J. geol. Soc. London, Vol. 140, 1983, pp. 939-944, 2 figs. Printed in Northern Ireland.

The mechanism of Surtseyan

B. P. Kokelaar

SUMMARY: During eruption, the vent of a Surtseyan volcano is occupied by a highly mobile slurry of , hyaloclastite and water. As ascends rapidly through the slurry, mixing occurs due to velocity shear, acceleration of the fluid-fluid contacts and mass incorporation. Consequent expansion of the admixed water causes the jetting and continuous up-rush activity that characterizesSurtseyan volcanism. There is no evidence of the widely accepted fuel-coolant interaction mechanism.

Basaltic volcanic eruptions occurring at an air-water thus permanently secured the vents from the influx of interface are characterized by violent explosive activity sea water. Further Surtseyan explosive activity occur- which clearlyresults from interaction of the magma red with the temporary emergence of volcanic piles at and water. Such activity occurred during the emerg- the satellite vents of Syrtlingur (May to October 1965) ence, in 1963, of thesubaqueous volcanowhich and Jolnir (December 1965 to August 1966). During formedthe new island named (63'18'N the activity of Surtur 1, Surtla, the most distal satellite 20"36'W), in the Westmann Island group off southern vent, built a pile to within 5 or 6m of sea level . The activity has since been termed Surtseyan. (December 1963 toJanuary 19641,with onlyminor There have been several attempts (Saemundsson1967; ejections of tephra above sea level. From an examina- Jones 1970; Bennett 1972; Tasieff 1972; Peckoveret al. tion of Surtla,Kokelaar & Durant (1983) deter- 1973) to explain the mechanism of the magma-water mined eruption processestransitional from purely interaction, but these are irreconcilable with descrip- subaqueous to truly Surtseyan. tionsand films of Surtseyan activity (Thorarinsson 1965, 1966, 1967; Thorarinsson et al. 1964; Knudsen, undated a, b). Here, the course of events and salient Characteristics of Surtseyan features of the type eruption are described, previously volcanism proposed mechanisms are critically appraised, and an alternative mechanism consistent with observations is The following observations derive primarily from the proposed. activity atSurtsey, but are applicable toother Surtseyaneruptions, suchas those atCapelinhos The Surtsey eruption (, 1957-8) (Machado et al. 1962;Machado & Forjaz 1968). Theemergence of Surtsey,during the night of Surtseyan explosive activity results when water gains November 14-15 1963, was preceded by approximate- access tothe top of thevent. When water is ly 1 week of submarine volcanic activity when completely and lastingly excluded the explosions soon predominantly clastic deposits(Jakobsson & Moore become comparatively mild and due solely to escape of 1980) gradually built up from Iceland's insular shelf, volatiles from within the magma. The initial exclusion 130 m below sea level. Initial Surtseyan activity, along of water results from enclosureof the vent by a ring of a short NE-trending fissure, built an elongate islet of tephra,although suchenclosure may not result in tephraabout mediana line of submarinevents. immediateexclusion. Film of theeruption of Jolnir However, activity soon becamecentred and a (Knudsen, undated 6) shows water flooding through a horseshoe-shaped island was formed.The vent, named permeable, albeit narrow, tephra embankment. Clear- Surtur l, was almost continuously open to the sea in ly suchaccess would be less likely through older the SW and Surtseyan explosions occurred there until andior wider embankments.Temporary enclosure, January 31 1964. On February 1, a second vent, Surtur resulting in restricted access or temporary exclusion of 2, was initiated 500 mNW of Surtur 1 and, as it too water, is commonly coincident with increased effusion wasmostly opentothe sea, Surtseyan activity rate and increased violence of eruption. Where there continued. However, on April 4 1964, the growingpile is considerable marine erosion, as at Surtsey, a high of tephra persistently enclosed the vent and seawater rate of effusion is required to maintain the tephra ring was atlast completely excluded. In response, the and hence the exclusion of water. violentlyexplosive Surtseyan activity gave way to Surtseyanexplosions arecharacterized by jets of comparatively mild fountaining of incandescent mag- tephra.These are mostlyshort-lived, occurring at ma(Hawaiian activity). Lowviscosity , flowing various frequencies, commonly up to 1 per second, or from a small lava , formed a carapaceto theclastic can combine into a continuous up-rush. At Syrtlingur deposits around the southern part of the islandand (Thorarinsson 1966) the phases of continuous up-rush 0016-764C'~7/110C-0939$02.00 63 1983 The Geological Society

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/140/6/939/4887762/gsjgs.140.6.0939.pdf by guest on 23 September 2021 940 B. P. Kokelaar

were preceded by increasing frequency and violence of withhyaloclastites, grading from fine atthe top to discretejets. Discrete jets are almost silent and coarseatthe base, underlain by water-enriched continuousup-rush emits a muted rumbling noise. magma and then normal magma. This sequenceis held Violentjets can be accompanied by basesurges in a condition of subdued boiling until the magma next (Moore 1967; Waters & Fisher 1971). movesupwards. Bennett considered that, when this Commonly,successive jets of tephra showa wide occurs, the resulting decrease in hydrostatic pressure variation in direction,and displayaspreading or causes part or allof thesequence to boil violently, non-spreading form (likened to cocks’ tails and cypress thus resulting in aneruption with sequentiala treesrespectively). Bombs, with black,comet-like appearance of ejecta. However, this model does not tails, shoot out of eachblack, tephra-laden jet, and explain the high frequency jets or continuous up-rush, then the whole turns grey and white as the associated as the processes it requireswould not have time to steamcondenses. At night, the columns associated occur. Also, it would be impossible to obtain a column with continuous up-rush activity can be seen to contain of ‘clean’ water in contact with the magma in a vent, alarge proportion of incandescentmaterial. Coarse becausepoorly sorted clastic deposits would always tephra can be ejected to heights of over 1 km and fine slump into such a hole, so a layered sequence could tephra can becarried far higher by convective not develop in the manner proposed. Theoretically, a processes,particularly during a continuous up-rush. graded clastic sequence could result from sorting in a However, a considerable proportion of the ejecta falls steam fluidized system within a well-defined cylindrical within thecrater rim and, by directfall, by mass- pipe, but the vent is surrounded by unstable deposits, gravity flow, or by sliding, is returned to the vent. is not well defined and, again, the process would take In the quiescent intervals during intermittent jetting moretime than is generallyavailable. Whatever the activity, thevent area is mostlyoccupied by avery mechanism, it must be capable of high frequency and shallow lagoon, violentlyagitated near its centre by continuousaction. Close to sea level, magma will the release of bubbles of steam. The steam condenses always have to pass through a poorly-sorted slurry of abovethe lagoon to form a dense, white billowing tephra, hyaloclastite and water. column.Such columns are commonly generated Tazieff (1972) considered that, following an initial continuously during jetting activity. subaqueous magmatic explosion, asuccession of steam explosionsresults fromthe entrapment of steam within or beneath fragments of incandescent magma The Mechanism (someassuming an ‘umbrellashape’) as they are hurled upwards through water. However, in Surtseyan Saemundsson (1967) and Jones (1970) suggested that ventsthere is no simplemagma-water interface as Surtseyanexplosive activity mightbe dueto water envisaged and, where incandescent basaltic magma is beingdrawn intothe volcano’splumbing system, explosively eruptedinto water, the steam initially through the subaqueousflanks of the pile, by a process produced is more likely to form an envelope around akin to Venturi action. If this was so, the characteristic the fragments and the vent, and tends to isolate the activity would continue irrespective of the exclusion of magma from further explosive interaction with water. water from the top of the vent, which is not the case. In support of this postulation, at Surtla, surfaces of a Althoughthe pile is permeableand water-saturated contortedslab of basaltic spatterare bread-crusted (Jakobsson & Moore 1980),this water evidently has and agglutinated with smaller lapilli of (Koke- littleor no role in Surtseyanactivity, and at deeper laar & Durant, 1983). levels of the pile is probably mostly isolated from the The mostfavoured published explanation of Surt- magma by chilled basalt lining the conduit(s). Also, in seyan activity was partly derived from an attempt to clastic deposits close to any long established conduit, solvea problem caused, coincidentally, by another continuous vaporization of water will cause precipita- eruption in the Westmann Island group, that of Eldfel tion of salts, mostly NaCl, which will markedly reduce on the island of Heimaey(Colgate & Sigurgeirsson the permeability. 1973). There, in 1973, a subaerial Strornbolian cinder Fromexamination of still photographs,Bennett cone partly buried the town and was breached by lava (1972) developeda model for Surtseyan activity, to flows which extended into the sea and threatened to accountfor ‘distinctive colourdifferences in the block the entrance to the harbour. It was thought that expanding cloud indicativeof contact surfaces between the submarine flow might be halted if explosives were differentsubterranean fluids initially in the volcanic used to fragment the lava, thus allowing sea water to pipe’. In this model,water enters the volcanicpipe cool it rapidly.Shortly before taking this courseof (vent)and encounters stationary magma at several action, it was realized that if the instantaneous heat pipe diameters below theambient water surface. exchange, between the lava (still incandescent beneath Here, asresulta of variousprocesses, including a thin skin) and water, caused steam explosivity a little quench granulation, turbulent boiling and convective more violent thanthe initial explosion,then a self heattransfer, a crudely layered sequence develops sustaining mixing of lava and water could take place

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/140/6/939/4887762/gsjgs.140.6.0939.pdf by guest on 23 September 2021 Mechanism of Surtseyan volcanism 941 wherein the thermal energy of the whole submarine accelerating upwards, is confronted by, and has to pass portion of the lava might largely be released explosive- through,a highlymobile fluid. Atcontacts between ly. possibleA explosion of 2-4megatons was the two fluids mixing takes place, as aresult of velocity calculated and the project was abandoned. shear(Helmholtz instability), acceleration of the Such self-sustaining mixing and explosivity (known fluid-fluid boundaries(Rayleigh-Taylor instability) to occurin industrial accidents) is referred to as and mass incorporation as the magma follows irregular fuel-coolant interaction,and Peckover et al. (1973) pathsthrough its turbulent host. Inevitably, some proposed that this is the likely mechanism of Surtseyan liquid water is enclosed with the clastic deposits in the volcanism. However, explosionsresulting from fuel- hot ascending magma, where it rapidly expands as it coolant interactions develop extremely rapidly (a few flashes tosteam. Rapid expansion of incorporated tens of milliseconds in metal-water explosions (Bucha- steamresults mainly from pressure relief during nan & Dullforce 1973)), and are characterized by loud ascent,but also from heating. There is noinstan- detonations, which are difficult to reconcile with the taneous explosion, but rather a violent and continuous almostsilent jets of Surtseyan activity. Also, the expansion that is manifested as a jet of tephra. As the mechanismrequires that magmacomes intocontact jets are erupted through a highly mobile medium they with liquid water, ideally at least in the ratio = liquid can be variably directed, although spreading jets are waterhagma (Colgate & Sigurgeirsson 1973; see also probably initiated at higher levels than non-spreading Sheridan & Wohletz1981). In intermittent jetting, jets. The first material to appear in a jet is the fine eachpulse of magmaascends through a mixture of tephra, which is more readilyaccelerated by the basalt clasts, steam and liquid water, and it is unlikely expandingsteam. However, as the rate of steam that there is enough liquid water for interaction of the expansiondiminishes abovethe vent, the coarser fuel-coolant type. During the more violent continuous tephra, withits greatermomentum, overtakes the up-rush activity, liquid water is probably even scarcer. finer, thus producing the characteristicspiky cock’s tail Furthermore,self-sustaining fuel-coolant interaction and cypressoidplumes. This sequence results from needsinitiation by anevent that causes mixing of sorting during jetting, and not sorting before eruption magma andwater such that heat is exchanged as envisaged by Bennett (1972). extremely rapidly. Such efficient exchange requires the With a high rate of magma supply, this mechanism magma to have a particle diameter of around 0.12 cm operatescontinuously, producing the continuous up- or less and, although it is cited (Colgate & Sigurgeirs- rush. As a resultof instability of the vent sides and the son 1973) that Surtseyan tephra is of this grade, this is return of tephra, it is impossible to clear the vent of theresult of theexplosion, and it isby nomeans clastic material,and whilewater is present this certain that the early disruptionby magmatic volatiles, materialremains ahighly fluid medium whichis quench granulation and steam generation, taking place continuouslyincorporated into the magma rushing within thelower part of thevent, are sufficient to through it. initiateself-sustaining fuelkoolantinteraction. Thus Since so much tephra returns to the vent, it seems there is no direct evidence of fuel-coolant interaction likely that much of the fine comminution of clasts, in Surtseyanactivity, and observations are either characteristic of Surtseyan tephra, (Walker & Croas- irreconcilable with the mechanism or indicate that the dale1972) results from repeated involvement in required conditions are unlikely to exist. explosive events. Figs 1 and2illustrate an alternative model of hasIt been stated (Thorarinsson et al. 1964; conditions and processes in Surtseyan volcanism. It is Williams & McBirney 1979) that when access of water envisaged that the lagoon within the ring of tephra is tothe vent is impeded,the activity changesfrom the topof a funnel-shaped ventfilled with a potentially intermittent jetting to continuous up-rush. The com- mobile slurry of basaltic clasts (tephra and hyaloclas- moncoincidence of the two phenomena is not tite) and water. The funnel walls are ill-defined and disputed,but the possibleimplication of causeand susceptible to collapse. effect is questionable.Records (Thorarinsson 1966; Pulsatingeruption is characteristic of low rates of Thorarinsson et al. 1964) showthat continuous up-rush explosive effusion of basalt, and the frequency of jets followsa shortperiod of increasingfrequency and in Surtseyan activity reflects therate of supply of violence of discretejets, and is notablefor rapid magma into the vent funnel. Between successive jets accumulation of tephra. Such activity commonly duringintermittent activity, steamgenerated around ceasesabruptly and is followed by perioda of the magma, temporarilylying at some depthwithin the quiescence. This pattern must be a direct response to vent,streams turbulently tothe surface. Thus, the an increasing and then terminated supply of magma. slurryabove the magma is probablyat least partly The increasingeffusion rate causesring closure and fluidizedand the surroundingliquid water must be impeded influx of water,rather than the reverse. close tothe boiling temperatureappropriate to the Although the violence of Surtseyan activity must be a depth.As a result, the nextpulse of lowviscosity function of a number of variables, it is likely that, with magma,rapidly expanding dueto vesiculation and other variables held constant, violence will be greatest

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/140/6/939/4887762/gsjgs.140.6.0939.pdf by guest on 23 September 2021 942 B. P. Kokelaar

-I m I I a z 0 U 0 a I 0 W a 0 U U m + 5 Q 0 n

Ua

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/140/6/939/4887762/gsjgs.140.6.0939.pdf by guest on 23 September 2021 Mechanism of Surtseyan volcanism 943

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/140/6/939/4887762/gsjgs.140.6.0939.pdf by guest on 23 September 2021 944 B. P. Kokelaar when all the water in the vent is close to the boiling normally effectively trapped and because of this the temperature, and diminished where energy is lost in associated volcanic explosions can be extremely vio- heatingsurplus cold water. Thus, increased violence lent. If nofresh magma is erupted, such violent results when wateraccess is impeded. With permanent explosions are correctlyreferred to as phreatic or exclusion of water,the slurry in the vent gradually steam blast eruptions. When juvenile magma is also driesand the eruptive stylechanges to Hawaiian ejected, the eruption can be termed phreatomagmatic. fountaining of incandescent magma. Because the wateris trapped, the eruption mechanism It is apparentthat at present ‘phreatomagmatic and effects are quite different from those of Surtseyan eruption’ is widely considered to be synonymous with eruptions, which therefore should not be referred to as ‘Surtseyan eruption’ (e.g. Williams & McBirney 1979), phreatomagmatic. but it should not be. ‘Phreatic’ denotes ground water (in the strictest use, meteoric in origin), but the water in the vent filling slurry clearly originates by flooding ACKNOWLEDGMENTS.The author is grateful to Drs M. F. over the topof the vent and can hardly be described asHowells and W. J. Phillips for their critical appraisal of the such.More importantly however, ground water is manuscript, and also to Jennifer Larkin for typing it.

References BENNET, F. D. 1972.Shallow submarine volcanism. J. 1973. Fuel-coolant interactions in submarine vulcanism. geophys. Res. 77, 5755-9. Nature, London, 245, 307-8. BUCHANAN,D. J. & DULLFORCE,T. A. 1973. Mechanism for SAEMUNDSSON,K. 1967.In discussion of Kjartansson G. vapour explosions. Nature, London, 245, 32-4. 1967. Volcanic forms at the sea bottom. In: BJORNSSON, COLGATE,S. A. & SIGURGEIRSSON,T. 1973. Dynamic mixing S. (ed.). Iceland and mid-oceanridges. SocietasScien- of lava and water. Nature, London, 244, 552-5. tiarum Islandica, 38, 65-6. JAKOBSSON,S. P. & MOORE,J. G. 1980.Through Surtsey: SHERIDAN,M. F. & WOHLETZ,K. H. 1981.Hydrovolcanic Uniquehole shows how volcano grew. Geotimes, 25 explosions:TheSystematics water-pyroclastof (April), 14-16. equilibration. Science, 212, 1387-9. JONES, J. G. 1970. Intraglacial volcanoes of the Laugarvatn TAZIEFF,H. 1972. About deep-sea volcanism. Geol. Rdsch. region,southwest Iceland, 11. J. Geol.Chicago, 78, 61, 470-80. 127-40. THORARINSSON, S. 1965.The Surtsey eruption: Course of KNUDSEN,0. undated a. Birth of an island (sound-motion events and the development of the new island. Surtsey film), V6k-Film, Reykjavik. Research Progress Report, 1, 51-5. - undated b. Sequel to Surtsey (sound-motionfilm), - 1966. The Surtsey eruption: Course of events and the V6k-Film, Reykjavik. development of Surtsey and other new islands. Surtsey KOKELAAR,B. P. & DURANT,G. P. 1983. The Research Progress Report, 2, 117-23. submarineeruption and erosion of Surtla(Surtsey), -1967. The Surtsey eruption: Courseof events during the Iceland. J. Volcanol. Geotherm. Res. (in press). year 1966. Surtsey Research Progress Report, 3, 84-92. MACHADO,F. & FORJAZ, V. H. 1968. Actividade Vulcdnica -, EINARSSON,T., SIGVALDASON,G. & ELISSON,G. 1964. da IIha do Faial (1957-58): Horta, AGores. Commisao de Bull. Volcanol. 27, 435-45. Turismo da Horta. WALKER,G. P. L. & CROASDALE,R. 1972. Characteristics of -, PARSONS,W. H., RICHARDS,A. F. & MULFORD,J. W. some basaltic pyroclastics. Bull. Volcanol. 35, 303-17. 1962. eruption ofFayal Volcano,Azores, WATERS,A. C. & FISHER,R. V. 1971. Base surges and their 1957-58. J. geophys. Res. 67, 3519-29. deposits:Capelinhos and TaaL Volcanoes. J. geophys. MOORE,J. G. 1967. Base surge in recent volcanic eruptions. Res. 76, 5596-614. Bull. Volcanol. 30, 337-63. WILLIAMS,H. & MCBIRNEY,A. R. 1979. Volcanology. PECKOVER,R. S., BUCHANAN, D. J. & Asnwt, D. E. T. F. Freaman,Cooper & Co., SanFrancisco, CA 94133, USA.

Received 25 January 1983; revised typescript received 16 May 1983. B.PETER KOKELAAR, School of EnvironmentalSciences, Ulster Polytechnic Newtownabbey, BT37 OQB.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/140/6/939/4887762/gsjgs.140.6.0939.pdf by guest on 23 September 2021