Magma chamber of the Campi Flegrei supervolcano at the time of eruption of the Campanian Ignimbrite Paola Marianelli ⎤ Alessandro Sbrana ⎥ Monica Proto ⎦
Dipartimento di Scienze della Terra, Universita` degli Studi di Pisa, via Santa Maria 53, 56126 Pisa, Italy
ABSTRACT A supereruption that occurred in the Campi Flegrei area, Italy, ca. 39 ka had regionaland global-scale environmental impacts and deposited the Campanian Ignimbrite (CI). We attempt to shed light on critical aspects of the eruption (depth of magma chamber, intensive pre-eruptive magma conditions) and the large-volume magma plumbing system on the basis of information derived from analyzing melt inclusion (MI) data. To achieve these aims, we provide new measurements of homogenization temperatures and values of dissolved H2O within phenocryst-hosted MIs from pumices erupted during different phases of the CI eruption. The MI data indicate that a relatively homogeneous overheated trachytic magma resided within a relatively deep magma chamber. Dissolved water contents in MIs indicate that prior to the eruption the magma chamber underwent radical changes related to differential upward movement of magma. Decompression of the rising trachytic magma caused a decrease in water solubility and crystallization, and trachytic bodies were emplaced at very shallow depths. The proposed eruptive model links portions of the main magma chamber and apophyses with specific eruptive units. Keywords: Campanian Ignimbrite, volatiles, melt inclusions, feeding systems. INTRODUCTION Ca. 39 ka (De Vivo et al., 2001), the Campi Flegrei supervolcano (Sparks et al., 2005) underwent a supereruption that produced the Campanian Ignimbrite (CI). More than 150 km3 dense rock equivalent (DRE) of trachytic magma was erupted, and the event had a volcanic explosivity index of 7. This explosion was the largest magnitude eruption to occur in the Mediterranean region during the late Quaternary; it resulted in the formation of a 14 km caldera (Rosi and Sbrana, 1987), and had regional- and global-scale impacts. The CI eruption has been the subject of several studies (e.g., Barberi et al., 1978; Di Girolamo et al., 1984; Orsi et al., 1996; Rosi et al., 1996; Pappalardo et al., 2002, and references therein) that have investigated deposit successions, eruption dynamics, and petrogenesis. The eruption began with a Plinian phase, during which a SE-distributed pumice fallout was emplaced. This was followed by a succession of pyroclastic density currents that deposited ashes and pumice flows and densely welded ignimbrites that covered the Campanian Plain and surrounding hills (Barberi et al., 1978). The lithic-rich Breccia Museo Unit is present at the top of the eruption deposits along the caldera margins and is interpreted as a proximal deposit related to the final calderaforming phase (Rosi and Sbrana, 1987; Rosi et al., 1996). The objective of this paper is to determine the pre-eruptive conditions of the CI in terms of pressure, temperature, and composition by studying melt inclusions (MIs), and to shed light on the relationship between eruption dy-
namics and magma feeding systems. Previous studies suggest that the CI eruption was fed by a slightly zoned (trachyte to trachyticphonolite) magma chamber (Barberi et al., 1978; Civetta et al., 1997, and references therein), having an upper volatile-rich magma layer (Signorelli et al., 2001) and located at ⬃4 km depth (Pappalardo et al., 2002). These previous interpretations are not supported by the findings derived from new data collected in this study. The MI approach provides distinct advantages over studies involving bulk rock samples, and places firm constraints on the pre-eruptive composition, volatile content, temperature, and pressure of melts. In this paper we demonstrate that such data are critical for reconstructing processes that occurred in the CI feeding system, such as the pre-eruptive rise of magma. RESULTS Juvenile fractions of the CI deposits were selected on the basis of cooling rate to provide suitable samples for MI studies. Sample collection was therefore restricted to pumice from fallout deposits, proximal lithic-rich breccias, and thin, relatively well quenched distal ignimbrite deposits (see Data Repository Appendix 11 for sample mineralogy and petrography). 1GSA Data Repository item 2006209, Appendix 1, Figures DR1–DR3, and Tables DR1–DR6, sampling strategy and characterization of juvenile fraction, analytical methods, bulk rocks, mineral chemistry, glassy matrix, and melt inclusion analyses, is available online at www.geosociety.org/pubs/ ft2006.htm or on request from editing@geosociety. org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.
Bulk rock samples and the glassy groundmass have approximately homogeneous trachyticphonolitic compositions (Fig. 1; Tables DR3 and DR4 [see footnote 1]). Well-quenched primary glassy MIs (20–60 m) are found in clinopyroxene. Isolated rare MIs occur randomly distributed within the crystals or concentrated along diopside-salite interfaces. Most MIs are two phase with brown glass and a shrinkage bubble, or are glassy and bubble free (Fig. 2A). The lack of shrinkage bubble(s), mainly observed within samples from the fallout, is due to the high quenching rate (⬎0.001 ⬚C s⫺1; Wallace et al., 2003) of the analyzed samples. Only rare MIs are characterized by multiple bubbles (Fig. 2B). The MIs are all of similar trachytic to phonolitic composition, even for MIs from different units, although some MIs from the lithic-rich breccia deposits have slightly more differentiated compositions with respect to matrix glasses (Fig. 1; Table DR5 [see footnote 1]). The analysis of MIs hosted in diopside was not possible because of the small size (⬍15 m) of such inclusions. The H2O content of MIs is highly variable; most data indicate 2–3 wt% and maximum values of 6 wt% (Fig. 3; Appendix 1; Fig. DR2; Table DR6 [see footnote 1]). It is noteworthy that higher water contents are recorded in samples from the ignimbritic units. Microthermometric experiments were carried out on melt inclusions hosted in clinopyroxene (Appendix 1; see footnote 1). Melt inclusions in salitic crystals homogenized between 950 ⬚C and 1080 ⬚C (Fig. 3); the higher values correspond to MIs trapped at the diopside-salite interface. Rare inclusions in diopsidic cores recorded higher temperatures, in excess of 1100 ⬚C, whereas few MIs hosted in Fe-rich clinopyroxene homogenized at temperatures as low as 870 ⬚C. DISCUSSION Compositional variability within the studied samples is more pronounced in bulk rocks, and less marked in glassy matrix (Fig. 1). This suggests that the chemical differences mainly reflect variable crystal content within juvenile fragments (Appendix 1; see footnote 1). The minor variation recorded in the composition of trapped melts (MIs) mainly reflects the crystallization of clinopyroxene and sanidine, as described in detail by Webster et al. (2003). The major element composition of most MIs
䉷 2006 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or
[email protected]. Geology; November 2006; v. 34; no. 11; p. 937–940; doi: 10.1130/G22807A.1; 5 figures; Data Repository item 2006209.
937
Figure 4. H2O contents of melt inclusions (MI) with respect to their compositions (SiO2 values normalized to anhydrous compositions).
Figure 1. Upper diagrams: symbols refer to compositions of Campanian Ignimbrite (CI) bulk rock samples. Compositional ranges of Campi Flegrei volcanics (dotted area; data from Rosi and Sbrana, 1987) and CI glasses (gray field; glassy matrix and melt inclusions [MIs], this work) are plotted for comparison. Dashed area is enlarged in lower plots. Lower diagrams: glassy matrix. Each symbol represents average of 20 points (standard deviation included in symbol size) within single pumice specimens. MIs: symbols represent average of three points within each inclusion.
is similar to that of the bulk pumice and the glassy matrix (Fig. 1). This trend could indicate that only minor differentiation of melts occurred between entrapment and eruption. The MI characteristics probably closely reflect the pre-eruptive conditions of the magma chamber. The more SiO2-rich MIs (Fig. 1) could reflect local differentiation of trapped melts. The MIs are characterized by variability in
Figure 2. Microphotographs of melt inclusions hosted within clinopyroxene. 938
water content and relatively homogeneous major element melt compositions. The abundance of H2O appears to be independent of increasing degree of melt evolution. In partic-
ular, the higher H2O contents do not correspond to more highly differentiated MIs (SiO2 65 wt%; Fig. 4). This observation is contrary to the hypothesis of Signorelli et al. (2001), who interpreted their analyzed MIs as water undersaturated, and suggested an increase in water content within more evolved melts (even though not recorded by MIs). The H2O solubility model of Di Matteo et al. (2004) for trachytic melt was applied in this study to calculate saturation pressures for the different MIs, and pressures of 40–60 to 150 MPa were obtained (Fig. 3). These pressures were converted to depths via a model based on measured densities in deep geothermal wells (Rosi and Sbrana, 1987). According to the hypothesis of water saturation conditions for the trapped melt, the data presented in this paper indicate that MIs formed over a relatively wide range of pressure within a magmatic storage region located at 2–6 km depth. The interpretation of the MI data necessitates a complex account of the processes affecting the feeding system of the CI eruption. Variation in apparent pressure can represent movement of the magma body. Alternatively, an influx of other volatiles (C, S) from mafic magmas could have lowered the water pressure and induced crystallization without mag-
Figure 3. Time-variance of saturation pressure and H2O wt% dissolved in melt inclusions [MIs] and homogenization temperature of MIs (filled circles: salite-hosted MIs; stars: MIs at diopside-salide boundary; open circles: diopsidic core-hosted MIs) for samples from different phases of Campanian Ignimbrite eruption. Eruption time based on deposit succession. Vertical scale is qualitative. GEOLOGY, November 2006
ma ascent. Gardner et al. (1995) described changes in water content and a shift from water-saturated to water-undersaturated conditions in dacites from Mount St. Helens, USA, caused by the addition of basaltic magma that diluted the water content of hybrid dacite and added CO2. Such a model for the CI is not supported by analysis of the MIs, because CO2 is below the detection limit (Appendix 1; see footnote 1) and S shows no clear relationship to the H2O content of MIs (Fig. DR3; see footnote 1). The explanation that we propose for the range of recorded pressure (or dissolved water content) is that the different values correspond to different time periods prior to eruption. Following this interpretation, we suggest that the CI trachytic magma, at ⬃1000 ⬚C and with 5– 6 wt% H2O dissolved in the melt, was stored in a large magma chamber located at 6 km depth and a saturation pressure as high as 150 MPa. The liquidus temperature of these trachytes was ⬃970 ⬚C, as estimated using PELE software (Ghiorso and Sack, 1994). This temperature is lower than that obtained by microthermometric experiments, which yield a mean magma temperature of ⬃1000 ⬚C (Fig. 3). Under such conditions the magma would have been near liquidus, and this is one possible explanation for the very low crystal content and the scarcity of MIs within eruption products. High temperatures of trapping and/or crystallization are indicated by MIs at the diopside-salite compositional interface, the sharp transition from diopsidic cores to salitic rims, and diopsidic cores that contain MIs with homogenization temperatures as high as 1100 ⬚C. All of these factors suggest the presence of hotter and presumably less evolved magmas at the base of the chamber; these magmas were the cause of the overheated thermal state of the chamber. The existence of such a hot magma body is also supported by the high-Mg reheated MIs reported by Webster et al. (2003). The estimated 6 km depth of this very large magma chamber is slightly less than depths of ⬃8 km estimated for the upper portions of basaltic crustal reservoirs, the existence of which in the Campanian area has been demonstrated for Vesuvius (Marianelli et al., 2005; Auger et al., 2001) and hypothesized for the Phlegrean Volcanic District (Cecchetti et al., 2005). Data presented here suggest that trachytic melts stored at depths of ⬃6 km represent the differentiation of the upper portions of a huge basaltic reservoir. This trachytic magma, with a volume of 150 km3, could have originated following differentiation mainly by crystal fractionation, in which 70 wt% of solids were removed from at least 500 km3 of basaltic magma within an enormous chamber intruded at crustal levels (Fig. 5A). Melt inclusions with a relatively low water GEOLOGY, November 2006
Figure 5. Evolution of Campanian Ignimbrite feeding system. A: Poorly evolved (basaltic?) magma intrusions (black) from which trachytes (gray) evolved by differentiation processes at mid-crustal depths. B: Rise of trachytes and storage at shallow depth; formation of thermometamorphic aureole. C: Opening of eruption with Plinian phase fed by shallow apophyses and deposition of fallout tephra. D: Commencement of caldera collapse and withdrawal of trachytic magma from deep reservoir that fed ignimbritic phase; hachured regions represent frac-
content of 2–3 wt% indicate that the trachytic magma underwent large changes in pressure from 150 MPa to 50 MPa prior to the eruption, assuming that the total pressure was represented by the water pressure. Variations in pressure are interpreted to represent a change in depth from 6 km to 2–3 km, and thus movements of portions of the reservoir cap within the overlying crust over time. Such movement could have involved the formation of apophyses to the main magma chamber (Fig. 5B). Magma rise may have resulted from a pressure buildup in the main magma storage region, possibly following an intrusion of more primitive volatile-rich magma (Webster et al., 2003). If the overpressure exceeds the tensile strength of the crust, fracturing of surrounding rocks and the movement of the upper parts of magma storage regions are promoted. During ascent, the hydrous trachytic body possibly underwent a slight decrease in temperature at its margins because of heat loss to the surrounding rocks, as recorded by the small numbers of low-temperature MIs within Fe-rich pyroxene. The associated melt would have lost some water due to a decrease in water solubility. Under these new pressure-temperaturecomposition conditions, the difference between magma temperature (⬃980 ⬚C) and the liquidus temperature (Tliquidus ⬎ 1020 ⬚C for ⬍3 wt% of dissolved H2O) represents an undercooling driving force (Nishimura et al., 2005) that induced magma crystallization and a suddenly increased growth rate, enabling the easy trapping of MIs and a slight differentiation in trapped melts. It has also been demonstrated for dacites from Mount Pinatubo, Philippines, that, for a decrease of 70 MPa over the short time interval of three weeks, crystallization began immediately via the growth of existing crystals, resulting in the extensive trapping of MIs (Hammer and Rutherford, 2002). This pre-eruptive stage that affects the CI feeding system is clearly recorded by the features (Fig. 2A) of the majority of inclusions that are trapped in the range 40–60 MPa at ⬃980 ⬚C. In any case, the generally low crystal content of the CI products suggests that this step could represent a shortlived pre-eruptive storage phase. Following this hypothesis, magmas residing at these lower pressures fed the Plinian phase of the eruption (Fig. 5C) and were involved in the final caldera-forming phases (Fig. 5E). The ignimbrite-forming stage withdrew magma from the 6-km-deep chamber, evidenced by the high H2O content of MIs from ignimbritic units (Figs. 3 and 5D). Otherwise, the
N tured country rocks. E: Main calderaforming phase, involving simultaneous withdrawal of magma from shallow apophyses and main deep reservoir; deposition of lithic-rich breccia. 939
occurrence of rare MIs characterized by multiple bubbles (Fig. 2B) within Plinian lower fallout samples, which are derived from the first magma to be erupted, suggests that these deposits could represent portions of magma affected by a vesiculation process. Such an interpretation is in agreement with the hypothesis of fluid saturation of magma at a depth of ⬃2 km, as proposed by Signorelli et al. (2001). The high temperature of trachytic apophyses and associated degassing would strongly favor the occurrence of heat transfer to rocks around and above the magma bodies. This supposition is consistent with the occurrence of widespread thermometamorphic rocks (Rosi and Sbrana, 1987) that have been sampled from deep (2 km) geothermal wells within the Campi Flegrei Caldera. Experimental phase equilibria on younger Phlegrean trachytes (Agnano Monte Spina eruption) indicate magma (pressure, temperature) conditions and a reequilibration of magmas prior to the eruption at very shallow depth (M. Rutherford, 2006, personal commun.) in perfect agreement with the findings of this work, based on MI data. CONCLUDING REMARKS On the basis of our study, we propose a new model for the CI feeding system. A very large magma chamber, emplaced ca. 39 ka at ⬃6– 9 km depth, far in excess of the previously assumed 4 km depth of the shallow Phlegrean magma chamber (Pappalardo et al., 2002, and references therein), was present beneath the Campi Flegrei Volcanic District. The upper part of the chamber contained overheated (1000 ⬚C) and water-rich (to 6 wt%) trachytes that were slightly zoned in composition. The high temperature indicated for the erupted magma reflects the presence within the lower part of the system of hot, less evolved (basaltic?) magmas from which the trachytes were derived by differentiation processes. Prior to the CI eruption, the magma chamber underwent radical changes because of upward movements, stagnation, and degassing of trachytic melts that formed apophyses in a diffuse range of depths. These apophyses fed the initial Plinian phase of the eruption that generated 20 km3 DRE of fallout deposits of pumices that were more crystal rich than those erupted during the ignimbritic phase. From the main, deeper reservoir the overheated subaphyric and water-rich trachytic magma fed the main phase of the eruption, in which ⬃130 km3 DRE of magma was emitted as pulses of ignimbrites spread over 3000 km2; the underlying basalts were never erupted. The partial emptying of the magmatic feeding system led to the collapse of a caldera
940
that was 14 km in diameter. Widespread lithicrich breccias mark this final phase of the eruption that simultaneously involved crystal-rich trachytes from the low-pressure apophyses and subaphyric trachytes from the deep main magma chamber (Fig. 5). The results of our work contribute to improving our understanding of such cataclysmic, high-magnitude and high-intensity events (so-called supereruptions). Furthermore, improvements in knowledge of this impressive late Quaternary eruption are particularly important, considering its occurrence within one of the most densely populated and high-risk volcanic areas of the world. ACKNOWLEDGMENTS This work was funded by grants from Italian Ministero Istruzione Universita` e Ricerca and Istituto Nazionale di Geofisica e Vulcanologia. Constructive comments and suggestions by J. Hammer, J. Lowenstern, and M. Rutherford greatly improved the quality and clarity of the manuscript. REFERENCES CITED Auger, E., Gasparini, P., Virieux, J., and Zollo, A., 2001, Seismic evidence of an extended magmatic sill under Mt. Vesuvius: Science, v. 294, p. 1510–1512, doi: 10.1126/science.1064893. Barberi, F., Innocenti, F., Lirer, L., Munno, R., Pescatore, T., and Santacroce, R., 1978, The Campanian Ignimbrite: A major prehistoric eruption in the Neapolitan area (Italy): Bulletin Volcanologique, v. 41, p. 10–31. Cecchetti, A., Marianelli, P., and Sbrana, A., 2005, Neapolitan active volcanoes: A study of the medium-high pressure feeding systems through melt inclusions, in XVIII European Conference on Research on Fluid Inclusions, Siena, Italy, University of Siena, abs. 03. Civetta, L., Orsi, G., Pappalardo, L., Fisher, R.V., Heiken, G., and Ort, M., 1997, Geochemical zoning, mingling, eruptive dynamics and depositional processes—The Campanian Ignimbrite, Campi Flegrei caldera, Italy: Journal of Volcanology and Geothermal Research, v. 75, p. 183–219, doi: 10.1016/S0377-0273(96) 00027-3. De Vivo, B., Rolandi, G., Gans, P.B., Calvert, A., Bohrson, W.A., Spera, F.J., and Belkin, H.E., 2001, New constraints on the pyroclastic eruptive history of the Campanian volcanic plain (Italy): Mineralogy and Petrology, v. 73, p. 47–65, doi: 10.1007/s007100170010. Di Girolamo, P., Ghiara, R., Lirer, L., Munno, R., Rolandi, G., and Stanzione, A., 1984, Vulcanologia e petrologia dei Campi Flegrei: Bollettino della Societa` Geologica Italiana, v. 103, p. 349–413. Di Matteo, V., Carroll, M., Behrens, H., Vetere, F., and Brooker, R.A., 2004, Water solubility in trachytic melts: Chemical Geology, v. 213, p. 187–196, doi: 10.1016/j.chemgeo.2004.08. 042. Gardner, J.E., Rutherford, M., Carey, S., and Sigurdsson, H., 1995, Experimental constraints on pre-eruptive water contents and changing magma storage prior to explosive eruptions of Mount St Helens volcano: Bulletin of Volcanology, v. 57, p. 1–17. Ghiorso, M.S., and Sack, R.O., 1994, Chemical mass transfer in magmatic processes IV. A re-
vised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid-solid equilibria in magmatic systems at elevated temperatures and pressures: Contributions to Mineralogy and Petrology, v. 119, p. 197–212. Hammer, J., and Rutherford, M., 2002, An experimental study of the kinetics of decompressioninduced crystallization in silicic melt: Journal of Geophysical Research, v. 107, 2021, doi: 10.1029/2001JB00281. Marianelli, P., Metrich, N., Sbrana, A., and Cecchetti, A., 2005, The deep feeding system of Vesuvius involved in recent violent strombolian eruptions: Geophysical Research Letters, v. 32, L02306, doi: 10.1029/2004GL021667. Nishimura, K., Kawamoto, T., Kobayashi, T., Sugimoto, T., and Yamashita, S., 2005, Melt inclusion analysis of the Unzen 1991–1995 dacite: Implications for crystallization processes of dacite magma: Bulletin of Volcanology, v. 67, p. 648–662, doi: 10.1007/s00445-0040400-8. Orsi, G., de Vita, S., and Di Vito, M., 1996, The restless, resurgent Campi Flegrei nested caldera (Italy): Constraints on its evolution and configuration: Journal of Volcanology and Geothermal Research, v. 74, p. 179–214, doi: 10.1016/S0377-0273(96)00063-7. Pappalardo, L., Civetta, L., de Vita, S., Di Vito, M., Orsi, G., Carandente, A., and Fisher, R.V., 2002, Timing of magma extraction during the Campanian Ignimbrite eruption (Campi Flegrei Caldera): Journal of Volcanology and Geothermal Research, v. 114, p. 479–497, doi: 10.1016/S0377-0273(01)00302-X. Rosi, M., and Sbrana, A., 1987, Phlegrean Fields: Roma, CNR, Quaderni de La Ricerca Scientifica, v. 9, p. 114. Rosi, M., Vezzoli, L., Aleotti, P., and De Censi, M., 1996, Interaction between caldera collapse and eruptive dynamics during Campanian Ignimbrite eruption, Phlegrean Fields, Italy: Bulletin of Volcanology, v. 57, p. 541–554. Signorelli, S., Vaggelli, G., Romano, C., and Carroll, M.R., 2001, Volatile element zonation in Campanian Ignimbrite magmas (Phlegrean Fields, Italy): Evidence from the study of glass inclusions and matrix glasses: Contributions to Mineralogy and Petrology, v. 140, p. 543–553. Sparks, S., Self, S., Grattan, J., Oppenheimer, C., Pyle, D., and Rymer, H., 2005, Super eruptions: Global effects and future threats: London, Geological Society, Report of a Geological Society of London Working Group, p. 1–24. Wallace, P., Dufek, J., Anderson, A.T., and Zhang, Y., 2003, Cooling rates of Plinian-fall and pyroclastic-flow deposits in the Bishop Tuff: Inferences from water speciation in quartzhosted glass inclusions: Bulletin of Volcanology, v. 65, p. 105–123. Webster, J.D., Raia, F., Tappen, C., and De Vivo, B., 2003, Pre-eruptive geochemistry of the ignimbrite-forming magmas of the Campanian Volcanic Zone, southern Italy, determined from silicate melt inclusions: Mineralogy and Petrology, v. 79, p. 99–125, doi: 10.1007/ s00710-003-0004-6. Manuscript received 13 March 2006 Revised manuscript received 31 May 2006 Manuscript accepted 7 June 2006 Printed in USA
GEOLOGY, November 2006
