The presence of heavy metals in air particulate at Vulcano island (Italy

Ž .The Science of the Total Environment 212 1998 1]9

The presence of heavy metals in air particulate at VulcanoŽ .island Italy

G. Dongarraa,b,U, D. Varricaa`aIstituto di Mineralogia, Petrografia e Geochimica, Via Archirafi 36, 90123 Palermo, Italy

bIstituto di Geochimica dei Fluidi, Via Ugo La Malfa 153, 90146 Palermo, Italy

Received 25 September 1997; accepted 12 November 1997

Abstract

Here we evaluate the presence and the dispersion pattern of heavy metals in air particulates at Vulcano island,Italy. To assess the impact of the volcanic source the metal content of a set of 42 lichen samples of the speciesParmelia was determined. Soil contribution in lichens was taken into account by the integrated use of datanormalized to an element of definite crustal origin and by enrichment factors. The elements Pb, Br, Sb, As, Cu, Znand Au appeared to be enriched in lichens with respect to local crustal material. The distribution maps of theenrichment factors clearly show how the entire island is subject to fallout by Pb, Br and Zn while the other elementsare preferentially enriched along a E]SE direction, following the main wind direction. Q 1998 Elsevier ScienceB.V.

Keywords: Biogeochemistry; Particulate matter; Heavy metals; Volcanoes; Lichens

1. Introduction

Volcanoes release significant quantities ofaerosol to the atmosphere and it is thereforeplausible that they are at least partly responsiblefor the presence of elements in air. It is estimated

U Corresponding author.

that volcanoes release 107]108 tons per year ofparticulate matter with a mean particle diameter

Ž .of-40 mm Mroz and Zoller, 1975 , i.e. 1]20%of the total quantity produced by natural pheno-mena. We may recall the famous ‘ete sans soleil’´ ´

Ž .of Napoleon Stommel and Stommel, 1979 , whichwas due to the eruption of the volcano Tambora

Ž .in 1815 on the island of Sumbawa Indonesia , theŽeffects of the eruptions of Krakatoa Schneider,

0048-9697r98r$19.00 Q 1998 Elsevier Science B.V. All rights reserved.Ž .P I I S 0 0 4 8 - 9 6 9 7 9 8 0 0 3 2 3 - 9

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9`2

.1983; Fiocco and Visconti, 1985 and, more re-cently, Pinatubo and the lively debate on thecauses of the great mass extinctions, particularlythat of the Cretaceous, in which dust appears to

Žhave played a fundamental role Alvarez et al.,.1980; Ganapathy, 1980 .

In the low troposphere, the effects of volcanismmingle with those produced by anthropicprocesses, not only as regards the quantities ofdust emitted into the atmosphere, but also thetype of substances, which are sometimes poten-tially toxic. For some elements, anthropic con-tributions are greater than those from volcanic.

Ž .Lantzy and Mackenzie 1979 showed that theamount of trace elements released to the atmo-sphere by anthropic activities exceed those fromnatural sources by many orders of magnitude.

Ž .Buat-Menard and Arnold 1978 believe that, if´an exception is made for Se and Pb, the emissionof metals from Mt. Etna is irrelevant with respectto which is produced by man’s activities in theentire Mediterranean area. However, if this istrue on a large scale, it may not be confirmed ona more limited one.

Volcanic activity and the consequent release ofsubstances into the air does not only occur duringeruptive periods, but also during normal fu-marolic activity. The presence of active volcanoesnear urban and industrial centres therefore opensup a scenario in which verification of the pres-ence of metals in air and the true contribution ofthe various sources become important.

The literature reports data on trace elementsin volcanic gases obtained by means of varioustechniques: analysis of fumarole condensate; in-crustations forming at the mouths of fumaroles;or aerosol samples collected by direct aspirationof gases from the plume. Each of these methodshas advantages and disadvantages. The presentstudy reports data obtained by a biomonitoringtechnique carried out using the metal content ofepilithic lichens collected on the island of Vul-

Ž .cano Italy .The method is based on the preferential rela-

tionship of lichens for air and the particular mor-phology and physiology of their thalli, which mayaccumulate trace elements in quantities higherthan those of the growing environment. Analysis

of thalli therefore provides a preliminary methodof assessing the environmental conditions of acertain area, with reference to a lengthy period ofobservation. This type of screening is more effec-tive, quicker and cheaper than conventionalmethods and allows an identification of not onlywhich metals are released into the air, but alsotheir dispersion in areas adjacent to the emissionsource.

2. The island of Vulcano

Vulcano is one of the seven islands making upthe Aeolian Archipelago and is related to thecollision between the African and European con-

Ž .tinental plates Fig. 1 . It is the most southerly ofthe Aeolian arc and, with a surface area of 22km2, the third in order of size, after Lipari andSalina. The morphology of the island is a trun-cated cone, due to the collapse of the central partof the stratovolcano which occurred about 100 000years ago and gave rise to the ‘Caldera del Piano’.The ‘Cratere della Fossa’, currently active, is 391m a.s.l., is located in the centre of the caldera and

Fig. 1. The island of Vulcano and location of the samplingsites.

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9` 3

has a base diameter of 1 km. Volcanic activitystarted during Upper Pleistocene and the islandis entirely made up of volcanic rocks. According

Ž .to Keller 1980 the following rock types can bedistinguished: trachybasalts; trachyandesites; tra-chytes; leucite tephrites; alkali-rich trachytes;latites; and rhyolites.

Ž .Since the last eruption 1888]1890 until today,volcanic activity has been limited to fumaroleoccurrences along a N]S alignment and to theoutflowing of thermal waters. The main fumarolefields lie on the internal flank and north-easternpart of the rim of the Cratere della Fossa andtheir activity is variable. A first increase of thetemperature fumaroles was recorded at the

beginning of this century, reaching approx. 2008CŽ .in 1913 and 6158C in 1923 Sicardi, 1941 . Renew

heating phase was recorded in 1935, volcanic ac-tivity increased again in late 1977, just before aseismic crisis of magnitude 5, when the fumaroletemperatures rose from 1908C to 3008C in 1980.A new crisis in 1988 increased the temperaturefrom 4708C to 6908C in 1993.

3. Materials and analytical methods

This study is based on a set of 42 lichen sam-ples of the species Parmelia collected exclusivelyfrom rock surfaces. In order to guarantee thestatistical significance of the samples, various dif-

Table 1Results of chemical analyses on the lichens from Vulcano island

Min. Max. Mean S.D. Geometric Analytical Detectionmean method limit

Au 0.006 0.361 0.05 0.068 0.032 INAA 0.0001As 1.1 13 4.6 2.5 4 INAA 0.01Ba 30 450 190 104 159 INAA 5.0Br 15 120 37 18 34 INAA 0.01Ca 2950 57 050 14 667 10 772 12 366 INAA, ICP 0.01%Co 2.5 19 6.6 3.6 5.9 INAA 0.1Cr 3.4 54 17.4 9.1 15.5 INAA 0.3Cs 0.4 5.6 2 1.3 1.6 INAA 0.05Fe 736 167 00 4912 4249 3373 INAA 0.01%Hf 0.3 3.6 1.3 0.8 1.1 INAA 0.05K 78 1690 7319 2516 6935 INAA 0.01%Mo 0.9 5.6 2.2 1.2 2 INAA 0.05Na 616 12 500 3664 2717 2751 INAA 1.0Rb 10 100 38 22 32 INAA 1.0Sb 0.2 2.3 0.6 0.4 0.5 INAA 0.005Sc 0.7 15 3.6 2.7 2.9 INAA 0.01Th 0.7 19 5.5 4.2 4.1 INAA 0.1U 0.4 5 2 1.2 1.6 INAA 0.01Zn 56 221 114 33 114 INAA, ICP 1.0Cu 34 843 124 137 94.4 ICP 1.0Pb 9 174 39 34 30 ICP 5.0Ni 2 40 8 6 6 INAA, ICP 1.0Mn 46 900 273 193 209 ICP 1.0Sr 22 653 192 154 134 ICP 1.0V 6 122 39 24 31 ICP 2.0P 730 1890 1251 302 1215 ICP 0.00%Mg 800 15 700 4288 2995 3354 ICP 0.01%Ti 100 2200 926 521 745 ICP 0.01%Al 3100 21 400 9052 4516 15 758 ICP 0.01%Y 3 64 15 11 12 ICP 2.0

Ž .Mean, maximum, minimum and geometric mean are given in ppm. Analytical techniques and detection limits ppm are also listed.

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9`4

Table 2Correlation matrix for selected elements

Ba Sr Th U Mn V Mg Ti Al Hf

Ba 1Sr 0.78 1Th 0.76 0.54 1U 0.88 0.67 0.92 1Mn 0.80 0.57 0.35 0.57 1V 0.79 0.64 0.45 0.56 0.85 1Mg 0.78 0.63 0.41 0.55 0.84 0.98 1Ti 0.77 0.57 0.52 0.61 0.72 0.87 0.86 1Al 0.90 0.69 0.69 0.78 0.82 0.92 0.91 0.91 1Hf 0.90 0.68 0.92 0.95 0.58 0.62 0.59 0.64 0.82 1

ferently orientated thalli were collected from eachsampling station. No Parmelia thalli were foundnear the crater area. Each sample, after beingdried naturally or in an oven at 408C, was care-fully separated from the substrate particles stick-ing on it, using wooden sticks under a low magni-fication stereo microscope. This was made in or-der to reduce soil contribution to the metal con-tent in lichens. Samples were then pulverized and

Ž .analyzed by the Activation Lab. LTD Canadausing INAA and ICP: certified reference materi-als NBS 1572 and NBS 1632 B were used asstandard reference materials. The analytical tech-nique and detection limits for each element areindicated in Table 1. The level of possible error,considering the sample homogeneity and the ana-lytical procedure, is less than 20% for trace ele-ments and much less for the main components.Sampling sites are shown in Fig. 1.

4. Results and discussion

Table 1 presents a statistical evaluation of theresults of chemical analyses on the lichens fromVulcano island. Eight elements, Al, Ca, Fe, Mg,K, Na, Ti and P, make up more than 95% inweight of all elements analysed. Some elementsshow statistically significant linear correlationsŽ .Table 2 , indicating a common origin. Their ra-tios, which are constant and comparable withthose of the surrounding rocks, also indicate thatthose rocks were the primary source of thetrapped material.

The abundance of REE in the analyzed sam-

ŽFig. 2. Average REE distribution in the main lithotypes sub-.strate of Vulcano and in the lichen samples, versus atomic

number. Data are normalized to REE abundances in chon-drites.

ples provides further information to validate suchhypothesis as their abundance ratios reflect thepetrogenetic characteristics of the rocks. This isdue to the fact that petrological and mineralogi-cal processes may fractionate these elements, onefrom the other, despite their similarity in chemi-

Ž .cal behaviour Henderson, 1984 .REE’S identify a particular rock type and may

be used as markers to determine from whichgeological source or geographic area the mostabundant trapped material in lichens might come.The underlying assumption is that the soil]rocksystem has a unique composition regarding theREE, which is not always true.

On this base, Fig. 2 compares the average dis-tribution of rare earths in some of the mainlithotypes of Vulcano and in the lichen samples.It is clear that the particulate content in thelichens is the result of trapping local dust andthat any difference in the content of an elementwith respect to the substrate must be attributedto a source different from the material making upthat substrate. Data interpretation must thereforetake into account the fact that crustal materialhas a considerable influence on the total contentsof elements in lichens.

As we are dealing with a process of bioaccumu-lation, absolute concentration is not a particularlysignificant variable. In this paper metal contentsin lichens were examined by transforming the

Ž .original data into enrichment factors EFs bymeans of the following algorithm:

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9` 5

Table 3Statistical parameters of the EFs, together with the percent-age of samples with EF)3

Enrichment factors

Min Max Mean S.D. %

Au 0.6 35.3 8.1 9.3 33As 0.5 8.7 2.8 1.9 33Ba 0.5 2.0 1.0 0.3 0Br 11.6 205 67.0 56 100Ca 0.4 9.7 1.7 1.7 21.4Co 0.6 3.7 1.2 0.6 2.4Cr 0.4 2.5 0.9 0.5 0Cs 0.4 5.5 1.5 0.8 2.3Fe 0.4 1.8 0.5 0.5 0Hf 0.6 2.6 1.3 0.5 0K 0.6 4.7 1.5 0.7 4.7Na 0.2 1.2 0.5 0.2 0Ni 0.4 13.6 2.0 2.0 11.9Rb 0.3 2.4 1.1 0.5 0Sb 1.1 81.9 9.2 15.3 66Sc 0.4 2.5 0.9 0.5 0Sr 0.3 5.2 0.7 0.7 2.3Th 0.4 1.9 0.9 0.4 0U 0.5 2.5 1.2 0.5 0Zn 1.8 31.3 7.7 6.1 88Cu 1 21 4.2 3.9 45Pb 2 85 13.0 14.7 85Mn 0.7 2.6 1.1 0.4 0V 0.6 1.2 0.9 0.2 0Mg 0.4 1.1 0.8 0.2 0Ti 0.7 2.0 1.1 0.3 0Y 0.9 7.6 2.8 1.6 26P 1.2 21.5 5.2 4.6 54.7Mo 0.2 28.4 3.8 5.0 33.3

Ž . Ž . Ž .EFs XrAl r XrAl 1lich sub

where X is the concentration of the element ofinterest and Al the reference element for normal-ization. The enrichment factor represents theconcentration ratio of normalized elements in thelichens and in the substrate of their area ofgrowth. In theory, an EFs1 indicates no enrich-ment, while any value over one should indicateenrichment of element X in the lichen with re-spect to its presence in soil. In order to minimize

Ž .uncertainty in the denominator of Eq. 1 , theorigin of the particulate matter accumulating inlichens has to be clearly established.

Considering the possible analytical error and

the uncertainty inherent in the denominator ofŽ .Eq. 1 , we prefer to consider as globally en-

riched, with respect to soil, those elements bothwith average EFs)3 and, for a greater signifi-cance, with Efs)3 in at least 30% of the ana-lyzed samples. Table 3 shows some statisticalparameters of the EF, together with the percent-age of samples with EF)3. Bromine, Pb, Sb, Zn,Au and Cu resulted enriched in a high percentageof samples exceeding the cut-off value. The aver-age EF of As is very close to three, but there are

Ž .14 samples 33% of total which exceed three,hence there is really no doubt that As may beconsidered enriched in the Vulcano samples. HighEF values for the various elements were observedin the same lichen sample, indicating a simultane-ous enrichment.

The EFs for Br and Pb are particularly rele-vant, being among the highest of all elementsmeasured. The mean concentrations of Pb and Brare, respectively, 38"34 ppm and 37"18 ppm.Their frequency distribution is shown in Fig. 3;the data clearly fit better in a log-normal distribu-tion. The distribution maps of the enrichment

Ž .factors Fig. 4 show how the entire island issubject to fallout by these two elements. On Vul-cano island, Pb was found in some sublimateswhich formed between 1924 and 1926, when thefumaroles reached a temperature of approx. 6008C

Ž .and recently 1990 onwards again as a conse-quence of a rise in fumarole temperatures to

Ž500]6008C Garavelli et al., 1993; Garavelli, 1994;.Ferrara et al., 1995 . Minerals identified in the

Ž .sublimates include cannizzarite Pb Bi S ,46 54 127Ž .galeno-bismuthite PbBi S and cotunnite2 4

Ž . Ž . Ž .PbCl Martini et al., 1988 . Martini et al. 19882estimate the release of Pb from the fumaroles ofCratere della Fossa at 10y2 tonsryear.

Bromine was found in NH Cl sublimates4ŽCoradossi and Maleci, 1972; Coradossi et al.,

.1985, 1996 and in fumarole condensates. Brominehas no significant anthropic sources, except forthat which, together with lead, is produced ingasoline combustion. Fig. 5a shows the relation-ship between the PbrBr ratio and the total Pbcontent in Vulcano lichens. It also shows, forcomparison, the areas identifying the typical

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9`6

Ž .Fig. 3. Frequency distributions of Pb and Br in the analysed lichen samples data in ppm .

PbrBr ratios in combustion products and in gasesemitted from other volcanoes in the world. It isclear that, as well as from the soil, contributionfrom the crater fumaroles is significant. The an-thropic contribution due to burning of gasoline islimited to the samples collected along the main

Ž .roads Fig. 5b .Average absolute concentrations of Sb and As

are 0.6"0.4 ppm and 4.6"2.5 ppm, respectively.Fig. 6 show that these elements are preferentiallyaccumulated in an E]SE direction, following pre-vailing winds. In their native state, arsenic andantimony have boiling temperatures of 7008C and15008C, respectively, but they are highly volatileas chloride compounds, with boiling temperatures

Ž . Ž .of 1308C AsCl and 2198C SbCl . Their3 3volatilization as chlorides was suggested by Pic-

Ž .cardi et al. 1979 on the basis of a study on thefumarolic gases of Vulcano. Although the AsrBrcorrelation suggests that As may also have beentransported as a bromide, the role played bybromides may have been only subordinate, sincethe ClrBr ratio is amply in favour of chloride. Itshould also be recalled that As and Sb are volatileas sulphides.

Among the most common trace elements in theparticulate matters from volcanic plumes and hightemperature fumaroles are Au, Cu and Zn. Yet,these elements show high enrichment factors inthe Vulcano lichens. Specifically, a diffuse accu-mulation of Zn was observed, while gold andcopper were prevailed to the east and south-eastŽ .Fig. 7 . In their study on soils and vegetation of

Ž .Vulcano and Stromboli, Bargagli et al. 1991found anomalies in Cu and Zn, pointing to thevolcanic activity as the only source of these met-als.

5. Concluding remarks

The data presented here, although derived froman indirect method of monitoring, demonstratehow volcanic activity is responsible for the releaseof Pb, Br, Sb, As, Cu, Zn and Au. It is worthnoting that these trace elements are the same ofthose observed in fumarolic gas from differenttype of volcanoes around the world.

Gas emissions from the Cratere della Fossafumaroles essentially come from two sources: adegassing magma; and the overlying hydrother-

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9` 7

Fig. 4. Graphical representations of the enrichment factors ofPb and Br.

mal system. Acid gases such as HCl, HF, H S,2SO and CO , whether produced by the water]2 2rock interaction or directly by degassing magma,may leach metals from rocks of the volcano. Mi-gration of metal-bearing compounds finds an ef-ficient support in the hydrothermal system, as anaccumulation reservoir and a source of chloride

Ž . Ž .Fig. 5. a Relationship between the PbrBr ratio ppmrppmŽ .and the total Pb content ppm in Vulcano lichens; also

shown, for comparison, the areas identifying the typical PbrBrratios in combustion products and in gases emitted from other

Ž .volcanoes in the world. b Pb]Br scatter diagram. The straightline indicates their most common ratio in gasoline. Dotsindicate the same sample points after correction for sea water

Ž .contribution by the equation: Br sBr y BrrNa Na .corr tot sea tot

species, which facilitate enrichment and volatiliza-tion and also as solutions which leach metalsfrom the host rocks.

The data also confirm that volcanoes are animportant source of release of trace elements tothe atmosphere, not only during the eruptivephases, but also during normal fumarolic activity.

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9`8

Fig. 6. Graphical representations of the enrichment factors ofAs and Sb.

In particular, if emissions occur at relative lowaltitudes, in conditions of atmospheric instability,some metals may accumulate in vegetation andsoil, representing a potential hazard for grazinganimals, agriculture and groundwaters.

If the preceding considerations are confirmedby future studies on the content and speciation ofmetals in volcanic emissions, such metals could beused to monitor variations in magma degassing

Fig. 7. Graphical representations of the enrichment factors ofCu, Zn and Au.

( )G. Dongarra, D. Varrica r The Science of the Total En®ironment 212 1998 1]9` 9

and for the volcanic surveillance. The possibilityof positioning a monitoring network at safe dis-tance from the point of emission would also beuseful.

These data may also serve to model the disper-sion into the surrounding environment of ele-ments and compounds contained in volcanic gases,improving current models, the formulation ofwhich is extremely difficult due to the reactionand dilutions which gaseous species undergo.Metals, instead, may constitute excellent markers.

Lastly, these results show how useful lichensmay be in examining the airborne particulatefound in volcanic areas. They have definite advan-tages in this case and are equally useful in studieson the environmental impact of industrial emis-

Ž .sions Dongarra et al., 1995 and in geochemical`Ž .prospecting Dongarra et al., 1996 .`

Acknowledgements

This research was supported by the Italian Na-Ž .tional Council of Research CNR } Istituto di

Geochimica dei Fluidi, which the Authors wish tothank. Financial support was also provided by

Ž .MURST funds 60% .

References

Alvarez LW, Alvarez W, Asaro F, Michel HW. Extraterres-trial cause for the Cretaceous-Tertiary extinction. Science1980;208:1095]1108.

Bargagli R, Barghigiani C, Siegel BZ, Siegel SM. Trace metalanomalies soils and vegetation on two active island vol-

Ž .canoes: Stromboli and Vulcano Italy . Sci Total Environ1991;102:209]222.

Buat-Menard P, Arnold M. The heavy metal chemistry of´atmospheric particulate matter emitted by Mount Etnavolcano. Geophys Res Lett 1978;5:245]248.

Coradossi N, Maleci M. Il rapporto BrrCl in alcuni prodottidi esalazione vulcanica. Rend Soc Ital Mineral Petrol1972;28:53]76.

Coradossi N, Pinarelli L, Bertolini GL. Fluttuazioni del rap-Žporto BrrCl nelle esalazioni fumaroliche di Vulcano Isole

.Eolie ed in alcuni prodotti delle loro reazioni. Rend SocItal Mineral Petrol 1985;40:299]309.

Coradossi N, Garavelli A, Salamida M, Vurro F. Evolution ofBrrCl ratios in fumarolic salammoniac from VulcanoŽ .Aeolian Island, Italy . Bull Volcanol 1996;58:310]316.

Dongarra G, Ottonello D, Sabatino G, Triscari M. Use of`lichens in detecting environmental risk and in geochemicalprospecting. Environ Geo 1995;26:139]146.

Dongarra G, Ottonello D, Sabatino G, Triscari M. Disper-`sione di metalli in traccia in prossimita di centrali ter-`moelettriche. Acqua Aria 1996;10:873]878.

Ferrara G, Garavelli A, Pinarella L, Vurro F. Lead isotopecomposition of the sublimates from the fumaroles of Vul-

Ž .cano Aeolian Island, Italy : inferences on the deep fluidcirculation. Bull Volcanol 1995;56:621]625.

Fiocco G, Visconti G. Gli effetti climatici delle eruzionivulcaniche. Le Scienze Quaderni 1985;93:38]43.

Ganapathy R. A major meteorite impact on the Earth 65million years ago: evidence prom the Cretaceous-Tertiaryboundary clay. Science 1980;209:921]923.

Garavelli A. Mineralogia e geochimica di fasi vulcaniche con-densate. I sublimati dell’isola di Vulcano tra il 1990 ed il1993. PhD Thesis, Univ. Bari, Italy, 1994.

Garavelli A, Garbarino C, Laviano R, Vurro F. Solfuri esolfosali tra i prodotti di deposizione dell’area craterica di

Ž .Vulcano } Isole Eolie ME . Plinius 1993;10:161]162.Henderson P, editor. Rare earth element geochemistry. N.Y.:

Elsevier, 1984.Keller J. The island of Vulcano. Rend Soc It Mineral Petrol

1980;36:369]414.Lantzy RJ, Mackenzie FT. Atmosferic trace metals: global

cycles and assessment of man’s impact. Geochimica Cos-mochim Acta 1979;43:511]525.

Martini M, Capaccioni B, Giannini L. Past and present influ-ence of volcanic systems on the surface environments:Vulcano and Lipari, Eolian islands, Italy. Boll. GNV1988:383]391

Mroz EJ, Zoller WH. Composition of atmosferic particulatematter from eruption of Heimaey, Iceland. Science1975;190:461]464.

Piccardi G, Martini M, Legittimo Cellini P. On the presenceof Cu, Zn, Cd, Sb, Bi and Pb in the fumarolic gases of

Ž .Vulcano Aeolian Island . Soc Ital di Min e Petr1979;35:627]632.

Schneider SH. Volcanic dust veils and climate:how clear is theconnection?, an editorial. Climatol Crouge 1983;5:111]113.

Sicardi L. Il recente ciclo dell‘attivita fumarolica dell’isola di`Vulcano. Bull Vulcanol 1941;7:85]140.

Stommel H, Stommel E. The year without a summer. Sci Am1979;240r6:134]140.