Abstract
Geophysical, structural, geochronological and geomorphological data indicate that the Psatha, East Alkyonides, Skinos and Pisia faults are Holocene-active structures whereas the status of the West Alkyonides, Strava, Perachora and Loutraki faults is less certain. We see no evidence for significant lateral surface fault growth. New data for late Pleistocene footwall uplift of the Psatha fault are comparable with previously estimated Holocene rates. Pre-Holocene stratigraphic sequences in the Alkyonides Gulf allow calculation of vertical displacement on the Skinos fault of 1.42–1.60 km over a period of >0.6 Ma. Previous palaeoseismological studies indicate comparable displacement rates extrapolated to 0.61–2.20 Ma, whereas extrapolation of previous geodetic data indicate a range of 0.17–0.46 Ma. The latter is too short given the evidence of the stratigraphic record, signifying either that these data may not be representative of longer-term rates, or that significant deformation has taken place elsewhere, for example, on offshore antithetic faults. A case is established for uniform late Quaternary (post-MIS 7) uplift of the Perachora peninsula at rates of c. 0.2–0.3 mm a−1. The lack of regional tilting over Perachora–Corinth–Isthmia is in marked contrast to the situation in the Alkyonides–Megara basins to the east.
The normal faulting and crustal deformation in the eastern Gulf of Corinth (Fig. 1) were first brought to light by field investigations of the effects of the 1981 earthquakes (Jackson et al. 1982a). Since then the area has received widespread scrutiny from the point of view of seismology (e.g. King et al. 1985; Taymaz et al. 1991; Abercrombie et al. 1995; Hubert et al. 1996; Hatzfeld et al. 2000), palaeoseismology (Ambraseys & Jackson 1997; Collier et al. 1998), marine geology and geophysics (Perissoratis et al. 1986; Sakellariou et al. 1998; Leeder et al. 2002; Stefatos et al. 2002), tectonics (Jackson et al. 1982b; King et al. 1985; Vita-Finzi & King 1985; Jackson & White 1989; Collier et al. 1992; Armijo et al. 1996; Leeder et al. 2003; McNeill & Collier 2004), structural geology (King et al. 1985; Vita-Finzi & King 1985; Roberts & Stewart 1994; Roberts 1996; Roberts & Koukouvelas 1996; Morewood & Roberts 1999, 2002; Cowie & Roberts 2001), sedimentology (Bentham et al. 1991; Papatheodorou & Ferentinos 1993; Leeder et al. 1998), sequence stratigraphy (Gawthorpe et al. 1994), basin analysis (Leeder et al. 2002) and tectonic geomorphology (Vita-Finzi & King 1985; Leeder et al. 1991, 2002). As a result of all this attention, the normal faults in the eastern Gulf of Corinth are undoubtedly some of the most studied examples in the world. Yet the cited literature contains a divergence of views concerning the status, definition and history of the faulting and vertical crustal deformation, central themes of this study.
General location and tectonic summary maps for the context of the study area in the eastern Gulf of Corinth, Central Greece, within its Aegean context (inset: plate velocities after McCluskey et al. 2001). Topography and onshore and offshore faulting after Jackson et al. (1982a), Higgs (1988), Papatheodorou & Ferentinos (1993), Goldsworthy & Jackson (2000), Leeder et al. (2002) and Stefatos et al. (2002). C, Corinth city; X, Xylocastron; Is, Isthmus of Corinth; P, Perachora peninsula. The area that includes Corinth, Isthmia and the Perachora peninsula enclosed by the bold dashed line is thought to have been uniformly uplifting for the past 240 ka (see text for discussion). Dashed line A–B in the Alkyonides Gulf marks the approximate line of the sketch section in Figure 3.
Previous work on fault activity and tectonics
Two major parallel faults, the Pisia and Skinos faults, were recognized (Jackson et al. 1982a) from surface breaks of the 1981 earthquakes on the south side of the Gulf of Corinth (Fig. 2; we do not concern ourselves further here with the antithetic faults on the northern side). Each fault comprised surface rupture strands 0.25–7 km long, separated by short (up to 1 km) gaps. Surface breaks with up to 1.5 m throw and slip vectors generally to the NNE were mostly coincident with basal slopes to major linear escarpments. Bornovas et al. (1984) subsequently mapped five further western fault ruptures ascribed to the 1981 earthquakes (A–E in the inset of Fig. 2). Seismological investigations of the 1981 earthquakes and their aftershocks initially suggested that ruptures along the Skinos and Pisia faults were responsible for the second of the main earthquakes, and that the first was due to rupture along an offshore fault inferred to lie along a linear bathymetric gradient on the northern margin of the Perachora peninsula (Strava fault (Fig. 2) of Papatheodorou & Ferentinos 1993; Hubert et al. 1996; Perachora fault of Stefatos et al. 2002). Subsequent work (Taymaz et al. 1991; Abercrombie et al. 1995) led to reappraisal of epicentral locations, and Taymaz et al. assigned movement of the two segments of the Pisia fault to the first earthquake with mean displacements of 0.9 m. The second earthquake was associated with the Skinos fault and a mean displacement of 0.3 m.
Map to show disposition of normal faulting in the Alkyonides Gulf and adjacent onshore areas, eastern Gulf of Corinth. Onshore 1981 fault breaks after Jackson et al. (1982a), with eastern extent of Skinos fault after Collier et al. (1998). Offshore faults mapped by Papatheodorou & Ferentinos (1993), Leeder et al. (2002) and Stefatos et al. (2002), from whom the fault names are taken. Numbered faults: 1, Pisia fault; 2, Skinos fault; 3, East Alkyonides fault; 4, Psatha fault; 5, West Alkyonides fault. Filled circles 6–8 are the favoured locations for the epicentres to the three 1981 earthquakes (after Taymaz et al. 1991). The data on status of uplifting and subsiding coastline are from Jackson et al. (1982a), Hubert et al. (1996) and Leeder et al. (2002, 2003). The dashed line labelled SPK 25 is the seismic reflection ‘sparker’ line shot in 1996, used subsequently in this paper to determine stratigraphic controls on tectonic tilting. Inset shows faults A–E in the western Perachora peninsula mapped by Bornovas et al. (1984); see also Fig. 6b.
Fault modelling by Hubert et al. (1996) supported the view of Taymaz et al. (1991) that the offshore Strava fault did not rupture in 1981. Hatzfeld et al. (2000) reported results of microseismicity recorded in the summer of 1993 together with reprocessing of 1981 aftershocks. Activity was concentrated in the footwall of the Strava fault and the hanging walls of the Pisia and Skinos faults, profiles of which indicate moderate dips of c. 45°, with no evidence for any dip decrease with depth.
Roberts (1996) proposed that the faults bounding the southern margin of the Alkyonides Gulf defined a ‘South Alkyonides Fault Segment’, with the 1981 ruptures affecting only part of this. Morewood & Roberts (1999, figs 11 and 12) proposed that the surface rupture trace of the stationary deep western tip to the South Alkyonides Fault Segment has gradually propagated westwards over the past 300 ka, displacing marine terraces and producing wind gaps as drainages were truncated in the footwall of the propagating rupture. Lateral fault growth has also been postulated for the eastern tip to the South Alkyonides Fault Segment, along the Psatha fault (Morewood & Roberts 2002).
Geodetic surveys indicate up to 100 years of aseismic crustal strain in central and southern Greece (Clarke et al. 1997; Davies et al. 1997; Briole et al. 2000). Separation across the Gulf of Corinth varied, although not as a smooth latidudinal gradient, from 12 ± 2.6 mm a−1 in the west to 6 ± 2.7 mm a−1 at the longitude of the mouth of the Alkyonides Gulf in the east. Collier et al. (1998) completed the first conclusive palaeoseismological investigation in the Gulf of Corinth at a site located on an alluvial fan cut by the Skinos fault (Fig. 2; Vramvrakies trench site) where a 0.6 m 1981 displacement was superimposed upon older scarps of unknown age. Average recurrence intervals of around 350 years were determined from 14C dates of hanging-wall stratigraphic successions, with mean vertical displacement rates in the range 0.7–2.5 mm a−1.
Leeder et al. (1991) mapped uplifted shoreline deposits in the footwalls to the active Skinos and Psatha faults. U-series dating of corals associated with deposits at c. 35–40 m gave a maximum age of MIS 5, probably 5e, and a mean rate since then of c. 0.3 mm a−1. The age of initiation of footwall uplift by the active coastal faults was extrapolated as 1–1.5 Ma. In modelling crustal deformation associated with the faulting along the Alkyonides margin, Armijo et al. (1996) determined that a post-MIS 5e slip of 220 m fitted the uplifted 40 m inner edge of the MIS 5e terrace, giving a mean throw rate of some 1.3 mm a−1 along the 45°-dipping fault. Novel diagenetic evidence for footwall uplift along the coastal faults was presented by Roberts & Stewart (1994). Those workers recorded marine dolomite at altitudes up to 650 m above sea level and, making use of the uplift rate determined by Leeder et al. (1991), inferred that the uplift had taken place over the past 2.2 Ma.
Deformation and faulting: offshore and onshore perspectives
Key evidence for the evolution and activity of faulting is located offshore. Early marine geophysical surveys (Perissoratis et al. 1986) supported the existence of major offshore faults. In the first detailed study, Papatheodorou & Ferentinos (1993) documented faulting and its role in submarine mass movements in the western approaches to the Alkyonides Gulf and defined the Strava graben as a major eastwards-narrowing structure bounded by the north-throwing Strava fault (Fig. 2). Sakellariou et al. (1998) and Leeder et al. (2002) presented further results on offshore faulting in the gulf.
Psatha fault
This WSW–ENE structure (Fig. 2) was mapped 1.5 km offshore (Leeder et al. 2002) as the Holocene-active western continuation of a 7.7 km long onshore scarp (Morewood & Roberts 2002). Seismic coverage suggests, but line-spacing is too coarse to prove, that the fault tip ends before the East Alkyonides fault. Onshore, the Mesozoic limestone footwall scarp features strongly oblique kinematic markers. King et al. (1985) proposed that the fault lost displacement rapidly to the east, an observation not supported by recent mapping (Morewood & Roberts 2002). King et al. (1985) interpreted 1981 aftershock data in terms of a tip-zone model for the oblique slip, a hypothesis later developed by Roberts (1996) and, in more detail, by Morewood & Roberts (2001).
The base of the fault footwall is cut by a prominent 2 m high uplifted Holocene marine solution notch, a feature previously used (Leeder et al. 1991) to estimate a mean footwall uplift rate for the Psatha fault of c. 0.3 mm a−1 since the local Holocene marine highstand abutted the Psatha fault scarp at c. 7 ka. We stress that this rate is a minimum value, as the notch would have required considerable time to form following the local highstand; hence we may be looking at more recent Holocene displacements of considerable proportions. New evidence for footwall uplift occurs in the form of raised beach gravels preserved in partially exhumed fissure fills (neptunean dykes), locally cemented onto the fault surface. The deposits comprise well-sorted and rounded polymict (serpentinite, chert, limestone) pebbles cemented by fibrous low-Mg calcite that we have 14C-dated to 24 690 ± 170 years (sample OxA-10101), approximately the Last Glacial Maximum (LGM). The cement has a δ18O of −5.1‰ VPDB, a composition consistent with precipitation from meteoric water. The highest gravel is at +10.2 m above mean sea level (MSL), about 2 m beneath a discontinuous ledge present along the western margin of the fault footwall that may represent a small palaeocliff. We cannot date the gravel directly but the youngest plausible age for its formation is OIS 5a at c. 83 ka, when sea level was about 10 ± 4 m lower than at present (Cutler et al. 2002; see later discussion and Table 3). This gives a maximum estimate for a mean footwall uplift rate from 83 ka to the present as 0.24 ± 0.05 mm a−1, a value similar to the mean Holocene rate.
The onshore trace of the Psatha fault is noteworthy for the existence in its hanging wall of several ‘beheaded’ drainage catchments (Gawthorpe et al. 1994; Morewood & Roberts 2002) that once flowed southwards into the Megara basin when this feature was an active fault-bounded depocentre. These drainages end in closed, flat-bottomed depressions, the southern sides of which are downfaulted against the steep footwall scarp of the Psatha fault. The latter thus clearly postdates activity on the Megara basin faults. Morewood & Roberts (2002) proposed that the eastward decline in throw documented for the Psatha fault was due to lateral propagation of the fault tip in the same direction, although no direct evidence for this is available.
East Alkyonides fault
This is an active (Leeder et al. 2002), 10 km long, WSW–ENE-trending structure (Fig. 2), mapped for most of its length along the base of a prominent 250 m high linear submarine escarpment with a maximum slope of 40°.
West Alkyonides fault
This east–west-trending fault (Fig. 2) is at least 15 km long. Leeder et al. (2002) presumed it to be largely inactive relative to the immediately onshore Skinos fault, although some degree of late Quaternary slip cannot be ruled out.
Skinos fault
This onshore structure has already been described in conjunction with the Jackson et al. (1982a) mapping of the 1981 fault breaks. We note here that Collier et al. (1998) located the eastern tip of the fault close to their trenching site in the Vramvrakies alluvial fan (Fig. 2), thereby considerably shortening, to c. 6 km, the surface extent of the active structure defined by Jackson et al. (1982a), Taymaz et al. (1991) and Hubert et al. (1996). We estimate the total throw across this fault in the area of its intersection with the East Alkyonides fault below.
Pisia fault
Concerning the western extensions to the 1981 faulting (the latter discussed above) proposed by Bornovas et al. (1984), only some traces of the ruptures may be seen today, because of subsequent building and agricultural activity. Observations and measurements of cracks and small (centimetre-scale) surface scarps throwing north were made by Bornovas et al. (1984) and, more recently, by N. Morewood (pers. comm.) and G. Roberts (pers. comm.). Although accepting the case for the existence of these 1981 ruptures, we see no field evidence for pre-1981 Holocene activity on these western fault strands. Thus the ruptures: (1) do not significantly displace Holocene raised shoreline markers in the most western examples (A and B in Fig. 2); (2) do not occur along well-defined precursor linear fault scarps (A, B, C and E in Fig. 2); (3) do not appear as penetrative, through-going displacive structures in good vertical exposures of Holocene alluvium (scarps D and E in Fig. 2). We contrast this situation, for example, with that of the 1981 Skinos fault and precursor scarps, which have been mapped across, and excavated through, Holocene alluvium (Collier et al. 1998). However, we stress that our arguments are negative ones: further systematic mapping and palaeoseismic trenching in this critical area is clearly necessary to resolve the issue. Hubert et al. (1996) also extrapolated the 1981 ruptures mapped by Jackson et al. (1982a) westwards, in this case to a prominent linear escarpment in Mesozoic limestone bedrock on the northern margin of Lake Vouliagmeni. We have seen no evidence that this degraded fault-line scarp is Holocene active.
Strava fault
This trends east–west offshore a few kilometres north of the Perachora peninsula (Fig. 2; Papatheodorou & Ferentinos 1993; Stefatos et al. 2002) and lies along a prominent bathymetric gradient that defines the southern faulted margin to the Strava graben. The eastern and western extent of the fault are poorly defined by available seismic reflection lines, and seismic coverage is insufficient to assess the degree of Holocene to Late Quaternary activity on the fault.
Perachora fault
This NE–SW-trending structure (Fig. 2; Papatheodorou & Ferentinos 1993; Stefatos et al. 2002) also has poorly defined tip zones. The occurrence of uplifted Holocene to Late Quaternary marine shorelines on the northern Perachora peninsula coastline might at first sight be taken as evidence for activity along the structure. Although we cannot rule this out, as we discuss below, the raised shorelines are part of a peninsula-wide phenomena and show no southward backtilt, a feature that might be expected if it was due entirely to footwall effects from the Perachora fault.
Loutraki faults
Two faults occur along the SW flank to the Gerania Range north and NW of Loutraki (see Figs 2 and 6a). The southernmost trends east–west and features as an active structure in some previous structural maps of the Perachora peninsula (Armijo et al. 1996, fig. 2). A similar structure has been extrapolated offshore along the NW–SE-trending southern coastline of the peninsula to Cape Heraion (e.g. Roberts & Jackson 1991). The two onshore faults clearly cut the surface across Mesozoic limestones, but to the east the faults are obscured amongst poor exposures of ophiolitic basement. Natural exposures of the two fault footwall scarps reveal highly irregular, eroded and karstified limestone free faces, with quarry exposures of the slickenslided fault planes in contact with Quaternary alluvial fan and Holocene talus cone deposits in the hanging wall. A thick deposit of hydrothermal aragonite obscures the southernmost fault on the coast immediately NW of Loutraki town. The field evidence suggests to us that both the Loutraki faults are probably Holocene-inactive, and that further studies are needed to assess the status of the northern fault.
Summary of faulting
Of the eight normal faults considered as past major contributors to the tectonic evolution of the study area only four, the Psatha, East Alkyonides, Skinos and Pisia faults, yield direct evidence that they are Holocene-active. Further studies are needed to assess the status of the remainder. The surface expressions of the various faults making up the South Alkyonides Fault Segment are generally distinct, but that does not prove that the South Alkyonides Fault Segment behaves as a kinematically coherent (soft-linked) array as previously proposed (e.g. by Roberts 1996; Morewood & Roberts 2001).
Long-term deformation rates from Alkyonides hanging-wall basement tilting
Previous attempts to determine long-term deformation rates (Roberts 1996; Collier et al. 1998) along the active south Alkyonides faults have used approximations for vertical stratigraphic separation between hanging-wall and footwall cutoff of pre-rift basement against fault surfaces (see Figs 3 and 4 for summaries and definitions). Approximations were necessary because of the lack of datum for directly determining cutoff. Two chief problems in estimating separation remain unsolved. First, the contribution of footwall uplift well above the datum provided by late Quaternary to Holocene marine shoreline indicators cannot be precisely estimated. Second, calculation of the fault slip rate requires knowledge of the time elapsed since fault and basin initiation. This has been possible only by linear extrapolation of the footwall uplift rate determined for the MIS 5e marine terrace at Alepochori (Leeder et al. 1991; Collier et al. 1992) as applied to the maximum elevation of the Megara basin floor, c. 400 m. The youngest age thus computed, c. 1 Ma, has been used to constrain estimates for timing of initiation of the Gulf of Corinth rift and modelling of Aegean tectonics (e.g. Armijo et al. 1996; Duermeijer et al. 1999), although arguments from the western Gulf have also been made for this age (Goldsworthy & Jackson 2001).
Schematic section along the line A–B in Figure 1 across the eastern part of the Alkyonides Gulf and Megara basin to show major morphological features useful in parameterization of rates of uplift and subsidence associated with displacements along the active Psatha–East Alkyonides faults.
We make use of tilt-induced stratigraphic features imaged by offshore geophysical surveys to estimate fault throw (Figs 3 and 4). Tilt-block geometry (Jackson & McKenzie 1983; Barr 1987; Jackson & White 1989) requires normal faults of dip θ to rotate horizontal pre-stretch basement to angle ψ with the understanding that at time t0 ψ=0 and θ=θ0. We develop a solution to determine ψ from offshore stratigraphic and sedimentological data using the unpublished sparker dip line SPK 25 (Fig. 5), shot slightly oblique to the surface intersection of the active Skinos and East Alkyonides faults. The seismic data reveal a strong tilted reflector, interpreted from wider seismic mapping and spatial alignment with onshore basement ridges as pre-rift Mesozoic basement. This basement is overlain by more than six well-defined pre-Holocene stratigraphic sequences, which we interpret as the record of the periodic, low-frequency (c. 100 ka), global sea-level cycles (e.g. Chappell & Shackleton 1986; Gallup et al. 1994; Cutler et al. 2002; Siddall et al. 2003; Fig. 5) developed over the last 600 ka. The evidence for this assignation comes in the form of high-amplitude reflectors interpreted as transgressional to highstand events, which progressively onlap the tilted hanging-wall basement. These reflectors each overlie major packages of prograding and offlapping shoreline deltaic clinoforms interpreted as lowstand deposits (see Fig. 5 and Collier et al. 2000, fig. 1). It is probable that a sedimentary record of higher-frequency sea-level change exists within these sequences, but an exploration of this aspect is beyond the scope of the present study. We assume that tilting has been achieved by progressive rigid-body rotation of the upper crust on the active Skinos fault. The effective tilt radius λ (Fig. 4) is given by the 17.5 km fulcral distance from the active Kaparelli–Livadostros antithetic faults (Fig. 2) at the northern gulf shoreline to the outcrop of the eastern strand of the Skinos fault. We propose that the relatively smooth hanging-wall basement surface has been formed progressively as horizontal to gently sloping shoreline topography, produced by marine shoreface erosion along a ravinement surface. We also make the assumption that tilt has not been substantially reduced by the antithetic faulting.
Definition diagram for determining tilting across normal faulted tilt blocks. Symbols defined in text.
Using water depths and sediment thicknesses computed along SPK 25 under mapped survey points of known spacing, determined from accurate differential global positioning system (GPS) navigation, we derive 10 determinations of ψ, with a mean value 3.9° (2σ=±0.34). We use these values (Table 1) to estimate the maximum downthrow of the Skinos fault as 1.12 km (range 1.02–1.20 km) from basic trigonometry. Although we have no direct knowledge of the extent of footwall uplift at the intersection of the Skinos and East Alkyonides faults (see discussion above), we make the reasonable assumption that it is comparable with the 0.4 km estimate for the main part of the East Alkyonides fault a few kilometres east, around Alepochori (Leeder et al. 1991). We compute a total fault throw of some 1.52 km (range 1.42–1.60 km).
Data derived from seismic line SPK 25
We next enquire, using independently determined rates of fault displacement from geodetic and palaeoseismic studies, how long the basin may have been active and whether our calculated long-term rates of displacement are compatible with these and with the limited subsurface stratigraphic controls available. As discussed above, geodetic data for crustal extension at the longitude of the mouth of the Alkyonides Gulf (Clarke et al. 1997) give rates of 6 ± 2.7 mm a−1 over the past 100 years. Assuming this extension is taken up only by vertical displacement along 45°-dipping normal faults in the upper crust (see the evidence for this value given by Taymaz et al. 1991; Hatzfeld et al. 2000) along the southern gulf margin, ages of deformation appropriate to the mean total 1.52 km vertical displacement are the range 0.17–0.46 Ma, with a mean of 0.25 Ma. These times are too short given both the known post-MIS 5e rate of footwall uplift and the stratigraphic record of the Alkyonides Gulf sediments. It seems that (1) GPS-determined displacement for the western limit to the Alkyonides Gulf must either decline eastwards (by >50%) into the Alkyonides Gulf, (2) deformation rates over the past 100 years are significantly greater than those averaged over the main part of the Holocene and Late Pleistocene, or (3) there is significant displacement taken up by antithetic faulting (see Fig. 2) along the northern gulf shoreline.
Adopting displacement rates of 0.7–2.5 mm a−1 from Holocene palaeoseismology (Collier et al. 1998), the calculated long-term displacement requires a time in the range of 0.61–2.20 Ma for the total duration of fault activity. The latter is a reasonable upper range estimate, given the stratigraphic constraints available.
The Perachora peninsula: part of a uniformly uplifting crustal terrane?
Introduction and previous work
The Perachora peninsula lies in a pivotal geographical position, defining a large-scale left step between the active faults of the main Corinth and Alkyonides sub-basin (Fig. 1). Vita-Finzi & King (1985), on the basis of field observations and interviews with local inhabitants, suggested that much of the peninsula underwent subsidence as a result of the effects of the 1981 earthquakes, a conclusion at odds with later observations in the area (Hubert et al. 1996). Regarding Holocene to Late Quaternary deformation, several workers (e.g.Vita-Finzi 1993; Pirazzoli et al. 1994; Dia et al. 1997; Leeder et al. 2003) subsequently proved persistent Holocene to Late Quaternary uplift by radiometric dating of marine fossil assemblages in raised marine shoreline deposits.
Raised shorelines
We have mapped beachface deposits, terrace platform inner edges, notched and Lithophaga-bored shorelines along palaeocliffs and sea stacks at 1:5000 scale, aided by local levelling, altimetric determinations and SPOT digital elevation model (DEM) images to give palaeo-sea levels to an accuracy of a few metres or less. Preliminary results of some of our field studies were reported by Leeder et al. (2003). We have found palaeo-shoreline deposits up to maximum altitudes of 220 m. Palaeo-sea-level elevations are correlated with eustatic sea-level curves, underpinned by previously published and numerous new U-series dates on coral samples accurately located with respect to local palaeo-shoreline elevation (Fig. 6b; Table 2). We have also mapped areas surrounding shoreline features to search for active fault traces that may have locally altered terrace elevations, by displacement and/or tilting.
Summary of new and published Perachora and Corinth radiometric determinations from C. cuspitosa corals: sample numbers, palaeo-shorelines, U and Th activity ratios and 230Th ages
Interpreted trace of seismic line SPK 25 to show hanging-wall dip-slope gradient and major reflectors used to define highstand onlap and lowstand offlap deposits. Gradient changes are much emphasized because of the great vertical exaggeration (×26) of the section. Inset shows schematic definition sketch to outline the geometric relations of highstand and lowstand shoreline deposits over a tilt-block hanging-wall surface in relation to the late Quaternary glacio-eustatic sea-level curve for the last 400 ka. TWT, two-way travel time.
(a) Detailed topography and sampling sites of raised shoreline features over the western Perachora peninsula assembled from field slips originally mapped at 1:5000 scale with the additional aid of altimetric observations, levelling and satellite imagery (after Portman 2004). (b) Generalized topography of the Perachora peninsula to show faulting, interior drainage basins and location of Panorama catchment (Fig. 7).
The coastal terraces of the Perachora peninsula are predominantly constructional. Their geometry is the expression of a complete sea-level cycle (sea-level rise, highstand and fall). The lower terrace bounding surface typically takes the form of an unconformity, formed as an erosive ravinement surface during the initial marine transgression, overlain by a coarse-grained transgressive lag deposit. The major part of the deposit comprises progradational highstand to early falling stage deposits. Unfortunately, terrace deposits per se cannot be used to locate palaeo-shorelines without identification of the inner edge, the highest and oldest position reached by the early marine transgression, which demonstrates palaeo-sea level to within a metre or so. This position can be achieved approximately using a simple topographic profile calculated from a detailed, accurate contour map, but in the field is most accurately recognized by the occurrence of unequivocal shoreline indicators such as notches, intertidal to very shallow subtidal trottoir (cemented red algal–vermetiform gastropod–bryozoan carbonate deposits) and Lithophaga borings. Four clusters of palaeo-shoreline elevations have been mapped around the Perachora peninsula: +8–12 m, +25–30 m, +35–42 m and +50–55 m.
Concerning intertidal notches, these are formed by combinations of chemical dissolution, physical abrasion and biogenic abrasion by boring, rasping and encrusting organisms (see Pirazzoli 1986; Trenhaile 1987). In areas of the gulf facing the most exposed wave action (north- and west-facing coastlines) notches have up to 1 m vertical amplitude relative to MSL. Recognition of uplifted notches is difficult at these locations, as the modern notch appears to be simply a composite of uplifted notches resulting from general high-energy coastline abrasion. This may indicate either steady uplift, or coseismic uplift insufficient to remove the base of the notch from the zone of erosion (i.e. ≤1 m). Because more sheltered areas around the peninsula record small-amplitude, multiple uplifted notches, we infer that uplift must have occurred periodically. The overall result is continuously undercut cliffs on exposed shorelines and vertical cliffs with multiple shallow notches on sheltered shorelines.
Pirazzoli et al. (1994) identified four notches at Cape Heraion. Kershaw & Guo (2001) identified three notches in the Cape Heraion area. We recognize three notches around Makrugoaz Ridge and Cape Sterna (Fig. 6b) at similar elevations. The highest notch is also well preserved at around 3 m along the coast NW of Loutraki. The elevation of all notches varies over short distances by up to 30 cm, probably as a result of contrasting local exposure. The notch at Mylokopi occurs at two significantly different heights, +1.7 m and c. +3 m, as a result of Holocene faulting (Pirazzoli et al. 1994). Field observations indicate that the elevation of the lower notch increases towards Cape Sterna, where it reaches +3 m elevation.
A chronology for Late Quaternary to Holocene raised shorelines
We make use of the Pirazzoli et al. (1994) 14C-dating of Lithophaga in life position at the level of the highest notches (c. 3 m) at both Mylokopi and Cape Heraion. The resultant corrected ages were c. 4400 bc and c. 3100 bc, respectively. The older age gives the minimum rate of uplift, 0.43 mm a−1, since the onset of the c.7 ka Holocene highstand. The lower notches dated from biogenic material obtained at Mylokopi by Pirazzoli et al. (1994) do not yield satisfactory results, exemplified by 14C ages of ad 400–540 and ad 1450–1830 from Vermetus triqueter at 0.8 m and Lithophaga sp. at 1.1 m, respectively. Chthamalus valves sampled by Pirazzoli et al. (1994) from the lower notch at Cape Heraion gave an age of ad 190–440, indicating minimal uplift rate of 0.06 mm a−1.
For dating Pleistocene shoreline deposits we have used U-series dating of Cladocora caespitosa. At present ubiquitous throughout the Mediterranean, including the Gulf of Corinth, this coral has a preferred bathymetry 5–25 m below sea level (Aguirre & Jimenez 1998; Peirano et al. 1998), the exact level being strongly influenced by local factors including wave energy, sedimentation rate, predation and substrate. Assessing the palaeobathymetry of this coral at a particular locality therefore requires a detailed palaeoenvironmental reconstruction. Dia et al. (1997), when calculating uplift rates in the Corinth area, estimated the palaeobathymetry of the corals as 10 ± 10 m, which, although within the likely living depth of the colonies, was not correlated in the field with terrace inner-edge elevations or other palaeo-sea-level indicators. Importantly, error ranges of ±10 m become large when applied to low sample elevations.
The results of U-series dating of corals by previous and current workers are shown in Table 2. UEA analyses are by alpha spectrometry, others are by thermal ionization mass spectrometry (TIMS). The difficulties of interpreting coral dates from the eastern Mediterranean area have been referred to by Collier et al. (1992) and Dia et al. (1997). The dates are organized into five groups based on sample locations, as follows.
(1) Cape Heraion A. Two samples are from marine sediments above prominent microbial bioherms (Richter et al. 1979; Portman 2004) at about +8 m MSL. The bioherms here are fractured and sometimes rotated landward. Both they and coralliferous sediments correlate with identical deposits 15 m higher in cliffs 200 m NE, indicating that post-depositional faulting has occurred. As the higher deposits may be correlated with the +25–30 m palaeo-shoreline, the lower deposits must also. UEA 798 gave an age of 136.5 +13.6/−12.1 ka, corresponding to early MIS 5e or late MIS 6, and UEA 842 gave an age of 181.7 +16.6/−14.3 ka, which corresponds to early MIS 6 or late MIS 7a.
(2) Cape Heraion B. Three new samples and one from Vita-Finzi (1993) were obtained from marine sediments overlying the bioherms that correlate with the 25–30 m palaeo-shoreline. Three samples had ages compatible with MIS 6: 137.6 +8.2/−7.4 ka (UEA 824), 150.8 +10.5/−9.4 ka (UEA 825) and 159.2 +9.9/−8.9 ka (UEA 826), although UEA 824 is also within error of MIS 5e. Sample VF 89/2 gave a clear MIS 5e age (128 ± 3 ka).
(3) Makrugoaz Ridge. Seven samples were dated from this site, which correlates with the 25–30 m palaeo-shoreline. All these samples (six from Leeder et al. (2003), one from Vita-Finzi (1993)) provide an MIS 5e age, with the exception of V-35-4b, which indicates MIS 5d.
(4) NW Lake Vouliagmeni. This coralliferous deposit correlates with the 35–42 m palaeo-shoreline and occurs by the road from the junction west of Lake Vouliagmeni to Cape Heraion. It includes three samples from this work and two from Dia et al. (1997). Three samples are within error of MIS 7a: 194.5 ± 0.8 ka (CTP2), 189.5 ± 4.8 ka (CTP2(2)) and 186.2 ± 15.7 ka (UEA 834); the last could also be MIS 6. UEA 823 gave an age of 304 ± 67.8 ka, which is within error of MIS 7e, MIS 8 and MIS 9, and UEA833 gave no age and had a 230Th/234U activity ratio of 1.020 ± 19.
(5) New Corinth. A final sample UEA 290601 originates from the New Corinth terrace (C, Fig. 1), and corresponds to a palaeo-shoreline c. +25 m above sea level. The sample yielded an age of 132.1 +8.2/−7.5 ka, within MIS 5e.
Perhaps the most telling conclusion from the above is that ages calculated by mass spectrometry from Vita-Finzi (1993) and Leeder et al. (2003), a technique requiring less coral for analysis and thereby allowing higher standards of sample selectivity, have consistently produced dates corresponding to MIS 5e from the 25–30 m palaeo-shoreline. From the mass spectrometry dates alone, it is reasoned therefore that the 25–30 m palaeo-shoreline corresponds to the MIS 5e highstand. Of the alpha-spectrometry dates, three samples (UEA 798, UEA 824 and 290601) are within error of MIS 5e and may also be correlated with the 25–30 m palaeo-shoreline (albeit indirectly for 290601).
Coral dates from deposits associated with the 35–42 m palaeo-shoreline include three ages (two of them TIMS dates from Dia et al. (1997)) that are within error of MIS 7a, the highstand preceding MIS 5e. All the corals from this stratigraphic level contain lower concentrations of U than those from 25–30 m, although all except one (UEA833, for which no age could be calculated) lie within the observed living and fossil coral U content distribution (Burnett & Veeh 1992). This may indicate some degree of diagenetic U loss, but equally it may simply reflect a lower marine uranium concentration in the nearshore zone of the Gulf of Corinth during MIS 7 relative to later interglacials. Late Holocene corals we have collected from dredged samples taken in the shallow gulf have a U content of 2.336 ppm. The fact that the two TIMS ages overlap at 1 SD and that stratigraphically the samples derive from a palaeo-shoreline topographically higher than the MIS 5 shoreline lead us to the conclusion that the dates are reliable.
Late Quaternary uplift rates
Estimation of uplift rates (Fig. 6b; Table 3) requires accurate knowledge of (1) elevation of palaeo-shoreline inner edge; (2) level of formative sea-level highstand relative to present sea level; (3) timing of onset of highstand. The highest Holocene notch is at c. +3 m with a minimum age of 6.4 ka, thus yielding a minimum mean uplift rate of 0.43 mm a−1. The best established Late Quaternary palaeo-shoreline is that of MIS 5e. For a +30 m palaeo-shoreline, a highstand some 6 m above present and an age of 130 ka for onset of highstand (Gallup et al. 1994), the mean uplift rate is 0.18 mm a−1. The 8–12 m palaeo-shoreline is undated, but as it is inset into downfaulted MIS 5e deposits it must be younger. The only highstands realistically placed to explain this are MIS 5c and MIS 5a, which both reached similar elevations (Chappell & Shackleton 1986; Gallup et al. 1994; Siddall et al. 2003) so that we are unlikely to be able to discriminate between them. An age of 100 ka for the combined MIS 5a–c highstand, with a palaeo-sea level of 10 ± 4 m below the present level (Cutler et al. 2002), gives a likely mean uplift rate in the range 0.2 ± 0.04 mm a−1.
Uplift rates: mean rates have varied between 0.18 and 0.43 mm a−1 since MIS 7e
Maximum elevation of the closely spaced MIS 7a–c palaeo-shorelines is 42 m. Sea level at those times, c. 200 ka, was possibly about ±10 m relative to present sea level (Gallup et al. 1994; Siddall et al. 2003), giving c. 30–50 m of uplift at a mean rate of c. 0.25 ± 0.06 mm a−1. The final palaeo-shoreline, 50–55 m above the present sea level, probably correlates with the MIS 7e highstand. Using an age of 240 ka and assuming the MIS 7e sea level was similar to the present sea level (Chappell & Shackleton 1986), a mean uplift rate of 0.23 ± 0.05 mm a−1 is calculated.
Discussion of uplift rates
Using the above data we deduce the following. (1) There are numerous distinctive and correlatable palaeo-shoreline features and terrace inner edges around the Perachora peninsula, from Cape Sterna in the north to Loutraki in the south. (2) Ages and elevations of mapped shoreline features correlate well with global sea-level highstands. (3) Rates of Late Pleistocene uplift in the western peninsula are reasonably consistent when likely errors caused by determination of highstand sea-level elevation are taken into account, with median values in the range c. 0.18–0.25 mm a−1. (4) There is no systematic evidence for displacement or warping of Holocene and late Quaternary shoreline deposits in our study area (Fig. 6a; see also Leeder et al. 2003, fig. 3B), apart from that caused by minor faulting such as around Cape Heraion (see also Kershaw & Guo 2001), Milokopi (see Pirazzoli et al. 1994) and on the SE margins of Agriliou Bay (Fig. 6a and unpublished mapping by the authors). This lack of evidence for warping leads us to doubt the hypothesis for late Quaternary surface lateral growth of the subsurface tip to the Pisia fault as proposed by Morewood & Roberts (1999). (5) The well-dated MIS 5e shoreline remains at 25–30 m over a 200 km area to the south of Loutraki that includes the Bay of Corinth, the Isthmus (where extensive minor faulting is superimposed on this uplift; Collier et al. 1992) and New Corinth (Leeder et al. 2003; Fig. 1).
Drainage, fault propagation and uplift
Preliminary geomorphological studies of drainage in the Perachora peninsula are aided by quantitative analysis of SPOT imagery and DEMs by RiverToolsTM software. The outstanding feature relevant to the present study is the presence of a number of closed interior drainage basins (Fig. 6b) with conspicuously level floors infilled by fine-grained continental sediment. Cols eroded in Mesozoic basement are present around the periphery of the depressions; an example at the head of Panorama gorge (Fig. 6b) issuing from Perachora interior basin was postulated by Morewood & Roberts (1999) to represent a wind-gap formed by a drainage subsequently truncated by westward lateral propagation of the surface trace to the South Alkyonides Fault Segment. Some of the cols must indeed mark formerly active valleys that led westwards and southwards out of the basin prior to active infill, but we have found no evidence that Panorama gorge was shut off by footwall uplift during fault propagation. Morewood & Roberts (1999) also interpreted fluvial deposits in Panorama gorge as depositional remnants of the pre-faulting drainage. We reinterpret these important deposits as mostly of fan delta origin (Fig. 7). They overlie and interdigitate with beach conglomerates that lie on a prominent littoral abrasion surface at about 150–155 m elevation. The shoreline sequence is buried and onlapped by south-prograding clinoform sets of fan delta beachface origins, comprising often well-sorted and rounded gravels of mixed (chert, limestone, serpentinite) provenance (Fig. 7).
General and detailed sketches from photomosaics to summarize the arrangement of ?Plio-Pleistocene sediments in a narrow strike-slip pull-apart basin in the central portion of Panorama gorge (location on Fig. 6b). The topmost fan delta and beachface deposits indicated in the detailed sketch erosively overlie the underlying deposits and their bounding faults and may date from c. 600 ka (see text for discussion).
In the context of drainage capture by possible fault propagation, the Panorama drainage catchment has evidently fed a fan delta and beachface like the present one for a very long time in the Late Quaternary, with shoreline and fan delta deposits of mixed provenance preserved as discontinuous remnants from sea level up to +220 m. Outcrops in the existing catchment south of the Perachora interior drainage basin have the necessary basement lithologies to supply the mixed sediment provenance. Any drainage closure at upper levels that may have taken place has clearly accompanied sedimentary infill of the Perachora basin and cannot be specifically ascribed to shut-off by fault propagation. Regarding the age of the gorge, our mean late Quaternary uplift rates for the peninsula indicate that the higher elevation marine deposits are possibly as old as 0.7–1.1 Ma. The gorge has progressively developed by incision as a result of uplift since that time and remains an active, although ephemeral, bedrock channel system.
Conclusions
(1) Stratigraphic interpretation of offshore geophysical data, onshore field mapping and geochronology, together with previously published data, indicate that the Psatha, East Alkyonides, Skinos and Pisia faults are Holocene-active structures, whereas the status of the West Alkyonides, Strava, Perachora and Loutraki faults is less certain.
(2) New data are presented for the maximum rate of late Pleistocene footwall uplift on the Psatha fault, 0.24 ± 0.05 mm a−1 over the past c. 83 ka. This is comparable with the Holocene footwall uplift rate on the same structure and within limits of a previously computed value for the adjacent East Alkyonides fault over the past 130 ka.
(3) More than six well-defined pre-Holocene stratigraphic sequences imaged from shallow seismic reflection profiles are interpreted as the record of periodic, low-frequency (c. 100 ka), global sea-level cycles that developed over the last 600 ka.
(4) The calculated tilt of pre-rift basement in the Alkyonides Gulf yields 1.42–1.60 km of vertical displacement across the 45°-dipping Skinos fault. Displacement rates from a published palaeoseismological study indicate that deformation may have continued for 0.61–2.20 Ma. In contrast, previous geodetic data give durations with the range of 0.17–0.46 Ma; either the geodetic rate may not be representative of longer-term mean rates or significant extra deformation has taken place on other offshore faults.
(5) Existing and new U-series geochronology establishes the case for uniform late Quaternary (post-MIS 7) uplift of the Perachora peninsula, generally at rates of c. 0.2–0.3 mm a−1. The tectonic style of the uplifting Perachora–Isthmus–New Corinth terrane seems distinct from the back-tilted crustal blocks of the Alkyonides Gulf and Megara basin to the east.
(6) Our studies provide no evidence to support published hypotheses for lateral surface growth of the relatively young and highly active eastern Gulf of Corinth faults. This may be because (a) we have misread the field evidence, (b) surface growth has not occurred or (c) growth was so rapid during the early stages of fault evolution (e.g. Cowie 1998) that it is now difficult to detect from field studies of the more evolved surface faulting.
Acknowledgements
We thank the University of East Anglia, Royal Society, Mobil North Sea Ltd, NERC for Research Grant NER/C/S/2000/00588 to R.L.G. and Studentship NER/S/A/2000/03324 to C.P., and IGME for logistical and financial support during our continuing research in Greece. Seismic line SPK 25 was shot during a joint University of Leeds–University of Patras research cruise in 1996. Radiocarbon date Ox-A-10101 was produced by the Oxford Radiocarbon Accelerator Unit. We thank referees N. Morewood and G. Roberts for numerous helpful criticisms of an earlier version, but this acknowledgement does not mean that they necessarily agree with all of our conclusions.
- © 2005 The Geological Society of London
References
Normal faulting and crustal deformation, Alkyonides Gulf and Perachora peninsula, eastern Gulf of Corinth rift, Greece
