubo eruptionsbut alsothe compensatingincrease (of typical effective radius of 0.5 •xm, but have a single- slightlysmaller amplitude) in the diffuseradiation. scatter albedo of 1, they strongly interact with solar The effect on solar radiation is so strongthat it can 38, 2 / REVIEWSOF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 195

'*'" X

lit 196 ß Robock:VOLCANIC ERUPTIONSAND CLIMATE 38, 2 / REVIEWSOF GEOPHYSICS

Plate 2. Photographof radiation instrumentsat the Mauna Loa Observatoryon the Island of Hawaii, United States,looking north toward Hualalai, on March 27, 1992. Observationsin Figures1 and 2 and photographin Plate3 are takenfrom theseand similarinstruments. Photograph by A. Robock.

Plate 3. Photographof skysurrounding the Sun on March 27, 1992,less than 1 year after the Pinatubo eruption,taken at MaunaLoa Observatoryby layingthe cameraon the supportshown in Plate2 andhaving a portionof that supportblock out the directsolar radiation. The milkyappearance is the enhancedforward scattering(Figure 2), clearlyvisible to the nakedeye. When the stratosphereis clear,the normalappearance is a deepblue producedby molecularRayleigh scattering. Photograph by A. Robock. 38 2 / REVIEWS OF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 197

Plate 4. Sunsetover Lake Mendota in Madison, Wisconsin,in May 1983, one year after the E1 Chich6n eruption. Photographby A. Robock.

easily be seen by the naked eye, making the normally [Houghtonet al., 1996, p. 109]. For aerosolswith a blue skymilky white (forwardscattering effect) (Plate 3). nonuniform vertical and horizontal distribution, Volcanic aerosolclouds are clearlyvisible from spacein Stenchikovet al. [1998] showedthat a completeformu- solarwavelength images [e.g., Robock and Matson, 1983] lation of radiative forcing must include not only the and in spaceshuttle photographs (backscattering). The changesof net fluxes at the tropopause,but also the reflection of the settingSun from the bottom of strato- verticaldistribution of atmosphericheating rates and the sphericvolcanic aerosol layers (called "dust veils" by change of downward thermal and net solar radiative Lamb [1970]) produces the characteristicred sunset fluxes at the surface. Using a detailed data set they (Plate 4) usedby Lamb as one meansof detectingpast developedfrom satelliteand ground-basedobservations, eruptions. The famous 1893 Edvard Munch painting, they calculated the aerosol radiative forcing with the "The Scream," showsa red volcanic sunsetover the Oslo ECHAM4 (European Center/Hamburg)general circu- harbor producedby the 1892 Awu eruption. The timing lation model (GCM) [Roeckneret al., 1996]. of the reportsof red sunsetswas usedby Symons[1888] An exampleof the radiative heatingfrom Pinatubois and Lamb [1970] to calculatethe height of the aerosol shown in Plate 5. At the top of the aerosol cloud, the layer, taking into accountthe geometryof the setting atmosphereis warmed by absorptionof solar radiation Sun. Robock [1983a] also providesa diagram showing in the near infrared (near-IR). This effect dominates how these red sunsets can be observed after the Sun has over the enhanced IR cooling due to the enhanced set. emissivity because of the presence of aerosols.An- To evaluate the effects of a volcanic eruption on dronovaet al. [1999] recentlyrepeated these calculations climate, the radiative forcing from the aerosolsmust be with a more detailed radiation model and confirmed the calculated.Stenchikov et al. [1998] presenteda detailed importanceof near-IR abosorption.In the lower strato- study of the radiative forcing from the 1991 Mount sphere the atmosphereis heated by absorptionof up- Pinatuboeruption. Although this was the most compre- ward longwaveradiation from the troposphereand sur- hensivelyobserved large eruption ever [e.g., Russellet face. Hence this warming would be affected by the al., 1996],they still neededto make severalassumptions distributionof cloudsin the troposphere,but Stenchikov to compensatefor gapsin the observations.For globally et al. [1998] found that this effect (on changedupward smoothradiative perturbations, such as changinggreen- longwaveflux) was random and an order of magnitude housegas concentrations, radiative forcing is definedas smaller than the amplitude of the warming. In the tro- the changein the net radiative flux at the tropopause posphere, there are small radiative effects, since the 198 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWSOF GEOPHYSICS

August, 1991 January• 1992 e) IO

30. 50 5O

lOO O0

300 O0

500 O0 700 O0 900' O0 505 305 EQ 3ON 5ON 6O5 30S EQ 30N 60N Visible b)

5o 5o !

100 100-

300 300

500 500 700- 700 900 go0 605 30S EO 30N 60• 60S 30S EQ 30N 60N Near IR c) g)

-,3

- 1

30

5o •o

ioo •oo

300' ß 0'05, .• ]oo

500 5o0 7OO 7oo 9O0 ., 90o 6OS 3OS EO 30N 6ON 60S 30S EQ 3ON 60N IR d) h) 10J ' ' ' 'I ,X•-n'n•1•• ,,•u•1 3O •o •o, Z' "• •" - 5o

I00

.01 3o0 500

500 50O I -0.01• '1I' 700 700 c•.y., • •-of• 900 .... 6OS 30S EO 3ON 60N 6OS 3OS EQ 30N 60N Total

Plate 5. Radiativeheating from Pinatubofor three differentwavelength bands, from Figure 10 of Stenchikov et al. [1998].Shown are monthly average, zonal average, perturbations of theradiative heating rates (K d-•) causedby the Pinatuboaerosols for (a) visible(X -< 0.68 txm),(b) near-IR (0.68 txm< X < 4 txm),(c) IR (X -> 4 txm),and (d) total, for August1991, and (e) visible,(f) near-IR, (g) IR, and (h) total, for January1992. 38, 2 / REVIEWSOF GEOPHYSICS Robock' VOLCANIC ERUPTIONS AND CLIMATE ß 199

540

520

500

480

460

440

420

400

38O

,360 Direct

54O 22O 2OO Diffuse 180 Radiation 160

140

120

1 O0

8O

6O

4O

2O

0 1978 1980 1982 1984 1986 1988 1990 '1992 1994 1996 1998

Figure 2. Direct and diffuse broadband radiation measurementsfrom the Mauna Loa observatory,mea- suredwith a trackingpyrheliometer and shadedisk pyranometer on morningswith clear skiesat solarzenith angleof 60ø , equivalentto two relative air masses[Dutton and Christy,1992]. The reductionof direct radiation and enhancementof diffuseradiation after the 1982E1Chich6n and 1991 Pinatuboeruptions are clearlyseen. Years on abscissaindicate Januaryof that year. Data courtesyof E. Dutton. reduceddownward near-IR (producingless absorption throughtropopause folds [Sassenet al., 1995], the global by water vapor in the troposphere)is compensatedby effect has not been quantified. the additional downward longwaveradiation from the aerosol cloud. At the surface the large reduction in direct shortwave radiation overwhelms the additional 4. INDICES OF PAST VOLCANISM downward diffuse shortwave flux and the small addi- tional downward longwave radiation from the aerosol To evaluate the causesof climate changeduring the cloud, except in the polar night, where there is no past centuryand a half of instrumentalrecords or during sunlight,which was first shownby Harshvardhan[1979]. the past 2000 years,including the MedievalWarming and This net cooling at the surface is responsiblefor the the "Little Ice Age," a reliable record of the volcanic well-known global cooling effect of volcanic eruptions. aerosol loading of the atmosphereis necessary.Five These calculationsagree with observationsof surface such indices(Table 2) have been compiled,based on flux changesmade at Mauna Loa by Dutton and Christy different data sourcesand criteria, but none is perfect. [1992].They alsoagree with observationsof Minnis et al. Robockand Free [1995, 1996] describethese indicesin [1993] made with Earth Radiation Budget Experiment detail and compare them, and here I summarizethem. satellitedata [Barkstrom,1984], if the effectsof strato- Another index describedby Robock and Free [1995], sphericwarming are considered. beingused by someclimate modeling groups at the time, Plate 5 also shows,for Pinatubo, that the lower strato- was never publishedand so is not includedhere. Pollack sphericheating is much larger in the tropicsthan at the et al. [1976] also compileda record of volcanicoptical poles.It is this latitudinal gradient of heatingwhich sets depth,but it was limited to 1880-1925 and 1962-1972. up a dynamicalresponse in the atmosphere,resulting in A perfect index would convey the radiative forcing the winter warming of NH continental regions,due to associatedwith each explosiveeruption. The radiative advectiveeffects, which dominate over the radiative ef- forcing is most directly related to the sulfur content of fects in the winter. emissionsthat reach into the stratosphereand not to the The possibleeffect of the aerosolson seedingcirrus explosivityof the eruption. However, all indirect indices cloud formation [Mohnen,1990] is indicatedin Plate 1. are either incompletein geographicalor temporal cov- While evidence exists for individual cases of cirrus cloud erage or are a measure of some property of volcanic formationby volcanicaerosols entering the troposphere eruptionsother than their stratosphericaerosol loading. 200 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWSOF GEOPHYSICS

TABLE 2. Indices of Past Volcanic Eruptions

Name Units How Calculated Reference

Dust veil index (DVI) Krakatau = 1000 Sapper[1917, 1927], sunsets,eruption, and radiation Lamb [1970, 1977, 1983] observations

Mitchell aerosol mass basedon H. H. Lamb (personalcommunication, 1970) Mitchell [1970] Volcanic explosivity Krakatau = 6 explosivity,from geologicand historicalreports Newhalland Self [1982] index (VEI) Simkinet al. [1981] Simkinand Siebert[1994] Sato -r (X = 0.55 txm) Mitchell [1970],radiation and satelliteobservations Sato et al. [1993] Ice core volcanic -r (X = 0.55 txm) averageof ice core acidityor sulfatemeasurements Robock and Free index (IVI) [1995, 1996]

Direct radiation measurements would be the best effects of the volcanic aerosols, and Lamb in addition technique, and combinationsof surface, aircraft, bal- used reports of atmosphericeffects. Still, up to the loon, and satellitemeasurements have clearlyquantified present,all the indicesmay misssome Southern Hemi- the distributionsand optical propertiesof the aerosols sphere (SH) eruptions,as they may not be reported. from the 1982 E1 Chich6n [Robock, 1983a] and 1991 Even in the 1980s, the December 1981 aerosolsfrom the Pinatubo[Stenchikov et al., 1998]eruptions (see also two eruption of Nyamuragirawere observedwith lidar but specialsections in GeophysicalResearch Letters: "Climat- were reported as the "mysterycloud" for severalyears ic Effectsof the Eruption of E1 Chich6n,"10(11), 989- until the source was identified by reexamining the 1060, 1983;and "The Stratosphericand Climatic Effects TOMS satelliterecord [Kruegeret al., 1996].As late as of the 1991 Mount Pinatubo Eruption: An Initial As- 1990, volcanicaerosols were observedwith Stratospheric sessment,"19(2), 149-218, 1992).The brightnessof the Aerosoland GasExperiment H (SAGE I1), but it hasnot Moon during lunar eclipsescan be used as an index of been possibleto identify the source[Yue et al., 1994]. stratosphericturbidity, but suchobservations can only be Before 1978,with no satelliteor lidar records,there may made about once per year, sometimesmissing the max- be importantmissing eruptions even in the NH averages. imum aerosol loading [Keen, 1983]. Even these most This problem does not exist for individual ice core recent large eruptions, however, have deficienciesin records,because they are objectivemeasures of volcanic their observations[Stenchikov et al., 1998]. In the past, sulfuric acid, except that the farther back in time one however, such measurementsare lacking, and indices goeswith ice cores, the fewer such recordsexist, and have had to use the available surface radiation measure- each ice core record is extremely noisy and may have ments combinedwith indirect measuressuch as reports otherproblems. Plate 6 showsthe five indices(Table 2) of red sunsetsin diaries and paintings,and geological for the NH for the past 150 years.Here they are briefly evidence.Geological methods, based on examinationof described. the depositsremaining on the ground from eruptions, canprovide useful information on the total masserupted 4.1. I.amb's Dust Veil Index and the date of the eruption,but estimatesof the atmo- Lamb [1970, 1977, 1983] createda volcanicdust veil sphericsulfur loading are not very accurate.This "pet- index (DVI), specificallydesigned for analyzingthe ef- rologic method" dependson the assumptionthat the fects of volcanoes on "surface weather, on lower and difference in sulfur concentrationbetween glassinclu- upper atmospherictemperatures, and on the large-scale sionsin the depositsnear the volcanoand the concen- wind circulation"[Lamb, 1970,p. 470].Lamb [1970] and trationsin the depositsthemselves are representativeof Pollack et al. [1976] suggestedwith data analysesthat the total atmosphericsulfur injection,but this has been volcanismwas an important causeof climate changefor shownnot to work well for recenteruptions for whichwe the past500 years,and Robock [1979] used Lamb's index have atmosphericdata [e.g.,Luhr et al., 1984]. to force an energy-balancemodel simulation of the For all the indicesthe problem of missingvolcanoes Little Ice Age, showingthat volcanicaerosols played a and their associateddust veils becomesincreasingly im- major part in producing the cooling during that time portant the farther backin time they go. There may even period. The methodsused to createthe DVI, described have been significantvolcanic aerosol loadings during by Lamb [1970] and Kellyand Sear [1982],include his- the past centurythat do not appearin any or most of the torical reports of eruptions,optical phenomena,radia- volcanicindices. Volcanoes only appear in most of the tion measurements(for the period 1883 onward),tem- indices if there is a report of the eruption from the perature information, and estimatesof the volume of ground.For recent eruptions,Lamb [1970] and Sato et ejecta. al. [1993] used actual measurementsof the radiative Lamb's DVI has been often criticized[e.g., Bradley, 38, 2 / REVIEWSOF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 201

1988] as havingused climatic information in its deriva- 1982], it had a negligiblestratospheric impact [Robock, tion, therebyresulting in circularreasoning if the DVI is 1981a]. used as an index to comparewith temperaturechanges. In fact, for only a few eruptionsbetween 1763 and 1882 4.4. Sato Index was the NH averaged DVI calculatedbased solely on Sato et al. [1993] produced monthly NH and SH temperatureinformation, but for severalin that period averageindices. Their index, expressedas optical depth the DVI was calculatedpartially on the basisof temper- at wavelength0.55 p•m,is basedon volcanologicalinfor- ature information. Robock [1981b] created a modified mation about the volume of ejectafrom Mitchell [1970] version of Lamb's DVI which excluded temperature from 1850 to 1882, on optical extinctiondata after 1882, information. When used to force a climate model, the and on satellite data startingin 1979. The seasonaland results did not differ significantlyfrom those using latitudinal distributionsfor the beginningof the record Lamb's original DVI, demonstratingthat this is not a are uniform and offer no advantagesover the DVI and seriousproblem. in fact showless detail than the latitudinallydependent 4.2. Mitchell Index indexof Robock[1981b], who distributedthe aerosolsin latitude with a simple diffusivemodel. The more recent Mitchell[1970] also produced a time seriesof volcanic part of the record would presumablybe more accurate eruptions for the period 1850-1968 using data from than the DVI or VEI, as it includes actual observations Lamb. As discussedby Robock[1978, 1981b]and Satoet of the latitudinal and temporal extent of the aerosol al. [1993], the Mitchell volcaniccompilation for the NH clouds. is more detailed than Lamb's, because Lamb excluded all volcanoeswith DVI < 100 in producing his NH annual averageDVI. Mitchell provided a table of the 4.5. Ice Core Volcanic Index order of magnitude of total mass ejected from each Robockand Free [1995] examinedeight NH and six volcano, which is a classificationsimilar to the volcanic SH ice core recordsof acidity or sulfate for the period explosivityindex. 1850 to the presentin an attempt to identify the volcanic signalcommon to all records.They explainin detail the 4.3. VolcanicExplosivity Index possibleproblems with using these records as volcanic A comprehensivesurvey of past volcanic eruptions indices,including other sourcesof acidsand bases,other [Simkinet al., 1981;Simkin and Siebert,1994] produced sourcesof sulfate, dating, local volcanoes,variable at- a tabulationof the volcanicexplosivity index (VEI) [Ne- mosphericcirculation, the stochasticnature of snowfall whalland Self,1982] for all knowneruptions, which gives and dry deposition,mixing due to blowing snow, and a geologicallybased measure of the power of the volca- uncertaintiesin the electrical conductivitymeasures of nic explosion.This index has been used without any the ice. For the NH, although the individual ice core modificationin many studies[see Robock, 1991] as an records are, in general, not well correlated with each index of the climatologicalimpact of volcanoes.A care- other or with any of the indices, a compositederived ful readingof Newhalland Self[1982],however, will find from averagingthe cores, the ice core volcano index the following quotes:"We have restrictedourselves to (IVI), showedpromise as a newindex of volcanicaerosol considerationof volcanologicaldata (no atmospheric loading.This new index correlatedwell with the existing data)..." (p. 1234) and "Sincethe abundanceof sulfate non-ice-corevolcanic indices and with high-frequency aerosolis importantin climateproblems, VEI's must be temperature records.Still, it is clear that high-latitude combinedwith a compositionalfactor before usein such volcanoesare given too much weight, and it is only studies"(pp. 1234-1235).In their Table 1, Newhalland possibleto adjustfor them if their signalscan unambig- Self list criteria for estimatingthe VEI in "decreasing uouslybe identified. For the SH the individual ice cores order of reliability," and the very last criterion out of 11 and indices were better correlated. The SH IVI was is "stratosphericinjection." For VEI of 3, stratospheric againhighly correlated with all indicesand individualice injectionis listed as "possible,"for 4 it is "definite,"and coresbut not with high-frequencytemperature records. for 5 and larger it is "significant."If one attempts to Robockand Free [1996] attemptedto extendthe IVI work backwardand use a geologicallydetermined VEI farther into the past. They compared all the ice cores to give a measure of stratosphericinjection, serious availablefor the past 2000 years with the DVI and the errors can result. Not only is stratosphericinjection the VEI for this period. An IVI constructedfor the period least reliable criterion for assigninga VEI, but it was 453 A.D. to the presentshowed little agreementwith the never intended as a descriptionof the eruption which DVI or VEI. They determinedthat exceptfor a very few had a VEI assignedfrom more reliable evidence.Nev- eruptions,the ice core record currently available is in- ertheless,Robock and Free [1995] found the VEI posi- sufficientto delineate the climatic forcing by explosive tively correlatedwith other indices,but imperfectly.For volcanic eruptionsbefore about 1200 for the NH and example,the Mount St. Helens eruption of 1980 has a before about 1850 for the SH. They also point out, large VEI of 5, and while it had a large local tempera- however, that the record of past volcanism remains ture impact [Robockand Mass, 1982;Mass and Robock, buried in the ice of Greenland and Antarctica, and more 202 ß Robock' VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWSOF GEOPHYSICS

TABLE 3. Effects of Large Explosive Volcanic Eruptions on Weather and Climate

Effect Mechanism Begins Duration

Reduction of diurnal cycle blockageof shortwaveand emissionof longwave immediately 1-4 days radiation Reduced tropical precipitation blockageof shortwaveradiation, reduced evaporation 1-3 months 3-6 months Summer cooling of NH tropics and blockageof shortwaveradiation 1-3 months 1-2 years subtropics Stratosphericwarming stratosphericabsorption of shortwaveand longwave 1-3 months 1-2 years radiation 1 Winter warming of NH continents stratosphericabsorption of shortwaveand longwave • year one or two radiation, dynamics winters Global cooling blockageof shortwaveradiation immediately 1-3 years Global coolingfrom multiple eruptions blockageof shortwaveradiation immediately 10-100 years Ozone depletion, enhancedUV dilution, heterogeneouschemistry on aerosols 1 day 1-2 years

deep coresthat analyzethe sulfur or acid content have other eruptions[Robock et al., 1995;Robock and Free, the promiseof producinga reliable record of the past. 1995;Self et al., 1997].Volcanic eruptionscan still have a large local effect on surfacetemperatures in regions near the eruption for severaldays, as Robockand Mass 5. WEATHER AND CLIMATE RESPONSE [1982]and Mass and Robock[1982] showed for the 1980 Mount St. Helens eruption. In this section I briefly Volcanic eruptionscan affect the climate systemon summarizethese climatic effects, starting from the short- manytimescales (Table 3). The greatestknown eruption est timescale. of the past 100,000years was the Toba eruptionof about 71,000years B.P. [Zielinskiet al., 1996],which occurred 5.1. Reductionof Diurnal Cycle intriguinglyclose to the beginningof a majorglaciation, The Mount St. Helens eruption in May 1980, in and while Rampino and Self [1992] suggesteda cause WashingtonState in northwesternUnited States,was a and effect relationship,it has yet to be established[Li very powerful lateral blast which produceda huge local and Berger, 1997]. Many papers, as discussedin the troposphericloading of volcanicash. In Yakima, Wash- introduction, have suggestedthat volcanic aerosolscan ington, 135 km to the east, it was so dark that automatic be importantcauses of temperaturechanges for several streetlightswent on in the middle of the day. This thick years following large eruptionsand that even on a 100- aerosol layer effectivelyradiatively isolated the Earth's year timescale,they can be important when their cumu- surfacefrom the top of the atmosphere.The surfaceair lative effects are taken into account. The effects of temperature in Yakima was 15øCfor 15 straighthours, volcanic eruptions on climate are very significant in independentof the normal diurnal cycle (Figure 3). analyzingthe globalwarming problem, as the impactsof Robockand Mass [1982] and Mass and Robock [1982] anthropogenicgreenhouse gases and aerosolson climate examined the errors of the model output statistics must be evaluated against a backgroundof continued (MOS) forecastsproduced by the National Weather natural forcingof the climatesystem from volcanicerup- Service. As the MOS forecasts did not include volcanic [ions,solar variations, and internalrandom variations aerosolsas predictors,they were able to interpret these from land-atmosphereand ocean-atmosphereinterac- errors as the volcaniceffect. They found that the aero- tions. solscooled the surfaceby as much as 8øCin the daytime Individual large eruptionscertainly produce global or but warmed the surfaceby as much as 8øC at night. hemisphericcooling for 2 or 3 years [Robockand Mao, The reduction of the diurnal cycleonly lasted a cou- 1995], and this signalis now clearer. The winter follow- ple of days,until the aerosolcloud dispersed.The effect ing a large tropical eruption is warmer over the NH wasalso observed after the Krakatau eruptionin Batavia continents, and this counterintuitive effect is due to (now know asJakarta), Indonesia [see Simkin and Fiske, nonlinearresponse through atmospheric dynamics [Rob- 1983, Figure 58]. While the Mount St. Helens eruption ock and Mao, 1992; Graf et al., 1993;Mao and Robock, had a large local effecton temperature,no other impact 1998;Kirchner et al., 1999]. Volcanic aerosolsprovide a was identified on precipitationor atmosphericcircula- surface for heterogeneouschemical reactions that de- tion. Its stratosphericinput of sulfurwas very small,and stroy ozone, and observationsfollowing Pinatubo have hencethis very explosiveeruption had a minimal impact documentedmidlatitude ozone depletion causedby a on global climate [Robock,1981a]. volcaniceruption [Solomonet al., 1996;Solomon, 1999]. While the large 1982-1983 E1 Nifio amplifiedjust after 5.2. SummerCooling the 1982 E1 Chich6n eruption in Mexico, there is no It has long been known that the global averagetem- evidenceof a causeand effectrelationship for this or any perature falls after a large explosivevolcanic eruption 38, 2 / REVIEWS OF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 203

c _.1._. I

i iii II I

o -T-

,,i • o

o

ix) o ..-

..,.... 204 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

M•Y 1980 and there have been so few large eruptionsin the past 17 18 19 20 century.For severalrecent eruptions,Angell [1988],Ni- Y A • cholls [1988], and Mass and Portman [1989] demon- K I I0 stratedthat the ENSO signalin the past climaticrecord M A partially obscuresthe detectionof the volcanicsignal on S gO P a hemisphericannual averagebasis for surfaceair tem- 0 • •o perature. N More recently,Robock and Mao [1995] removedthe GF • ENSO signal from the surfacetemperature record to RA œL extractmore clearlythe seasonaland spatialpatterns of A k •0 TS the volcanicsignal in surfacetemperature records. This

0 observedsignal matches the general cooling patterns I S • found from GCM simulations[Hansen et al., 1988;Rob- E •0 ock and Liu, 1994]. Robock and Mao found, by super-

12 0 12 0 12 0 12 posing the signalsof Krakatau, Santa Maria, Katmai, LST Agung, E1 Chich6n, and Pinatubo, that the maximum cooling is found approximately 1 year following the Figure 3. Time seriesof surfaceair temperaturefor Yakima eruptions.This cooling,of 0.1ø-0.2øC,follows the solar and Spokane,Washington; Great Falls, Montana; and Boise, declinationbut is displacedtoward the NH (Figure 4); Idaho, for May 17-20, 1980, under the plume of the 1980 Mount St. Helens eruption,from Figure 3 of Robockand Mass the maximumcooling in NH winter is at about 10øNand [1982].Time of arrivalof the plume is indicatedwith an arrow. in summeris at about 40øN. This pattern is becauseof LST is local standard time. The plume never passedover the distribution of continents:Land surfacesrespond Boise,which is includedas a control. Note the dampingof the more quicklyto radiationperturbations and thusthe NH diurnal cycleafter the arrival of the troposphericaerosol cloud. is more sensitive to the radiation reduction from volca- nic aerosols. Robock and Liu [1994] analyzedthe Hansen et al. [e.g.,Humphreys, 1913, 1940;Mitchell, 1961].The direct [1988] GCM simulationsand found reduced tropical radiativeforcing of the surface,with a reductionof total precipitation for 1-2 years following large eruptions. downward radiation, cools the surface. For example, The global hydrologicalcycle is fueled by evaporation, Hansenet al. [1978],using a radiative-convectiveclimate and the coolingafter the eruptionsproduced this effect. model, successfullymodeled the surface cooling and They evenfound a reductionin Sahelprecipitation that stratosphericwarming after the 1963Agung eruption.In matched the observed enhancement of the Sahel the tropics and in the midlatitude summerthese radia- drought following the E1 Chich6n eruption, but this tive effectsare larger than most other climaticforcings, result needs further confirmation before it can be con- as there is more sunlightto block. In somelocations in sidered robust. the tropics,however, even the effectsof a large eruption like E1 Chich6n can be overwhelmedby a large E1 Nifio, 5.3. StratosphericHeating as was the case in 1983. The radiative-convective model It haslong been knownthat the stratosphereis heated studyby Vupputuriand Blanchet[1984] also simulated after injection of volcanic aerosols[e.g., Quiroz, 1983; coolingat the surfaceand warming in the stratosphere. Parker and Brownscombe,1983; Angell, 1997b]. As ex- Energy-balancemodels [Schneiderand Mass, 1975; Ol- plained above, this heating is causedby absorptionof iver, 1976;Bryson and Dittberner,1976; Miles and GiM- both near-IR solar radiation at the top of the layer and ersleeves,1978; Robock, 1978, 1979; Gilliland, 1982; Gilli- terrestrial radiation at the bottom of the layer. Figure 5 land and Schneider,1984] have all showncooling effects and Plate 7 showsobservations of lower stratospheric for up to severalyears after major eruptions.An early temperaturefor the past20 years.Two strongsignals can zonally averageddynamic climate model [MacCracken be seen.After the 1982E1 Chich6neruption the globally and Luther, 1984] and GCM studiesby Hunt [1977], averaged stratospherictemperature rose by about IøC Hansenet al. [1988, 1992, 1996],Rind et al. [1992], and for about 2 years. After the 1991 Pinatubo eruption a Pollacket al. [1993] alsoshowed volcanic cooling effects. warmingof equallength but abouttwice the amplitudeis The problem of detectionand attributionof a volca- clearlyvisible. These large warmingsare superimposed nic signal in the past is the same as the problem of on a downward trend in stratospherictemperature identifyinga greenhousesignal in the past:identifying a caused by ozone depletion and increased CO2 [Ra- unique fingerprintof the forcingand separatingout the maswamyet al., 1996;Vinnikov et al., 1996].This cooling, effects of other potential forcings. It is necessaryto at 10 timesthe rate of troposphericwarming for the past separatethe volcanicsignal from that of other simulta- century,is a clear signalof anthropogenicimpacts on the neous climaticvariations because the climatic signalof climate system. volcaniceruptions is of approximatelythe same ampli- Plate 7 showsthe stratospherictemperatures sepa- tude as that of E1 Nifio-Southern Oscillation(ENSO) rated by latitude bands.After the 1982 E1 Chich6n and 38 2 / REVIEWS OF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 205

60N

ß-0 4 \ ,',",.... 0 2 .-, ' "'V' ' ' '•' ": ..... " ' : ' ø ' ;"" ; ' 50N :,,.,', :-o.2.,''" ' IO,.2/ 40Nlo:, i '"½.,' 4½, • o-' i / ¾,/i' !..,•',.i ",ii ø:':•/ i } :'.o.:•. SONj : : : : : ß : : . : •ON,•.

20S-!.-o., "' : :";...... ' : : "': :• : : 0 ...... •"'' ß",,,',, .' D :o•: ""ø?!•-""': : : ::"'/ - ,'.";"::•: ;': ;: ::::: -',/'/::'œ:2-', : 6•2 -• -\;'.-03,;: •)•",::; ' -0 '• •:--:'-',;•'ß ::;f ::/'::. '.. ,,...,:)?:,'.:,-.•'•....ß -'= ...... ::,,...... ,v,.',...... -œ0.2 :::,3,'--- ß ':-0,...... 1 •',•'., "•, ;"," ,'...... ,.:---o-,,.,.-, .... ;,- - ...... --. '-'e,-,.. -, 50S- .,:' ;,• ,'.'•' .. ,,.,.,...... , -.... -5 -4 -,.3 -2 -1 0 1 2 3 4 5 Log (years)

Figure 4. Zonal averagesurface air temperatureanomalies, averaged for the six largestvolcanic eruptions of the past century.Anomalies (øC) are with respectto the mean of the 5-year period before eacheruption, with the seasonalcycle removed. Values significantlydifferent (95% level) from 0øC are shaded.Contour interval is 0.1øC,with the 0øC interval left out and negativecontours dashed. From Figure 6 of Robockand Mao [1995]. '

1991Pinatubo eruptions the tropicalbands (30øS-30øN), 5.4.1. Observations. It was first suggestedby shownby the greenand black curves,warmed more than Groisman [1985], with reference to previous Russian the 30øN-90øNband (blue curve), producingan en- studies, that warm winters over central Russia were a hanced pole-to-equatortemperature gradient. The re- consequenceof large volcaniceruptions. Using surface ß sultingstronger polar vortex producesthe tropospheric air temperaturedata for stationsin Europe (including winter warming describednext. the European part of Russia)and northeasternNorth America for averagesof two or three winters after the 5.4. Winter Warming volcanic years of 1815, 1822, 1831, 1835, 1872, 1883, Robock [1981b, 1984b] used an energy-balancecli- 1902, 1912, and 1963, he showeda significantwarming mate model to examine the seasonal and latitudinal over the central European part of Russiaand insignifi- responseof the climate systemto the Mount St. Helens cant changesin the other regions,including cooling over and E1 Chich6n eruptionsand found that the maximum northeasternNorth America. In an update, Groisman surfacecooling was in the winter in the polar regionsof [1992] examinedthe winter patterns after the 1982 E1 both hemispheres.This was due to the positivefeedback Chich6n and 1991 Pinatuboeruptions and found warm- of sea ice, which lowered the thermal inertia and en- ing over Russia again but found warming over north- hancedthe seasonalcycle of coolingand warmingat the eastern North America. poles[Robock, 1983b]. Energy-balance models, however, By averagingover severalwinters and only examining are zonally averagedand parameterizeatmospheric dy- part of the hemisphere,Groisman [1992] did not dis- namicsin terms of temperaturegradients. Thus they do cover the completewinter warmingpattern but did in- not allow nonlinear dynamical responseswith zonal spire other studies.His explanationof the warming over structure. While the sea ice/thermal inertia feedback is Russia as due to enhancedzonal winds bringing warm indeed part of the behavior of the climate system,we maritime air from the north Atlantic over the continent now know that an atmosphericdynamical response to was correct,but he explainedthese enhancedwinds as large volcanic eruptions dominatesthe NH winter cli- due to a larger pole-to-equatortemperature gradient mate response,producing tropospheric warming rather causedby polar cooling resulting from the eruptions. than an enhancedcooling over NH continents. However, if the larger pole-to-equatortemperature gra- 206 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38 2 / REVIEWS OF GEOPHYSICS

Global Avg. Lower StratosphericTemp. Anomalies with respect to 1984-1990 mean

1.4

1.2

1

• 0.8

-• o.6

o 0.4

• 0.2

(D 0

I-- -0.2

-0.4-

-0.6-

-0.8 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Figure 5. Globalaverage monthly stratospheric temperatures from microwavesounding unit satelliteobserva- tions,channel 4 [Spenceret al., 1990] (updatedin 1999).Anomalies (øC) are with respectto the 1984-1990 nonvolcanicperiod. Times of 1982E1 Chich6nand 1991Pinatubo eruptions are denotedwith arrows.Years on abscissaindicate January of that year.

Lower Stratospheric Temp. Anomalies - Latitude Bands with respect to 1984-1990 meon

1.5

.•- 0.5

o • 0--

-2 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 38, 2 / REVIEWS OF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 207 dient wascaused by polar cooling,why were the temper- sponsein the climate systemto the radiative forcing atures in the polar region over Eurasia warm and not from the eruptions.The direct radiativeresponse in the cool? Why was northeasternNorth America cool after winter would be a smallcooling, as it is winter becauseof some eruptionsand warm after others? the low insolation.At the pole the surface radiative Lough and Fritts [1987], usingtree-ring data and se- forcingis actuallywarming, as there is a smallincrease in lecting 24 volcanicyears between 1602 and 1900, found downwardlongwave radiation. winter warming over western North America for the 5.4.2. lheory. The winter warmingpatterns de- averageof years 0-2 after the eruptionyears. The pat- scribed above are closelyrelated to troposphericand tern they found is quite similar to that during ENSO stratosphericcirculation. Both Perlwitzand Graf [1995] years, and they made no attempt to correct for this and Kodera et al. [1996] examinedthe observationsof factor. They also found spring and summer cooling in NH winter stratosphericand troposphericcirculation for the central United Statesand summerwarming over the the past 40 years and found that the dominant mode of United Stateswest coast.They did not use anyvolcanoes circulationof the stratosphereis a strong polar vortex south of 10øS,believing them not important for NH (polar night jet), which occurssimultaneously with a temperature variations. 500-mbar pattern with a low anomaly over Greenland Robock and Mao [1992] were the first to systemati- and high anomaliesover North America, Europe, and cally examinethe global surfaceair temperaturerecord east Asia. The associatedsurface air temperature pat- for the NH winter warming phenomenon.They exam- tern is exactlythat shown in Plate 8. Perlwitz and Graf ined the winter surface air temperature patterns after call this pattern the "baroclinicmode." The same pat- the 12 largesteruptions of the pastcentury, removed the tern had previouslybeen identified as the North Atlantic ENSO signal,and found a consistentpattern of warming Oscillation(NAO) [Hurrell,1995, and referencesthere- over the continentsand coolingover the oceansand the in] and is now also called the Arctic Oscillation(AO) Middle East,when combiningthe firstwinter after trop- [Thompsonand Wallace,1998, 2000a, b]. ical eruptionsand the secondwinter after high-latitude Both studies also found a second mode with weak eruptions.Robock and Mao [1995] showed that this stratosphericanomalies but a strongmidtropospheric pattern is mainly due to the tropicaleruptions. pattern of wave anomalies generated in the tropical The winter warming pattern is illustrated in Plate 8, Pacific and propagating across North America. This which showsthe global lower tropospherictemperature secondbarotropic mode [Perlwitzand Graf, 1995] has anomaly pattern for the NH winter of 1991-1992, fol- previouslybeen identified as the PacificNorth America lowingthe 1991 Mount Pinatuboeruption. This pattern teleconnectionpattern associatedwith ENSO [Wallace is closelycorrelated with the surface air temperature and Gutzler,1981]. It is confinedto the Western Hemi- pattern where the data overlap, but the satellite data sphere and only influences extratropical regions over allow global coverage. The temperature over North North America. Kitoh et al. [1996] found the same pat- America, Europe, and Siberia was much warmer than terns in a GCM experiment,with the baroclinicmode normal, and that over Alaska, Greenland, the Middle dominating and no relationship between the ENSO East, and China was cold. In fact, it was so cold that mode and stratosphericcirculation. winter that it snowedin Jerusalem,a very unusual oc- The picture that emergesis one of a characteristic currence. Coral at the bottom of the Red Sea died that baroclinicmode, or NAO or AO circulationpattern, in winter [Genin et al., 1995], becausethe water at the the troposphere(Figure 6) that producesanomalously surfacecooled and convectivelymixed the entire depth warm surfaceair temperaturesover the continentsin the of the water.The enhancedsupply of nutrientsproduced NH winter. It is a natural mode of oscillation of the anomalouslylarge algal and phytoplankton blooms, winter atmosphericcirculation, so external stratospheric which smotheredthe coral. This coral death had only forcing apparently can push the systeminto this mode happened before in winters following large volcanic without too much trouble. As Perlwitzand Graf [1995] eruptions[Genin et al., 1995]. explain in detail, the theoretical explanationis that the While the tropical regionscool in all seasons[Robock strongpolar vortextraps the verticallypropagating plan- and Mao, 1995] (Figure 4), in the winter following a etary waves,which constructivelyinterfere to produce large eruption,the zonal averagetemperature change is this stationary wave pattern. When there is a strong small,but the wave pattern of anomaliesproduces large polar night jet, it preventssudden stratospheric warm- warm and cool anomalies, indicating a dynamical re- ings[Mcintyre, 1982] later in the winter and perpetuates

Plate 7. (opposite)Stratospheric temperature anomalies for four equal-areazonal bands, shown as 5-month runningmeans, from microwavesounding unit satelliteobservations, channel 4 [Spenceret al., 1990](updated in 1999).Anomalies (øC) are with respectto the 1984-1990nonvolcanic period. Times of 1982E1Chich6n and 1991Pinatubo eruptions are denotedwith arrows.Note that after theseeruptions the tropicalbands warmed more than the 30ø-90øNband, producingan enhancedpole-to-equator temperature gradient. Years on abscissaindicate January of that year. 208 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

itself throughout the winter, keeping the polar lower stratospherecold in the isolatedvortex center.When the pattern occursnaturally without externalforcing, it more frequently breaks down due to these sudden strato- sphericwarmings. The winter warmingcirculation response occurs after all large tropical eruptions that have occurred since radiosondeobservations began, Agung in 1963, E1 Chi- ch6n in 1982, and Pinatubo in 1991 [Kodera,1994]. It shouldbe noted, however, that other forcingscan also work in the same direction. One is the l 1-year solar cycle, as the l 1-year cycle in ultraviolet flux and ozone amount combine to produce anomalous stratospheric heating and circulation [Kodera et al., 1991; Robock, 1996a;Haigh, 1996;Shindell et al., 1999].These circula- tion changeshelp to explain the solar cycle-climate relations discoveredby Labitzke and van Loon [1988] and vanLoon and Labitzke[1990], but their patternsare still not completelyunderstood. Another forcing is sim- ply globalwarming, which increasesthe thicknessof the tropical tropospheremore than at higher latitudes,also enhancing the pole-to-equator gradient [Graf et al., 1995]. Ozone depletion in the lower stratospherehas producedmore coolingat the poles [Ramaswamyet al., b 1996], also enhancingthe pole-to-equatorgradient and producinga trend in the strengthof the polar vortex [Graf et al., 1995]. These global warming and ozone depletiontrends help explain the observedtrend in the NAO [Hurrell,1995; Hurrell and vanLoon, 1997]and the AO [Thompsonand Wallace,1998, 2000b] of the past 30 several decades and in the related cold ocean-warm land .% pattern of Wallaceet al. [1995, 1996].Nevertheless, after a large tropicalvolcanic eruption, the forcing is so large that it overwhelms these more subtle trends and domi- nates the NH winter circulation of the succeedingyear. The connectionbetween stratospheric circulation and surface air temperature anomalies in the NH winter after a large tropical volcanic eruption is illustrated schematicallyin Plate 9. The heating of the tropical lower stratosphereby absorptionof terrestrialand solar near-IR radiation(Plate 5) expandsthat layer and pro- ducesand enhancedpole-to-equator temperature differ- ence. The strengthened polar vortex traps the wave energyof the troposphericcirculation, and the station- ary wave pattern known as the NAO or AO dominates Figure 6. (a) Observed 500-mbar geopotential height the winter circulation,producing the winter warming. (meters)anomaly pattern for the 1991-1992NH winter (DJF) 5.4.3. Modeling. To testthe abovetheory, Graf et followingthe 1991 Mount Pinatuboeruption. Data are from al. [1993] presentedresults from a perpetual January the National Centersfor EnvironmentalPrediction reanalysis [Kalnayet al., 1996], and anomaliesare with respect to the GCM calculationof the effectsof stratosphericaerosols period1986-1990. (b) Generalcirculation model (GCM) sim- on climate that showedwinter warmingover both north- ulationsfrom Kirchneret al. [1999]. Shownis the difference ern Eurasia and Canada and cooling over the Middle between the two GCM ensembles with and without strato- East, northern Brazil, and the United States.They used sphericaerosols. In both panels,regions significantly different forcing with a stratosphericaerosol loading in the pat- from 0 at the 80% level are shaded,with positivevalues shaded tern of the E1 Chich6n volcano and the low-resolution lighter and negativevalues shaded darker. The simulations (T21) ECHAM2 GCM, and the circulationresponse in reproducedthe observedpattern of circulationanomalies, but Januaryfollowing volcanic eruptions was well simulated not perfectly. [Kirchnerand Graf, 1995].Kirchner et al. [1999]repeated this calculationin more detail with an improvedmodel, 38, 2 / REVIEWSOF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 209 in an experimentwith better definedaerosol parameters stratosphericheating during the winter of 1783-1984 [Stenchikovet al., 1998], interactivecalculation of the and a reduced pole-to-equator temperature gradient. aerosol radiative effects, and an improved GCM This theory suggeststhat the opposite phase of the (ECHAM4) [Roeckneret al., 1996], includingthe full pattern discussedabove was produced, with a large seasonalcycle. They were successfulin reproducingthe negativeNAO anomaly.This suggestionis now being observedcirculation (Figure 6) and surfacetemperature testedwith GCM experiments. patterns (Figure 7) after the 1991 Pinatubo eruption when usingclimatological sea surfacetemperatures. Al- 5.5. tittle Ice Age thoughthe simulatedpatterns did not exactlymatch the Sincevolcanic aerosols normally remain in the strato- observations,the positiveheight anomaliesover North sphere no more than 2 or 3 years, with the possible Americaand Europe and negativeanomalies over Baffin exceptionof extremelylarge eruptionssuch as that of Island and the Middle East (Figure 6) are in the right Toba, approximately71,000 years ago [Bekki et al., location,which correspondto warm and cold anomalies 1996], the radiative forcing from volcanoesis interan- overthe samelocations (Figure 7). The simulationswere nual rather than interdecadal in scale. A series of volca- not as successfulwhen using the actual observedsea nic eruptions could, however, raise the mean optical surface temperatures,which included a moderate E1 depth significantlyover a longerperiod and therebygive Nifio. Thus GCMs have not yet demonstratedan ability rise to a decadal-scalecooling. If a period of active to successfullysimulate the complete responseof the volcanismends for a significantinterval, the adjustment climate systemto surfaceand stratosphericforcing. This of the climate systemto no volcanicforcing could pro- is an area of active current research,including the new duce warming. This was the casefor the 50 years from Pinatubo Simulation Task of the GCM-Reality Inter- 1912 to 1963, when global climate warmed. Further- comparisonProject for the Stratosphere(GRIPS) [Paw- more, it is possiblethat feedbacksinvolving ice and son et al., 2000]. ocean,which act on longer timescales,could transform Mao and Robock [1998] conductedanother model the short-termvolcanic forcing into a longer-termeffect. test of the winter warming pattern. Patternssimilar to As a result, the possiblerole of volcanoesin decadal- that in Plate 8 appeared over North America in the scaleclimate changeremains unclear. In particular, the winters following the 1982 E1 Chich6n and 1991 Pina- current centuryis the warmestof the past five, with the tubo eruptions,which were both during E1 Nifios. This previousfour centuriesearning the moniker of the Little led someto suggestthat the temperaturepatterns were Ice Age due to its coldness.This period has never been producedby the PacificNorth American teleconnection satisfactorilyexplained. What was the role of volcanism pattern connectedwith the E1 Nifio. As both forcings in this climate change? were actingat the sametime, observationsalone cannot Until recently,Schneider and Mass [1975] andRobock distinguishthe causeof the pattern. Mao and Robock [1979]were the only modelsto usevolcanic chronologies took advantageof the designof the AtmosphericModel to investigateperiods before the midnineteenthcentury. IntercomparisonProject (AMIP) [Gates, 1992], which At the same time, they examined the hypothesisthat was conductedfor the period 1979-1988. Thirty differ- solarconstant variations linked to the sunspotcycle were ent GCMs simulatedthe weather of this 10-yearperiod, also an important causeof climate changeon this time- forced only with observed sea surface temperatures scale. Schneider and Mass used a zero-dimensional en- (SSTs), which included two E1 Nifios, 1982-1983 and ergy-balancemodel and found agreement of several 1986-1987. No volcanicaerosol forcing was used in the large-scalefeatures of their simulationswith climate simulations, so the results of the simulation should re- records. They concluded that while volcanoes had a flect ENSO patternsonly. Mao and Robock found that weak relationshipto climate,the solar-climateeffect was the North American patterns simulatedby the models not proven.Robock used a latitudinallyresolved energy- were very similar for both E1 Nifio winters. They were balance model with volcanic [Mitchell, 1970] and solar very close to the observedpattern for 1986-1987, a forcing (proportional to the envelope of the sunspot winter with only E1 Nifio forcing and no volcanicaero- number) and found the volcanic forcing explained a sols,but did not resemblethe observedpattern of 1982- much larger share of the temperaturevariability since 1983 (Plate 8) at all. Therefore they concludedthat the 1620 than did the solar series. Both the Schneider and major warming over the North American and Eurasian Mass and Robock models used a simple mixed-layer continents in the 1982-1983 NH winter is not an ENSO- ocean. dominantmode, but rather a pattern associatedwith the Three new studies[Crowley and Kim, 1999;Free and enhancedstratospheric polar vortex by the larger equa- Robock, 1999;DMrrigo et al., 1999] have now made use tor-to-pole temperaturegradient produced by volcanic of new volcanochronologies, new solar constantrecon- sulfate aerosolsin the stratosphere. structions,and new reconstructionsof climate change Franklin [1784] noted that the winter in Europe fol- for the Little Ice Age to addressthis problem again. lowing the 1783 Lakagigareruption was extremelycold They used upwelling-diffusionenergy-balance climate rather than warm. BecauseLakagigar was a high-lati- models to simulate the past 600 years with volcanic, tude eruption, it could have produced high-latitude solar, and anthropogenicforcings and compared the 210 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

75N

60N

45N

30N

15N

EQ 180

Figure 7. Temperaturepatterns for NH winter (DJF) 1991-1992 followingthe Mount Pinatuboeruption. (a) Observedanomalies (with respectto the averagefor 1986-1990) from channel2LT of the microwave soundingunit satellite observations(same as in Plate 8), representativeof the lower troposphere.(b) Anomaliessimulated by a GCM forcedwith observedPinatubo aerosols [Kirchner et al., 1999] expressedas GCM simulatedchannel 2LT temperatures.Shown are differencesbetween the two GCM ensembleswith and without stratosphericaerosols. Contour interval is IøC, and shadingis as in Figure 6. The locationsof anomalouswarm and cold regionsare quite closeto thoseobserved.

resultswith paleoclimaticreconstructions, mainly based Ropelewski,1992]. Owing to the coincidenceof the be- on tree rings.They conclude,subject to the limitationsof ginningof the ENSO eventand the E1 Chich6neruption, the forcing and validationdata sets,that volcanicerup- severalsuggestions were made as to a causeand effect tions and solarvariations were both important causesof relationship,even going so far as to suggestthat mostE! climate changein the Little Ice Age. Furthermore, the Nifios were causedby volcaniceruptions. warming of the past century cannot be explained by How could a volcaniceruption produce an E1 Nifio? natural causesand can be explainedby warming from Would the mechanisminvolve the stratosphericaerosol anthropogenic greenhouse gases. Further work is cloud, troposphericaerosols, or the dynamicalresponse needed in this area, however,linking ocean-atmosphere to surfacetemperature changes?What is the evidence interactionswith better volcanicand solar chronologies. based on the past record of volcanic eruptions and. ENSO events? Was the simultaneous occurrence of the E1 Chich6n eruption and large E1 Nifio just a coinci- 6. DO VOLCANIC ERUPTIONS PRODUCE dence,or was it an exampleof a mechanismthat works EL NINOS? after many other large eruptions? Hirono [1988] proposed a plausible mechanism The April 1982 E1 Chich6nvolcanic eruption was the whereby troposphericaerosols from the E1 Chich6n largestof the centuryup to that time. Beginningin April eruption would induce an atmosphericdynamical re- 1982, as indicatedby the SouthernOscillation Index, an sponseproducing a trade wind collapse,and the trade unpredictedand unprecedentedE1 Nifio began, result- wind reduction would produce an oceanographicre- ing in the largestwarm ENSO eventof the centuryup to sponsewhich affected the timing and strength of the that time, with recordwarm temperaturesin the eastern resultingE1 Nifio. He suggestedthat large tropospheric equatorial Pacific Ocean and remote temperature and aerosolsfalling out of the E1 Chich6n eruption cloud precipitationanomalies in distantlocations [Halpert and into the regionwest of Mexico in the eastPacific in the 38, 2 / REVIEWSOF GEOPHYSICS Robock:VOLCANIC ERUPTIONSAND CLIMATEß 211

z z z o (/• (/) o o o ta.I o o 212 ß Robock' VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

======.. .•.•.. Aerosol .::i:!:!:i:i:i:i:!:i:i:i:i:!:i:i:i:!:i:i:i:!:i:!:i:i:!:•:i:i:i:!:i:!- '.>:.:.:.:.:.>:.:.:..'.:.:.:.:.:-:.:->:.:-:-:-:.:-:.:q•-:-:-:-:-:-' 50mb before heating ...... >¾:¾•¾..-..:..:..-.>:.-.---' ß . -. -. -' -' -' Tropopause Jet stream ams

i i Aerosolheating increasesEquator-Pole temperaturegradient, increasing heightgradient, and makingjet stream (polar vortex) stronger.

II North Pole Tropics

Strongervortex amplifies "North Atlantic Oscillation" circulation pattern,producing winter warming. III I II

Winter1991-92 (DJF) Surface Air TemperatureAnomalies (øC) •! - .

Plate 9. Schcmaticdiagram of howtropical lower stratospheric heating from volcanicaerosols produces the winter warming temperaturepattern at the surface.The map at the bottom is surfaceair temperature anomalies (with respect to 1961-1990) for the 1991-1992 NH winter (DJF) following the 1991 Mount Pinatuboeruption. The green curvesindicate the anomaloustropospheric wind patternsresponsible for horizontaladvection that producedthese temperature anomalies. This temperaturepattern is very similarto the one shownin Plate 8 but is for the surface.Data courtesyof P. Jones. 38, 2 / REVIEWSOF GEOPHYSICS Robock:VOLCANIC ERUPTIONS AND CLIMATE ß 213 weeks after the eruption would absorb terrestrial and Robock and Free [1995] calculated the correlation solar radiation, heating the atmosphereand inducinga betweenthe SouthernOscillation Index [Ropelewskiand low-pressureregion, changingthe troposphericcircula- Jones,1987], the best availableindex of past E1 Nifios, tion. The wind spiralinginto the low would reducethe and everyvolcanic index describedabove, as well as each northeasttrade winds, producingan oceanographicre- individual ice core record that was available, and in no sponsethat wouldcause an E1Nifio. Robocket al. [1995] casesfound significantcorrelations. Self et al. [1997] used the LawrenceLivermore National Laboratoryver- recently examinedevery large eruption of the past 150 sion of the Community Climate Model, Version 1 years and showedthat in no caseswas there an E1 Nifio (CCM1) GCM to investigatethis mechanismwith radi- that resultedas a consequence.Therefore, as concluded ative forcing from the observed aerosol distribution. by Robocket al. [1995], the timing and locationof the E1 Indeed, there was a trade wind collapseinduced in the Chich6n eruption and the large ENSO event that fol- atmosphere,but it wastoo late and in the wrongposition lowed were coincidental, and there is no evidence that to have producedthe observedE1 Nifio. large volcaniceruptions can produce E1 Nifios. It is interestingto note that when this study was beginning,E. Rasmusson,one of the world'sexperts on E1 Nifios and whose office was next to mine at the 7. EFFECTS ON STRATOSPHERIC OZONE Universityof Maryland, the week before the 1991 Pina- tubo eruption excitedlywalked down the hall announc- Volcanic aerosolshave the potential to change not ing that a new E1 Nifio was beginning.Thus althoughin only the radiative flux in the stratosphere,but also its 1991 there was again a large volcaniceruption at the chemistry.The most important chemicalchanges in the sametime as a large E1 Nifio, the E1 Nifio occurredfirst. stratosphereare related to O3, which has significant In fact, Rampino et al. [1979] once suggestedthat cli- effectson ultraviolet and longwaveradiative fluxes.The matic changecould cause volcanic eruptions by changing reactionswhich produce and destroyO3 depend on the the stress on the Earth's crust as the snow and ice UV flux, the temperature,and the presenceof surfaces loading shifts in responseto the climate. More recent for heterogeneousreactions, all of which are changedby evidenceof relativelylarge changesin the Earth's rota- volcanicaerosols [Crutzen, 1976; Tabazadehand Turco, tion rate after E1 Nifios, in responseto the enhanced 1993; Tie and Brasseur,1995; Tie et al., 1996; Solomon et atmosphericcirculation [Marcus et al., 1998], couldpro- al., 1996]. The heterogeneouschemistry responsible for duce an additional lithosphericstress. These specula- the ozone hole over Antarctica in October each year tions have not been proven,however, and further study occurson polar stratosphericclouds of water or nitric is limited by availablelong-term records. acid, which only occur in the extremely cold isolated Schattenet al. [1984] and Strong[1986] suggestedthat springvortex in the SouthernHemisphere. These reac- the atmosphericdynamic response to the stratospheric tions make anthropogenicchlorine availablefor chemi- heating from the volcanic aerosolswould perturb the cal destructionof O3. Sulfate aerosolsproduced by vol- Hadley cell circulationin the tropicsin sucha way as to caniceruptions can also provide these surfacesat lower reducethe trade winds and trigger an E1 Nifio. Schatten latitudes and at all times of the year. et al.'s model, however,was zonally symmetricand did Solomon [1999] describesthe effects of aerosolson not explain why the responsewould be so rapid and ozone in great detail. Here only a brief mention of some mainly in the Pacific. of the issues related to volcanic eruptions is made. Handler[1986] suggested that the climaticresponse to Quantifying the effects of volcanic aerosolson ozone coolingover continentsproduces a monsoon-likecircu- amount is difficult, as chemical and dynamical effects lation that somehowproduces E1 Nifios, but he did not occursimultaneously and the effectsare not muchlarger clearly explainthe physicalmechanism or demonstrate than natural variability. Nevertheless, attempts were that a model can producethis effect.His claim to find a made after the 1991 Pinatubo eruption to estimatethe statisticallink between occurrenceof volcaniceruptions effectson ozone.Column O3 reductionof about 5% was and E1 Nifios is flawed in severalways. As there have observedin midlatitudes [Zerefoset al., 1994; Coffey, been many volcaniceruptions in the past few centuries 1996],ranging from about2% in the tropicsto about7% and E1 Nifios occurevery 4-7 years,it is possibleto find in the midlatitudes [Angell, 1997a]. Therefore ozone an eruption closein time to each E1 Nifio. However, to depletion in the aerosol cloud was much larger and establisha cause and effect relationship,the eruption reachedabout 20% [Grantet al., 1992;Grant, 1996].The must be of the proper type and at the proper time to chemicalozone destructionis lesseffective in the trop- producethe claimedresponse. Handler's studies did not ics,but lifting of low ozoneconcentration layers with the use a proper index of stratosphericsulfate loading. He aerosolcloud [Kinneet al., 1992] causesa fast decrease selectedvery small eruptionsthat could not have had a in ozone mixing ratio in the low latitudes. Similarly, climaticeffect, and he was not carefulwith the timing of subsidenceat high latitudesincreases ozone concentra- the eruptionshe chose,claiming eruptions that occurred tion there and masks chemical destruction. after particularENSO eventsor long before them were Decrease of the ozone concentration causes less UV their causes. absorptionin the stratosphere,which modifiesthe aero- 214 ß Robock:VOLCANIC ERUPTIONSAND CLIMATE 38, 2 / REVIEWSOF GEOPHYSICS

sol heating effect [Kinne et al., 1992; RosenfieMet al., matic responserepresent a large perturbation to the 1997]. The net effect of volcanicaerosols is to increase climate systemover a relatively short period, observa- surfaceUV [Vogelmannet al., 1992].The subsequent03 tions and the simulatedmodel responsescan serve as depletionallows through more UV than is backscattered important analogs for understandingthe climatic re- by the aerosols. sponseto other perturbations.While the climatic re- Kirchneret al. [1999] calculateda January1992 global sponseto explosivevolcanic eruptions is a usefulanalog average70-mbar heatingof about 2.7 K from the aerosol for some other climatic forcings,there are also limita- heating following the 1991 Pinatubo eruption as com- tions. paredwith the observedheating of only 1.0 K. The GCM The theory of "nuclearwinter," the climatic effectsof calculationsdid not include considerationof the quasi- a massiveinjection of soot aerosolsinto the atmosphere biennialoscillation (QBO) or 03 effects.They then con- from fires followinga globalnuclear holocaust [Turco et ducted another calculation to see the effects of the al., 1983, 1990;Robock, 1984c, 1996b], includes upward observed Pinatubo-induced 0 3 depletion on strato- injectionof the aerosolsto the stratosphere,rapid global spheric heating and found that it would have cut the dispersalof stratosphericaerosols, heating of the strato- calculatedheating by 1 K. If the GCM had considered sphere, and cooling at the surface under this cloud. the observedQBO coolingand this 03 effect, the calcu- Becausethis theory cannot be tested in the real world, lated heatingwould agree quite well with observations. volcaniceruptions provide analogsthat support these The volcaniceffect on 03 chemistryis a new phenom- aspectsof the theory. enon, dependent on anthropogenicchlorine in the Even though a climate model successfullysimulates stratosphere.While we have no observations,the 1963 the responseto a volcaniceruption [e.g.,Hansen et al., Agung eruption probably did not deplete 03, as there 1992], this does not guaranteethat it can accurately was little anthropogenicchlorine in the stratosphere. simulatethe responseto greenhousegases. The ability of Becauseof the Montreal protocoland subsequentinter- the samemodel to respondto decadal-and longer-scale national agreements,chlorine concentration has peaked responsesto climate forcingsis not tested, as the inter- in the stratosphereand is now decreasing.Therefore, for annual time-dependentresponse of the climate system the next few decades,large volcaniceruptions will have dependsonly on the thermal inertia of the oceanicmixed effects similar to Pinatubo, but after that, these 0 3 layer combinedwith the climate model sensitivity.The effectswill go away and volcaniceruptions will have a long-termresponse to globalwarming depends on accu- stronger effect on atmosphericcirculation without the rate simulation of the deep ocean circulation, and a negativefeedback produced by 03 depletion. volcanoexperiment does not allow an evaluationof that portion of the model. If a volcano experiment [e.g., Kirchneret al., 1999] simulatesthe winter warming re- 8. SUMMARY AND DISCUSSION sponseof a climate model to volcanicaerosols, then we can infer the ability of a model to simulate the same Large volcaniceruptions inject sulfur gasesinto the dynamicalmechanisms in responseto increasedgreen- stratosphere,which convert to sulfate aerosolswith an housegases, ozone depletion,or solarvariations. e-folding residencetimescale of about 1 year. The cli- As climate models improve, through programslike mate responseto large eruptionslasts for severalyears. the GCM-Reality Intercomparison Project for the The aerosol cloud producescooling at the surfacebut Stratosphere(GRIPS) of the StratosphericProcesses heating in the stratosphere.For a tropical eruption this and their Relation to Climate (SPARC) program[Paw- heating is larger in the tropicsthan in the high latitudes, son et al., 2000], they will improve their ability to simu- producing an enhanced pole-to-equator temperature late the climatic responseto volcanic eruptions and gradient and, in the Northern Hemisphere winter, a other causesof climate change. strongerpolar vortex and winter warming of Northern Hemispherecontinents. This indirect advectiveeffect on temperatureis strongerthan the radiative coolingeffect that dominates at lower latitudes and in the summer. GLOSSARY The volcanic aerosolsserve as surfacesfor heteroge- neous chemical reactions that destroy stratospheric 500-mbar pattern: Wind circulationin the middle ozone,which lowersultraviolet absorptionand reduces of the troposphere,typically at a heightof about 5.5 km, the radiativeheating in the lower stratosphere.Since this which is indicativeof the generaldirection of transport chemical effect dependson the presenceof anthropo- of energy and moisture. Average surface pressureis genic chlorine, it has only become important in recent about1000 mbar, equal to l0s Pa. decades.There is no evidencethat volcaniceruptions Advectiveeffect: Changesproduced by horizontal produceE1 Nifio events,but the climaticconsequences transportof energyby windsas opposedto thosecaused of E1 Nifio and volcaniceruptions are similar and must by radiation. be separatedto understandthe climaticresponse to each. Aerosol: A liquid droplet or solid particle sus- Becausevolcanic eruptions and their subsequentcli- pended in a gas. 38, 2 / REVIEWS OF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 215

Apparenttransmission: Amount of total solarradi- measures backscattered ultraviolet radiation in several ation transmittedto the surfacecorrected for geometry wavelengths,from which the total columnamount of 03 and time of day of measurements. and SO2 can be derived. December-January-February(DJF): Winter in the Northern Hemisphere. e-foldingdecay time: The amountof time it takes ACKNOWLEDGMENTS. I thank Georgiy Stenchikov, for ananomaly to decayto e- • of itsinitial perturbation. Ellsworth Dutton, Steven Self, and the reviewers for valuable This is typicallytaken to be the definition of the time- commentson the manuscript.My work on volcaniceruptions scaleof a logarithmicprocess. and climate has been supportedby NSF and NASA for more Glass inclusions: Pocketsof liquid or gas inside than 20 years, most recentlyby NSF grant ATM 99-96063 and NASA grant NAG 5-7913. I particularly thank Jay Fein, Ken glassformed when molten rock cools after a volcanic Bergman,and Lou Walter for their supportas program man- eruption. agers.I thank Clifford Mass, Michael Matson, JianpingMao, Heterogeneouschemical reactions: Chemicalre- Yuhe Liu, Georgiy Stenchikov,Karl Taylor, Hans Graf, Ingo actionsthat occurwith interactionsinvolving more than Kirchner, Melissa Free, and Juan Carlos Antufia for their one phase. Gas reactionsthat occur on the surface of collaborationswith me that made these results possible.I aerosolparticles are an example. thank John Christy and Phil Jones for temperature data. I Interdecadal: Varying from one decade (10-year thank Ellsworth Dutton for radiation observations used in period) to another. Figures 1 and 2. Figures 1, 2, and 4-7 and Plates 5-8 were Lithic: Pertainingto stone. In the contexthere, it drawn using GRADS, created by Brian Doty. I thank Juan meansthat the aerosolparticles are solid and consistof CarlosAntufia for helpingto draw Figures5 and 7 and Plates 7 and 8. crustal material. Little Ice Age: The period 1500-1900A.D., which Michael Coffey was the Editor responsiblefor this paper. He thanks Glynn Germany for the cross-disciplinaryreview was cooler than the period before or after. and O. B. Toon and R. Stolarski for the technical review. Magmaticmaterial: From the magma,the hot liq- uid from the interior of the Earth that often emerges during a volcaniceruption. MedievalWarming: Relativelywarm period, 900- REFERENCES 1200 A.D. Mt: Mcgaton= 106T = 109 kg = 10•2g = tcra- Ahn, M. H., A new SO2 retrieval algorithm using total ozone mappingspectrometer radiance and its applications,Ph.D. gram. dissertation,192 pp., Dep. of Meteorol., Univ. of Md., Optical depth: A measureof the amountof extinc- College Park, 1997. tion of radiation along a path through the atmosphere, Andronova,N. G., E. V. Rozanov,F. Yang, M. E. Schlesinger, proportionalto the amount of material in the path. and G. L. Stenchikov,Radiative forcingby volcanicaerosols Pyrheliometer: Radiation measuring instrument from 1850 to 1994, J. Geophys.Res., 104, 16,807-16,826, that measures the direct solar flux. 1999. Angell, J. K., Impact of E1 Nifio on the delineationof tropo- Pyroclasticflow: A hot, ash-ladeneruption cloud sphericcooling due to volcaniceruptions, J. Geophys.Res., from a volcano that descendsdown the mountainside, in 93, 3697-3704, 1988. spite of its temperature, due to the weight of the sus- Angell, J. K., Estimated impact of Agung, E1 Chich6n, and pended aerosols. Pinatubo volcanic eruptions on global and regional total Quasi-biennialoscillation (QBO): Areversal of the ozone after adjustmentfor the QBO, Geophys.Res. Lett., 24, 647-650, 1997a. wind direction in the tropical stratospherethat occurs Angell, J. K., Stratosphericwarming due to Agung, E1 Chi- approximatelyevery 27 months and influencestemper- ch6n, and Pinatubo taking into accountthe quasi-biennial ature and chemistry. oscillation,J. Geophys.Res., 102, 9479-9485,1997b. StratosphericAerosol and Gas ExperimentII: A Asaturov, M. L., M. I. Budyko, K. Y. Vinnikov, P. Y. Grois- satellite instrumentin space since 1984 that measures man, A. S. Kabanov, I. L. Karol, M.P. Kolomeev, Z. I. Pivovarova, E. V. Rozanov, and S.S. Khmelevtsov, Volca- the amount of radiation in four visible and ncar-IR nics,Stratospheric Aerosol and Earth'sClimate (in Russian), frequenciesthat is transmittedhorizontally through dif- 256 pp., Gidrometeoizdat,St. Petersburg,Russia, 1986. ferent altitudesby limb scanning.The amount and size Barkstrom, B. R., The Earth Radiation Budget Experiment distribution of volcanic aerosols can be derived from (ERBE), Bull. Am. Meteorol.Soc., 65, 1170-1185, 1984. these data. Barnes, J. E., and D. J. Hoffman, Lidar measurements of stratosphericaerosol over Mauna Loa Observatory,Geo- Sulfate: Containingoxidized sulfur, in the form of phys.Res. Lett., 24, 1923-1926, 1997. SO4. Sulfat½aerosols consist of H20 and H2SO4. Bekki, S., J. A. Pyle, and D. M. Pyle, The role of microphysical Teleconnectionpattern: Patternthat is correlated and chemicalprocesses in prolongingthe climate forcing of but not contiguous. the Toba eruption,Geophys. Res. Lett., 23, 2669-2672, 1996. Teragram:Equal to 10TM g = 109 kg = 106T = Bluth, G. J. S., S. D. Doiron, S.C. Schnetzler,A. J. Krueger, and L. S. Walter, Global tracking of the SO2 cloudsfrom megaton. the June 1991 Mount Pinatubo eruptions, Geophys.Res. •otal Ozone MappingSpectrometer frOMS): An Lett., 19, 151-154, 1992. instrument that has flown on satellites since 1979 and Bradley,R. S., The explosivevolcanic eruption signal in North- 216 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

ern Hemisphere continental temperature records, Clim. eruption of Italian and Icelandicvolcanoes in 1783, Qua- Change,12, 221-243, 1988. ternaire, 9, 25-35, 1998. Bryson,R. A., and G. J. Dittberner, A non-equilibriummodel Groisman, P. Y., Regional climate consequencesof volcanic of the hemisphere,J. Atmos. Sci., 33, 2094-2106, 1976. eruptions(in Russian),Meteorol. Hydrol., No. 4, 39-45, Coffey, M. T., Observationsof the impact of volcanicactivity 1985. on stratosphericchemistry, J. Geophys.Res., 101, 6767- Groisman,P. Y., Possibleregional climate consequences of the 6780, 1996. Pinatubo eruption:An empiricalapproach, Geophys. Res. Crowley, T. J., and K.-Y. Kim, Modeling the temperature Lett., 19, 1603-1606, 1992. responseto forced climate changeover the last six centu- Haigh, J., The impact of solar variability on climate, Science, ries, Geophys.Res. Lett., 26, 1901-1904, 1999. 272, 981-984, 1996. Crutzen, P. J., The possibleimportance of CSO for the sulfate Halpert, M. S., and C. F. Ropelewski,Surface temperature layer of the stratosphere,Geophys. Res. Lett., 3, 73-76, 1976. patternsassociated with the SouthernOscillation, J. Clim., D'Arrigo, R. D., G. C. Jacoby,M. Free, and A. Robock, 5, 577-593, 1992. Northern Hemispheretemperature variability for the past Handler, P., Possible associationbetween the climatic effects three centuries: Tree-ring and model estimates, Clim. of stratosphericaerosols and sea surfacetemperatures in Change,42, 663-675, 1999. the easterntropical Pacific Ocean, J. Climatol.,6, 31-41, 1986. Dutton, E.G., A coherencebetween the QBO and the ampli- Hansen, J. E., W.-C. Wang, and A. A. Lacis, Mount Agung tude of the Mauna Loa atmospherictransmission annual providesa test of a global climatic perturbation,Science, cycle,Int. J. Climatol.,12, 383-396, 1992. 199, 1065-1068, 1978. Dutton, E.G., and J. R. Christy, Solar radiative forcing at Hansen, J., I. Fung, A. Lacis, D. Rind, S. Lebedeff, R. Ruedy, selected locations and evidence for global lower tropo- G. Russell,and P. Stone,Global climatechanges as forecast sphericcooling following the eruption of E1 Chich6n and by the Goddard Institute for Space Studiesthree-dimen- Pinatubo, Geophys.Res. Lett., 19, 2313-2316, 1992. sional model, J. Geophys.Res., 93, 9341-9364, 1988. Dutton, E.G., J. J. DeLuisi, and A. P. Austring,Interpretation Hansen, J., A. Lacis, R. Ruedy, and M. Sato, Potential climate of Mauna Loa atmospherictransmission relative to aero- impactof Mount Pinatuboeruption, Geophys. Res. Lett., 19, sols,using photometric precipitable water amounts,J. At- 215-218, 1992. mos. Chem., 3, 53-68, 1985. Hansen, J., et al., A Pinatubo climate modeling investigation, Ellsaesser,H. W., Isolating the climatologiceffects of volca- in The Mount PinatuboEruption Effects on theAtmosphere noes,Rep. UCRL-89161, 29 pp., LawrenceLivermore Natl. and Climate, NATO ASI Set., Set. I, 42, 233-272, 1996. Lab., Livermore, Calif., 1983. Harington, C. R. (Ed.), The Year Withouta Summer?World Forsyth,P. Y., In the wake of Etna, 44 B.C., ClassicalAntiq.,7, Climate in 1816, 576 pp., Can. Mus. of Nature, Ottawa, 49-57, 1988. Ont., Canada, 1992. Franklin, B., Meteorological imaginations and conjectures, Harshvardhan,Perturbations of the zonal radiation balanceby Manchr.Lit. Philos.Soc. Mem. Proc.,2, 122, 1784.(Reprint- a stratosphericaerosol layer, J. Atmos.Sci., 36, 1274-1285, ed in Weatherwise,35, 262, 1982.) 1979. Free, M., and A. Robock, Global warming in the contextof the Hirono, M., On the triggerof E1 Nifio-Southern Oscillationby Little Ice Age, J. Geophys.Res., 104, 19,057-19,070,1999. the forcing of early E1 Chich6n volcanicaerosols, J. Geo- Gates, W. L., AMIP: The AtmosphericModel Intercompari- phys.Res., 93, 5365-5384, 1988. son Project,Bull. Am. Meteorol.Soc., 73, 1962-1970, 1992. Holton, J. R., P. H. Haynes,M. E. Mcintyre, A. R. Douglass, Genin, A., B. Lazar, and S. Brenner, Vertical mixing and coral R. B. Rood, and L. Pfister, Stratosphere-troposphereex- death in the Red Sea following the eruption of Mount change,Rev. Geophys.,33, 403-439, 1995. Pinatubo, Nature, 377, 507-510, 1995. Houghton,J. T., L. G. Meira Filho, B. A. Callander,N. Harris, Gilliland, R. L., Solar, volcanic, and CO2 forcing of recent climatic changes,Clim. Change,4, 111-131, 1982. A. Kattenberg, and K. Maskell (Eds.), Climate Change Gilliland, R. L., and S. H. Schneider,Volcanic, CO2, and solar 1995raThe Scienceof Climate Change,584 pp., Cambridge Univ. Press, New York, 1996. forcingof Northern and SouthernHemisphere surface tem- peratures,Nature, 310, 38-41, 1984. Humphreys,W. J., Volcanic dust and other factors in the Graf, H.-F., I. Kirchner, A. Robock, and I. Schult, Pinatubo productionof climaticchanges, and their possiblerelation eruptionwinter climate effects:Model versusobservations, to ice gases,J. Franklin Inst., Aug., 131-172, 1913. Clim. Dyn., 9, 81-93, 1993. Humphreys,W. J., Physicsof theAir, 676 pp., Dover, Mineola, Graf, H.-F., J. Perlwitz, I. Kirchner, and I. Schult, On the inter- N.Y., 1940. relationshipbetween recent climate trends, ozone change and Hunt, B. G., A simulationof the possibleconsequences of a increasedgreenhouse gas forcing, in AtmosphericOzone As volcaniceruption on the general circulationof the atmo- a Climate Gas, edited by W. C. Wang and I. S. A. Isaksen, sphere,Mon. WeatherRev., 105, 247-260, 1977. pp. 163-179, Springer-Verlag,New York, 1995. Hurrell, J. W., Decadal trends in the North Atlantic Oscilla- Graf, H.-F., J. Feichter, and B. Langmann, Volcanic sulfur tion' Regionaltemperatures and precipitation,Science, 269, emission:Estimates of sourcestrength and its contribution 676-679, 1995. to the global sulfate distribution,J. Geophys.Res., 102, Hurrell, J. W., and H. van Loon, Decadal variations in climate 10,727-10,738, 1997. associated with the North Atlantic Oscillation, Clim. Grant, W. B., Tropical stratosphericozone changesfollowing Change,36, 301-326, 1997. the eruption of Mount Pinatubo,in The Mount Pinatubo Kalnay, E., et al., The NCEP/NCAR 40-yearreanalysis project, EruptionEffects on theAtmosphere and Climate,NATO ASI Bull. Am. Meteorol. Soc., 77, 437-471, 1996. Set., Set. I, 42, 161-175, 1996. Keen, R. A., Volcanicaerosols and lunar eclipses,Science, 222, Grant, W. B., et al., Observations of reduced ozone concen- 1011-1013, 1983. trations in the tropical atmosphereafter the eruption of Kelly, P.M., and C. B. Sear, The formulation of Lamb's dust Mount Pinatubo, Geophys.Res. Lett., 19, 1109-1112, 1992. veil index, in AtmosphericEffects and Potential Climatic Grattan, J., M. Brayshay,and J. Sadler, Modelling the distal Impact of the 1980Eruptions of Mount St. Helens,edited by impactsof pastvolcanic gas emissions: Evidence of Europe- A. Deepak, NASA Conf. Publ. 2240, 293-298, 1982. wide environmentalimpacts from gasesemitted during the Kelly, P.M., P. D. Jones,A. Robock, and K. R. Briffa, The 38, 2 / REVIEWSOF GEOPHYSICS Robock:VOLCANIC ERUPTIONS AND CLIMATE ß 21 7

contributionof Hubert H. Lamb to the study of volcanic Mass, C., and A. Robock, The short-term influence of the effects on climate, Weather, 53, 209-222, 1998. Mount St. Helensvolcanic eruption on surfacetemperature Kinne, S., O. B. Toon, and M. J. Prather, Buffering of strato- in the northwest United States, Mort. Weather Rev., 110, sphericcirculation by changingamounts of tropicalozone: A 614-622, 1982. Pinatubocase study, Geophys. Res. Lett., 19, 1927-1930,1992. Mcintyre, M. E., How well do we understandthe dynamicsof Kirchner, I., and H.-F. Graf, Volcanos and E1 Nifio: Signal stratosphericwarmings?, J. Meteorol.Soc. Jpn., 60, 37-65, separationin Northern Hemispherewinter, Clim. Dyn., 11, 1982. 341-358, 1995. Miles, M. K., and P. B. Gildersleeves, Volcanic dust and Kirchner, I., G. L. Stenchikov,H.-F. Graf, A. Robock, and J. C. changesin Northern Hemispheretemperature, Nature, 271, Antufia, Climate model simulationof winter warming and 735-736, 1978. summercooling following the 1991 Mount Pinatubovolca- Minnis, P., E. F. Harrison, L. L. Stowe, G. G. Gison, F. M. nic eruption,J. Geophys.Res., 104, 19,039-19,055, 1999. Denn, D. R. Doelling, and W. L. Smith Jr., Radiative Kitoh, A., H. Koide, K. Kodera, S. Yukimoto, and A. Noda, climate forcing by the Mount Pinatubo eruption, Science, Interannualvariability in the stratospheric-troposphericcir- 259, 1411-1415, 1993. culation in a coupled ocean-atmosphereGCM, Geophys. Mitchell, J. M., Jr., Recent secular changesof the global Res. Lett., 23, 543-546, 1996. temperature,Ann. N.Y. ,4cad. Sci., 95, 235-250, 1961. Kodera, K., Influence of volcaniceruptions on the troposphere Mitchell, J. M., Jr., A preliminary evaluationof atmospheric throughstratospheric dynamical processes in the Northern pollutionas a causeof the globaltemperature fluctuation of Hemispherewinter, J. Geophys.Res., 99, 1273-1282, 1994. the past century,in Global Effectsof EnvironmentalPollu- Kodera, K., M. Chiba, and K. Shibata, A general circulation tion, edited by S. F. Singer, pp. 139-155, D. Reidel, Nor- model study of the solar and QBO modulation of the well, Mass., 1970. stratosphericcirculation during the Northern Hemisphere Mohnen, V. A., Stratosphericion and aerosol chemistryand winter, Geophys.Res. Lett., 18, 1209-1212, 1991. possiblelinks with cirrus cloud microphysics--A critical Kodera, K., M. Chiba, H. Koide, A. Kitoh, and Y. Nikaidou, assessment,J. ,4tmos. Sci., 47, 1933-1948, 1990. Interannual variability of the winter stratosphereand tro- Newhall, C. G., and S. Self, The volcanic explosivityindex pospherein the Northern Hemisphere,J. Meteorol. Soc. (VEI): An estimate of explosivemagnitude for historical Jpn., 74, 365-382, 1996. volcanism,J. Geophys.Res., 87, 1231-1238, 1982. Kondratyev, K. Y., Volcanoes and climate, WCP-54, WMO/ Nicholls,N., Low latitude volcaniceruptions and the E1 Nifio- TD-166, 103 pp., World Meteorol. Org., Geneva, 1988. Southern Oscillation, J. Climatol., 8, 91-95, 1988. Kondratyev,K. Y., and I. Galindo, VolcanicActivity and Cli- Oliver, R. C., On the responseof hemisphericmean temper- mate, 382 pp., A. Deepak, Hampton, Va., 1997. ature to stratosphericdust: An empiricalapproach, J. Appl. Krueger, A. J., C. C. Schnetzler,and L. S. Walter, The De- Meteorol., 15, 933-950, 1976. cember1981 eruption of NyamuragiraVolcano (Zaire) and Parker, D. E., and J. K. L. Brownscombe,Stratospheric warm- the origin of the "mysterycloud" of early 1982,J. Geophys. ing followingthe E1 Chich6nvolcanic eruption, Nature, 301, Res., 101, 15,191-15,196, 1996. 406-408, 1983. Labitzke, K., and H. van Loon, Association between the l 1- Pawson,S., et al., The GCM-Reality IntercomparisonProject year solarcycle, the QBO, and the atmosphere,part I, The for SPARC (GRIPS): Scientificissues and initial results, troposphereand the stratospherein the Northern Hemi- Bull. Am. Meteorol. Soc., 81,781-796, 2000. spherein winter, J. Atmos. Ten'.Phys., 50, 197-207, 1988. Perlwitz, J., and H.-F. Graf, The statistical connection between Lamb, H. H., Volcanic dust in the atmosphere,with a chro- troposphericand stratosphericcirculation of the Northern nology and assessmentof its meteorologicalsignificance, Hemispherein winter, J. Clim., 8, 2281-2295, 1995. Philos. Trans. R. Soc. London, Set. A, 266, 425-533, 1970. Pinto, J.P., R. P. Turco, and O. B. Toon, Self-limitingphysical Lamb, H. H., Supplementaryvolcanic dust veil index assess- ments, Clim. Monit., 6, 57-67, 1977. and chemicaleffects in volcaniceruption clouds, J. Geophys. Res., 94, 11,165-11,174, 1989. Lamb, H. H., Update of the chronologyof assessmentsof the volcanic dust veil index, Clim. Monit., 12, 79-90, 1983. Pollack,J. B., O. B. Toon, C. Sagan,A. Summers,B. Baldwin, Li, X. S., and A. Berger, Simulating climatic impact of the and W. Van Camp, Volcanic explosionsand climatic Toba eruption with the LLN 2-D climate model, Sci. Rep. change: A theoretical assessment,J. Geophys.Res., 81, 1997/14, Inst. d'Astron. et de Geophys.G. Lemaitre, Univ. 1071-1083, 1976. Cath. de Louvain, Louvain-la-Neuve,Belgium, 1997. Pollack, J. B., D. Rind, A. Lacis, J. E. Hansen, M. Sato, and R. Lough, J. M., and H. C. Fritts, An assessmentof the possible Ruedy, GCM simulationof volcanicaerosol forcing, part I, effects of volcanic eruptions on North American climate Climate changes induced bv steady-stateperturbations, usingtree-ring data, 1602 to 1900 A.D., Clim. Change,10, J. Clim., 6, 1719-1742, 1993. 219-239, 1987. Quiroz, R. S., The isolation of stratospherictemperature Luhr, J. F., I. S. E. Carmichael,and J. C. Varekamp, The 1982 changedue to the E1 Chich6n volcaniceruption from non- eruptionsof E1 Chich6nvolcano, Chiapas, Mexico: Miner- volcanicsignals, J. Geophys.Res., 88, 6773-6780, 1983. alogy and petrology of the anhydrite-bearingpumices, J. Ramaswamy,V., M.D. Schwarzkopf,and W. J. Randel, Fin- Volcanol. Geotherm. Res., 23, 69-108, 1984. gerprint of ozone depletion in the spatial and temporal MacCracken,M. C., and F. M. Luther, Preliminaryestimate of pattern of recent lower-stratosphericcooling, Nature, 382, the radiative and climatic effectsof the E1 Chich6n erup- 616-618, 1996. tion, Geofis.Int., 23, 385-401, 1984. Rampino, M. R., and S. Self, Sulphur-richvolcanic eruptions Mao, J., and A. Robock, Surface air temperature simulations and stratosphericaerosols, Nature, 310, 677-679, 1984. by AMIP general circulationmodels: Volcanic and ENSO Rampino, M. R., and S. Self, Volcanicwinter and accelerated signalsand systematicerrors, J. Clim., 11, 1538-1552, 1998. glaciationfollowing the Toba super-eruption,Nature, 359, Marcus, S. L., Y. Chao, J. O. Dickey, and P. Gegout, Detection 50-52, 1992. and modeling of nontidal oceaniceffects on Earth's rota- Rampino, M. R., S. Self, and R. W. Fairbridge, Can climatic tion rate, Science, 281, 1656-1659, 1998. change cause volcanic eruptions?,Science, 206, 826-828, Mass, C. F., and D. A. Portman,Major volcaniceruptions and 1979. climate: A critical evaluation, J. Clim., 2, 566-593, 1989. Rind, D., N. K. Balachandran,and R. Suozzo,Climate change 218 ß Robock: VOLCANIC ERUPTIONS AND CLIMATE 38, 2 / REVIEWS OF GEOPHYSICS

and the middle atmosphere,part II, The impact of volcanic Tahiti-Darwin Southern Oscillation Index, Mon. Weather aerosols,J. Clim., 5, 189-208, 1992. Rev., 115, 2161-2165, 1987. Robock,A., Internally and externallycaused climate change,J. Rosenfield, J., D. Considine, P. Meade, J. Bacmeister,C. Jack- Atmos. Sci., 35, 1111-1122, 1978. man, and M. Schoeberl,Stratospheric effects of Mount Pina- Robock, A., The "Little Ice Age": Northern Hemisphere av- tuboaerosol studied with a coupledtwo-dimensional model, J. erage observationsand model calculations,Science, 206, Geophys.Res., 102, 3649-3670, 1997. 1402-1404, 1979. Russell, P. B., et al., Global to microscale evolution of the Robock,A., The Mount St. Helensvolcanic eruption of 18 May Pinatubo volcanic aerosol derived from diverse measure- 1980:Minimal climaticeffect, Science, 212, 1383-1384,1981a. mentsand analyses,J. Geophys.Res., 101, 18,745-18,763,1996. Robock,A., A latitudinallydependent volcanic dust veil index, Sapper,K., Beitrfigezur Geographieder tfitingenVulkane, Z. and its effect on climate simulations,J. Volcanol. Geotherm. Vulkanol., 3, 65-197, 1917. Res., 11, 67-80, 1981b. Sapper, K., Vulkankunde,Englehorn Verlag, Stuttgart, Ger- Robock,A., The dustcloud of the century,Nature, 301,373- many, 1927. 374, 1983a. Sassen,K., D. O. Starr, G. G. Mace, M. R. Poellot, S. H. Melfi, Robock, A., Ice and snow feedbacks and the latitudinal and W. L. Eberhard, J. D. Spinhirne, E. W. Eloranta, D. E. seasonaldistribution of climatesensitivity, J. Atmos. Sci., 40, Hagen, and J. Hallett, The 5-6 December 1991 FIRE IFO 986-997, 1983b. II jet streamcirrus case study: Possible influences of volca- Robock, A., Review of VolcanoWeather: The Storyof 1816, the nic aerosols,J. Atmos. Sci., 52, 97-123, 1995. Year Withouta Summer,Natl. WeatherDig., 9, 5-6, 1984a. Sato, M., J. E. Hansen, M.P. McCormick, and J. B. Pollack, Robock, A., Climate model simulations of the effects of the E1 Stratosphericaerosol optical depths, 1850-1990, J. Geo- Chich6n eruption, Geofis.Int., 23, 403-414, 1984b. phys.Res., 98, 22,987-22,994, 1993. Robock,A., Snowand ice feedbacksprolong effects of nuclear Schatten,K. H., H. G. Mayr, I. Harris, and H. A. Taylor, A winter, Nature, 310, 667-670, 1984c. zonallysymmetric model for volcanicinfluence upon atmo- Robock, A., Volcanoes and climate, in Climate and Geo- sphericcirculation, Geophys. Res. Lett., 11,303-306, 1984. Sciences:A Challengefor Scienceand Societyin the 21st Schneider,S. H., and C. Mass, Volcanic dust, sunspotsand Century,NATO ASI Ser., Ser. C, 285, 309-314, 1989. temperaturetrends, Science,190, 741-746, 1975. Robock,A., The volcaniccontribution to climate changeof the Self, S., M. R. Rampino, J. Zhao, and M. G. Katz, Volcanic past 100 years, in Greenhouse-Gas-InducedClimatic aerosolperturbations and strongE1 Nifio events:No gen- Change:A Critical Appraisal of Simulationsand Observa- eral correlation, Geophys.Res. Lett., 24, 1247-1250, 1997. tions, edited by M. E. Schlesinger,pp. 429-444, Elsevier Shelley, M., Frankenstein,Bantam Books, New York, 1818. Sci., New York, 1991. (Classicedition, 209 pp., BantamBooks, New York, 1981.) Robock, A., Review of Year Without a Summer?World Climate Shindell, D., D. Rind, N. Balachandran, J. Lean, and P. Lon- in 1816, Clim. Change,26, 105-108, 1994. ergan, Solar cycle variability, ozone, and climate, Science, Robock, A., Stratosphericcontrol of climate, Science,272, 284, 305-308, 1999. 972-973, 1996a. Simarski,L. T., Volcanismand climate change,special report, Robock, A., Nuclear winter, in Encyclopediaof Weatherand 27 pp., AGU, Washington,D.C., 1992. Climate, vol. 2, edited by S. H. Schneider,pp. 534-536, Simkin, T., and R. S. Fiske, Krakatau 1883: The Volcanic Oxford Univ. Press, New York, 1996b. Eruptionand Its Effects,464 pp., Smithson.Inst., Washing- ton, D.C., 1983. Robock, A., and M.P. Free, Ice cores as an index of global Simkin,T., and L. Siebert,Volcanoes of the World,2nd ed., 349 volcanismfrom 1850 to the present,J. Geophys.Res., 100, pp., GeosciencePress, Tucson, Ariz., 1994. 11,549-11,567, 1995. Simkin, T., L. Siebert, L. McClelland, D. Bridge, C. G. New- Robock, A., and M.P. Free, The volcanic record in ice cores hall, and J. H. Latter, Volcanoesof the World,232 pp., Van for the past 2000 years,in Climatic Variationsand Forcing Nostrand Reinhold, New York, 1981. Mechanismsof the Last 2000 Years,edited by P. D. Jones, Solomon,S., Stratosphericozone depletion:A review of con- R. S. Bradley,and J. Jouzel,pp. 533-546, Springer-Verlag, ceptsand history,Rev. Geophys.,37, 275-316, 1999. New York, 1996. Solomon, S., R. W. Portmann, R. R. Garcia, L. W. Thomason, Robock, A., and Y. Liu, The volcanic signal in Goddard L. R. Poole, and M.P. McCormick, The role of aerosol Institute for SpaceStudies three-dimensional model simu- variations in anthropogenicozone depletion at northern lations, J. Clim., 7, 44-55, 1994. midlatitudes,J. Geophys.Res., 101, 6713-6727, 1996. Robock, A., and J. Mao, Winter warmingfrom large volcanic Spencer,R. W., J. R. Christy,and N. C. Grody, Global atmo- eruptions,Geophys. Res. Lett., 19, 2405-2408, 1992. spheric temperature monitoring with satellite microwave Robock, A., and J. Mao, The volcanicsignal in surfacetem- measurements:Method and results 1979-1984, J. Clim., 3, perature observations,J. Clim., 8, 1086-1103, 1995. 1111-1128, 1990. Robock, A., and C. Mass, The Mount St. Helens volcanic Stenchikov, G. L., I. Kirchner, A. Robock, H.-F. Graf, J. C. eruptionof 18 May 1980:Large short-termsurface temper- Antufia, R. G. Grainger, A. Lambert, and L. Thomason, ature effects, Science, 216, 628-630, 1982. Radiative forcing from the 1991 Mount Pinatubovolcanic Robock,A., and M. Matson, Circumglobaltransport of the E1 eruption,J. Geophys.Res., 103, 13,837-13,857,1998. Chich6n volcanicdust cloud, Science,221, 195-197, 1983. Stommel,H., and E. Stommel,Volcano Weather.' The Storyof Robock,A., K. E. Taylor, G. L. Stenchikov,and Y. Liu, GCM 1816, the Year Withouta Summer,177 pp., SevenSeas Press, evaluationof a mechanismfor E1 Nifio triggeringby the E1 Newport, R. I., 1983. Chichdnash cloud, Geophys. Res. Lett., 22, 2369-2372, 1995. Stothers,R. B., The great Tambora eruption in 1815 and its Roeckner,E., K. Arpe, L. Bengtsson,M. Christoph,M. Claus- aftermath, Science,224, 1191-1198, 1984. sen, L. Duemenil, M. Ecsh, M. Giogetta, U. Schlese,and Stowe,L. L., R. M. Carey,and P. P. Pellegrino,Monitoring the U. Schulzweida,The atmosphericgeneral circulation model Mount Pinatubo aerosol layer with NOAA 11 AVHRR ECHAM4: Model descriptionand simulationof present- data, Geophys.Res. Lett., 19, 159-162, 1992. day climate, Rep. 218, 90 pp., Max-Planck-Inst.ffir Meteo- Strong, A. E., Monitoring E1 Chich6n aerosol distribution rol., Hamburg, Germany, 1996. using NOAA-7 satellite AVHRR sea surfacetemperature Ropelewski, C. F., and P. D. Jones, An extensionof the observations,Geofis. Int., 23, 129-141, 1984. 38, 2 / REVIEWS OF GEOPHYSICS Robock: VOLCANIC ERUPTIONS AND CLIMATE ß 219

Strong,A. E., The effect of E1 Chich6n on the 82/83 E1 Nifio van Loon, H., and K. Labitzke, Association between the l 1- (abstract),Eos Trans.AGU, 67(44), 880, 1986. year solarcycle, the QBO and the atmosphere,part IV, The Symons,G. J. (Ed.), TheEruption of Krakatoa,and Subsequent stratosphere, not grouped by the phase of the QBO, Phenomena,494 pp., Triibner, London, 1888. J. Clim., 3, 827-837, 1990. Tabazadeh,A., and R. P. Turco, Stratosphericchlorine injec- Vinnikov, K. Y., A. Robock, R. J. Stouffer, and S. Manabe, tion by volcaniceruptions: HC1 scavengingand implication Vertical patterns of free and forced climate variations, for ozone, Science, 260, 1082-1086, 1993. Geophys.Res. Lett., 23, 1801-1804, 1996. Thompson,D. W. J., and J. M. Wallace, The Arctic Oscillation Vogelmann, A.M., T. P. Ackerman, and R. P. Turco, En- signature in the wintertime geopotential height and tem- hancementsin biologically effective ultraviolet radiation perature fields,Geophys. Res. Lett., 25, 1297-1300, 1998. followingvolcanic eruptions, Nature, 359, 47-49, 1992. Thompson,D. W. J., and J. M. Wallace, Annular modesin the Vupputuri, R. K. R., and J.P. Blanchet,The possibleeffects of extratropicalcirculation, part I, Month-to-month variabil- the E1Chich6n eruption on atmosphericthermal and chemical ity, J. Clim., 13, 1000-1016, 2000a. structureand surfaceclimate, Geofis. Int., 23, 433-447, 1984. Thompson,D. W. J., J. M. Wallace, and G. C. Hegerl, Annular Wallace, J. M., and D. S. Gutzler, Teleconnections in the modes in the extratropical circulation, part II, Trends, geopotentialheight field duringthe Northern Hemisphere J. Clim., 13, 1018-1036, 2000b. winter, Mon. Weather Rev., 109, 784-812, 1981. Tie, X., and G. Brasseur,The responseof the stratospheric Wallace, J. M., Y. Zhang, and J. A. Renwick,Dynamic contri- ozone to volcanic eruptions: Sensitivityto atmospheric bution to hemisphericmean temperature trends, Science, chlorine loading,Geophys. Res. Lett., 22, 3035-3038, 1995. 270, 780-783, 1995. Tie, X., G. P. Brasseur, C. Granier, A. De Rudder, and N. Wallace, J. M., Y. Zhang, and L. Balik, Interpretation of Larsen,Model studyof polar stratosphericclouds and their interdecadaltrends in Northern Hemisphere surface air effect on stratosphericozone, 2, Model results,J. Geophys. temperature,J. Clim., 9, 249-259, 1996. Res., 101, 12,575-12,584, 1996. Yue, G. K., R. E. Veiga, and P.-H. Wang, SAGE II observa- Toon, O. B., Volcanoesand climate, in AtmosphericEffects and tions of a previously unreported stratosphericvolcanic PotentialClimatic Impact of the 1980 Eruptionsof Mount St. aerosolcloud in the northern polar summerof 1980, Geo- Helens, edited by A. Deepak, NASA Conf. Publ. 2240, phys.Res. Lett., 21,429-432, 1994. 15-36, 1982. Zerefos, C. S., K. Tourpali, and A. F. Bais, Further studieson Toon, O. B., and J. B. Pollack, Atmospheric aerosolsand possiblevolcanic signal to the ozone layer,J. Geophys.Res., climate, Am. Sci., 68, 268-278, 1980. 99, 25,741-25,746, 1994. Trepte, C. R., and M. H. Hitchman, Tropical stratospheric Zhao, J., R. P. Turco, and O. B. Toon, A model simulation of circulation deduced from satellite aerosol data, Nature, 355, Pinatubovolcanic aerosols in the stratosphere,J. Geophys. 626-628, 1992. Res., 100, 7315-7328, 1995. Trepte, C. R., R. E. Veiga, and M.P. McCormick, The pole- Zielinski, G. A., P. A. Mayewski, L. D. Meeker, S. Whitlow, ward dispersalof Mount Pinatubovolcanic aerosol, J. Geo- M. S. Twickler, and K. Taylor, Potential atmosphericim- phys.Res., 98, 18,563-18,573, 1993. pact of the Toba mega-eruption---71,000 years ago, Geo- Turco, R. P., O. B. Toon, T. P. Ackerman, J. B. Pollack, and phys.Res. Lett., 23, 837-840, 1996. C. Sagan,Nuclear winter: Global consequencesof multiple nuclear explosions,Science, 222, 1283-1292, 1983. Turco, R. P., O. B. Toon, T. P. Ackerman, J. B. Pollack, and C. A. Robock,Department of EnvironmentalSciences, Rutgers Sagan,Nuclear winter: Climate and smoke:An appraisalof University,14 CollegeFarm Road, New Brunswick,NJ 08901- nuclear winter, Science,247, 166-176, 1990. 8551. ([email protected])