New Stratigraphic and Geochemical Data for the Megalo Vouno Complex: a Dominating Volcanic Landform in Minoan Times

This can be concluded from published eruption ages of intercalated pyroclastic deposits. Field observations also indicate that the MV complex must have been a prominent volcanic high up to Minoan times. The occurrence of an inferred Graben structure between the MV complex and the southern basement complex (Reck 1939; Heiken and McCoy 1984) has strongly influenced the volcanic activity and the topographic development of Santorini. On the basis of field observations and published stratigraphic and tectonic data we present a palaeotopographic reconstruction of the evolution of the composite volcanic field.

GEOLOGY OF SANTORINI

The geology of Santorini has been described by Fouqué (1879), Reck (1936) and Pichler and Kussmaul (1980). More specifically, the lavas have been described by Nicholls (1971), Mann (1983), Barton et al. (1983), Barton and Huijsmans (1986), Wyers (1987), Huijsmans et al. (1988) and Huijsmans and Barton (1989), whereas the pyroclastic deposits of Santorini have been described by Druitt and Sparks (1982), Heiken and McCoy (1984), Druitt (1985) and Druitt et al. (1989).

In this paper we have adopted the stratigraphy of Druitt et al. (1989) for the Thera Pyroclastic Formation (TPF), although we do not always agree with their stratigraphy in relationship to the lavas. We have also adopted their mapped distribution of members of the TPF on the basis of isopachs (Druitt et al. 1989, Fig. 6). One of the most important discrepancies between our observations and the work of Druitt et al. is the fact that they conclude that US1 deposits pre-date eruption of the Skaros lavas. Our observations in the Skaros sequence below Merovigli (Fig. 1) indicate that US1 deposits occur between cycle III and cycle IV of Skaros activity (cf. Huijsmans and Barton 1989), i.e. US1 is intercalated in the Skaros sequence (see also addendum). Below Phira US1 is exposed below Skaros lavas from the fourth cycle, whereas in a northern direction this deposit forms a discontinuous marker horizon, filling old depressions on the 54 Kyr Skaros shield. A chemical analysis of the upper part of US1 below Merovigli (i.e. SH 151 from Huijsmans 1985) is similar to an analysis given by Druitt et al. (1989, Table 4, analysis no. 7). North of Merovigli lavas of cycle IV wedge out and a local US1 deposit directly underlies US2 deposits. In this exposure of US1 many imprints of fossil leaves can be found.

Huijsmans (1985, Fig. 9. 2) recognized that a large scale cyclic eruption pattern can be recognized in the pyroclastic sequence of the TPF in the caldera wall between Phira and Athinios. Druitt et al. (1989) have interpreted this as two major mafic-to-silicic cycles of volcanism which occur on Santorini. The first cycle includes the eruptions of Cape Therma 1(CT1; < 940 Kyr), Cape Therma 2 (CT2), Cape Therma 3 (CT3), and Lower Pumice (LP1 and LP2; 100 Kyr), whereas the second cycle includes Cape Thera (CTR), Middle Pumice (MP; 79 Kyr), Vourvoulos (V), Upper Scoriae 1 (US1; 54 Kyr), Upper Scoriae 2 (US2; 37 Kyr), Cape Riva (CR; 18 Kyr) and the Minoan (Min). The relationship between lavas and pyroclastics in terms of this cyclicity has been described by Huijsmans and Barton (1989) for the second cycle.

STRUCTURE OF THE VOLCANIC FIELD

Fouqué (1879) and Reck (1936) have described the structural framework of Santorini. More recently, Heiken and McCoy (1984) have discussed the structural framework in relation to caldera development during the Minoan eruption. In general, all these authors recognize the importance of a NE-SW trending fault system or Graben structure. This structural lineament connects the loci of volcanic activity of the Christiana islands, Santorini and the submarine Kolombos volcano or Kolombos Bank. It also controls the orientation of pre-volcanic faults in the basement, the orientation of dykes in the northern caldera wall and the orientation of volcanic vents.

Apart from a NE-SW trending system Fouqué (1879) in his detailed study of the 1866 eruptions on Nea Kameni also recognized a subordinate NW-SE trending system. Reck (1936) and Neumann van Padang (1936) mapped similar fault systems in the pre-Minoan volcanics and in the basement complex of Mount Profitis Ilias.

Fig. 2 illustrates tectonic features in combination with the on-land topography and the bathymetry of the surrounding waters down to a level of approximately 40 metres. Also indicated are slump structures and the traces of inferred fault or Graben structures. The figure shows that the focus of present day volcanism (Kameni islands) and probably of most explosive pre-Minoan volcanism (Druitt et al. 1989) is situated in the area of the intersection of the two above-described fault systems. In addition, pre-Minoan eruption centres in the northern part of Santorini, which nowadays form topographic highs, are situated along this Graben structure, as well as the maar volcano of Cape Kolombos and the submarine, volcanic Kolombos Bank.

The breached character of the caldera wall with associated slump structures can also be explailed by the presence of the above-described fault systems. Palaeo-valleys or long-lived depressions in the topography, such as the long-lived depression between MPI and Phira (filled with the lavas of the Skaros volcano); the palaeo-valley between Akrotiri and Cape Therma (filled with a brown-coloured welded ignimbrite from the Cape Riva eruption); and the long-lived depression between MV and MPI (filled with a sequence of TPF deposits), are other expressions of the fault systems depicted in Fig. 2. The bathymetry of Santorini shows the occurrence of two basins in the caldera subdivided by a ridge of historic Kameni lavas and domes. The southern basin shows a uniform depth of approximately 280 m, whereas the northern basin is deeper (up to nearly 400 m). The - 40 m contour in Fig. 2 shows that most of the area between Akrotiri and Aspronisi is very shallow. Fault-bound (?) slump structures (S1 and S3; Fig. 2) in this area are restricted to the inner part of the caldera. The shallow area is continued on the western side of the deep slump structure (S3) between Aspronisi and Therasia. The S3 deep itself (nearly 400 m) lines up in a NE-SW direction with the northern basin. A submarine ridge between S3 and the northern basin is probably related to historic activity of the Kamenis.

STRATIGRAPHY OF THE MEGALO VOUNO VOLCANIC COMPLEX

The stratigraphy of the MV complex has been described by Pichler and Kussmaul (1980). A detailed description of the stratigraphy of the caldera wall between Cape Perivola and Cape Heptapedio was given by Huijsmans (1985). In the summer of 1986 fieldwork was carried out again in the caldera wall section between Cape Heptapedio and MPI. A schematic diagram of the northern caldera wall, drawn on the basis of photographs, is given in Fig. 3. The MV complex is characterized in part by the abundant occurrence of intermediate and silicic lava compositions, as will be discussed below. This is in contrast with the volcanic centres of MPI and Skaros (Barton and Huijsmans 1989), which mainly produced basalts and basaltic andesites.

The core of the MV complex (MV1 and MV2) consists of an old, 150 m high lava shield, which is exposed in the caldera wall between Cape Perivola and the old MPI volcano. Lavas are basaltic andesites in the lower part and low-silica andesites in the upper part of the shield. Between the MV complex and MPI a faulted zone occurs. In depressions on the irregular surface of the old MV shield LP deposits (100 Kyr) are found. These deposits are in turn overlaid by large, flow-banded domes and lavas of high-silica andesite. Below LP a conspicuous red scoriae layer is present, which at the base consists of a buff pumice fall deposit. This outstanding horizon is also present in the contact zone between MV and MPI (Fig. 3) and represents the lower part of LP deposits (LP1) or (Druitt pers. comm. 1990) CT3 deposits. It should be stressed here that LP deposits in the northern caldera in most cases are only preserved on places where the silicic domes and lavas occur. This implies eruption of domes and lavas shortly after eruption of LP. A subsequent period of erosion must have removed most LP deposits from the northern shields. Depositional dips of lavas from the old core complex, LP deposits and associated domes show a southward component, i.e. into the present caldera, while the younger products dip outward. We conclude: 1) that the old core complex of MV must have been an independent volcanic edifice for a long time; and 2) that the lower part of the present caldera wall of MV shows a cross-section through the southern flank of the pre-100 Kyr MV lava-shield.

The discontinuous marker horizon of LP deposits and silicic domes is embedded in pyroclastic deposits, which represent an incomplete sequence of younger (< 100 Kyr) TPF deposits. A prominent ledge of partly welded, red scoriae rags fills a palaeo-valley at Cape Heptapedio. This deposit is very similar in appearance to the densely welded Upper Scoriae 2 (US2) unit of Druitt et al. (1989) on Cape Tourlos and we tentatively suggest that this is a US2 (37 Kyr) deposit. Below this ledge we also find indications for the presence of remnants of Upper Scoriae 1 deposits (US1; 54 Kyr). Between US1 and US2, basaltic to low-silica andesitic lava flows are regurarly intercalated between unspecified yellow and brown coloured pyroclastic deposits. This lava and pyroclastics sequence thickens considerably in an eastern direction (Fig. 3). Evidence for the presence of US1 further eastwards in the MV caldera walls is hard to find, partly because the cliffs become very inaccessible. US2 only fills the palaeo-valley described above and is absent east of Cape Heptapedio. From our observations in the Skaros sequence below Merovigli we conclude that these lavas represent or are similar in age to the youngest (< 54 Kyr - > 37 Kyr) lavas of Skaros (cycle IV). 'Skaros' flows overlying US1 deposits are also present in the faulted zone between MV and MPI.

The sequence of 'Skaros' lavas with associated pyroclastics gradually passes into grey to black cinder deposits of the present Megalo Vouno basaltic cinder cone (MV3). These grey-black cinders also drape the fault controlled topography (Fig. 3) between the MV complex and MPI. Within the grey cinders of MV3 numerous, irregular flows and domes (basaltic-andesite to high-silica andesite) occur (MV5), which disappear towards the summit of the cinder cone. On top of Megalo Vouno red, basaltic cinders are found, which became erupted from another cinder cone, i.e. Kokkino Vouno (MV4). Flank activity of the Kokkino Vouno cinder cone (black and red cinders) produced basaltic-dacitic lava flows (MV5). One of these lava flows overlies ashes from the relatively young maar volcano of Cape Kolombos.

At Cape Heptapedio MV3 cinders and intercalated MV5 flows and domes are covered by basaltic lava flows (MV6), which overlay the red scoriae deposit (US2) below Oia (Fig. 3). Interestingly, at Cape Perivola the MV6 lavas are in turn covered by a thin red scoriae deposit, which, chemically, is very similar to the US2 below MV6. Lavas, which were mistakenly classified by Huijsmans as MV2C, form the base of the MV6 lava sequence between Cape Perivola and Cape Heptapedio and no pyroclastic deposits occur between MV2C and MV6 (Huijsmans 1985, Pl. 1). These lavas at the base of MV6 are low-silica andesites. On top of the MV complex, Cape Riva and Minoan deposits occur. The former deposits especially occur in the fault-controlled depression between Megalo Vouno and MPI.

GEOCHEMISTRY

The volcanics from the MV complex form a medium to high-K calc-alkaline suite, and, as such, are similar in chemistry to volcanics from other eruption centres, e.g. MPI and Skaros (Huijsmans et al. 1988; Huijsmans and Barton 1989). The observed compositional trend (basalt-dacite) can, for most elements, be explained in terms of fractional crystallization, involving removal of the observed phenocryst phases. However, all MV1 and MV2 lavas, which pre-date the 100 Kyr LP eruption and some of the younger (< 54 Kyr) volcanics, which pre-date the Minoan eruption, show high Sr-contents (300-450 ppm) (Fig. 4). The composition of these lavas is in general low-silica andesite, although minor basaltic andesites occur in this group. In addition, these lavas are strongly phyric in character and they show abundant disequilibrium and reaction textures. Some of the high Sr-lavas are also high in Ni.

High Sr-contents may partly reflect time-dependent geochemical differences, which are observed between successive volcanic centres on Santorini (Huijsmans et al. 1988). MPI lavas are higher in Sr and some incompatible trace elements in comparison to Skaros lavas, whereas Skaros lavas are again more enriched compared to the youngest Kameni lavas. Huijsmans et al. (1988) attributed these variations to processes which occur in the upper mantle, i.e. a progressive depletion of the upper mantle source region with time.

Most of the younger (< 100 Kyr) MV volcanics show trace element characteristics that are similar to the geochemistry of the Skaros lavas (trend I in Fig. 4), whereas the oldest lavas (MV1 and MV2) show Sr-concentrations which are even higher than Sr- concentrations observed for MPI lavas (d. trends II and III in Fig. 4). The decrease in Sr with time for MV lavas could again be explained on the basis of the above-described scenario of a progressive depletion of the upper mantle. However, some of the younger lavas (e.g. basal lavas of MV6; < 34 Kyr) have high Sr- contents (up to 360 ppm) as well, i.e. they plot in trend II or III of Fig. 4). Apart from these lavas being relatively young, they are also intercalated in trend I lava sequences, i.e. lavas with Skaros characteristics. On the basis of unpublished petrological and mineralogical data it is concluded that some of the high Sr-lavas from MV are the product of mixing between relatively evolved liquid compositions (high-silica andesite or dacite) and a Sr-rich component. The latter could be a plagioclase-phyric basaltic liquid or a plagioclase-rich crystal mush. A similar mixing scenario for Santorini basaltic andesites has been described by Pyle et al. (1988). A detailed description of the petrology and geochemistry of the MV lavas will be given elsewhere.

DISCUSSION

       Stratigraphic relationship of MV complex with other eruption centres:

The stratigraphic relationship between MV and MPI is difficult to establish. The lavas of the core complex of MV are separated from the lava sequence of MPI by a zone of major and minor faults (Fig. 3), which are related to the NE-SW trending Graben system (Fig. 2). One of these faults (designated with 'ff' in Fig. 3) cuts off the complete stratigraphic sequence of MPI. In this faulted contact zone a stratigraphy is hard to establish, but some units from the TPF can be recognized. The assumed LP1 or CT3 deposit in the MV complex also occurs as an irregular, off-set horizon between MV and MPI. This conspicuous marker horizon covers lavas designated with 'P' (unspecified MPI lavas) in Fig. 3. In addition, pink coloured pyroclastics from the Cape Thera eruption, as well as MP deposits can be recognized. In the upper part of the sequence US1 with Skaros-like flows (cycle IV?) and US2 on top is also present.

This is based on the fact that no LP2 deposits can be observed in the MPI lava pile. However, a major period of erosion has followed the LP eruption (see above) and LP deposits may have been eroded from the steep slopes. Nowadays Minoan deposits are also mostly absent from the MPI steep slopes, due to erosion! Druitt et al. (1989) also report the occurrence of Middle Pumice deposits just below the summit of MPI (P4; see Fig. 3) and a lava from the summit has been dated at 79 Kyr (Pyle et al. 1988). If we assume the presence of MP below the summit lavas, we emphasize that most of the MPI lavas can still be old and that we have no possibility to discriminate in age between the core complex of MV and the largest part of the MPI lava sequence.

Prior to eruption of the Skaros lavas, MPI was separated by water from the erosional surface through TPF deposits below Phira. This can be concluded from field observations in the western caldera walls between MPI and Phira and from the geological map of Pichler and Kussmaul (1980). Subaerial Skaros activity (Sk1) may have started with the growth of an andesitic-dacitic dome in the middle of this sea. The occurrence of US1 in the Skaros sequence places constraints on the stratigraphic relationship between MV and Skaros. Field observations in the northern caldera wall of the MV section indicate that US1 and US2 deposits are present. Between these deposits a section of Skaros-like flows and pyroclastics occurs (Fig. 3). It is concluded that these lava flows are related to eruptive activity of cycle IV of the Skaros volcano, which implies an extension of the Skaros lava shield throughout much of the northern part of the present caldera at the end of the Skaros eruptive activity.

This picture for Skaros is also supported by observations on Therasia. Druitt et al. (1989) suggest that the Th2 lavas (Pichler and Kussmaul 1980; Huijsmans 1985) of Therasia represent a feather edge of the Skaros shield. Th2 lavas also occur west of Oia at sea-level (Fig. 3). According to Druitt et al. (1989, Fig. 14) Th2 lavas occur between US1 and US2 deposits, which implies that these lavas also represent cycle IV activity of Skaros. The older Th1 lavas of Therasia (Huijsmans 1985, Pl. 1), which occur at sea-level, are high-silica andesites. These lavas show typical circular flow-banding patterns, and most outcrops in the caldera wall are cross-sections through westward dipping lava flows. Th1 lavas are very similar in appearance and composition to the high-silica andesitic domes, which occur in the MV complex, just above the LP deposits. We conclude that: 1) the Th1 lavas are similar to the silicic domes in the MV sequence; and 2) the Th1 lavas represent the post-100 Kyr coast of the MV lava shield.

       Pre-Minoan caldera collapse:

Acording to Heiken and McCoy (1984) there is a distinct change in slope at an elevation of 200 m for the caldera wall profile of MV. According to these authors, slopes above 200 m are about 27° and have a poorly developed soil. The lower part of the caldera wall is much steeper (45°) and rocks look more fresh. The caldera wall profile of MPI, as well as the caldera wall profile below Phira (section through TPF) shows similar characteristics. Heiken and McCoy (1984) concluded that the northern caldera walls were only partly truncated by Minoan caldera collapse. This conclusion is supported by the occurrence of Minoan deposits which occur on the upper, caldera-side slopes of MV. Recently, a primary Minoan pumice-fall deposit was also found below Phira (Cape Apanophira; Francaviglia pers. comm., 1988). Again, this is support for the hypothesis of partial, pre-Minoan collapse of the northern area.

CONCLUSIONS

Taking into account available field observations and the present topography of the island, the following reconstruction for the development of the volcanic field of Santorini is envisaged:

Pyroclastics of the Cape Therma 1-3 eruptions were deposited in an area in the southern part of the present caldera. Ponding of these deposits in the topographic depression between the oldest Akrotiri volcanoes and the basement resulted in partial welding. In the northern part of the present caldera a complex, 150-200 m high, volcanic shield was formed above sea-water. The shield was dissected by an active fault system, which forms the northern edge of the major NE-SW Graben system (Fig. 2). We tentatively suggest that the dissected shield forms the present cores of MV and MPI. The MV shield was formed before eruption of Cape Therma 3 deposits (Druitt et al. 1989).

Deposits from the subsequent LP1 and LP2 eruptions draped the topography in the south and also covered and enlarged parts of the volcanic shield in the north. Subsidence of the southern area may have occurred during these eruptions (Fig. 5a). Volcanic activity was continued on MV by the eruption of silicic flows and domes on the MV shield. These flows and domes preferentially occur in depressions in the palaeotopography, which are partially filled with LP deposits. Some of these silicic flows may have flowed to the western edge of the MV shield, where they form the present Th1 lavas on Therasia. A period of volcanic inactivity must have followed the eruption of the silicic lavas. This resulted in considerable erosion of the LP deposits on the flanks of the northern lava shields (MV and MPI).

Subsequent volcanic activity from several, major pyroclastic eruptions (Cape Thera, Middle Pumice and Vourvoulos eruptions) enlarged the volcanic shield in the north considerably. Vent systems related to these eruptions were situated along the southern coast of the northern shield complex, in the vicinity of the present Kameni islands. The location of the vents and the nature of the vents, i.e. situated on the intersection of two active fault systems, enabled magma to interact with sea-water. Subsequently, the Skaros silicic dome (Sk1; Huijsmans and Barton 1989)appeared above sea-level between MPI and Phira. Sk1 activity was followed by eruption of the Sk2 lavas of cycles I, II and III, which have been described in detail by Huijsmans and Barton (1989). This volcanic activity connected the northern and southern part of Santorini (Fig. 5a). The lavas of Sk2 (basalts and basaltic andesites) also partly embedded the MV and MPI shield. After eruption of US1, which may have resulted in a small caldera or collapse structure in the Skaros shield, Skaros activity continued (cycle IV). Outpouring of these lavas enlarged the Skaros shield considerably, i.e. the western edge of the shield (present Th2 lavas of Therasia) and spill-over of these lavas over the MV complex (MV3; Fig. 3). Collapse structures as a result of US2 eruption, may have developed on the Skaros shield. The latter eruption was followed by the formation of a silicic dome complex in the northern part of Santorini (Skaros-Therasia dome complex of Druitt 1985), which capped the Skaros shield.

Finally, the Cape Riva eruption occurred, with an inferred vent position between Oia and Therasia. The effect of this eruption could well have been: 1) caldera collapse around the cape Riva vent (S2; see Fig. 2); and 2) more important, partial collapse of the Skaros shield along a system of ring-dykes, related to the existing fault systems (Fig. 5b). The upper parts of the present caldera walls in the north with the poorly developed soils (Heiken and McCoy 1984) could be related to this event. After collapse the partially dissected MV complex with two cinder cones on top stood out as a dominating volcanic landmark on the northern edge of the non-flooded caldera structure, up to Minoan times.

The Minoan eruption itself caused the break-up of most of the Skaros-Therasia dome complex, further collapse and flooding of the northern area. Many of the black, glassy blocks in the Minoan deposit are presumably fragments of the silicic dome complex, which is also supported by microscopic observations and chemical analyses.

Addendum

Outcrop of US1 in the Skaros lava sequence between Phira and Merovigli (Fig.6). Upper-right corner shows cycle IV lavas of Skaros, lower left-corner shows cycle III lavas. US1 deposits (US1A of Druitt et al. 1989) are intercalated. Cape Tourlos visible in the background.

----------------------------------------------------

------------------------------------------------