Journal of Volcanology and Geothermal Research 102 (2000) 237–270 www.elsevier.nl/locate/jvolgeores Effusion rate trends at Etna and Krafla and their implications for eruptive mechanisms A.J.L. Harrisa,b,*, J.B. Murraya, S.E. Ariesa, M.A. Daviesa, L.P. Flynnb, M.J. Woosterc, R. Wrighta, D.A. Rotherya aDepartment of Earth Sciences, The Open University, Milton Keynes, MK7 6AA, UK bHawaii Institute of Geophysics & Planetology (HIGP)/School of Ocean & Earth Science Technology (SOEST), University of Hawaii, 2525 Correa Road, Honolulu, HI 96822, USA cDepartment of Geography, King’s College London, Strand, London, WC2R 2LS, UK Received 23 July 1999; received in revised form 23 February 2000; accepted 23 February 2000 Abstract Using effusion rates obtained from ground- and satellite-based data we build a data set of 381 effusion rate measurements during effusive activity at Etna and Krafla between 1980 and 1999. This allows us to construct detailed effusion rate curves for six fissure-fed eruptions at Etna and Krafla and four summit-fed eruptions at Etna. These define two trends: Type I and II. Type I trends have effusion rates that rise rapidly to an initial peak, before declining more slowly, resulting in an exponential decrease in eruption rate and declining growth in cumulative volume. Type II trends are characterised by steady effusion and eruption rates, and hence a linear increase in cumulative volume. The former is typical of fissure eruptions and can be explained by tapping of an enclosed, pressurised system. The latter are typical of persistent Etnean summit eruptions, plus one persistent effusive eruption at Stromboli (1985–1986) examined here, and can be explained by overflow of the time-averaged magma supply. We use our effusion rate data to assess the magma balance at Etna (1980–1995) and Krafla (1975–1984). Between 1980 and Ϫ 1995, Etna was supplied at a time-averaged rate of 6:8 ^ 2:3m3 s 1 of which 13% was erupted. At Krafla 817 ^ 30 × 106 m3 was erupted and intruded during 1975–1984, and the ratio of erupted to intruded volume was 0.3. At Etna there is evidence for intrusion of the unerupted magma within and beneath the edifice, as well as storage in the central magma column. At Krafla unerupted magma was intruded into a rift zone, but an increasing proportion of the supply was erupted from 1980 onwards, a result of the rift zone capacity being reached. Magma intruded prior to an eruptive event may also be entrained and/or pushed out during eruption to contribute to the initial high effusion rate phases of Type I events. The detail in our effusion rate curves was only possible using a thermal approach which estimates effusion rates using satellite data. We look forward to analysing satellite-derived effusion rate trends in real-time using data from current and soon-to-be- launched sensors. ᭧ 2000 Elsevier Science B.V. All rights reserved. Keywords: effusion rate; Etna; Krafla 1. Introduction Detailed time series of lava effusion rates can serve * Corresponding author. Tel.: ϩ1-808-956-3157; fax: ϩ1-808- 956-6322. three main purposes. First, they help constrain E-mail address: [email protected] (A.J.L. Harris). relationships between effusion rate and dimensional 0377-0273/00/$ - see front matter ᭧ 2000 Elsevier Science B.V. All rights reserved. PII: S0377-0273(00)00190-6 238 A.J.L. Harris et al. / Journal of Volcanology and Geothermal Research 102 (2000) 237–270 or morphological characteristics, such as flow length removes the effects of decompression and exsolving and area (e.g. Walker, 1973; Malin, 1980; Pieri and gases from bulk effusion rate estimates (Thordarson Baloga, 1986; Pinkerton and Wilson, 1994), and Self, 1993). By providing regular measurements 0 0 pahoehoe– a a transitions (Rowland and Walker, that consider flow activity over the entire flow field, 1990), and tube formation and size (Hallworth et al., including all branches of a bifurcating channel or flow 1987; Peterson et al., 1994; Calvari and Pinkerton, fed by multiple vents, satellite-based measurements 1998). Second, the shape of an effusion rate curve also provide a means of constraining time-varying can be used to infer eruptive mechanisms during effu- effusion and eruption rates. This is especially valuable sive activity (Wadge, 1981). Third, by quantifying the in morphologically complex cases where it may be amount, rate and variation of the erupted magma flux, difficult to measure discharge at all channels/vents effusion rate data contribute to the definition of the simultaneously. mass balance between supplied, intruded and erupted Relationships between effusion rate, flow dimen- magma (e.g. Dzurisin et al., 1984; Dvorak and sions and morphological characteristics have Dzurisin, 1993; Denlinger, 1997). This in turn contri- previously been thoroughly analysed and debated butes to an improved understanding of the rate at (e.g. Malin, 1980; Guest et al., 1987; Pinkerton and which magma is balanced by recycling or is supplied Wilson, 1994; Calvari and Pinkerton, 1998). Here we to conduits and shallow reservoirs, eruption sites, and focus on building detailed effusion rate time-series to intrusions (e.g. Francis et al., 1993; Allard, 1997; deduce eruption mechanisms and mass balances. Harris and Stevenson, 1997; Oppenheimer and Specifically we use a combination of satellite-derived Francis, 1998). However, to date there have been and ground-collected data to produce effusion rate few eruptions for which there have been a sufficient time-series to: (1) define any common and character- number of effusion rate measurements to allow effu- istic effusion rate trends; (2) infer eruptive mechan- sion rate curves to be defined (Wadge, 1981). Here we isms that can explain observed trends; and (3) use ground-based and satellite-data-derived effusion quantify erupted mass fluxes and magma supply rate measurements to produce effusion rate curves rates, and hence examine the relative importance of for as many eruptions as possible at Etna (Sicily) extrusion, intrusion and recycling. and Krafla (Iceland) volcanoes between 1980 and 1999. Between 1980 and 1999, there were 24 and 6 major 2. Effusive activity at Etna and Krafla effusive eruptions at Etna and Krafla, respectively. These resulted in the emplacement of 450 ^ 79 × Etna has been the site of frequent effusive eruptions 106 and ϳ 198 × 106 m3 of lava over a total of 1190 of basaltic (typically hawaiite) lava throughout the and 39 days, at the two volcanoes respectively. 30,000–40,000 year evolution of the currently active Although ground-based estimates made during these Mongibello edifice (Chester et al., 1985; Calvari et al., eruptions are too sparse to allow effusion rate curves 1994; Coltelli et al., 1994). As of 1994 the edifice was to be constructed, a notable exception being Frazzetta 3321 m high. Geophysical and geochemical studies and Romano (1984), use of frequently collected satel- indicate that magma is supplied from a sub-crustal lite thermal data has allowed effusion rate curves to be reservoir at a depth of 16–24 km (Sharp et al., constructed for a number of these eruptions (e.g. 1980; Tanguy et al., 1997), where SO2 emissions Harris et al., 1997a,b). We distinguish between effu- between 1975 and 1995 indicate supply to the upper sion rates and eruption rates. Effusion rate is the plumbing system at a rate of 4.5–8.0 m3 sϪ1 (Allard, instantaneous volume flux of erupted lava that is feed- 1997). In historic times effusive eruptions have ing flow at any particular point in time. Eruption rate typically occurred from the summit craters or from is the time-averaged volume flux obtained by dividing fissures on the volcano flanks (Guest and Murray, the volume of erupted lava at any particular point in 1979; Romano and Sturiale, 1982), producing flow 0 0 an eruption by the time since the eruption began. By fields dominated by a a surfaces (Chester et al., using dense rock equivalent volumes in the calcu- 1985). Data for effusive eruptions between 1868 and lation of eruption rate, consideration of eruption rate 1995 (J.B. Murray, unpublished data) indicate average A.J.L. Harris et al. / Journal of Volcanology and Geothermal Research 102 (2000) 237–270 239 flow-field lengths, areas and bulk volumes of 3.4 km, ning of the rifting episode and to events for which, 1.5 km2 and 16 × 106 m3; respectively. These data although we have no effusion rate data, volumes can show that between 1868 and 1949 an effusive event be obtained from published sources. occurred on average about once every three and a half years, with a time-averaged bulk eruption flux of Ϫ 0.15 m3 s 1. The period from 1950 onwards, however, 3. Effusion rate calculation: data, methods and saw an increase in effusive activity (Murray, 1990). results Between 1950 and 1995, an effusive event occurred about once every 9 months, and a flank eruption every To build effusion rate time series of sufficient 3.3 years. The time-averaged bulk eruption flux temporal detail to define variations in effusion during Ϫ during 1950–1995 was 0.86 m3 s 1, increasing a persistent eruption (i.e. trends occurring over a time throughout the period. Details of the main effusive scale of days to weeks), we have used two sources: (1) eruptions at Etna between 1980 and 1999 are given satellite thermal data; and (2) ground-based measure- in Table 1. ments. Our satellite-based data consist of thermal data The Krafla volcanic system is one of five fissure from three satellite-borne radiometers: the Advanced swarm-central-volcano complexes within the axial Very High Resolution Radiometer (AVHRR), Along- rift zone of N.E.