Caldera-forming eruptions are characterized by their large scale and low frequency. The question of how to prepare for such a rare, large-scale eruption poses a serious problem for the general public, administrative agencies, and volcanologists. The most commonly employed method is to investigate the past. Volcanologists seek to understand past events and to predict future developments by examining the history of eruptions. However, we face a high hurdle in preparing for such large eruptions. Using current prediction techniques, it is possible to detect the anomalous signals that immediately precede an eruption, by carefully monitoring volcanic activity. It is difficult, however, to judge whether such signals will lead to a caldera-forming eruption. In addition, it may be necessary to continue monitoring for thousands of years, until such an eruption occurs. From this reason, caldera-forming eruptions are not considered as part of the prediction program in Japan. Rather than waiting for such an eruption, it would be better to identify the processes that precede caldera-forming eruptions. This paper presents a hypothesis regarding the diversity of volcanic activity, as a basis for considering caldera-forming eruptions. The hypothesis holds that volcanic activity occurs as two end-member patterns: eruption dominant (ED) volcanism, in which magma is readily transported to the surface; and geothermal activity dominant (GD) volcanism, in which magma stagnates in the subsurface and induces geothermal activity. A spectrum of activity patterns exists between these two end members, related to the tendency of magma to ascend or to stagnate in the crust. Basic magma is stored repeatedly after intrusive events at GD volcanoes, eventually evolving to silicic magma. After repeated events, basic magma from deeper levels interacts with pre-existing magma in the chamber, causing heating, bubble formation and a huge eruption. According to this hypothesis, large amounts of magma will be found beneath volcanoes at which a caldera-forming eruption will occur in the future. It is also important to know where and why magma stops rising in the crust. The first step in understanding the preparation process of a caldera-forming eruption is to examine the hypothesis proposed in this study.
Historical documents that describe caldera-forming eruptions in Indonesia are reviewed to identify precursory events that occurred immediately before caldera-forming eruptions, and to determine the characteristics of the eruptive history before caldera-forming eruptions. The review revealed that small- to intermediate-scale eruptions occurred several months before climactic eruptions. The distribution of eruption sites or fumarole sites expanded by several kilometers during the precursor period. The long-term eruption rate remained high for around 100 ky, although the following changes have occurred in the past 5,000-10,000 years: (1) a decrease in the long-term eruption rate, (2) an increase in explosive eruption events, and (3) the migration or restriction of the distribution of eruption sites on volcano flanks. Larger pre-caldera volcano edifices tend to result in larger calderas. Finally, the range of caldera morphologies is discussed.
The deposition of large-scale pyroclastic flows (LPF) as part of caldera collapse is a key factor in understanding Crater Lake caldera, Oregon (8×10 km), one of the best examples of crater collapse in the world. The caldera, which formed at 6845 yr B.P., is intermediate in size between a smaller central conduit and a larger ring conduit system. Analyses of lithic components in the ejecta confirm that the climactic ignimbrite was ejected from multiple conduits along a ring fracture. This finding demonstrates that in a caldera such as Crater Lake, the structural resurgence may be too small to act as a shallow stress source; nevertheless, such calderas may have a ring conduit system. Two types of deformation, ooze-outs and squeeze-outs of fiammes, occur in the Wineglass Welded Tuff (WWT) on the northeast caldera margin, produced by concentric normal faulting and landsliding at the time of caldera collapse. Calculations regarding the conductive cooling history of the WWT and the preclimactic Cleetwood lava flow (CLF) yield the following results: (a) a maximum interval of 9 days from emplacement of the WWT to caldera collapse, and (b) less than 100 years from effusion of the CLF to the onset of the climactic eruption. The size of LPF units relative to the total eruptive volume appears to be unrelated to caldera size. The flow unit is not necessarily easily identified in the case of an LPF. It is possible that during caldera collapse, the decompression rate of the magma reservoir increases in the case of a smaller number of flow units resulting from a higher eruption rate. We propose that a Valles-type caldera is the most likely candidate in terms of generating this feedback process, since its ring conduit system was stable over time and requires less kinematic energy during caldera collapse than does a funnel caldera.
The Miocene Miyanotani composite dike, exposed in the central part of Kii Peninsula, southwest Japan, was examined to infer the processes of magma mixing and dike emplacement. This composite dike consists mainly of marginal basaltic andesite and central rhyolite, with an intervening andesitic facies. The central rhyolite contains irregularly shaped mafic enclaves, particularly along its margins. Based on field occurrences, petrography, and bulk-rock compositions, the following processes of magma mixing and dike emplacement are inferred: (1) the injection of mafic magma into a felsic magma chamber during cooling, and further mafic dike emplacement; (2) magma mixing in the felsic magma chamber to form intermediate magma; and (3) the injection of felsic and intermediate magma into a previously emplaced mafic dike to form a composite dike. Many composite dikes, including the Miyanotani dike, develop along arcuate faults and pyroclastic dikes or conduits at Odai caldera. The ages of the composite dikes and conduits indicate that they were emplaced contemporaneously. In addition, the pyroclastic dikes include juvenile fragments with mafic and intermediate compositions. These findings strongly suggest a genetic relationship between the composite dikes and the formation of Odai caldera. Many of the irregularly shaped granite enclaves included within the felsic center and andesite of the Miyanotani dike, which show micrographic texture indicative of emplacement at shallow levels in the crust, may have been derived from the felsic magma chamber during cooling, from which the Odai caldera eruption was fed. Therefore, the granite enclaves may yield important information in understanding the relationship between composite dike formation and the caldera eruption.
The Nakaoku tuffite dike (NTD) is a large, arcuate pyroclastic conduit located along the margin of Odai caldera in central Kii Peninsula, SW Japan. The NTD is one of the most likely candidates as a conduit of an explosive eruption during caldera formation at 13-15 Ma. We reappraised the origins of lithological units at the Hibora-dani section of the NTD and discussed how magma passed through the conduit to form the NTD facies. Our detailed observations and grain-size analyses revealed that three types of lithofacies occur from the margin to the center of rhyolitic pyroclastic dikes: welded lapilli-tuff, welded crystal-tuff, and welded vitric-tuff. In the central part of dikes, the welded vitric-tuff contains spindle-shaped pumice clasts stretched parallel to the dike wall, representing a foliation defined by aligned fiamme in a matrix with eutaxitic texture. The aspect ratio of stringy pumice clasts is 1 : 50, indicating rheomorphic deformation in a viscously flowing body. These findings suggest that the vitric tuff could have agglutinated as a laminar flow unit parallel to the dike wall at the time of “hot” emplacement. The welded foliation of fiamme suggests that the closing pressure of the dike wall compressed a vesiculated magma body. These rheomorphic deformations in hot magma could have resulted from the secondary viscous flow of welded tuffs in the dike interior, as with the drain-back process that occurs during closure of a conduit. Based on the present results and those of previous studies, we conclude that the distinct lithofacies at the Hibora-dani section of the NTD were produced during the final stages of a ring eruption at the Odai caldera.