Nanyang Technological University, Singapore.
Historical bathymetric charts and the evolution of Santorini submarine volcano, Greece
Watts, A. B.; Nomikou, P.; Moore, J. D. P.; Parks, M. M.; Alexandri, M.
2015
Watts, A. B., Nomikou, P., Moore, J. D. P., Parks, M. M., & Alexandri, M. (2015). Historical bathymetric charts and the evolution of Santorini submarine volcano, Greece.
Geochemistry, Geophysics, Geosystems, 16(3), 847‑869.
https://hdl.handle.net/10356/86280 https://doi.org/10.1002/2014GC005679
© 2015 The Authors. This is an open access article under the terms of the Creative Commons Attribution‑NonCommercial‑NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‑commercial and no modifications or adaptations are made.
Downloaded on 05 Nov 2021 01:07:00 SGT
RESEARCH ARTICLE
10.1002/2014GC005679
Historical bathymetric charts and the evolution of Santorini submarine volcano, Greece
A. B. Watts
1, P. Nomikou
2, J. D. P. Moore
1, M. M. Parks
3, and M. Alexandri
41Department of Earth Sciences, University of Oxford, South Parks Road, Oxford, UK,2Faculty of Geology and Geoenviron- ment, University of Athens, Panepistimioupoli Zografou, Athens, Greece,3Nordic Volcanological Center, Institute of Earth Sciences, Reykjavik, Iceland,4Institute of Oceanography, Hellenic Center for Marine Research, Anavyssos, Greece
Abstract Historical bathymetric charts are a potential resource for better understanding the dynamics of the seafloor and the role of active processes, such as submarine volcanism. The British Admiralty, for example, have been involved in lead line measurements of seafloor depth since the early 1790s. Here, we report on an analysis of historical charts in the region of Santorini volcano, Greece. Repeat lead line surveys in 1848, late 1866, and 1925–1928 as well as multibeam swath bathymetry surveys in 2001 and 2006 have been used to document changes in seafloor depth. These data reveal that the flanks of the Kameni Islands, a dacitic dome complex in the caldera center, have shallowed by up to 175 m and deepened by up to 80 m since 1848.
The largest shallowing occurred between the late 1866 and 1925–1928 surveys and the largest deepening occurred during the 1925–1928 and 2001 and 2006 surveys. The shallowing is attributed to the emplacement of lavas during effusive eruptions in both 1866–1870 and 1925–1928 at rates of up to 0.18 and 0.05 km
3a
21, respectively. The deepening is attributed to a load-induced viscoelastic stress relaxation following the 1866–
1870 and 1925–1928 lava eruptions. The elastic thickness and viscosity that best fits the observed deepening are 1.0 km and 10
16Pa s, respectively. This parameter pair, which is consistent with the predictions of a shal- low magma chamber thermal model, explains both the amplitude and wavelength of the historical bathymet- ric data and the present day rate of subsidence inferred from InSAR analysis.
1. Introduction
Santorini is one of Europe’s most active volcanoes and the site of some of its most destructive earthquakes.
The Late Bronze Age (1600 BC) Minoan event [Friedrich et al., 2006], for example, developed into a Plinian caldera-forming eruption that deposited a pumice-fall up to 6 m thick [Bond and Sparks, 1976] and a tsu- nami that fatally damaged the thriving Minoan culture on Thera and adjacent islands, including Crete [Bruins et al., 2008; Novikova et al., 2011].
Santorini has a long historical record of volcanic activity [Druitt et al., 1989]. The Minoan eruption, for example, culminated in caldera formation, which may have been the source of Plato’s (428–347 BC) myth of Atlantis [Marinatos, 1939; Galanopoulos, 1960]. Strabo, in 197 BC described an island erupting from the sea in flames between Therasia and Thera. Named Iera (Hiera), the island is thought to have subsequently subsided below sea level, forming a shoal known as Banco. Today, the center of the caldera is dominated by the Kameni Islands, which formed by a mix of small, frequent effusive eruptions that comprise lava domes up to a few hundred meters high, interspersed with moderate explosive activity. These dome-forming eruptions occurred during 46—47 AD, 726, 1570–1573, 1707–1711, 1866–1870 [Fouqu e, 1879], 1953, 1925–1928 [Washington, 1926; Ktenas, 1927], and 1938–1941 AD. The most recent volcanic activity was in 1950 [Georgalas, 1953] on Nea Kameni and involved the opening of a vent and extrusion of a very small lava flow (the Liatsikas lavas) punctuated by sporadic explosive activity which generated an ash plume in the atmosphere up to 1 km high.
Until recently, most information on the eruption history of Santorini has come from geological field map- ping and sampling onshore. Little was known of offshore regions other than from lead line soundings recorded on British, German, Dutch, and Greek government nautical charts and widely spaced single-beam echo sounders, single-channel seismic reflection and gravity and magnetic profiling, and seafloor sampling using dredge and piston cores on academic research ships (e.g., R/V Chain and R/V Jean Charcot in 1966 and 1978, respectively).
Key Point:
Repeat bathymetry measurements reveal changes in Santorini volcano over the past 150 years
Correspondence to:
A. B. Watts, tony@earth.ox.ac.uk
Citation:
Watts, A. B., P. Nomikou, J. D. P. Moore, M. M. Parks, and M. Alexandri (2015), Historical bathymetric charts and the evolution of Santorini submarine volcano, Greece, Geochem. Geophys.
Geosyst., 16, 847–869, doi:10.1002/
2014GC005679.
Received 8 DEC 2014 Accepted 10 FEB 2015
Accepted article online 19 FEB 2015 Published online 25 MAR 2015
This is an open access article under the terms of the Creative Commons Attri- bution-NonCommercial-NoDerivs License, which permits use and distri- bution in any medium, provided the original work is properly cited, the use is non-commercial and no modifica- tions or adaptations are made.
Geochemistry, Geophysics, Geosystems
PUBLICATIONS
The first multibeam swath bathymetry survey offshore Santorini was carried out in 2001 using a 20 kHz hull- mounted SEABEAM 2120 swath and Trimble GPS system onboard R/V Aegaeo [Alexandri et al., 2003; Nomi- kou et al., 2012a, 2012b, 2013a, 2013b]. The swath system had an angular coverage of 150
owith 149 beams, which provides seafloor depth to an accuracy of up to 62 m while the Trimble system, which provides aver- age ship position, is probably accurate to 610 m. A second swath and GPS survey was carried out in May–
June 2006 onboard R/V Aegaeo and R/V Endeavor. Data from the swath surveys have been used for a num- ber of marine geological and geophysical studies and as a basis to locate sites for seafloor sampling by Remotely Operated Vehicles (ROVs) The results of these surveys have been described in Sigurdsson et al.
[2006]; Nomikou et al. [2012a]; Bell et al. [2013]; and Nomikou et al. [2014].
Despite these studies, there is still little known about the volumes, durations, and addition rates of submar- ine volcanic activity in Santorini. Such parameters are important for understanding a volcano’s past and future activity. Arguably, the world’s best-studied submarine volcano is Monowai, in the Tonga-Kermadec arc (southwest Pacific Ocean), which was repeatedly swath surveyed in 1998, 2004, and 2007 [Chadwick et al., 2008; Wright et al., 2008]. The surveys show that the volcano summit shallowed and deepened by up to 90 m and >50 m, respectively, over time periods as short as 6 years. At Vailulu’u volcano, east of Samoa, a 300 m depth change was recorded in just 5 years [Staudigel et al., 2006] as a volcanic dome built up in the center of a summit caldera. Finally, Watlington et al. [2002] and Lindsay et al. [2005] have described shallow- ing of up to several tens of meters in 2 years during 2001–2002 at the Kick’em Jenny submarine volcano in the Lesser Antilles. This volcano was resurveyed in 2013 and 2014 (S. Carey, personal communication, 2015), but no significant changes in seafloor depth have been reported since 2001–2002.
In 2011, a repeat swath bathymetry survey of Monowai during a single research cruise onboard R/V SONNE, together with hydroacoustic T-wave data recorded at an IRIS seismic station on Rarotonga (Cook islands), revealed that the summits of submarine volcanoes can shallow and deepen by up to 70 and 18 m, respec- tively, on time scales as short as 5 days [Watts et al., 2012]. The shallowing was interpreted in terms of the emplacement of a lava spine, the volume of which was reported to be 0.0085 km
3, implying a magma addi- tion rate of up to 0.63 km
3a
21(20 m
3s
21).
While there has been one further swath survey undertaken in the vicinity of Santorini since 2001 (in 2006), the surveys did not overlap and so the data acquired cannot, unfortunately, be used to document any changes that might have occurred in seafloor depth during 2001 and 2006. During a survey in 2012 [Nomi- kou et al., 2012c], two continuously recording pressure gauges, one of them with a tilt meter device attached, were successfully installed and recovered a year later to detect vertical deformation inside the northern part of the caldera. We are not aware, however, of any other attempts to monitor the seafloor sta- bility in the caldera. The only estimates of magma volumes through time have come from onshore studies of individual lava flows [Pyle and Elliott, 2006]. Such estimates refer to subaerial lavas, and do not consider any submarine lavas. Recently, Nomikou et al. [2014] used the 2001 and 2006 swath bathymetry data to map the spatial extent of the submarine flows on the seafloor. They estimated thickness, and hence volume, by fitting a smooth mathematical surface, however, between the edges of the submarine flows using the GMT algorithm surface [Wessel and Smith, 1991] with different assumed gridding intervals and amounts of tension.
Santorini has a long-maritime history and so a number of charts exist, some of which record seafloor depth.
The oldest dates back to 1848, which postdates the intracaldera eruption of 1707–1711 and predates the 1866–1870 eruption. The former eruption created the island of Nea Kameni in the center of the caldera. The purpose of this paper is to compare the historical bathymetry data to modern swath bathymetry data, quantify the differences, and determine the changes that have occurred in the Santorini region during the past 150 years. The results of the paper have implications for magma generation rates, volcano loading and crustal and mantle rheology and, more generally, the use of historical charts to better understand sub- marine volcanic activity in Greece, especially Santorini’s unpredictable neighbor of Kolumbo Bank, as well as at other active submarine volcanoes elsewhere in the world’s ocean basins.
2. Geological and Geophysical Setting
Santorini forms part of the Hellenic arc, deep-sea trench system, which marks the boundary between the
underthrusting African and overthrusting Aegean plates [McKenzie, 1972; Angelier, 1979; Le Pichon and
Angelier, 1979]. The rate of plate convergence (40–45 mm a
21) and the length of the seismically defined Benioff zone (160 km) imply that volcanism in the arc started 3–4 Ma [e.g., Papanikolaou, 1993; Royden and Papanikolaou, 2011].
Despite its convergent setting, the Hellenic arc, trench system has undergone significant amounts of exten- sion and crustal thinning. Receiver function data, for example, suggest that a 25–33 km thick crust underlies the arc, which is about a factor of 1.4 thinner than that beneath mainland Turkey and Greece [Zhu et al., 2006]. Geological studies reveal that arc volcanic rocks have been emplaced on a metamorphic core com- plex comprising Mesozoic blue schist facies limestones and Tertiary metapelites [van Hinsbergen et al., 2005]. There is no evidence, however, on Santorini of a detachment surface separating the core complex from underlying pre-Alpine Paleozoic basement rocks, although low-angle detachment faults have been described from other islands in the Cyclades such as Naxos and Paros [Lister et al., 1984; Papanikolaou, 2013].
Regional seismic reflection profile data suggest that Santorini is located in a NE-SW trending extensional graben, dubbed the Anyhydros basin. Bohnhoff et al. [2006] and Sakellariou et al. [2010] propose that the basin is bounded by two overlapping dextral strike-slip faults, which may have controlled the spatial distri- bution of the volcanic centers of the Santorini volcanic field. Direct evidence of such faults (e.g., flower structures) have not, however, been observed in multichannel seismic reflection profiles [H€ ubscher et al., 2015]. Nevertheless, prominent tectonic lineaments within the basin include the Kameni line, which extends from Thera to the Kameni Islands and was the focus site of seismicity during the 2011–2012 period of vol- canic unrest [Parks et al., 2012] and the Kolumbo line, which extends northeast of Cape Kolumbo on Thera and is presently the locus of an actively venting submarine hydrothermal system [Nomikou et al., 2012a, 2012b; H€ ubscher et al., 2015].
Santorini presently comprises five islands: Thera, Therasia, and Aspronisi flank a 300–390 m deep caldera that includes the islands of Nea Kameni and Palaea Kameni [Druitt et al., 1989]. The steep, subaerial, walls of the caldera expose pyroclastic flows, lavas, pumice, and ash that reflect two main mafic-to-silicic eruptive cycles. The cycles, which have been dated as 360–180 ka and 180–3.6 ka respectively, are characterized by one or more caldera forming events, the most recent of which was the Minoan eruption [Druitt and Fran- caviglia, 1992]. There is evidence, for example, that stratigraphic dips of the Minoan deposits on Therasia and Thera dip inward toward the center of the modern caldera [Heiken and McCoy, 1984; Druitt and Franca- viglia, 1990; Fabbro et al., 2013] suggesting that a caldera (associated with the so-called Cape Riva eruption) was already present prior to the Minoan eruption. The Kameni Islands, which comprise dacitic lavas and domes, now occupy much of the center of the present day caldera. The islands formed gradually during the last 3600 years (shortly after the last caldera-forming eruption) through a series of small dome-building eruptions. Santorini appears to alternate between two distinct types of activity—being characterized by both minor effusive eruptions and major explosive eruptions. Druitt et al. [1989] have estimated from stud- ies of the earliest volcanic centers (e.g., on Akrotiri Peninsula) that there have been at least 12 large explo- sive eruptions alternating with constructional intracaldera activity at Santorini during the past 650 kyr.
Figure 1 shows a bathymetry map constructed from the combined 2001 and 2006 swath survey data described in Nomikou et al. [2012a, 2013a, 2013b, 2014]. The caldera comprises a 300 m deep basin in the south, a 390 m deep basin in the north, and a 325 m deep basin in the west. The caldera margins are characterized by steep slopes (up to 50–60
o) and narrow shelves, and the coastline appears to have been
‘‘scalloped’’ in places by large-scale slope failures. A narrow, NNW-SSE trending, 350 m deep, downfaulted [Heiken and McCoy, 1984] channel that leads into a hour-glass-shaped bathymetric depression connects the caldera to the deeper waters of the Eastern Mediterranean sea and may have formed following a breaching of the caldera wall.
Seismic reflection profiles indicate that the south, north, and west basins of the caldera are underlain by
>200 m of stratified material which Sakellariou et al. [2012] have divided into six subunits. Sakellariou et al.
[2012] interpreted the uppermost 1–4 units as pyroclastic deposits of Minoan age. Johnston et al. [2014],
however, point out that reflectors within units 1–4 are continuous, unfaulted, and merge with a clastic fan
on the flank of the Kameni Islands and so are better interpreted as the products of the early submarine
stages of post caldera collapse central dome formation. They interpret the lowermost 5–6 units as the
upper portion of downfaulted Minoan deposits. Recent interpretations, however, reveal three distinct
volcaniclastic units [Johnston et al., 2015]. From top to bottom these units are: (i) present-day sedimentary infill, (ii) shallow marine phreatomag- matic volcanism associated with the relatively recent formation of the Kameni Islands, and (iii) downfaulted
‘‘Minoan’’ pyroclastic deposits which formed during caldera collapse toward the end of the Late Bronze Age event.
The Santorini caldera correlates with a Bouguer gravity anomaly ‘‘low’’ of
6–8 mGal [Budetta et al., 1984]
which reaches minimum values of
112 mGal [Lagios et al., 2013] over the north basin, north-northwest of Nea Kameni. The low can be inter- preted in terms of a low-density sedi- ment body, a shallow low-density partial melt body, or some combina- tion of these bodies. Other minima have been mapped over Aspronisi, between Fira and Athinios and on Akrotiri Peninsula, which have been attributed to faults [Paraskevas et al., 2014]. Magnetotelluric (MT) data indicate two main zones of high con- ductivity, one in northern Thera and the other between Akrotiri Peninsula and the channel between Thera- sia and Thera, suggesting underlying fluid pathways that are orthogonal to the trend of the Kameni and Kolumbo lines [Papageorgiou et al., 2010].
There is evidence of both seismicity and recent crustal movements in the Santorini region. Local network data recorded during 2002–2004, for example, suggest seismicity is focused along the Kolumbo line where there is active venting and hydrothermal activity on the seafloor [Bohnhoff et al., 2006; Dimitriadis et al., 2009]. In contrast, the caldera region appears to have been relatively quiet during the same period [Bohnh- off et al., 2006]. There are descriptions, however, of earthquakes that accompanied the 1866–1870 and 1925–1928 eruptions on the Kameni Islands [Fouqu e, 1879; Washington, 1926]. Interferometric Synthetic Aperture Radar (InSAR) analysis of satellite images acquired during 1993–2010 indicates a quiet period, with the south coast of Nea Kameni experiencing a mean subsidence rate of 6 mm/yr [Parks et al., 2015]. Dur- ing January 2011 and April 2012, however, the caldera experienced a period of unrest with seismicity (Mw > 2) focused along the Kameni line [Parks et al., 2012]. GPS and InSAR observations suggest that during January 2011 and April 2012 there was a radial movement of Therasia, the Kameni Islands, and Thera away from a point in the center of the caldera’s north basin, which has been interpreted as the locus of a shallow (3–6 km) magma chamber [Newman et al., 2012; Parks et al., 2012].
3. The 1848 British Admiralty Chart
During the late 1790s and early 1800s, the British Admiralty carried out a campaign of depth measurements using disarmed war ships that had been converted to surveying ships. By the mid-1800s, more than 1900 sounding charts of the coasts of western Europe, North, and South America, West and East Indies, Australia, and New Zealand, and the Pacific ocean islands had been published. To estimate seafloor depths, ships would typically lower a graduated, prestretched, hemp rope with a lead weight attached to it until it hit the seafloor.
Various techniques such as Burt’s ‘‘buoy and nipper‘‘ were available to ensure line verticality [Millar, 2013].
Although the main motivation of the Admiralty surveys appears to have been navigation, a number of workers have recognized their scientific importance. Laughton and colleagues at the UK National Institute
Bathymetry/Elevation (m) SB WB
25˚20' 25˚25' 25˚30'
' 0 2
˚ 6 3 '
0 2
˚ 6 3
' 5 2
˚ 6 3 '
5 2
˚ 6 3
' 0 3
˚ 6 3 '
0 3
˚ 6 3
−600 −400 −200 0 200
0 km 5
Thera
NB
SB WB Asp.
Therasia
KI
Figure 1. Bathymetry and topography map of Santorini based on the combined swath bathymetry surveys onboard R/VAegaeo of Nomikou et al. [2014] and referen- ces therein. The map shows a 20 3 20 m grid of the swath data. NB 5 North Basin, SB 5 South Basin, WB 5 West Basin, Asp. 5 Aspronisi, and KI 5 Kameni Islands. Illumi- nation is by an artificial sun at an azimuth and elevation of 035oand 20o,
respectively.
of Oceanography, for example, used lead line soundings data to compile the first bathy- metric maps of the Red Sea [Laughton, 1970]
and parts of the north- east Atlantic Ocean [Roberts et al., 1979].
Also recognized has been their significance for studies of temporal changes in seafloor depth. Sigurdsson et al.
[1991], for example, used Dutch and British admiralty soundings to document the changes that occurred in sea- floor depth before and after the 1883 Krakatau eruption. Other studies have used the soundings to determine changes in shelf edge location [Hedley, 1911], shoal, and channel migration [Van der wal and Pye, 2003] and the stability of sand banks [McCave and Langhorne, 1982].
Santorini was first surveyed using lead lines in 1848 by Captain Thomas Graves R.N. (1802–1856) onboard HMS Volage, a 28 gun sixth rate ‘‘ship of the line’’ fitted out for survey work in London during late 1846 to early 1847. Captain Graves was a specialist in hydrography and had previously been involved in surveys along the Atlantic coast of South America. Captain’s logs held at the UK National Archives in Kew reveal that the Volage sailed from Malta for Pireus, Greece in April, 1848, returning to Malta in December, 1848.
During May–November, 1848 Captain Graves made numerous soundings of the approaches to Pireus Port and offshore of many of Greek islands, including Paros, Syros, Skyros, and Cyprus. The logs do not detail how the soundings were made, but there are a number of references in the logs to loss of ‘‘leads and lines’’
suggesting traditional methods. Positions are recorded and were determined mostly by bearings on promi- nent landmarks. The logs also record celestial measurements of latitude using sextants and dead-reckoning, and ships chronometer measurements of longitude.
The charts based on Captain Graves’ soundings are remarkably detailed. Figure 2, for example, shows an enlarged part of the 1848 Santorini Admiralty chart in the vicinity of the Kameni Islands. The chart shows Nea Kameni when it comprised two islands: Nea Kameni and Mikra Kameni. Soundings are recorded in fath- oms. The map reveals a narrow shallow channel between Mikra Kameni and Nea Kameni and a wide, deep, channel between Nea Kameni and Palea Kameni. St. Giorgio Bay (St. George’s harbor) on the southwest coast of Nea Kameni and Voulcano Cove on the southeast coast of Nea Kameni were popular with sailing ships at the time.
In order to map Captain Graves’ data, we digitized the coastline, heights, and soundings recorded on the 1848 Admiralty chart. Three georeference points, shown in Figure 3, were selected on the basis of their simi- larity and, hence, stability with respect to the present day coastline and assigned a latitude and longitude using Google Earth. The coastline (total number of points 5 2402) was digitized using GraphClick (Version 3) and assigned a zero depth or height. The offshore depth soundings (total 51853) and onshore height measurements (total 5 11) were also digitized assuming they were located at the center of the number recorded on the chart.
Figure 3 shows a good general agreement between the 1848 and present-day coastlines. The agreement is particularly close at Aspronisi, which formed during the eruptions of 1707–1711, and at Palea Kameni, which predates Aspronisi. The similarity suggests that these islands have been stable since 1848. Other regions of agreement include the coast of northwest Nea Kameni, where lavas of the 1707–1711 eruption outcrop,
Figure 2. Enlarged part of British Admiralty Chart No. 2043 in the region of the Kameni Islands. The chart is based on soundings acquired in 1848 by Captain Thomas Graves and the officers and crew onboard the survey ship HMS Volage. The chart shows the coastline of the Kameni Islands when they comprised of two islands: Nea Kameni and Mikra Kameni. Numbers offshore show soundings in fath- oms (1 fathom 5 1.8288 m).
and along most of the caldera coast of Therasia and Thera.
except for some of the coves between Cape Simandiri and Cape Riva on Therasia and between Cape Skaro and Fira on Thera. The most striking differen- ces between the 1848 and present-day coastlines, however, are found in the west, east, and south coast of Nea Kameni.
Clearly, Nea Kameni has grown significantly since 1848. Other, smaller, differences occur along the east and south coast of Thera where the present day coastline appears to have moved both inward and outward since 1848.
The east coast of Thera, which is composed of pyroclastic flows of Phase 4 of the Minoan eruption [Druitt, 2014], is subject to strong east-northeasterly winds and waves and so it is quite plausible that the coastline (e.g., at Perissa and Kamari beaches) has moved inward since 1848 due to erosion. The south coast, in contrast, is composed of Minoan tuff of phase 4 with vertical cliffs and a narrow beach (e.g., at Red and Vlychada beaches) and it is difficult to explain the outward movement of the coastline here, sug- gesting errors in the 1848 coastline.
The 1848 coastline data were then combined with the lead line soundings and height data and smoothed and gridded using GMT’s blockmean and surface algorithms and a range of assumed grid intervals and tension factors. Grid intervals were selected after consideration of the spatial distribution of the individual soundings. Minimum distances between soundings were binned at 20 m intervals and the most frequently occurring value (i.e., mode) was determined to be 160–180 m for the entire data set and 60–80 m for the Kameni Islands region (Figure A1, inset). Based on these distances, we selected a range of smoothing and gridding intervals from 70 3 70 m to 280 3 280 m and higher and tension factors from 0.0 (less tension) to 0.35 (more tension), finding a 80 3 80 m grid interval and a tension factor of 0.0 to be optimal. This parame- ter pair was not too small that it produced too many unrealistic closed contours in the gridded bathymetry or too large that contours were too heavily smoothed.
Figure 4 shows that a 80 3 80 m grid of the 1848 soundings reproduces well the main bathymetric features of the Santorini region. The depth, width, and general shape of the north, south, and west basins, for exam- ple, closely resemble that derived from the modern swath surveys (Figure 1). The map also resolves some bathymetric details at the northern and southern entrances to the caldera, including the elongate feature north of the northern entrance. The most striking differences appear to be in the region of Nea Kameni, the submarine edifice of which has clearly increased in size to the east, south, and west since 1848. Other differ- ences occur offshore on the eastern and western flanks of the volcano, probably because of the paucity of height and sounding measurements in these regions. The 1848 soundings, for example, suggest the flanks of the volcano are quite smooth while the modern swath data (Figure 1) suggest it is complex with evi- dence of submerged terraces, erosional chutes, and downslope debris flow deposits.
4. Comparison of the 2001 and 2006 Swath Bathymetry with the 1848 Soundings
In order to quantitatively compare the 1848 soundings and swath data, we first sampled the swath grid of Nomikou et al. [2012, 2013a, 2013b, 2014] at the same location as Captain Graves’ 1848 depth soundings,
25˚20' 25˚25' 25˚30'
' 0 2
˚ 6 3 '
0 2
˚ 6 3
' 5 2
˚ 6 3 '
5 2
˚ 6 3
' 0 3
˚ 6 3 '
0 3
˚ 6 3
Geo-reference positions on the 1848 map 1848 coastline Present-day coastline
Changes in coastline position
Cape Messa Akroteri
Amouthi Cove
0 km 5
Cape Columbos
CS CSk
MC
CT CR
Fira
Athinios Thera Therasia
Aspronisi
Nea Kameni
Figure 3. Comparison the 1848 and the present-day coastline of Santorini. The black-filled arrows indicate apparent changes in coastline position since 1848. The three circled crosses show the location of the georeference points (which are assumed to have been stationary since 1848) used to digitize the lead line soundings, height, and coastline and construct grids of the 1848 soundings data. CSk 5 Cape Skaros, CS 5 Cape Simandiri, CR 5 Cape Riva, CT 5 Cape Tourlos, MC 5 Mousaki Cove.
and then subtracted the 1848 depths from the swath grid. The mean and standard deviation of the differen- ces are 15.2 and 62.8 m, respectively (Figure 5). A positive mean indicates a swath depth that is generally shallower than the lead line sounding. If we omit the Kameni Islands region, which coastline data (Figure 3) suggests has undergone the most significant changes since 1848, the mean and standard deviation decrease to 214.4 and 49.5 m, respectively. A negative mean indicates a lead line sounding that is generally shallower than the swath data. The Mean Absolute Deviation (MAD), which takes less account of data out- liers than the standard deviation, is 19.5 m for the entire data set and 13.3 m for the entire data set exclud- ing the Kameni Islands. These values compare quite favorably with the MAD of bathymetry crossover differences on intersecting ship tracks acquired on ocean going research vessels during 1955–1992, prior to GPS navigation, of 26 m [Smith, 1993].
Figure 6a shows the differences between the 1848 soundings and the modern swath data in map form. The differences were obtained by first resampling the swath data at the same grid interval and tension factor as was used to construct the 1848 grid and then subtracting the 1848 depth from the swath depth at each grid node. The 1848 grid is based on a grid interval of 80 3 80 m and tension factor of 0.0. Figure 6b dem- onstrates that the use of other grid intervals and tension factors, however, has little impact on the general pattern of the differences. In particular, the apparent shallowing of the depth of the seafloor in the region of the Kameni Islands since 1848 is a robust feature of all the grids tested, as is the apparent deepening of the seafloor at the edge of the north basin, offshore Therasia.
Differences between the 1848 and swath surveys reflect errors in the lead line or swath data or real changes in the depth of the seafloor through time. Errors in swath bathymetry data arise from uncertainties in the
25˚20' 25˚25' 25˚30'
' 0 2
˚ 6 3 '
0 2
˚ 6 3
' 5 2
˚ 6 3 '
5 2
˚ 6 3
' 0 3
˚ 6 3 '
0 3
˚ 6 3
−600 −400 −200 0 200
Bathymetry/Elevation (m)
0 km 5
Figure 4. Bathymetry and topography map based on the 1848 soundings, coastline, and height data. The map was constructed by digitiz- ing the 1848 soundings, coastline (which was assumed to be at zero depth and height) and heights and smoothing and gridding the com- bined data set using GMT’s blockmean and surface algorithms with a grid interval of 80 3 80 m and tension factor of 0.0. Contour interval 5 100 m. 200 m contours are indicated by a thick black line. Black-filled circles show location of individual soundings. The dashed box delineates the Kameni Islands region. Illumination is as in Figure 1.
slope correction, water sound velocity, and outermost beam returns. Schmitt et al. [2008]
estimates from ‘‘benchmark’’
statistical analysis that errors in GPS modern navigated swath bathymetry generally result in depth errors of 1–2 m or smaller, depending on sea state. The SEABEAM 2020 sys- tem used on R/V Aegaeo has a reported along-track and across-track error of 10 3 10 m, enabling swath data to be gridded down to intervals of 20 3 20 m and less. Errors in the 1848 soundings data are difficult to assess. Known errors include shoal bias (due to the preference of the lead line sur- veys to find the shallowest depth for navigational rea- sons), drag on the hemp line, and uncertainty in whether or not the lead line hit bottom.
The main errors, however, are probably navigation, which in 1848 relied on triangulation and line of sight measure- ments to prominent land- marks. Within the caldera, navigational errors were probably small because its steep walls would have been easily visible from the ship. However, on the flanks of the volcano, for example, offshore the low-lying east coast of Thera, navigational errors could have been significant. Other errors arise from the irregularly spaced nature of the soundings data. Statistical tests (Appendix A), however, in which a set of points are generated with similar spacing to the soundings and randomized geographical location, suggest that an 1848 type survey would have been capable of resolving the bathymetry to a MAD of 15.5 m (8.6 m in the Kameni Islands region).
The difference in seafloor depth between the 1848 and swath surveys mapped in Figure 6a ranges from 2175 to 1175 m, with the largest changes occurring at the edges of the North basin and in the region of the Kameni Islands. Regions of the seafloor that appear to be shallower at the present day than they were in 1848 are indicated by red shading while deeper regions are indicated by blue shading. The figure shows that the main regions of shallowing are in the region of the Kameni Islands, in what is now the southwest and southeast of Nea Kameni where the seafloor has shallowed by up to 175 m. This estimate is robust and, as Figure 6b clearly shows, does not depend on assumptions made in the smoothing and gridding. We therefore suggest that the shallowing reflects growth due to the addition of magmatic material on the sea- floor. The main regions of deepening are at the edges of the North Basin, offshore Cape Skaro and Cape Tourlos and Cape Perivola on Thera and between Cape Simandiri and Cape Tripiti on Therasia (Figure 6a).
The deepening, which reaches -175 m off Therasia, is also a robust feature of the difference maps (Figure 6b). The deepening occurs close to shore, however, where bathymetry gradients are steep and small navi- gational errors in either the soundings or swath data sets could give rise to some of the differences. Indeed, statistical tests (Appendix A) show some discrepancies between the observed and modeled bathymetry in the region of the caldera walls, but not along all of their length. We cannot therefore rule out the possibility that seafloor processes such as those associated with collapse due to large-scale slope failures and slump- ing and sliding have modified at least some segments of the edge of the North basin since 1848.
All data
All data less Kameni Islands
Difference in depth (Modern swath surveys - 1848 survey (m)
Frequency (%) Frequency (%)
0 5 10 15 20 25 30
−200 −150 −100 −50 0 50 100 150 200
N = 574 Mean = -14.4 σ = 49.5 0
5 10 15 20 25 30
N = 867 Mean = +5.2 m σ = 62.8 m
Figure 5. Histograms of the differences in seafloor depths measured in 1848 and during the combined swath bathymetry surveys. Differences were calculated by subtracting the 1848 soundings from the seafloor depths inferred from the 20 3 20 m swath bathymetry grid at the location of the 1848 soundings. The uppermost histogram (brown fill) shows the differ- ences for all the data. The mean difference is 15.2 m, which indicates the swath data are generally shallower than the 1848 soundings. The lowermost histogram (green fill) shows the differences for all the data, except those in the immediate vicinity of the Kameni Islands.
The mean difference is 214.4 m, which indicates that the 1848 soundings are generally shallower than the swath data. N 5 number of differences, r 5 standard deviation.
Figure 7 shows an enlargement of the difference map in the region of the Kameni Islands. The 1848 map (Figure 7a) is based on the same soundings data as was used to construct Figure 4. The combined 2001 and 2006 swath grid [Nomikou et al., 2014] has been modified, however, to include soundings from the 1994 British Admiralty Chart (No. 1541) so as to better represent seafloor depths in shallow water where it is diffi- cult to acquire good swath coverage. Figure 7b shows a 10 3 10 m grid of this combined data set. We then resampled this grid at the same grid interval as the 1848 soundings data (80 3 80 m). Figure 7c shows the differences in seafloor depth between the two grids, together with estimates of the volumes of shallowings and deepenings since 1848, which we interpret as growth and collapse, respectively. The figure shows well the growth of Nea Kameni to its west, south, and east flanks as well as the growth that has occurred at ‘‘NK north’’ and ‘‘Drakon,’’ two new offshore lavas flows that were identified recently by Nomikou et al. [2014]
using a similar swath grid as used in Figure 7b.
Difference (Swath - 1848 soundings) (m)
Distance along profile (km)
Thera Therasia
Kameni Islands
a)
b)
Bathymetry/T opography (m)
25˚20' 25˚25' 25˚30'
' 0 2
˚ 6 3 '
0 2
˚ 6 3
' 5 2
˚ 6 3 '
5 2
˚ 6 3
' 0 3
˚ 6 3 '
0 3
˚ 6 3
−200 −100 0 100 200
NW SE
Seafloor shallowing since 1848
0 km 5
−400
−200 0 200
−6 −4 −2 0 2 4
120 x 120 m (T=0.35) 200 x 200 m (T=0.35) 280 x 280 m (T=0.35) 80 x 80 m (T=0.35) 80 x 80 m (T=0)
1848
Swath
+2.5/−2.7 km3
Figure 6. Differences in seafloor depths measured in 1848 and during the combined swath surveys. (a) Difference map calculated by sub- tracting the 1848 80 3 80 m soundings grid (Figure 4) from an 80 3 80 m grid of the combined swath surveys of Nomikou et al. [2012a, 2013a, 2014]. Red regions show an apparent shallowing of the seafloor since 1848 while blue regions show an apparent deepening. The shaded regions reflect either errors in the 1848 or swath bathymetry data or geological processes occurring on the seafloor. Numbers above the horizontal scale bar are volumes of the apparent shallowing and deepening. Gray solid lines show the location of the profile in Figure 6b. The center line is the actual profile location. The outer two lines are offset 500 m from the center line and illustrate the regions where soundings data have been projected onto the profile. The ‘‘x’’ symbol indicates the origin of the profile. Present-day coastline out- lined in black. (b) Comparison of different grids of the 1848 soundings (T 5 tension factor) to the swath bathymetry along a NW-SE trend- ing profile that intersects Nea Kameni (see Figure 6a for location). Inset summarizes the different grid parameters. The shallowing of the seafloor of south and southeast of Kameni Islands and the deepening of the seafloor around the margin of the caldera offshore Therasia appear to be robust features, irrespective of which of the grids is used.
The differences between the seafloor depths, coastline, and onshore heights in 1848 and the present day in the Kameni Islands region are illustrated in the perspective plots in Figure 8. The most significant changes include the infilling of the deep channel between Palea Kameni and Nea Kameni and the shallow channel between Nea Kameni and Mikra Kameni. Other changes include the annexation of Banco to Nea Kameni and the enlargement of Nea Kameni to the south and east. The south and west flank of Palea Kameni and the shoals between Banco and Fira port, in contrast, appear to have been relatively stable and have changed little since 1848.
Figure 8 shows that the once separate submarine volcanic cones of Palea Kameni, Nea Kameni, Mikra Kameni, and Banco have, as a result of subsequent volcanism, joined into a single edifice. About 1750 m
25˚25'
36˚25'
1848
−600 −400 −200 0 200
25˚25'
25˚25'
36˚25'
1994+2001+2006
36˚25'
−200 −100 0 100 200
Bathymetry/Elevation (m)
Difference (m)
0 2
km St. George’s
Harbour
Vulcanos Cove
Aspironisi Palea Kameni
Mikri Kameni Nea Kameni
NK/MK channel PK/NK
channel
Banco
Shoals Phira
Aspironisi Palea Kameni
NK GK
PK/NK channel
Shoals Phira
c)
b) a)
NK north
NK east
Drakon
+0.54 /−0.25 km 3
Figure 7. Map of the differences in seafloor depths measured in 1848 and a combined ‘‘modern’’ grid that comprises Hydrographic Office 1994 soundings and swath bathymetry data from the 2001 and 2006 surveys of R/VAegaeo in the vicinity of the Kameni Islands (area delineated by a dashed box in Figure 4). The ‘‘x’’ symbol marks the origin of the profiles shown in Figures 6b and 11. (a) 1848 80 3 80 m grid and coastline. Filled black arrow shows the orientation of the perspective plot shown in Figure 8. (b) Swath bathymetry 10 3 10 m grid and present day coastline. (c) Difference grid obtained by subtracting the 1848 sounding and swath bathymetry grids at the same grid interval (i.e., 80 3 80 m). Numbers show estimates of the volumes of the seafloor shallowings, which we interpret as growth and the deepenings of the seafloor, which we interpret as collapse. Present-day coastline outlined in black. NK north, NK east, and Drakon locate the three new offshore lava flows identified by Nomikou et al. [2014] from geomorphological studies.
wide in 1848, Nea Kameni has expanded in width to 3000–3500 m by the time of the modern swath sur- veys. This so-called ‘‘buttressing’’ of a pre-existing cone is a feature of submarine volcanism and appears to occur on both short time scales (a few years) such as at Monowai where dome growth during 2004–2007 was followed in 2011 by flanking eruptions and summit widening [Watts et al., 2012] and, on long time scales (a few Myr) in the Tenerife, Canary islands where the relatively young Las Ca~ nadas volcano (<1.8 Ma) has filled the gap between the three older volcanoes (3.3–12.0 Ma) of Teno, Anaga, and Roques del Conde in the northwest, northeast, and south part of the island, respectively [Ancochea et al., 1990].
5. Forqu e’s Maps and Other Historical Bathymetry Charts
In his seminal book, translated by A. R. McBirney in 1998 [Fouqu e, 1998], Fouqu e [1879] details the changes that occurred in the coastline of Nea Kameni during the volcanic event of March 1866 to September 1870.
The maps (Figure 9) reveal that it was during this time that the enlargement of Nea Kameni to the south and east, clearly visible in Figures 3, 7, and 8, actually took place.
Fouqu e [1879] showed, using field observations, that Nea Kameni increased its surface area during March 1866 to September 1870 by as much as 2–3 times, mainly through effusive volcanic activity that involved fissuring, dacitic dome construction and long runout blocky lava flows. The most prominent of these domes were Georgios, Aphroessa, and Reka. The Georgios dome was located on the southeast coast of Nea Kameni and first appeared as an island on 22 January 1866 in Voulcano Cove. The Aphroessa dome, which was located on the southwest coast of Nea Kameni, rose up from a seafloor depth of 27 m on 16 February
25˚22'
25˚23'
25˚24'
300 200 100 0
25˚24' 36˚24'
36˚24'
25˚22'
25˚23'
−300 −200 −100 0 100
Palea
Palea
Nea Mikri
Banco
1848
1994+2001+2006
25˚24' 36˚24'
36˚24' 300
200 100 0
Depth (m)Depth (m)
m
(7 m)
Shoals (35 m) Shoals
(36 m) Kameni
Georgios+Dafni+WW2+Liatsikas lavas
(1570-73) (197 BC?) (46-726 AD.)
(1701-11) Kameni
Nea Kameni
Kameni
Kameni
(Iera?)
Figure 8. Perspective view from the southeast (azimuth 155o, elevation 20o) of bathymetry the Kameni Islands region in 1848 and at the present day. The bathymetry has been constructed from a 80 3 80 m grid of the 1848 soundings and modern swath bathymetry data. The coastline is shown in black-filled circles. The red numbers are estimated ages. Banco is believed to be the oldest submarine volcanic cone on the 1848 map, but there are no samples or dates to confirm this. The age of the shoals between Banco and Fira, which are present in both the 1848 and present day maps, are unknown. The blue numbers are depths below sea level in meters.
1866 [Brine, 1866]. The Reka dome, which was located southwest of the Aphroessa dome, was 1–2 m high and 30–40 m wide dome on 10 March 1866 [Fouqu e, 1879]. By September 1870, the southeast and south- west coasts of Nea Kameni resembled quite closely those of the present day.
The accuracy of Fouqu e’s maps are difficult to assess, but the close resemblance between the coastlines on his March, May and June 1866 maps and Captain Graves’ 1848 coastline (compare Figures 2 and 9) suggests that Fouqu e used the latter maps as a base to determine the subsequent enlargement of Nea Kameni.
Indeed, Fouqu e [1879] makes reference in his book to the ‘‘English chart.’’ The Fouqu e maps demonstrate that the main changes in the surface area of Nea Kameni occurred during 1 June 1866 and September 1870. The maps do not, however, include soundings so they do not constrain the changes that occurred off- shore during this time.
There are though two additional soundings charts that provide information on seafloor depth changes during 1848 and the modern swath surveys. The first, published by Von Fritsch et al. [1867], dates from early 1866 and shows the configuration of Nea Kameni after the initial growth of the Georgios, Aphroessa, and Reka domes.
The chart was based on the 1848 map with modifications based on soundings from the M/V Reka (Austria), the M/V Principe Carignano (Italy), and Messrs. Schmidt and Palasca of the 1866 Greek commission and shows enlargement of land areas in the southwest and southeast of Nea Kameni and a new group of four islets—the May islands—that had emerged in the channel separating Palea Kameni and Nea Kameni. Comparison of the 1848 and early 1866 soundings show that St. George’s Harbor had also experienced uplift, shallowing by some 18 m. According to [Schmidt, 1868], the shallowing improved the harbor by raising its southern and western edges thereby making it safer for large sailing ships. Fouqu e [1879] refers in his book to the coastline on the early 1866 map, which quite closely resembles the coastline on his 1 June map.
The second, published by Reck [1936], dates from 1925 to 1928. This chart shows that the present-day coast- line of Nea Kameni was essentially complete by 1925–1928. The chart reveals the infilling of the channel between Nea Kameni and Mikra Kameni, the annexation of Banco (ancient Hiera?) to Nea Kameni and Mikra Kameni, and the subsidence of the May islands. The infilling of the channel between Nea Kameni and Mikra Kameni is described in detail by Kt enas [1925] and Washington [1926] and began with a dacitic lava flow emanating from a 175 m high dome (the Dafni dome) during 13–27 September 1925. The lava eventually filled the channel to a height of some 20–45 m and extended to the northwest and southeast of Mikra Kameni. The depth of the channel recorded on the early 1866 map was 4–22 m, which indicates a total flow thickness of some 24–67 m. The annexation of Banco to Nea Kameni implies even greater flow thick- nesses. The seafloor on the early 1866 map is up to 140 m deep between Mikra Kameni and Banco. The height of the annexed region on the 1925–1928 map is not well defined, but if we assume the lavas filled to a similar height as they did between Mikra Kameni and Nea Kameni, then the total thickness of volcanic material added during the annexation of Banco could have exceeded 140 1 20 5 160 m.
6. Results
In order to quantify the change in seafloor depth that has occurred between the 1848 and the modern swath surveys, we georeferenced and then digitized the soundings on the early 1866 and 1925–1928 maps of Von Fritsch et al. [1867] and Reck [1936], smoothed and gridded them.
March 1866
May 1st 1866
June 1st 1866
February 1867
September
1870 Present Day Mikra
Kameni Nea
Kameni
Aphroessa Georgios
1. 1925-1928 2. 1939-1941
1 1
2 2
Reka
Figure 9. Sketch maps showing the change in the coastline of Nea Kameni and Mikra Kameni during March 1866 and the present day as documented by Fouque [1879]. The 1 May 1866 to September 1870 maps show the enlargement of the southeast and southwest flank of Nea Kameni due to the emplacement of the Georgios, Aphroessa, and Reka domes and their associated lava flows. The present day map shows the enlargement of the north, west, and east flanks of Nea Kameni due to dome and lava emplacement in 1925–1928 and 1939–1941.
Figure 10 shows the differences in seafloor depth between 1848 and early 1866, the late 1866 and 1925–
1928, and the 1925–1928 and the modern (i.e., 2001 and 2006) swath bathymetry. The maps reveal periods of shallowing when the Kameni Islands grew (red shading) and deepening (blue shading) when they col- lapsed. They show that the main periods of growth occurred during 1848 and early 1866 in the channel between Palea Kameni and Nea Kameni and during late 1866 and 1925–1928 when the south, north, and east submarine flank of Nea Kameni was enlarged. The field observations of Fouqu e and Washington limit the growth to three main periods: (a) early 1866 when the channel between Palea Kameni and Nea Kameni was infilled, the May islands formed and St. Georges Harbor was infilled, (b) late 1866–1870 when the south and southeast coast of Nea Kameni was enlarged and Voulcano Cove (Note: the cove is referred to as the Bay of Exhaltations on the 1866 map) was infilled, and (c) 1925–1928 when the channel between Nea Kameni and Mikra Kameni was infilled, Banco was annexed to Nea Kameni and the north submarine flank of Nea Kameni was enlarged [Nomikou et al., 2014].
The maps also show evidence of collapse during 1848 and early 1866, late 1866, and 1925–1928 and 1925–
1928 and the present day. We do not know the precise dates of these collapses, but Schmidt [1868] reports that the early 1866 eruption started with subsidence and Brine [1866] notes that by 22 February 1866 that the southeast portion of Nea Kameni had ‘‘sunk considerably, and was still sinking.’’ Fouqu e [1879] makes frequent reference to subsidence during the late 1866 eruption from the southeast coast of Nea Kameni when ground floors of several houses were flooded. Finally, Schmidt [1868] reports that by 4 May 1866 only three of the four May islands remained and Washington [1926] reports their final disappearance during the 1925–1928 eruption.
25˚22' 25˚24'
1925-1928 1994+2001+2006
1925-1928 − Late 1866 1994+2001+2006 − 1925-1928
0.13/-0.15 0.36/-0.14
Difference (m) Bathymetry (m) 25˚22' 25˚24'
36˚24'
1848
−400 −300 −200 −100 0 25˚22' 25˚24'
Early 1866
36˚24'
Early 1866−1848
−200 −100 0 100 200
0.35/-0.18 NK
PK MK
25˚22' 25˚24'
1925-1928
1707 and older Late 1866-1870
1940-1941 1939-1940 1939
1950
Growth Collapse
a) b) c) d)
e) f) g)
Figure 10. Maps of the bathymetry, geology, and differences in seafloor depth in the Kameni Islands region. The ‘‘x’’ is defined in Figure 7a Bathymetry in 1848. (b) Bathymetry in early 1866. (c) Bathymetry in 1925–1928 showing the extent of the 1866–1870 and 1925–1928 lavas. (d) Modern bathymetry showing the extent of the 1939, 1939–1940, 1940–1941, and 1950 lavas. (e) Differences in seafloor depths between 1848 and early 1866. Thin dotted lines show the mask used to isolate the region of seafloor shallowing and estimate the volume of submarine lava associated with the early 1866 event. (f) Differences in seafloor depths between late 1866 and 1925–1928. Thin dotted lines show the mask used to isolate the volume of submarine lava associated with the late 1866–1870 (southernmost polygon) and the 1925–1928 (northernmost polygon) events. Thick-dashed lines show the minimum of the region of seafloor deepening. (g) Differences in seafloor depths between 1925–1928 and the modern bathymetry.
The changes in seafloor depth associated with the growth and collapse of the Kameni Islands during the past 150 years are seen in profile form in Figure 11. The profiles are oriented North-South and East-West and pass through an origin that is located approximately in the center of present day Nea Kameni. The pro- files shows a shallowing and, hence, growth, of up to 150 m and a deepening and, hence, collapse of up to 80 m, which is much larger than the seafloor displacements seen in the north and south basins which are generally less than 625 m. This suggests to us that the largest displacements in the seafloor since 1848 have indeed occurred in the region of the Kameni Islands and that in comparison, the seafloor in the north and south basin of the caldera has been relatively stable.
7. Discussion
7.1. Growth and the Volume and the Rate of Magma Addition Through Time
One implication of the difference maps in Figure 10 is that they can be used to estimate the volume of mag- matic material that has been added to the crust through time. When combined with field evidence of the timing and duration of individual magmatic events, then the volumes may be used to calculate the rate of magma addition through time.
We have estimated the volumes of the early 1866, the late 1866–1870, and 1925–1928 events from the dif- ference grids, which are plotted in map form in Figures 10e, 10f, and 10g. The numbers in the lower right corner of each map reflect the volume of the growth and collapse in km
3, respectively. These volumes reflect the growth and collapse in the entire map area and so in order to calculate the volumes of just the early 1866, the late 1866–1870, and 1925–1928 events we masked out the region (fine-dotted lines in Fig- ures 10e, 10f and 10g) deemed to have not been in these events.
−400
−200 0 200
−6 −5 −4 −3 −2 −1 0 1 2 3 4
−400
−200 0 200
−6 −5 −4 −3 −2 −1 0 1 2 3 4
−400
−200 0 200
−400
−200 0 200
N-S Profile
E-W Profile
1848 1866 1928 1994+2001+2006
Distance (km)
Elevation/Bathymetry (m)Elevation/Bathymetry (m)
Submarine Subaerial Submarine
Subaerial
Filling of navigational channel Early 1866 Growth of N flank 1925-1928
Growth of N flank Early 1866
Growth of E flank Late 1866-1867 Growth of W
flank Late 1866-1867
Growth of S flank Late 1866-1867
Nea Kameni
Figure 11. Bathymetry profiles showing the changes in seafloor depth on the flanks of Nea Kameni since 1848. The profiles are located in Figure 10. The ‘‘x’’ is defined in Figure 7. (a) North-south profiles. (b) East-west profiles. Red-dashed lines show 1848 heights above sea- level. Red-filled triangles show the actual 1848 height measurements projected onto the profiles. Blue-dashed lines show the modern heights above sea level.
The volumes are summarized in Table 1 where they are compared to the previously published estimates of Pyle and Elliott [2006] and Nomikou et al. [2014]. Errors in the volumes derived from historical data are difficult to assess, but we found variations of up to 60.007 km
3in the volume of the early 1866 event, depending on the different intervals and tension factors used to grid the 1848 soundings data. The vol- umes in Table 1 are 2–4 times larger than previous estimates. The Pyle and Elliot [2006] estimates are only based on the subaerial lavas and therefore do not include the submarine lavas. The Nomikou et al.
[2014] estimates include submarine lavas. They estimated the top surface of these lavas from modern swath data, but they use a gridding algorithm to estimate their base. The algorithm was based on GMT’s surface, which requires the assumption of a certain tension parameter that defines how a surface is rep- resented in a region of no data and this is unknown. We believe that historical bathymetric charts, which record changes in the actual seafloor depth through time, provide the best measure of the base on which any subsequent lava flows have been emplaced.
Figure 12 shows a log-log plot of the volumes of the early 1866, the late 1866–1870, and 1925–1928 events against duration of magmatic activity, along with estimates from other submarine volcanoes [Crisp, 1984;
Watts et al., 2006, 2012; Staudigel et al., 2006; Sparks et al., 1998]. The plot shows that volume increases with duration. The dashed lines in the figure show the rate of magmatic addition. The rate for intracaldera con- struction during the 1866–1870 event at Santorini is 0.18 km
3a
21, which Figure 11 shows is among the highest recorded for a submarine volcano.
The present-day volume of the Kameni Islands, as estimated from its height above the regional depth of the caldera, is 3.2 km
3[Pyle and Elliott, 2006]. If we assume that the islands were constructed in the caldera that formed following the Minoan eruption 1200 BC [Nomikou et al., 2014], then the implied aver- age magmatic addition rate is 0.001 km
3a
21. This rate, which has been used by Heiken and McCoy [1984]
to estimate the height of the island that was constructed in the center of the Cape Riva caldera during the
22 ka that elapsed between the Cape Riva and Minoan eruptions, is smaller than the rate deduced here, which suggests that magmatic addition during intra-caldera for- mation occurred in a number of high rate events. The volume of the Kameni Islands, for example, would require 18 events of the type that characterized the 1866–1870 erup- tion, an average of one such event every 200 year. Therefore, intra-cal- dera formation is associated with long periods of quiescence punctu- ated by rare effusive events involv- ing uplift and subsidence of up to 175 and 80 m, respectively every
200 year.
7.2. Collapse and the Viscoelastic Behavior of the Crust and Mantle Field-based observations [Fouqu e, 1879; Leycester, 1850; Brine, 1866;
Table 1. Summary of Volumes and Rates of Magmatic Addition in the Region of the Kameni Islandsa
Volume (km3) Duration (a) Rate (km3a21)
Early 1866 lavas (Palea Kameni-Nea Kameni channel) 0.05 0.33 0.15
Late 1866–1870 lavas (e.g., Giorgios dome) 0.25 1.30 0.19
Total 1866–1870 lavas 0.30/0.07/0.17
1925–1928 lavas (e.g., Dafni dome) 0.04/0.02/0.08 0.75 0.05
aThe volumes in red are estimates based only on subaerial lavas [Fouque, 1879; Pyle and Elliott, 2006]. The volumes in blue are based on subaerial and submarine lavas [Nomikou et al., 2014]. The volumes, duration and rates in black are estimates in this paper, which are based only on submarine lavas.
10
−410
−210
010
210
410
610
−410
−210
010
210
410
610
8Volume (km )3
Santorini Montserrat Monowai + seamounts Ocean islands Vailulu’u, Samoa
Duration of magmatism (a) 0.1 km /a
3
1000 km /a 3
0.00001 km /a 3
Figure 12. Plot of volume versus duration of magmatism for select submarine volca- noes in the Pacific and Atlantic oceans and the Mediterranean Sea. Large red-filled circles based on this paper. Large blue-filled circles based on Watts et al. [2006]. Small blue-filled circles based on Crisp [1984]. Other symbols are Monowai: blue-filled dia- mond [Watts et al., 2012], Vailulu’u: blue-filled square [Staudigel et al., 2006], Montser- rat: green-filled circles [Sparks et al., 1998]. Dashed gray lines show the rate of magmatic addition in km3a21.
Schmidt, 1868; Washington, 1926], together with the difference bathymetry maps in Figures 10e, 10f, and 10g, indicate that the Kameni Islands region is associated with both shallowing and deepening of the sea- floor since 1848. We have attributed the shallowing to growth and the build-up of magmatic material (e.g., submarine lava flows) on the surface of the preexisting seafloor.
The origin of the deepening is not as clear. It could be due to large-scale slope failure, thermal cooling and magma deflation, or some combination of these factors. Large-scale slope failures are a characteris- tic feature of volcanic oceanic islands and seamounts where the development of ‘‘head scars’’ and sliding on steep slopes generates debris flows with large (up to 1 km wide) blocks which, in turn, generate turbi- dites that are able to travel large distances across the seafloor [e.g., Hawaii - Moore et al., 1989; Tenerife - Watts and Masson, 1995]. Although landslide deposits have been reported on the submarine flanks of Santorini, east of Thera [Bell et al., 2013], we do not believe that the Kameni Islands have undergone sig- nificant mass wasting. The difference bathymetry maps do not show the characteristic upslope head scar and deepening or their associated downslope debris flow deposits and shallowing, typical of sector collapses on stratovolcanic cones of other arc volcanoes (e.g., Monowai). This is confirmed by the ROV dives carried out during 2010–2011 on board E/V Nautilus [Nomikou et al., 2012c, 2013a], although there is evidence of some downslope talus deposits and small, scattered, blocks all around the submarine edi- fice of Nea Kameni.
Post eruption cooling and contraction of lava flows and deflation of a shallow magma chamber can both create subsidence. Parks et al. [2015] show that cooling and contraction since the 1866–1870 loading event on the Kameni Islands could explain up to 50% of the present-day rate of subsidence of 6 mm a
21inferred from InSAR analysis and they propose that loading of historic flows and viscoelastic relaxation of the crust may account for the remainder.
The phenomenon of viscoelastic relaxation arises in oceanic crust (and lithosphere) because of its thermal cooling history and the strong dependence of viscosity on temperature [Watts and Zhong, 2000]. On vol- cano loading, bending stresses in the hot lowermost low-viscosity layer of the lithosphere would migrate upward into the cool uppermost high viscosity layer. The effective thickness of the lithosphere that sup- ports a volcanic load therefore changes with time from its initial short-term (seismic?) thickness to its long- term (elastic?) thickness. Evidence for this relaxation is seen in the rapid subsidence of newly formed vol- canic oceanic islands [Watts and Zhong, 2000] and the shift of the main depocenters in toward the load and stratigraphic offlap in their flanking flexural moats [Collier and Watts, 2001].
In order to apply the viscoelastic model to the Kameni Islands, we have used the volcanic loads esti- mated from masking the regions of growth in the difference maps in Figures 10e, 10f, and 10g and emplaced them on a simple flexural loading model. We follow Brotchie and Silvester [1969], who used a Green’s function approach to solve the general equation for the flexure due to a concentrated load on a 3-D plate and calculate the subsidence as a function of time for different shape loads and values of the thickness of an elastic plate and relaxation time of the underlying substrate. We use here a modified loading model in which the load is emplaced on the surface of an elastic plate that overlies a viscoelastic half-space. We assume that the density of the load, flexural infill, and half-space are 2800, 2800, and 3300 kg m
23, respectively. By comparing the predicted substance due to the total 1866–1870 load to the observed total subsidence between 1866–1870 and the present day (blue-shaded regions in Figures 10e, 10f, and 10g), together with the rates of subsidence inferred from InSAR [Parks et al., 2015], we have been able to constrain the effective elastic thickness, T
e, of the relatively strong upper layer and the vis- cosity of the relatively weak lower layer, g, that together are supporting the volcanic load.
Figure 13 shows in map form the subsidence and uplift expected at the present day for a volcanic load emplaced during 1866–1870 on an elastic plate (T
e5 0.6 km) that overlies a viscoelastic half-space (g 5 6.3 3 10
15Pa s). These parameters best explain both the magnitude and wavelength of the bathymetric- derived subsidence and the InSAR measurements (Figure B1). In particular, they account for the subsidence observed from field observations of the southeast coast of Nea Kameni [Fouqu e, 1879], the total subsidence of 80 m inferred from historical bathymetric charts offshore the southeast coast of Nea Kameni (Figures 10f and 10g), the lack of evidence of any subsidence from the shoals between Nea Kameni and Fira Port (Figures 8 and 13) and, the relatively slow rate of ‘‘background’’ subsidence derived from InSAR analysis of
6 mm a
21(Figure B1).
The best fitting parameters that we have derived for T
eand g are plausible, we believe, given the tectonic setting of the Kameni Islands. The intra-caldera location of an active arc volcano suggests a shallow magma chamber and ele- vated temperatures and, hence, low elastic thickness and low vis- cosities, of the order of values deduced at fast spreading mid- ocean ridges where T
eand g val- ues as low as 2.0 km [Cochran, 1979] and 10
14Pa s [Buck and Su, 1989], respectively, have been reported. We have confirmed this by calculating the temperature and viscosity structure associated with a simple model of a buried magma chamber of similar dimen- sions as inferred by Newman et al.
[2012] and Parks et al. [2012] from GPS and InSAR observations dur- ing the period of seismic unrest in 2011–2012 (Appendix B). The calculations show a thin, essentially elas- tic, upper layer that overlies a low-viscosity lower layer (actually a half-space) with a depth-integrated viscosity of 10
16Pa s and less within a radial distance of 3 km from the center of the magma chamber.
We note that our best T
evalue of 0.6 km is less than the seismogenic thickness, T
s, which during the 2011–2012 period of unrest along the Kameni line was between 1 and 6 km thick [Newman et al., 2012]. While both T
eand T
sare proxies for strength they are not the same: T
ereflects the integrated
long-term elastic thickness of the lithosphere, whereas T
sreflects the depth to which historical earthquakes have occurred [Watts and Burov, 2003]. Usually in oce- anic lithosphere T
eT
s, but both T
e< T
sand T
e> T
sare possible because T
ecan decrease with increasing curvature and bending stress and increase with the age of the lithosphere at the time of loading, in contrast to T
swhich simply reflects local stress levels.
A best fit viscosity, g, of 6.3 3 10
15Pa s implies a relaxation time, s , of the order of 4 year, assuming a shear modulus, m, of 5.0 3 10
7Pa (s 5 g/m). As Figure 14 shows (red solid line), most of the subsidence would therefore have occurred soon after the 1866–1870 loading event. This is because the relaxation time is short so the material underlying
1900 1950 2000
100 80 60 40 20 0
Year
V ertical displacement (m)
Early 1886 bathymetry map
1925-1928 bathymetry map
InSAR data 1866-70
Lava loading
2000 mm/yr
6 mm/yr
Figure 14. Plot of the maximum subsidence versus year. The subsidence has been computed for the same values of shear modulus and viscosity as assumed in Figure 13.
The red solid line shows the case of a one loading event at 1866–1870 while the blue solid line shows the case of two loading events at 1866–1870 and 1925–1928. The elas- tic thickness that best accounted for the magnitude and wavelength of the observed subsidence was 0.6 km for the one loading event and 1.0 km for the two loading event.
The plot shows that while both loading cases generally account for the rate of subsi- dence inferred by the InSAR-derived displacements, only the two load case is consistent with the relative proportion of subsidence observed between 1866–1870 and 1925–
1928 and between the 1925–1928 and the modern survey.
25˚22' 25˚23' 25˚24' 25˚25'
' 3 2
˚ 6 3 '
3 2
˚ 6 3
' 4 2
˚ 6 3 '
4 2
˚ 6 3
' 5 2
˚ 6 3 '
5 2
˚ 6 3
−80
−60 −40
−20 0
0
Shoals
1
1
Figure 13. Bathymetry map of the Kameni Islands region showing the predicted subsi- dence and uplift at the present day assuming the submarine 1866–1870 volcanic load was emplaced on the surface of a thin elastic plate overlying a viscoelastic half-space.
The subsidence and uplift have been computed for a shear modulus, m, of 5.0 3 107 Pa, viscosity, g, of 6.3 3 1015Pa s, and elastic thickness, Te, of 0.6 km. Black solid lines indicate subsidence (contour interval 5 220 m). Black-dashed lines indicate uplift (contour interval 5 11 m).
