Historical accounts provide insight on the geological evolution of the 20th century eruptions at Santorini volcano, Greece

Abstract

The 20th century eruptions of the Santorini volcano in Greece are the most recent activity of the volcano’s long lifespan. While the different eruptions taking place between 1925 and 1950 have traditionally been considered to exhibit similar eruptive styles, aspects of their evolution and precise information related to the individual eruption dynamics were poorly constrained. This study collates field reports and historical accounts, mainly from the Greek national scientific committee, which was assigned to study the volcanic activity in Nea Kameni Island with recent field campaigns. This analysis provides further insight into these eruptions and attempts to unravel the timing and style of explosive and effusive episodes that took place. Reconstruction of the recent geological evolution and of the eruptive history allow a more complete description of the eruption dynamics and associated unrest. These include fumarolic behaviour, explosion intensity, direction and volume of the lava flows, eruption duration, vent morphological changes (such as craters, domes, and horseshoe ramparts), textural characteristics and lava morphologies, as well as surface fracturing. Specific features related to first-hand accounts of the eruptions and associated products, in conjunction with our in situ post-eruptive geological study, allow an improved reconstruction of activity, both prior to and during the historical eruptions, which contributes to understanding the development of the eruption and enhances the forecast of potential future eruptions from patterns of precursory activity.

Introduction

Reconstructing a stratovolcano’s geological evolution is necessary to catalogue past eruptive history and forecast future activity (Sparks 2003; Sparks et al. 2012; Kilburn 2018; Gudmundsson et al. 2022; Kilburn and Bell 2022). Real-time and long-term forecasts can rely on holistic (multi-proxy) tools related to geophysical, geochemical, seismic and thermal signals (e.g. Polacci et al. 2017; Poland et al. 2020; Girona et al. 2021). However, this information is often based on models of the subsurface determined from geological or geophysical studies (Gudmundsson 2020; Geshi et al. 2020). To complement these signals and measurements, it is essential to hold good constraints of a volcano's eruptive history (Fedotov 1975; Staudacher 2010; Field et al. 2012; Andronico et al. 2021). Although there have been some successful attempts at forecasting eruptions (e.g. Linde and Sacks 1995; McNutt 1996; Harris et al. 2000), eruption forecasting remains uncertain, largely because of the inherent complexity of magma evolution through a volcanic system (Castro and Dingwell 2009) and poor knowledge (i.e. time-space) of past eruptive history (Karátson et al. 2018; Karátson et al. 2020; Biggs et al. 2021; Pyle and Barclay 2020).

Studying the frequency, duration and intensity of explosive or effusive eruptive activity from any volcano is necessary since the different behaviours can dynamically change the near and far-field hazards surrounding the eruption (Ripepe et al. 2009; Michon et al. 2013). Eruptive activity, even within a given eruption, can vary in character, suddenly or episodically (Andronico and Corsaro 2011; Walter et al. 2013; Cassidy et al. 2018). For example, pyroclastic flows (e.g. Merapi type or Peléan type) occurred during the predominantly dome building (effusive) 1991/92 Unzen volcanic eruption (Sato et al. 1992) and the formation of the Santiaguito dome complex in Guatemala in 2015/16 (Lamb et al. 2019). Constraints on the record of a volcano’s eruptive style in conjunction with its post-eruption activity are necessary for a robust, reliable understanding of its current and future state (Girona et al. 2015; Chiodini et al. 2016; Caudron et al. 2021). Nowadays, new technological advances such as high-resolution spaceborn remote sensing (Harris et al. 2000), unmanned aerial vehicle UAV imagery and data collection (De Beni et al. 2019), multiparametric monitoring (Cameron et al. 2018) and machine-learning based analysis (Malfante et al. 2018) provide valuable data for ongoing eruptions. In contrast, data collection in the past heavily relied on eyewitness accounts and reports (e.g. Gorée 1710; da Corogna 1867), a method that was highly qualitative in contrast to the recent quantitative methods.

Santorini’s stratovolcano is one of the volcanic centres that form the active Hellenic volcanic arc in the Aegean Sea. Back arc extension in the Aegean microplate began in the Miocene (Grasemann et al. 2012) but volcanic activity started in the Oligocene (Fytikas et al. 1984) and gradually migrated to the south (Jolivet and Brun 2010). Since the late Pliocene to Pleistocene, the volcanic activity on Santorini was concentrated along a NE-SW linear volcanotectonic feature, the Christiana–Santorini–Kolumbo volcanic field, which is associated with the regional Hellenic subduction system located south of Crete (Nomikou et al. 2019). Santorini’s last eruptions occurred in the 20th century between 1925 and 1950 on the volcanic island of Nea Kameni, located in the central part of an 8-km diameter, flooded caldera (Georgalas 1953). The island was formed after the last explosive caldera-forming eruptive activity (Druitt et al. 1999). In 2011/2012 the volcano underwent a period of unrest which produced rapid ground uplift (150 mm/year), an increased rate of diffuse gas emission and seismicity associated with the inflation of a shallow magma chamber located 4 km in depth (Newman et al. 2012; Parks et al. 2012, 2013; Lagios et al. 2003). The earthquake swarm associated with this unrest generated seismic activity along a NE-SW plane beneath the Nea Kameni Island. It was suggested that the unrest partly reactivated sections of the Nea Kameni volcanotectonic line (Newman et al. 2012; Papadimitriou et al. 2015; Karakonstantis et al. 2019). The geochemical unrest was corroborated with an increase in soil CO₂ flux (Parks et al. 2012; Rizzo et al. 2015) and anomalous gas measurements in the Nea Kameni fumaroles associated with increased H₂ and CO₂/CH₄ (Tassi et al. 2013). During the unrest period there was no evidence of magma transport nor any eruption (Browning et al. 2015; Rizzo et al. 2015).

It is currently unclear what would be the precise signals accompanying a dyke injection at Santorini, nor is it clear how any future eruption would manifest. Previous studies have attempted to assess possible future scenarios based on volcanic hazard assessments related to the eruptive history of Santorini volcano (Vougioukalakis et al. 2016) and empirical correlations between the eruption length and the eruption intervals for four prior large eruptions on the island (Pyle and Elliott 2006; Pyle 2018) or from probabilistic assessment of gas and ash hazards (Jenkins et al. 2015). Based on Pyle and Elliott (2006) and Pyle (2018), the most probable precursor would be general uplift around the island and discoloration of the sea waters some days to weeks before the eruption. The eruption would be characterised by effusive activity and lava domes followed up by irregular explosive activity and possibly vent formation. Based on recent chemistry of eruptive products, lava is expected to form a dome, which would grow to a height of 20 m in a day and offshore dome growth would last from 0.3 to 1 year for a 70 m height dome scenario or 1–3 years for more than a 100 m height. Similarly, based on older eruptions, the next eruption could be expected around Palea and Nea Kameni (northern part) and last from 2.7 to more than 4 years while for a duration longer than 2 months two or more eruptive events may be expected. To complement these previous studies, it is advantageous to extract more information from direct observations related to precursory and syn-eruptive activity to accurately reconstruct these historical events to ground forecast of future activity.

Here we analyse Nea Kameni’s eruptive behaviour derived from a comparative study of the last three eruptive episodes (1925–1928, 1939–1941, 1950). Our study combines a series of geological field observations from the 20th century activity with historical accounts from of the eruptions. More importantly, it explores the evolution of the volcano’s latest activity (1925–1950) and uses a method that relies on data extraction from historical resources similar to current volcanic activity reports. These data allow us to better constrain the short-term development of Nea Kameni volcano. The main aims of this work are then (1) to improve understanding of the short-time scale processes that occurred while the volcanic activity was in progress by reconstructing the evolution, frequency, intensity, duration and style of explosive/effusive activity; (2) collate information on the reported precursor activity preceding the three main eruptions (in 1925–1928, 1939–1941, 1950); and (3) revise previous terminology, revisiting acronyms, localities, and other sociological information that have been simplified/overlooked through time but that represent key parts of the geoheritage of Santorini volcano.

Geological setting

Santorini is a stratovolcano in the Aegean Sea which lies in front of the continental Aegean back-arc basin. It is part of the South Aegean Active Volcanic Arc (SAAVA), formed after the northward subduction of the African plate below the Eurasian plate (Meulenkamp et al. 1988; Jackson 1994). From a petrogenetic point of view, SAAVA is formed by partial melting of metasomatized mantle and slab-derived sediment due to dehydration of the subducting oceanic crust (Francalanci and Zellmer 2019). Slab rollback (Brun et al. 2017; Faccenna et al. 2003) created extension at the Aegean, a stress field that has remained active since the Miocene (Le Pichon and Angelier 1979). Several volcanic arc centres have been created, including Santorini, Soussaki, Methana, Aegina, Poros, Milos, Yali and Nisyros (Mitropoulos et al. 1987) (Fig. 1).

Fig. 1

A Simplified geological map of Santorini volcano modified from Druitt et al. (1999). The blue dashed rectangle box indicates our area of research. The relief is derived by the Copernicus DEM (https://land.copernicus.eu/) and the coordinates from Parks et al. (2012). B Map of Greece and the Aegean Sea modified from Konstantinou et al. (2006). C Map of Nea Kameni Island showing the locations—workpoints (WP) studied during the field campaigns. The coordinates are derived from Pyle and Elliott (2006). D Map of Nea Kameni Island showing the area that belongs to the three eruptive episodes in the 20th century. Images of Nea Kameni Island in parts C and D acquired by Google Earth ©2021, collected 6 June 2021 (data: Google, Maxar technologies). SAAVA: South Aegean active volcanic arc

Santorini is defined by the constellation of islands of Thira, Aspronisi, Thirasia, Palea and Nea Kameni islets (Heiken and McCoy 1984). It has been formed by sequences of Plinian eruptions (Druitt et al. 1999; Pfeiffer 2001; Gertisser et al. 2009), interplinian eruptions (Eriksen et al. 1990), caldera formation (Friedrich et al. 1988; Druitt and Francaviglia 1992; Druitt et al. 1999; Drymoni et al. 2022), effusive dome formation (Pyle and Elliott 2006; Nomikou et al. 2014), and dyke-fed eruptions (Drymoni et al. 2020). Three volcanotectonic lines, the Kolumbo and Kameni lines and the Christiana–Santorini–Kolumbo (CSK) rift zone, are right-lateral transtensional fault zones that trend in a NE-SW direction and cut the volcano at different points (Bohnhoff et al. 2006).

The oldest known volcanic activity generated the Akrotiri lavas (650 ka) in the island’s southern part (Druitt et al. 1999). The lavas lie above the non-volcanic basement, consisting of metamorphosed mudstones (pelites), recrystallised carbonates, cherts, metaconglomerates, metasandstones and metavolcanic rocks (Skarpelis and Liati 1990) found at the Athinios port, Profitis Ilias mountain, Mesa Vouno and Gavrillos localities. The stratigraphy of the island exhibits two explosive cycles (360–172 ka and 172–3.6 ka) that produced a variety of pyroclastic deposits and three lava constructs, i.e. Simandiri (172 ka), Skaros (70–54 ka) and Thirasia shields (48–24 ka) (Druitt et al. 1999). Several collapsed caldera events have interrupted the effusive and explosive activity sequence (Heiken and McCoy 1984), the last of which occurred during the Minoan era at 3.6 ka (Friedrich et al. 2006; Manning et al. 2014; Pearson et al. 2018; Kutschera 2020). The eruption discharged 30–60 km³ of material and deepened the pre-existing caldera morphology (Druitt et al. 1999 and references therein). Observations from the exposed parts of the pre-volcanic basement and the older activity of the Peristeria stratovolcano can be found at the northern part of the caldera wall (Druitt and Francaviglia 1992; Druitt et al. 1999). The volcano’s northern, western and southern parts are permeated by a radial dyke swarm hosting approximately 99 dyke segments (Drymoni 2020) which conditionally interact with faults in space and time (Drymoni et al. 2021, 2022).

Materials and methods

Historical analysis of the 20th century eruptions

The 20th century eruptions of the Santorini volcano are divided into three episodes interrupted by periods of repose. Information and data reported on the Nea Kameni eruptions in the international bibliography are in English, German and French (e.g. Fouqué 1879; Reck 1936; Hantke 1951; Newhall and Dzurisin 1988; Friedrich 2000), and several studies have used records from the historical eruptions (even before the 20th century) to reconstruct the submarine activity (Nomikou et al. 2014; Watts et al. 2015) or to assess future volcanic crises (Fytikas et al. 1990; Vougioukalakis et al. 2016; Pyle and Elliott 2006; Watt et al. 2007; Jenkins et al. 2015).

During this study, we focused on reports/books, mainly in Greek but also translated in other languages, from the National Academy of Athens presented by the national scientific committee, which was assigned to study the Nea Kameni islet during the 1925–1950 eruptions. This section deals with recounting the salient aspects of the 20th century eruptions as derived from information of the following resources (Akilas 1925; Ktenas 1925a; Ktenas 1925b; Ktenas 1925c; Ktenas 1926a; Ktenas 1926b; Georgalas 1925a; Georgalas 1925b; Kokkoros 1926; Ktenas 1927a, 1927b; Kokkoros and Ktenas 1928; Ktenas 1928; Critikos 1928; Ktenas and Kokkoros 1928; Georgalas and Liatsikas 1936a, 1936b; Georgalas and Kokkoros 1940; Liatsikas 1942; Liatsikas 1944; Georgalas 1950; Georgalas and Papastamatiou 1951; Georgalas 1953; Georgalas and Papastamatiou 1953; Georgalas 1959; Georgalas 1962). Our selection is based on the following criteria: (1) the reports we analysed were provided by the official scientists assigned to each eruptive phase, (2) the reports contained in situ observations from the scientists who visited the volcano themselves during its activity or whom collated data from civilians/local authorities, mainly during 1939–1941 due to World War II, (3) reports that contained data that could be compared directly with our field observations, and (4) reports or books that could be retrieved either online, or from the National Academy of Athens, or the L. & E. Bellonia Foundation Library on Santorini volcano or hardcopies collected by other scientists and provided to us.

By exploring these studies, we could better examine the evolution of the dynamic physical parameters during the eruptions. For example, they allowed us to constrain (if reported), more precisely, the style of the eruptive behaviour (frequency, duration and intensity of its explosive or effusive activity), the eruptive products, the duration of each eruptive period, some precursor phenomena, and specific features related to first-hand accounts of the morphology of the eruption products (craters, domes, horseshoe ramparts).

Geological analyses of the 20th century eruptions

Field campaigns were performed on Nea Kameni Island to confirm or support certain aspects of the erupted product morphologies and which, when combined with the historical records provided a more complete portrayal of the dynamics of the eruptions. The fieldwork focused on collecting volcanological, geomorphological and structural data from the last three eruptive episodes, e.g. 1925–1928, 1939–1941, 1950. In detail, the macroscopic study reported their textural characteristic, lava morphologies, and secondary surface fracturing. We studied surface morphologies (e.g. morphological lava fabrics), physical processes (e.g. signs of gas escape and oxidation), and structural data that could denote flow patterns and rheological features. The comparative data attempts to constrain dissimilarities in the evolution of the volcanological activity and volcanotectonic setting during the 20th century eruptions.

Results

This study combines historical data collected from previous scientific reports on the eruptive activity of Nea Kameni from 1923 to 1962 with field data on structural and macroscopic petrological analysis of the same products from recent geological campaigns. We interpret the reported geological observations and use them to reconstruct the dynamic morphological changes of the island and the explosive and effusive characteristics of the eruptions. We also expand our observations into broader concepts by analysing data such as the localities, orientation and exposure of surface fractures, and eruption profiles. A comparison of the 20th century eruptions provides insight into the short-term evolution of the last eruptions on the island.

Volcanological constraints and eruption styles based on data collated from historical reports

The eruptions of 1925–1928

First phase of the formation of the Kameni-Fouque (Daphne lavas)

Akilas (1925) recorded the presence of two new hydrothermal vents formed at least two years prior to the 1925 eruption. He postulated that these locations might be the sites of future eruptions. The first vent was located at the Georgios double crater (Fig. 2b), where sulphate deposits were seen to be crystallised in the crater’s surroundings, and gases were escaping from NW-SE active tension fractures around the crater. The second was an underwater vent within the bay between the Mikra and Nea Kameni islands (Fig. 2a) before they eventually merged in 1925. The vent was responsible for turning the seawater red, most likely due to the release of iron-rich fluids, and so the bay was named “Kokkina Nera” (i.e. red waters). Akilas (1925) noted that the water’s temperature was hot enough to destroy the fisherman’s boats, and as such, it was an area they largely avoided. However, the larger boats used the hot waters as a natural method for cleaning their hulls from microorganisms and flora. These reports agreed with previous reconnaissance campaigns between 1890 and 1921 from several scientists that reported hydrothermal activity from solfataras at various spots around Nea Kameni after the 1866–1870 eruption episode. The solfataras released CO₂ and SO₂ gases, and the offshore vents were acidic hot spots (40–60°C) as discerned from free HCl concentrations (Ktenas 1925a).

Fig. 2

Interpreted geomorphic maps and interpreted satellite image showing the evolution of the 1925–1928 lavas modified from Ktenas (Ktenas 1925a, Ktenas 1925b ,Ktenas 1925c). a Morphological evolution of the 1925 eruption according to Akilas (1925), modified after Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c). The map shows (dotted line) the initial morphology of the bay before the eruption, according to Akilas (1925). The underwater vent is denoted with a white dot. The Kameni Fouque and its gradual evolution is also observed: (I) the lava expansion 48 h after the initial eruption (13 August) from Akilas (1925), (II) the lava expansion 44 days after the initial eruption (24 September) from Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c), (III) the lava expansion 46 days after the initial eruption (26 September) from Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c) and the dome of Daphne with the rampart ejecta surrounding it. The yellow area corresponds to the water evaporation sites. b Kameni Fouque and its dome on 6 January 1926 according to Akilas (1925) and modified after Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c). The three branches (Liatsikas, Georgalas, Kokkina Nera) and their flow direction are shown on the map. The black spots show the fumarolic activity during that time. F denotes the surface fractures (active in red) observed. Light cherry red colour denotes the lava flow, dark cherry red colour denotes the newly formed levee and green the old levee. c The study area during the early stages of the eruption of 1928 modified from Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c). More information regarding this stage is shown in Fig. 3. d The 1925–1928 lavas nowadays as seen from above (Google Earth image ©2021 collected 6 June 2021, data: Google, Maxar technologies). The yellow arrows show the lava flows and the current annotated morphology. The blue dots show the location of our new field sampling points (with reference number) collected during our campaigns

The first phase of the 20th century eruptions began in August 1925 (Ktenas 1925a, Ktenas 1925b). Historical catalogues (Newhall and Dzurisin 1988), personal reports and publications (Akilas 1925) indicate that precursory phenomena were initiated just 10 days before the beginning of the activity at the location of a previous eruption in 1866. On 28 July 1925 and later between 7 and 9 August 1925, a mild earthquake swarm was felt in the main town of Santorini, Fira. The first reports of a volcanic eruption came on 11 August 1925. The explosive activity produced a plume that spread ash 1.5 km away (Newhall and Dzurisin 1988). Local fishermen reported a rise in the local sea temperature and excessive wave turbulence near Nea Kameni on 10 August 1925 (Reck 1936). The first phase of activity reached its peak 11 days after the first eruption. The eventual eruptive activity was a mixture between a Vulcanian and a Peléan type with characteristic sequences of effusive activity (domes and lava flows) frequently interrupted by explosive episodes releasing significant amounts of pyroclastic material (Ktenas 1925a; Ktenas 1925b). No casualties were reported at the beginning of the activity. However, a fisherman did have a close escape while successfully saving his two children left on the Mikra Kameni Island that day (Akilas 1925).

After the beginning of the August 1925 activity, three Greek navy ships, named Daphne, Nautilos (or Naftilos) and Iera, surrounded the island to observe the ongoing activity and conduct sonar surveys. Forty-eight hours after the first plume, as seen from Oia village, a lava dome 55m above sea level (a.s.l.) had reportedly formed, and two lava flows were flowing towards the north and the east surrounding the Mikra Kameni island (Akilas 1925) (Fig. 2a). Two days later (13 August), an off-centred crater cone (dipping 10° towards the northwest) appeared at the dome’s top. Its upper section had a diameter of approximately 60 m, while the base was almost perfectly circular with a diameter of 120 m. The cone changed shape and size throughout the eruption as it was destroyed and rebuilt periodically. Pyroclastic materials accumulated gradually on the cone’s north and northwest sides over the next 4 months, forming a horseshoe shape with a rampart (Ringwallberge in German terminology; Wolff 1914) of 150 m in diameter. During the first phase of activity (11–28 August 1925), the dome reached a height of 73 m (a.s.l.) with a growth rate of 0.34 m/h and a maximum growth rate of 1.15 m/h in the first 2 days. By the activity’s end, the inclined dome’s eastern section was dipping 20° towards the northeast and deflated to 40–45 m (a.s.l.). The northern area remained at around 55 m (a.s.l.), and its width was almost 40 m. The crater of Daphne, which lies with its top at 62 m a.s.l, produced several solfataras that reached heights of up to 62 m (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c).

There is a minor but interesting historical discrepancy regarding the name of the 1925/28 eruptions. All current maps and publications refer to the 1925–1928 activity and products as ‘Daphne lavas’. Akilas (1925) and Ktenas (Ktenas 1925b; Ktenas 1925c), however, report that the name Daphne was restricted only to the crater named after the navy ship, which arrived in the area soon after the activity on 11 August 1925. Indeed, the new effusive activity was named ‘Kameni Fouque’ after the French petrologist Ferdinand A. Fouque who had extensively studied the previous eruptive activity (1866–1870) at Nea Kameni volcano. Ktenas (Ktenas 1925b, Ktenas 1925c), in his reports, explains that his formal nomination was first indicated by two letters (16 and 19 August 1925) to the local newspaper ‘Estia’ since no name had been given to the new activity, neither by the scientific community nor by the Greek naval forces by that time. The National Academy of Athens received a formal notification about this nomination on 7 September 1925, and the Greek and the International committees accepted his nomination. Finally, Ktenas (1925b) denotes that only a reporter from the newspaper ‘Scrip’ used the name ‘Daphne’ for the activity in an article on 17/18 of August at that time. Still, they were referring to the volcano between Palea and Nea Kameni.

The lavas extruded from Daphne’s crater flowed in three directions forming three distinct branches, as shown in Fig. 2b (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c; Georgalas 1953). The northern branch, named after the Greek professor of petrology and mineralogy Georgios Georgalas, covered an area of 660,000 m². The western unit or Liatsikas flow was smaller, forming an area of 290,000 m², and took its name from the Greek geologist Nikolaos Liatsikas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c). The eastern branch was named after the local hot spot ‘Kokkina Nera’ and covered an area of 130,000 m² (Newhall and Dzurisin 1988). The quantitative data regarding the branches and the lava flows recorded by Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c) rely on topographic measurements made by himself and aerial photos taken by a Greek navy seaplane on February 1927.

This activity ceased on 22 May 1926 (Ktenas 1926), extruding a volume of lava equal to 108,239 m³, covering an area of 980,000 m² while its dome reached a maximum height of 89 m (Georgalas 1953). Since April 1926 the fumaroles close to Kameni Fouque dome were experiencing reductions in temperatures from 380°C on April 28 to 150°C on July 28, 1926 (Ktenas 1928). On 8 May till 20 September 1926, a new system of tension fractures with an N-S orientation formed 300 m S-SW from the Daphne dome at the Kokkina Nera flow (Reck 1936). Later, several fractures appeared in 1927–1933 on the Daphne crater walls, and its crater partially subsided, generating strong SO₂ and CO₂ emissions. However, the recorded fumaroles found around the craters on the 7^th of March 1927 had a maximum temperature equal to 184°C (Ktenas 1928).

Second phase of the formation of Kameni-Fouque: the parasitic eruption of 1928

The fumarolic activity increased again on 20 January 1928 (Reck 1936; Georgalas 1953). A minor earthquake swarm presaged the resurgence of the eruptive activity at the Kokkina Nera branch between 23 January and 17 March 1928 (Kokkoros and Ktenas 1928; Critikos 1928; Ktenas and Kokkoros 1928), characterised by four phreatic eruptions accompanied by effusive and explosive eruptions (Friedrich 2000). This phase was studied by close observation from M.P. Syrigos, a catholic clergyman who visited the volcano three times during its activity between 20 February and 15 March 1928, and M. Ioakimidis, who provided photographs of the volcano. Data from the explosive activity were provided by M. Aeginitis, the director of the National Observatory of Athens (Ktenas and Kokkoros 1928).

The activity originated from four craters and was found to follow a NE orientation in an area with a diameter that reached 100 m on the south-eastern part of the Kameni-Fouque dome (Figs. 2c and 3). In the early stages, the eruptions were characterised by gas phases and solfataras. The level of the explosive activity was qualitatively defined by the following parameters: (a) the plume height, (b) the duration of the activity and (c) the distance of the ejected solid products (Ktenas and Kokkoros 1928; Georgalas 1953). The second phase of the eruption was divided into four stages:

• 1. 23–28 January 1928: low-level explosive activity was produced by four craters A, B, C and D. Crater A was an elliptical (major axis 42 m, minor axis 28 m) crater with 4 m depth located 130 m S-SE of the Kameni Fouque central dome. Crater B was a circular 20 m (radius) crater with 4 m depth located 40 m SW of crater A. Both craters A and B were covered by products deposited during the 1925–1926 eruptions. Crater C located S-SW of crater B and 20 m away from crater Georgios A (1866 eruption). It was also elliptical with a diameter of 25 m covered by the 1925–1926 products. The chief scientist of the 1928 activity (P. Kokkoros) reported the existence of small circular features that allowed gas release on top of the crater. Crater D was developed along with a set of N-NE tension fractures and was surrounded by many others that centralised fumarolic activity during the 1925–1926 eruptions (Fig. 3a).

• 2. 29 January–5 February 1928: low-level explosive activity produced by craters A and C.

• 3. 6–26 February 1928: low-level explosive activity produced by crater C.

• 4. 28 February–15 March 1928: medium-level explosive activity.

Fig. 3

Schematic map modified from Ktenas and Kokkoros (1928) showing the morphology and location of craters A–D (no information about craters E–F could be found) during the parasitic eruption of 1928 (observations following M.P. Kokkoros). a A close-up section of Kameni Fouque and the surrounding craters which produced the long explosive activity between 23 January and 3 March 1928. b Kameni Fouque, after the effusive activity and the formation of Nautilos dome (4–15 March 1928) modified from Ktenas and Kokkoros (1928). The denoted red box marks the position of a

On 4 March 1928, lava was produced, and a parasitic dome was formed at the southern part of the Kokkina Nera flow covering the area of craters C and D. The dome had a diameter of 70–80 m and a height of 10–15 m and was cut by a series of crosscutting fractures (Ktenas and Kokkoros 1928). Three more domes were formed locally close to the central Kameni-Fouque dome, but very little information became available (Fig. 3b). This phase of the eruption and products was named after the navy ship ‘Nautilos’ (Newhall and Dzurisin 1988), and the eruptive volume was 40,000 m³ (Georgalas 1953). The morphological evolution of the volcano during these eruptions was critical since three more craters (craters E, F and G their exact location in Fig. 3 could not be retrieved) were formed in the surroundings of the Kameni-Fouque dome. The two of the craters were ellipsoidal and had a major axis of 27 m, located 35 m north of crater A. The third one was located 50 m north of crater A; it was deep and circular. Fumaroles were reported in all craters during this phase.

Compared to the previous eruption (1925–1926), several paroxysmal phases occurred almost hourly but with less explosivity, lower than in 1925. White plumes were formed and reached 600–800 m in height. The eruption is characterised as parasitic with a progressive change of explosive to effusive activity. Mineralogically the deposits were similar to the 1925 eruption but sometimes more crystalline (Ktenas and Kokkoros 1928). The ‘Kameni Fouque’ term did not survive in time and was eventually replaced by ‘Daphne lavas’. Ktenas (1928) has reported similarities between the 1925–1928 eruption and the volcanic activity of Tarumai (1909) volcano in Hokkaido Japan (Tanakadate 1924). Finally, he reported further similarities between the systems of the craters that formed the Kameni Fouque with the Auvergne domes in France (Ktenas 1926).

The ‘polycentric’ eruptions of 1939–1941

This episode was studied only by Greek scientists and particularly by local authority and residents because World War II largely prevented the international community from studying the ongoing activity. Some of the observers were Mr. Giannakas, Dr. Koveos (local medicine doctor) and Don Zacharia Douratso-Morozini (Georgalas 1959). On 20 August 1939, a series of precursory phenomena, such as rising sea temperature, shore subsidence, fumarolic activity and red-coloured sea waters, were again reported (Liatsikas 1942). This time, the red-coloured waters appeared on the St. Georgios Bay at the western coast, which is now covered from later flows. Two domes were formed during the eruption, Triton A and B (named after the fish-tailed sea-God in ancient Greek mythology), accompanied by small volume explosive activity. The domes gradually united to form a single one with a height of 13 m a.s.l. while a lava flow associated with the dome began a steam bay in the NW, covering an area of 9850 m² (Georgalas and Kokkoros 1940). Liatsikas (1942) also studied the domes on the 4th day of their formation. According to his field observations, Triton A was the central dome responsible for the effusive activity. At the same time, Triton B was restricted only to a weak explosive activity over the first four days. After 24 August 1939, the explosive activity ceased, but Triton A continued to grow. This competing account then suggests that the two domes did not coalesce as reported later (Georgalas 1953), but instead, only Triton A continued to form. Six days later, a 21 m wide and 14 m deep funnel-type structure, the ‘Warathron’, was created from ‘noiseless’ explosive eruptions at the southern foot of the Nea Kameni (1707) dome that produced ash clouds.

On 23 September 1939, new series of eruptive activities were initiated, and by 1 October 1939, a new dome formed 260 m to the NW (N 30° W) from the Georgios lavas top. Fresh lava flowed towards the SW and covered the previously mentioned activities, covering 132,500 m² (Fig. 4a). The dome was named Ktenas, after the geologist who studied the eruptions of 1925–1928. It reached a height of 89 m a.s.l. and the explosive ejecta formed a horseshoe rampart. The activity ceased on November 25th of the same year. The Greek Ministry of Education members have named the Triton and Ktenas domes. In addition, Fouque, Reck and Smith domes were proposed by the geologist N. Liatsikas, the director of the geological community and chief scientist during the reactivation of 1939–1941 (Liatsikas 1942).

Fig. 4

Annotated satellite maps (Google Earth images ©2021 collected 6 June 2021, data: Google, Maxar technologies) showing the distribution of lavas during the second and third eruptive episodes as described by Liatsikas (1942) and Georgalas (1953). a The location of Ktenas dome (1939) along with its horseshoe rampart; the direction of its lava flow is shown with a yellow arrow, the orange dotted line shows the nowadays residual deposit of Ktenas lavas. b The location of Fouque (1939/40) dome along with its horseshoe rampart; the direction of its lava flow is shown with the yellow arrows, the orange dotted line shows the nowadays traceable remnants. c The location of Smith dome (1940); the yellow arrow shows the direction of Smith effusive activity, and the location of the double explosive craters at the Georgios crater top. d The location of Reck dome (1940) and its crater; the direction of its lava flow is shown with the yellow arrows, the orange dotted line shows the nowadays distribution of lavas. e The location of the Niki dome (1941); the direction of its lava flow is shown with yellow arrows, the orange dotted line shows the nowadays distribution of lavas. f The Liatsikas dome formed during the 1950 eruption. The numbers indicate recorded GPS ground-controlled points

During the Ktenas activity, a new eruption occurred at the Warathron locality (could not be documented on a map) on 13 November 1939. Four days later (17 November), a new dome was formed at 75 m NE from the Ktenas dome. This new dome, the Fouque dome, formed three branches (Fig. 4b). The eastern branch covered the Nautilos dome (1928 eruption) and parts of the 1925–1926 and Georgios lavas. The northern branch overlapped the north slope of the Nea Kameni summit and some of the 1925–1926 lavas. Finally, the southwestern branch completely covered the Ktenas dome except for a small part close to the sea level and some parts of the 1870 lavas. The lavas covered a total area of 376,250 m² and produced 27 paroxysmal explosions. For this reason, it was deemed to be stronger than the previous activity. The lava branches have not been given specific names; hence, they are simply referred to as the Fouque lavas. This activity ceased on 9 July 1940, while its dome was 122 m tall and surrounded by ejecta that shaped a horseshoe rampart, open to the SW (Liatsikas 1942; Georgalas 1953).

The next eruptive activity was described in detail by Liatsikas (1942). It refers to the formation of Smith and Reck lavas, an activity that formed two domes north of the Georgios lavas (Fig. 4c, d). Smith activity initiated on 12 July 1940, along with a dome that evolved into a crater. Reck activity started 4 days later, accompanied by fumarolic activity. Both domes remained active simultaneously after 6 September 1940, when Smith ceased its activity while Reck continued until October. During the end of the Reck activity, Smith had a short continuance up to November 1940. Smith was characterised by frequent but weak activity, while Reck seldom erupted but with stronger explosions. Both eruptions did not form horseshoe craters, but their lavas flowed towards SW and E. The northern Reck’s eruptive volume was more significant than Smith’s. Their activity was more potent than Triton but weaker than Ktenas and Fouque.

During the activity of Smith and Reck domes (Fig. 4d), another explosive event took place between 28 August and 8 September 1940. Liatsikas (1942) reported it as the double crater explosive event (A and B) closely supervised by the geologist G. Aronis, a member of the geological committee during that time (Fig. 4c). The location of the craters overlapped with previous craters that were formed during the 1860–1870 lavas of Ag. Georgios. No effusive activity was reported.

The last event of the 1940/41 was the effusive and explosive activity of Niki’s lavas observed from its early signs and reported by Don Zacharia Douratso-Morozini who was keeping a daily calendar of his testimonies and observations. He was also visiting Nea Kameni to collect sea temperature measurements and macroscopic observations (Georgalas 1959). Again, the ongoing war did not permit other scientists to study the eruption site. Douratso-Morozini was a local catholic priest and a social benefactor who was keen on publishing reports about the history of Thira and its socioeconomic evolution from 1904 to 1936 (Kasapidis 2016). The activity was initiated on 24 November 1940 and ceased on 1 July 1941. The effusive activity covered the northeastern and eastern part of the Nea Kameni Island and the Nautilos dome and crater (Fig. 4e). An overview map showing the location and morphological features of the 1939–1941 activity is shown in Fig. 5.

Fig. 5

Schematic map of the Nea Kameni Island showing the eruptive products of 1570, 1707–1710, 1866–1870, 1925–1928 and the location of the craters and domes formed from 1929 up to1940 modified from Liatsikas (1942)

The ‘epigenetic’ (or Liatsikas) eruption of 1950

Georgalas (1953, 1959 and 1962) report that the eruption started on 10 January 1950 at 1 am local time, accompanied by a small earthquake and noise. Fifteen small eruptions and one paroxysmic followed the following days on the eastern side 17.5 m away from Georgios dome (or Kameni Georgios) (1866–1870) (Fig. 2b). Lava initiated from surface fractures at the eastern flanks of Georgios crater (1866–1870) and flowed towards the south-eastern flank of the Georgios dome, forming an ellipsoidal dome (coulee) with a base diameter of 40 m at 108 m (a.s.l) (Fig. 4f). Fumarolic activity was reported both in the western part (50–60°C) and the northwestern region (190–210°C), along with SO₂ and HCl emissions. The activity finally ceased on 3 February 1950. Due to its limited duration and volume, it was characterised by many scientists as a minor event (Georgalas and Papastamatiou 1951). However, Georgalas (1953) reported it as an epigenetic eruption which became later related to precursor deep seismicity in the Aegean (Papadopoulos 1986). The eruption was named after N. Liatsikas, who died at that time. The eruption should not though be coincided with the Liatsikas lava flow that is the western branch of the 1925 activity. Finally, during the Liatsikas eruption in 1950, the geologist Georgalas visited the volcano while it was erupting to provide qualitative and quantitative data for his and Liatsikas previous reports (Georgalas and Liatsikas 1936b; Liatsikas 1942, 1944). He provided a large dataset (not provided here) that reflects changes in the explosive activity of the 1950 eruption (Georgalas 1953, 1959, 1962).

The term ‘epigenetic’ is seldom used nowadays. An epigenetic eruption describes the magmatic process where a residual bunch of magma cannot extrude from the magma chamber due to vent blockage. Bring this into a modern notation, this can either reflect the cooling and crystallisation of the feeding fingers of the eruption around the central vent (Georgalas 1953) or possibly insufficient overpressure to drive the magma to the surface. As a result, the latter can finally erupt when the magmatic excess pressure can be built up later in the magma chamber and trigger a new dyke which can overcome the tensile strength (To) of the surrounding host rock (Gudmundsson 2020) and feed the eruption.

Effusive vs explosive eruption mechanisms during the three eruption episodes

It is possible to characterise the effusive activity by the volume of extruded lavas in each eruption. The historical observations from those witnessing and reporting the eruptions provided such constraints. These allowed us to compare the relative size of each eruption phase qualitatively as follows, where V is the erupted volume in m³. The smallest to the largest eruption, in terms of relative volume, is the Daphne parasitic eruption (1928) (followed successively by) < Triton (A and B) (1939) < Liatsikas (1950) < Reck (1939) < Ktenas (1939) < Smith (1939) < Fouque (1939) < Niki (1939) < Daphne (1925–26). The total erupted volume of the Daphne 1925–1926 phase is higher than all 1939–1941 eruptions combined. An estimation of the total Daphne eruption volume is 1,080,000 m³ in 1925–1926 and 40,000 m³ during just the parasitic eruption of 1928 (Ktenas 1925b; Ktenas 1925c).

It is again possible to derive data from the historical accounts to qualify the relative size and explosivity of the studied eruptive phases. Since a plethora of scientists and observers collected data during the studied eruptions, we synthesised their data to draw comparisons. The first phase of eruptions (1925–1928) was reported mainly by Akilas (1925), Ktenas (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c), Kteénas (1926, Ktenas, 1926) and Georgalas and Liatsikas (Georgalas and Liatsikas 1936a, Georgalas and Liatsikas 1936b). The descriptions in these studies were focused mainly on the geomorphological features and volcanological observations rather than the physical characteristics of the eruptions, such as the duration and frequency of the eruptions, the plume heights, and the vertical and lateral distance of the solid ejecta. However, their qualitative analysis did include correlations between the explosive styles of the two phases of Kameni Fouque.

We restored quantitative data regarding the height of the plumes, the frequency of the eruptions and the mean vertical and lateral distance of the solid ejecta reported below during the second (1939–1941) and the third episodes (1950), where available (see Table 1). WW2 crucially affected the data collection during the second eruption phase. The chief geologists, expert in Aegean eruptions, and their international colleagues could not visit the volcano (except during some special occasions). They had to rely mainly on observations from citizens living in Santorini. The collected data were mainly quantitative and restricted to observations that lack volcanological perspectives and terminology.

Table 1 Diary of the overall eruption characteristics of the 20th century eruptions from (Akilas 1925; Ktenas 1925a; Ktenas 1925b; Ktenas 1925c; Ktenas 1926a; Ktenas 1926b; Georgalas 1925a; Georgalas 1925b; Kokkoros 1926; Ktenas 1927a, 1927b; Kokkoros and Ktenas 1928; Critikos 1928; Ktenas and Kokkoros 1928; Georgalas and Liatsikas 1936a, 1936b; Georgalas and Kokkoros 1940; Liatsikas 1942; Liatsikas 1944; Georgalas 1950; Georgalas and Papastamatiou 1951; Georgalas 1953; Georgalas and Papastamatiou 1953; Georgalas 1959; Georgalas 1962)

These data in conjunction with the qualitative information reported allow comparisons to be derived using the following classification in terms of the eruption intensity of each eruption: Daphne parasitic eruption (1928) = Triton (A and B) (1939) < Liatsikas (1950) < Smith (1939) < Reck (1939) < Ktenas (1939) < Fouque (1939) = Niki (1939) < Daphne (1925). A combined qualitative (regarding eruption explosivity) and quantitative (regarding erupted magma volume) comparative illustration of the 20th century eruptions’ effusive and explosive style is shown in Fig. 6. Here we can note, generally, that those eruptions with the highest intensity produced the largest measured volume of erupted material. However, there are interesting caveats to this generalisation. For example, while the Smith and Reck eruptions produced very similar volumes of erupted material, the Reck eruption was observed as being somewhat more explosive. In terms of volume, the smallest eruption was, by far, the most recent Liatsikas eruption (1950) and the least explosive was the parasitic eruption of 1928.

Fig. 6

Schematic diagram showing the relative volumes of the 1925–1950 eruptions (depicted by the size of the edifice) with their relative intensity (depicted by the size of eruption column)

Moreover, the historical records allowed us to make qualitative comparisons that shed light on the change of the eruptive style for each of the 20th century eruptions. In Fig. 7 and Table 3, we analyse the duration and temporal evolution of the various stages of each eruption. The longest eruption in 1925 (phase A) lasted 266 days and the shortest in 1939 (Warathron) lasted just 5 days (with a short resurgence 2.5 months later). By demonstrating the time between the first explosive plume and the appearance of the first lava flow for each eruption, we find that the 1928 (phase B) and 1939 (Ktenas) eruptions generally had the longest periods of time (between 7 and 40 days) between their explosive and effusive behaviour. The other eruptions all presented less than 2 days until the change in eruptive style. Finally, there are two documented eruptions in 1939 (Warathron) and 1940 (double craters) that they were purely explosive.

Fig. 7

Eruption style for each ongoing eruption vs their duration (in days). For the absolute time of the eruptions, please refer to Table 3

Volcanological analysis based on field data collected from recent field campaigns

We performed two geological field campaigns focusing on the comparative study of the 20th century eruption products. We made geological observations at the localities shown in Fig. 1c and performed a detailed macroscopic and microscopic petrological examination of the different lava flows. Since this study focuses on joint qualitative analyses of historical and geological data, with some quantitative analysis related only to historical data, we report only the macroscopic comparisons. Both types of data rely on mesoscale observations.

During the field campaigns we used ground-controlled points to make sure that our studied points overlap (or are very close) to the localities where the historical data were retrieved to allow comparisons between the current and previous landscape. We studied the lava textures (scoriaceous or massive) and type (e.g. blocky, rubbly). We identified oxidised surfaces, enclaves (or xenoliths) (see Martin et al. 2006), signs of gas escape (vesicular textures), and flow surfaces.

The 1925–1928 eruption products

The Daphne lavas are predominantly blocky, with some individual blocks > 2 m wide. The exterior of the blocks is often oxidised and covers mainly the fresh internal part. Two main macroscopic textures can be recorded, firstly, a compound or massive surface and, secondly, a scoriaceous marker texture that usually covers the massive parts of the lava (Fig. 8). Several enclaves/xenoliths are seen in some parts of the lavas restricted to the core (interior) of the Georgios and Kokkina Nera flow (WP 030, Fig. 1c). A crater (Daphne crater) has been found.

Fig. 8

a Oxidised Daphne lavas covered by scoriaceous marker textures (WP 003, see Figs. 1c and 2). b Eastern view of the Niki (Nk) and Daphne (Df) lavas (vessels for scale). c Massive lavas (WP 030, see Figs. 1c and 2d). d A closer look at the blocky Daphne lavas and their enclaves/xenoliths (20 cm diameter) (the hammers are around 35 cm in length)

The 1939–1941 eruption products

The 1939–1941 lavas are characterised by six slow or fast transitioning explosive to effusive eruptions, or vice versa, and two entirely explosive eruptions covering the island’s central part. The oldest eruptions could not be studied during the field campaigns since Triton A and B, the Warathron locality and the Ktenas lavas (the part closer to the main path) are covered by later flows. In addition, the double craters that were activated between the Smith and Reck phase presented evidence of only explosive activity in the historical records but could not be retrieved in the field scale. Although all those eruptions were not studied in the field, their historical data are used for the final synthesis. In the next part, we report comparative macroscopic observations related to Fouque, Smith, Reck and Niki lavas.

The Fouque lavas (WP 025, Fig. 1c) (1939/40) show signs of gas escape, alteration, and surface oxidation on the exterior. In addition, rafted blocks of bedded scoria lavas and brecciated lavas show the complexity of the processes that formed the Fouque coulee and its collapsed horseshoe based on the historical data (Fig. 9). The historical data report two lava domes (Reck and Smith), which accompanied two lava flows towards the SW (Fig. 10a). Our field observations in the study area have identified the remnant of Smith dome (1940) at the southern contact with the Georgios lavas (1866/70). The lavas are blocky and consist of both massive and scoriaceous parts. In this flow, only the scoriaceous components are surrounded by oxidised surfaces (Fig. 10b). A remnant of stalactite lava was found (Fig. 10c), but no other morphological features were observed. Reck lavas were extruded after the explosive episode that generated the double craters (A and B, Fig. 5) within the Georgios crater. They form a distinctive rubbly texture, and the lava products are highly scoriaceous and extensively oxidised (Fig. 11), which has also been reported by previous studies (Holness et al. 2005). Niki lavas are the last flows during the 1939/41 phase. Their texture is blocky, with individual blocks > 2 m wide. The lavas are glassy and similar to obsidian texturally. They bear enclaves and xenoliths (Martin et al. 2006) (Fig. 12).

Fig. 9

a Collapse of the Fouque dome, b gas escape and oxidation, c oxidised crust (WP025, see Fig. 4b), d rafted blocks of bedded scoria lavas, e cellular textures, and f brecciated lavas (the hammers are around 35 cm in length)

Fig. 10

a Western view of the Smith (Sm) blocky flow (WP028, see Fig. 4c), (Gr) denotes the Georgios lavas (1866–1870), b massive and scoriaceous parts of the lava, and c stalactite lava (the hammers are around 35 cm in length)

Fig. 11

a The crater of Reck dome (WP024, see Fig. 4d) and b the contact of Reck (Rc) lavas with Smith (Sm) flow showing the difference between the blocky Smith texture and the rubbly Reck lava

Fig. 12

a Eastern view of the Niki dome which evolves into a levee towards the NE (see also Fig. 4e), b, c flow and folding patterns, d brecciated lavas, and e blocky lava texture (WP 014, see Fig. 4e) (the hammers are around 35 cm in length)

The eruption of 1950

Liatsikas lavas are the last activity of the Nea Kameni volcano that formed oxidised surfaces and flow patterns. The textures were also blocky, but no brecciated products or glassy textures could be observed, indicating that the volatile content could have been decreased during the eruption (Fig. 13).

Fig. 13

a Northern view of the Liatsikas dome (Lt) and its contact with the Georgios lavas (Gr) (1866–1870) and Niki lavas (Nk) (1940–1941). b Eastern view of the Liatsikas top flow showing the fractured and oxidised surfaces, c, d flow textures and layering, and e, f oxidised surfaces (WP017, see Fig. 4f) (the hammers are around 35 cm in length)

Discussion

Comparison of the textural characteristics and lava morphologies

Lava textures are found to be controlled by their temperature during emplacement, their viscosity, their volatile content, and the morphology of the deposits (Voight and Elsworth 1997; Huppert 1982). The main volcanic products of the three explosive periods described here were dacitic domes, lava flows and pyroclastics, which created classic surface morphologies commonly associated with viscous felsic lavas. During the field campaigns, we identified rafted blocks of bedded scoria (Fig. 9d), flow folds and layering (Figs. 12b, c and 13c, d), stalactite lavas (Fig. 10c), coulee (Fig. 9a) and levee (Fig. 3d), blocky lavas (Figs. 8, 9, 10, 12 and 13), rubbly lavas (Fig. 11) gas escape and oxidation markers (Figs. 9, 10, 11, 12 and 13), enclaves (Figs. 8, 9 and 12), ogives, blast pits and talus piles (Fig. 3d). The evolution of the Nea Kameni (1925–1950) activity in the historical accounts reports similar geomorphological characteristics with only very minor differences. Assessing all this information together suggests that there were changes occurring within the shallow magma feeding system.

The main changes and evolution of textural characteristics can be summarised as follows:

• 1. The typical texture of all the lavas is blocky, apart from the Reck eruption, which was predominantly rubbly.

• 2. The Kameni Fouque (1925/1926), Ktenas (1939) and Fouque domes (1939/40) formed horseshoe ramparts constructed with a large amount of solid ejecta and pyroclastic material.

• 3. The Kameni Fouque (1925/1926), Fouque (1939/40), and Reck (1940) and Niki (1941) domes evolved into craters.

• 4. The Fouque lavas (1939/40) exhibited several geomorphological features and petrological textures associated with increased degassing, such as rafted lava blocks and bedded scoriae, autobrecciation, pyroclastic flows.

• 5. Only the last two eruptions, the Niki lavas (1941) and the Liatsikas flow (1950), show signs of folding, flow patterns and layering, within the lavas, which could be associated with minor changes in lava viscosity.

• 6. The Daphne (1925–1928), Ktenas (1939), Fouque (1939/40), Reck (1940), Niki (1940/1941) and Liatsikas (1950) eruptions all created coulees and levees.

• 7. Macroscopic enclaves and xenoliths were observed in the Daphne (1925–1928), Fouque (1939/40) and Niki (1940/41) lava flows.

Eruption style, size and explosivity

Analysis of historical data provided both quantitative and qualitative data relating to each eruption’s relative explosivity and effusivity. The following common patterns can be inferred for all the eruptions:

• i) They were pre-empted by similar precursory phenomena as documented in Table 1, including (re)activation of surface fractures and increasing fumarolic activity.

• ii) During the eruptions the explosive activity formed white-coloured plumes which gradually changed to darker colours.

• iii) Interchangeable sequence of explosive and effusive activity.

• iv) Gradual decrease of the explosive and fumarolic activity, termination of the eruption and post-eruptive fumarolic activity which declines with time.

However, some eruptions deviate from these characteristics, such as the explosive activity of the Warathron and double crater eruptions (see Table 1) in between the Triton and Ktenas and Smith and Reck lavas, respectively, which themselves did not produce a lava flow. Also, the Reck eruption (1940) has similarities, according to the historical data, with the 1928 parasitic eruption of the Daphne lavas, and so it could have also been described as parasitic to the previous Fouque eruption although our statistical analysis does not support this notion. The joint geological and historical study provided a detailed timeline of each activity and its distinct style allowing us to get valuable insights on their eruption profiles as follows.

The 1925–1928 eruption profile

Geological perspectives

The 1925–1928 lavas were extensively studied at the time of emplacement by Georgalas and Liatsikas (Georgalas and Liatsikas 1936a,Georgalas and Liatsikas 1936b) and Ktenas, (Ktenas 1925a, Ktenas 1925b, Ktenas 1925c; Ktenas 1926a, Ktenas 1926b) who focused mainly on their geomorphological evolution, macroscopic textures, petrological indications and fumarolic activity. The lavas are blocky, they include xenoliths and enclaves (see Holness et al. 2005; Martin et al. 2006) and the recent field observations restricted to lava flows (coulee and levee) at the northern and mostly northeastern part of Nea Kameni within the crater of Daphne demonstrate typical geomorphological features related to felsic lavas such as ogives, a talus pile and a blast pit. Although the magma composition was constantly dacitic, the last effusive activity produced lava with more ropy textures than the previous products. These textures can be attributed to the higher temperature of the lava, perhaps because the magma was stored longer in the magma chamber, and a more considerable amount of contained volatiles could not escape (Ktenas 1925b; Ktenas 1925c).

Ktenas 1925b and Ktenas 1925c reported a series of tension fractures permeating the central dome at Mikra Kameni, while the orientations of the fractures were not reported our field reconnaissance notes what we believe to be the same fracture sets striking predominantly N30E (Fig. 2b). Before and during the eruption, fumaroles formed within and close to the fractures, shifted in the NW section of the dome to N40W fractures close to the Georgios A dome (Fig. 2b), striking parallel to the orientation of lava fractures previously formed in Nea Kameni. Solfataras of the 1925–1928 dome forming eruptions suggest fracturing closely followed the alignment of the present-day volcanotectonic lineaments (NE-SW) and vent distributions across the Kameni islands, a notion that was characterised by subsurface tectonic control (Fytikas et al. 1990; Druitt et al. 1999; Pyle and Elliott 2006). The existence of two perpendicularly striking fracture systems (NW-SE and NE-SW), as shown in Fig. 2b, could imply that the local volcanotectonic setting on Nea Kameni Island is anisotropic and directionally controlled, which is a topic that requires further investigation. Mikra Kameni dome was permeated by a NE-SW fracture set aligned with the previous Georgios A crater (1866–1870). Its surface fractures on the southern part (NW-SE) were slightly reactivated since a few fumaroles existed. The same NW-SE system permeated the Nea Kameni dome, but no fumarolic activity was observed.

Historical perspectives

The first phase (1925–1926) initiated with a specific sequence of precursor phenomena:

• 1. Onshore and offshore hydrothermal vents two years prior to eruption (long-term precursors)

• 2. Active tension fractures and gas escape

• 3. Red waters and temperature increase within the water and shore subsidence

• 4. Fumarolic activity

• 5. Earthquake swarm 10 days before the eruption (short-term precursor)

The 1925–26 activity was explosive but fast evolved into effusive in 2 days (Fig. 7). The activity initiated with plumes and intense explosive activity at a pre-existing vent. The vent became covered by a dome that experienced several rapid morphological changes during its bulging. The dome evolved fast into a crater accompanied by other smaller craters on its periphery that were formed during the last paroxysmal phase (17–22 March 1926), delineating a small subsidence area. The effusive activity covered the northern and northeastern part of the island, and the synchronous prolonged explosive and effusive activity formed a horseshoe rampart around the main crater that became open to the NE at the very final stages of the eruption. In May 1926, the explosive and effusive activity ceased and the temperature of the fumarolic activity decreased by around 40%.

The second phase of the eruption (1928) had limited precursor phenomena which was restricted to gradual increase of fumarolic activity a year prior to eruption and intense fumarolic activity just 3 days before the eruption. The activity initiated prolonged explosive activity until a dome was formed towards the end of the eruption; hence the local scientific community interpreted this behaviour as a parasitic eruption. The explosive activity was characterised by intensive phreatic eruptions, which formed several craters around the main crater of the 1925/26 eruption. The effusive activity was lower and restricted to a dome and a new eastward lava flow.

1939–1941

Geological perspectives

The eruptions of this episode are categorised as polycentric eruptions since explosive and effusive activity occurred synchronously from more than three centres during the 1939/1940 parts of the eruption. This term reflects the synchronous explosive or effusive activity of more than one crater or domes in a volcano, and it is rarely used now. Such a mechanism could correspond to either a lateral dyke that feeds the different centres as it moves towards the south or several vertical fingers that propagate at the same time from the shallow magma chamber. For that reason, a volcanotectonic study of this eruptive behaviour should be considered in the future. Similar activity on the island was only previously reported at the synchronous centres of the Georgios lavas and the Afroessa lava flow (1866–70) (Liatsikas 1942).

Our field studies identified very few geomorphological features compared to the actual processes described by historical scientific reports. In detail, our observations are restricted to the Niki dome (1941) that formed during the end of this episode and two craters namely Reck (1940) and Fouque (1939/40) and the remnant of the Smith (1940) dome. The Fouque crater is surrounded by a coulee (Fig. 8) covered by pyroclastic and solid rock deposits (horseshoe rampart). Oxidised surfaces covered the solid rock deposits with cellular textures, possibly associated with gas depressurisation and extensive hydrothermal alteration (Ball et al. 2015). Numerical studies by the same authors provided insights on collapsed domes associated with hydrothermal alteration proposing that shallow-seated domes can be formed at the contact between the core and the talus of the dome if there is a sufficient heat supply. As such large-scale alterations during dome collapse can form clay minerals (Opfergelt et al. 2006), the presence of alunite, kaolinite and smectites can locally increase the pore fluid pressure forming apparent lower-strength zones which are prone to fracture or promote explosive decompression, leading to collapse and slope failure (Voight and Elsworth 1997; Boudon et al. 1998; Ball et al. 2015). We propose that the Fouque coulee collapse may have been associated with similar mechanisms.

Historical perspectives

The 1939–1941 activity had similar precursory phenomena (hot spot, earthquake swarm, water temperature rise, shore subsidence, red waters, fumaroles) with the previous episode (phase A 1925/26), but the reports did not allow us to recover a timeline as for other eruptions. This eruption was characterised by the synchronous activity of multiple craters that experienced explosive and effusive activity. For that reason, the eruption was named as ‘polycentric’ by the scientific community. The brief sequence of the events reports two domes, Triton A and B (1939), accompanied by low explosive activity followed up by a noiseless explosive event, called Warathron (1939) that formed a funnel-type structure in the area. The activity continued with the formation of a new dome, the Ktenas dome (1939), which evolved into a horseshoe rampart crater. All the aforementioned activity was covered by the deposits of the Fouque lavas (1940). The Fouque lavas were emplaced along with the Warathron subsidence locality (fracture) and later filled the fractures after a period of active hydrothermal activity. The early stages of this dome were surrounded by a horseshoe rampart which was open to the SW. The activity was followed by two synchronous explosive and effusive eruptions that formed two domes, the Smith (1940) and Reck (1940), respectively. The second evolved into a crater. This synchronous activity was overlapped by a pure explosive event that formed two double craters namely A and B (1940). The activity completed with the last eruption that formed a dome named Niki (1940/41) later evolved into a crater. In total, the 1939–1941 eruptions are represented by six volcanic centres that formed both effusive and explosive activity, generating lava domes and craters and two volcanic centres produced only explosive activity and created a structure of double craters and a funnel-type structure, respectively.

1950

The eruption of 1950 is described as an epigenetic eruption since it was proposed to represent a residual magma batch stored in the shallow magma chamber but could not erupt during the Niki eruption. The historical data reports that the maximum eruption frequency was eight eruptions per hour. The solid projectiles reached 100–200 m vertical and horizontal distance from the crater source, and the plume height was 50–300 m (Georgalas 1953) (Table 2). Instead, our field observations indicate a coulee with signs of folding, flow patterns and layering, within the lavas.

Table 2 Quantitative data during the second (1939–1941) and third phase (1950) of the eruptions. Data reports mean values collected from (Georgalas and Liatsikas 1936a, 1936b; Georgalas and Kokkoros 1940; Liatsikas 1942, 1944; Georgalas 1950; Georgalas and Papastamatiou 1951; Georgalas 1953). The undocumented data are based on the aforementioned reports

1925–1950 eruptions vs 2011/12 unrest

After the eruptions of 1866–1870 Nea Kameni volcano remained active hosting a shallow hydrothermal system. This can be discerned from the hydrothermal activity and fumaroles reported to surround the craters continuously from 1866 till nowadays. The 1925–1950 volcanological reports and the in situ observations from the locals have corroborated a list of precursory phenomena (earthquake swarm, temperature rise, shore subsidence, red waters). Subject to their existence, either a vertical dyke managed to feed an eruption on the surface or multiple edifices were activated synchronously, and different dykes or fingers of the same dyke (since enclaves and xenoliths exist in both Fouque (1939/40) and Niki (1940/41) deposits) triggered mostly explosive events. In the vast majority of the eruptions the volcano released gas phases first and soon afterwards lava flows, in very short timescales (6-day interval, Table 3) which relates to the felsic composition of the magma.

Table 3 Diary of the eruption timelines during the three main studied phases collected from (Akilas 1925; Ktenas 1925a; Ktenas 1925b; Ktenas 1925c; Ktenas 1926a; Ktenas 1926b; Georgalas and Liatsikas 1936a; Georgalas and Liatsikas 1936b; Kokkoros 1926; Ktenas 1927a, 1927b; Kokkoros and Ktenas 1928; Critikos 1928; Ktenas and Kokkoros 1928; Georgalas and Liatsikas 1936a, 1936b; Georgalas and Kokkoros 1940; Liatsikas 1942; Liatsikas 1944; Georgalas 1950; Georgalas and Papastamatiou 1951; Georgalas 1953; Georgalas and Papastamatiou 1953; Georgalas 1959; Georgalas 1962)

During this time no earthquake activity or surface deformation activity was reported, in contrast to the unrest event that occurred in 2011/12 which accompanied substantial seismicity and deformation. We assume that the local scientists and international community would have reported these data, but they were difficult to collect compared to now (Pyle 2018), especially during WW2 when the condition of the monitoring system was rudimentary. This allows us to hypothesise that the 2011/12 incident was the first tangible event of unrest without eruption on the island during the last century. Subsequently, the differences between a dyke-fed eruption and an unrest on Santorini rely heavily on the type of precursory phenomena. A dyke-fed event combines increased hydrothermal activity, signs of surface deformation (shore subsidence) but also an earthquake swarm which spans locally around the future vent. The seismicity profile based on the reports relies on a dyke that propagates over a few days towards the surface. Hence, when the citizens of Fira in 1925 felt the earthquakes, that gave them around a 48-h warning for the dyke to make it to the surface. So, a dyke of similar overpressure based on the historical accounts presented here can practically feed an eruption in 8–10 days. Instead, the 2011/12 unrest profile reports a combination of high seismicity, geochemical unrest and surface deformation (uplift) for almost a year without eruption (Newman et al. 2012; Parks et al. 2012; Lagios et al. 2003; Rizzo et al. 2015). Still, it should be noted that the vast majority of dykes do not make it to the surface to form an eruption but instead become arrested in the crust (Gudmundsson 2020; Drymoni et al. 2020).

Thanks to the higher-quality qualitative data of the 20th era, we assess the eruptive behaviour in shorter timescales such as days and understand the shorter spatial and temporal evolution of the specific eruptions. Those indications allow us to infer that eruptions on the island are characterised by fast escalation, 10 days average from the first earthquake based on the 1925 eruption data (Tables 1 and 2) and the pre-eruptive signals from the day of their first appearance (Table 3). Subsequently, the lava flow or crater formation begins approximately 2 days (on average) after the first plume; exception to this rule is the second phase of the 1925–1928 lavas where the lava flow occurred 40 days after the first plume (Fig. 7). Finally, the assessment of the eruption profiles for the 11 studied eruptions (Table 1) based on the eruption episode shows a variety of eruptive behaviours:

• 1925–1928—an explosive to effusive transition accompanied, or not, by domes that evolve into craters.

• 1939–1941—strong and synchronous explosive and effusive activity accompanied by domes and craters interrupted by mainly effusive behaviours (Triton and Reck) or absolute explosive behaviours (Warathron and double craters)

• 1950—strongly explosive to low effusive transition

Based on the duration of the eruptions (Fig. 7) we can classify them into three categories:

• Long eruptions (≥94 days): a fast explosive to effusive transition delineated by a vent that transforms into a horseshoe rampart accompanied by paroxysmic events. The eruption recharges magma from the shallow magma chamber (as suggested by the presence of xenoliths and enclaves).

• Intermediate eruptions (53–93 days): a fast explosive to effusive transition or synchronous explosive and effusive activity.

• Short eruptions (10–23 days): either purely explosive (if less than 10 days) or a strong/weak explosive to low effusive transition with or without dome formation.

Under most circumstances, the state of the magma in the shallow reservoir controls the eruptive behaviour (Popa et al. 2021). Our observations show that the temporal evolution of the intensity and eruption volumes are not monotonous. Since they all occur in very short timescales, the different degassing behaviours between eruptions could possibly relate to magma ascent (Gonnermann and Manga 2007) and other external processes such as the geometry of the conduit, the decompression rate, the viscosity of the magma or the volatile gradients (Cassidy et al. 2018) as well as stress changes caused by the eruption itself (Tarasewicz et al. 2012) or overpressure build-up at the top of the magma column (Sparks and Melnik 1999). Previous studies have shown that transitions between explosive and effusive eruptions can either record changes in the magma flow rate for small magma reservoirs or temporal variations in the chamber pressure (Woods and Koyaguchi 1994). Recent studies from Volcan Chaiten in 2008 and Cordon Caulle in 2011–2012 have shown that hybrid eruptions (where explosive and effusive behaviours coincide) can occur by magma sintering within the conduit (Wadsworth et al. 2020). We report that explosive to effusive transitions on Santorini occur relatively quickly, so we suggest changes related mostly to magma ascent processes rather than magma chamber conditions. Magma–hydrothermal–ground/sea water system interactions can also produce local explosions, and more investigation is needed to determine such dynamics at Nea Kameni.

Conclusions

The main conclusions from our study can be summarised as follows:

• 1. The translation and re-examination of historical reports combined with field observations and volcanological analysis are a powerful tool for constraining details of the qualitative characteristics of past eruptive activity since their individual study is inconclusive. Our study used a small part of the available data and further studies are needed on restoring more data.

• 2. The eruptive styles of the 20th century eruptions at Santorini volcano are more diverse although compositionally monotonous. While the magma composition and conditions within the underlying magma chamber did not change significantly, the combination of explosive and effusive eruptions presented key differences to produce rather different surface styles and morphologies in the erupted lavas which led to varying degrees of surface fracturing. Although previous studies have proposed correlations between magmatic processes (magma generation, ascent, storage and crystallisation) in the area, critical aspects of the links between the magmatic and volcanotectonic conditions remain enigmatic. We recommend that future studies attempt to shed light on the current arrangement of the magmatic plumbing system beneath Nea Kameni.

• 3. Several pre-eruptive precursory signals were evident across all the eruptive periods, such as the appearance of fumaroles and elevated temperatures in parts of the shallow ocean. It is reasonable to assume similar precursory signals would arise in the days, weeks, and years before future eruptions at Nea Kameni. It is essential that continuous monitoring of the key locations described is carried out.

• 4. We recognise different types of eruption which could all be expected in future eruptive events based on the longevity, the intensity and the eruption style.