Characterisation of the Quaternary eruption record: analysis of the Large Magnitude Explosive Volcanic Eruptions (LaMEVE) database
© Brown et al.; licensee Springer. 2014
Received: 5 September 2013
Accepted: 7 March 2014
Published: 25 March 2014
The Large Magnitude Explosive Volcanic Eruptions (LaMEVE) database contains data on 1,883 Quaternary eruption records of magnitude (M) 4 and above and is publically accessible online via the British Geological Survey. Spatial and temporal analysis of the data indicates that the record is incomplete and is thus biased. The recorded distribution of volcanoes is variable on a global scale, with three-quarters of all volcanoes with M≥4 Quaternary activity located in the northern hemisphere and a quarter within Japan alone. The distribution of recorded eruptions does not strictly follow the spatial distribution of volcanoes and has distinct intra-regional variability, with about 40% of all recorded eruptions having occurred in Japan, reflecting in part the country’s efforts devoted to comprehensive volcanic studies. The number of eruptions in LaMEVE decreases with increasing age, exemplified by the recording of 50% of all known Quaternary eruptions during the last 20,000 years. Historical dating is prevalent from 1450 AD to the present day, substantially improving record completeness. The completeness of the record also improves as magnitude increases. This is demonstrated by the calculation of the median time, T50, for eruptions within given magnitude intervals, where 50% of eruptions are older than T50: T50 ranges from 5,070 years for M4-4.9 eruptions to 935,000 years for M≥8 eruptions. T50 follows a power law fit, suggesting a quantifiable relationship between eruption size and preservation potential of eruptive products. Several geographic regions have T50 ages of <250 years for the smallest (~M4) eruptions reflecting substantial levels of under-recording. There is evidence for latitudinal variation in eruptive activity, possibly due to the effects of glaciation. A peak in recorded activity is identified at 11 to 9 ka in high-latitude glaciated regions. This is absent in non-glaciated regions, supporting the hypothesis of increased volcanism due to ice unloading around this time. Record completeness and consequent interpretation of record limitations are important in understanding volcanism on global to local scales and must be considered during rigorous volcanic hazard and risk assessments. The study also indicates that there need to be improvements in the quality of data, including assessment of uncertainties in volume estimates.
KeywordsVolcanic hazards Database Explosive eruptions Magnitude Under-recording Record completeness Japan Quaternary
The LaMEVE (Large Magnitude Explosive Volcanic Eruptions) database is a product of VOGRIPA (Volcano Global Risk Identification and Analysis Project), a component of the Global Volcano Model (GVM). LaMEVE is an open access database available online at the British Geological Surveya. The objectives behind the database are to facilitate understanding of how explosive volcanism is distributed in space and time, to identify locations at high risk and gaps in knowledge, and to enable assessments of societal and environmental impacts of volcanism. Global datasets enable scientists and disaster managers to analyse hazard and risk within a global context of systematic information.
The Smithsonian Institution’s Global Volcanism Program (GVP) provides a global dataset documenting Holocene eruptions of all magnitudes (Volcanoes of the World 4.0 (VOTW 4.0)b). However, Deligne et al. (2010) showed that the Holocene is too short a time period to be statistically representative for M > 6.5 eruptions. A database of M ≥ 8 eruptions by Mason et al. (2004) permitted the study of the very largest eruptions recorded through the Ordovician to the present, with no eruptions of this size recorded by Mason et al. (2004) in the Holocene. The LaMEVE database incorporates and builds upon these and other data sources and analyses of global volcanic information in particular to fill a knowledge gap for eruptions of M > 6.5. Relevant and complementary large datasets include the 2012 version of the Collapse Caldera Database (Geyer and Martí 2008), which has been in part incorporated within LaMEVE. The Large Quaternary Caldera list of Decker (1990), a Preliminary List of Large-Volume Pleistocene Eruptions (Siebert et al. 2010) and two mature databases focussing on volcanism in Japan – the One-Million Year Tephra Database of Hayakawa (2010) and the Active Volcanoes of Japan Database (2008)) – are also important data sources for LaMEVE. In short, LaMEVE is a collaborative effort to integrate these datasets and move towards a comprehensive and systematic, open-access source of volcanic data on large magnitude explosive eruptions.
LaMEVE contains data on Quaternary (the last 2.58 Myr as defined in Gibbard et al. 2010) explosive eruptions of magnitude (M) 4 and above, where magnitude is a function of erupted mass (Pyle 2000). The database contains information on 3,130 volcanoes, of which 471 have had at least one M ≥ 4 eruption. We designate these 471 volcanoes as Quaternary Explosive Activity Recorded (QEAR) volcanoes. All volcano types are eligible for inclusion, as the sole physical criterion is the magnitude. There are 1,883 eruptions of M ≥ 4 in LaMEVE. M ≥ 4 eruptions are less frequent than smaller (M < 4) events (Siebert et al. 2010), but present the greatest risk at local to global scales with the potential for high death tolls, economic losses and societal disruption over large areas (Auker et al. 2013). The largest magnitude eruptions can cause global disturbances through the injection of ash and aerosols into the atmosphere, causing climate perturbations over several years (Robock 2000). However, even relatively small eruptions can have global economic ramifications, as illustrated by the M4 2010 Eyjafjallajökull eruption which created mass disruption across Europe by grounding air travel. In 1990, 455 million people were estimated to live within 100 km of a volcano known to have been active in the Holocene (Small and Naumann 2001). Auker et al. (2013) suggested that, with population growth, this has increased to at least 600 million people.
The largest known explosive eruption in the geological record is the M9.3 Guarapuava Tamarana-Sarusas rheoignimbrite eruption from the Paraná-Etendeka igneous province (Bryan et al. 2010), which occurred around 132 Ma, thus falling outside the temporal scope of the LaMEVE database. Volumetrically larger lava eruptions have been recorded in Large Igneous Provinces, but dominantly effusive products are not included in LaMEVE. The largest magnitude eruption currently in LaMEVE is the ~74 thousand years ago (ka) M8.8 eruption of the Younger Toba Tuff at Toba, Indonesia (Ninkovich et al. 1978; Chesner et al. 1991; Oppenheimer 2002; Petraglia et al. 2012; Storey et al. 2012). The Baegdusan-Tomakomai eruption at Changbaishan, China at 950 BP (Volcanoes of the World 4.0, 2013), and the Dakataua caldera-forming eruption in Papua New Guinea at 998 BP (Machida, 2002) are the two largest magnitude Holocene eruptions in LaMEVE (both M7.4). The Arequipa Ignimbrite of Nevado Chachani, Peru is the oldest event currently recorded in LaMEVE at 2.42 Ma.
Crosweller et al. (2012) described the LaMEVE database structure and content in detail, so only a summary is provided here. The name, synonyms, coordinates, region and volcano type are provided for each volcano in LaMEVE, derived from the Smithsonian Institution’s VOTW 4.0 to ensure compatibility between the two databases. Although 85% of the volcanoes in LaMEVE do not currently have associated eruptions listed, they are included to facilitate updates as new information becomes available. LaMEVE has been created to provide a sustainable, open-access catalogue of eruption data. As such, there is no single specified degree of uncertainty for dates, volumes or magnitudes throughout LaMEVE, as the data are derived dominantly from peer-reviewed studies published over the last 90 years. Where studies provided estimates of uncertainty, this has been included in LaMEVE and this information has informed evaluation of overall uncertainties in the dataset. Quality indices using simple criteria are given for ages and magnitudes as a first pass assessment of data reliability (Crosweller et al. 2012). References are given for all data derived from the literature; other entries are calculated from other known properties or ‘assumed’ on the basis of qualitative descriptions. The following key data are provided for each eruption, with additional data, such as uncertainty estimates, when available:
Bulk DRE (Dense Rock Equivalent)
VEI (Volcanic Explosivity Index)
Tephra fall volume (bulk, DRE)
Ignimbrite volume (bulk, DRE)
Intra-caldera deposit volume (bulk, DRE)
This paper provides an introduction to, and synoptic analysis of, the datasets compiled in the LaMEVE database to describe spatial and temporal trends in recorded volcanic activity, identify gaps in the existing knowledge, and discuss issues of under-recording. In this paper, the term ‘under-recording’ refers to the level of record completeness relative to the “true” but unknown record. Under-recording results from a variety of factors, including the failure of people to record events in the historical period (see Siebert et al. 2010), variable preservation of deposits and extent of scholarly study. The assessment of under-recording in this paper is based on the assumption that the global rate of explosive volcanic activity has been stationary throughout the Quaternary. Mason et al. (2004) identified clustering on time scales of the order of tens of millions of years, using this to support the hypothesis that rates of large eruptions are non-uniform and possibly controlled by regional or global tectonics. Although this paper attributes variations in the record of explosive volcanism largely to under-recording, we recognise at the outset that interpretations are confounded by the possibility of non-stationarity in rates of volcanism. We evaluate the assumption of stationarity a posteriori, having considered the evidence.
LaMEVE is a dynamic database and will be updated periodically to account for and incorporate new data. Our analyses use Version 2, released in July 2013. Some data contained within LaMEVE will become obsolete as further research is undertaken and better information emerges. This analysis uses the most up-to-date, published, peer-reviewed data available and therefore presents a review of the status quo. One purpose of the study is to understand the quality of existing data to identify knowledge gaps, issues of data quality and deficiencies in methodologies so that there can be improvement.
Global and regional variability in volcano and eruption distribution
Proportion of volcanoes, eruptions and Quaternary Explosive Activity Recorded (QEAR) volcanoes in LaMEVE by hemisphere
Regional breakdown of QEAR volcanoes and eruptions in LaMEVE
Number of QEAR volcanoes
Number of Quaternary eruptions
Eruptions/ QEAR volcanoes
Japan, Taiwan, Marianas
Japan, Taiwan, Marianas
New Zealand to Fiji
Mexico & Central America
Kamchatka & Mainland Asia
Japan, Taiwan, Marianas
N. America inc. Alaska
Mexico & Central America
Europe (Mediterranean and W.Asia)
Kamchatka and Mainland Asia
Europe (Mediterranean and W.Asia)
Iceland and Arctic Ocean
N. America inc. Alaska
Kamchatka & Mainland Asia
Africa and Red Sea
New Zealand to Fiji
Europe (Mediterranean and W.Asia)
Iceland and Arctic Ocean
Hawaii and Pacific Ocean
Melanesia and Australia
Mexico & Central America
New Zealand to Fiji
Melanesia and Australia
Iceland and Arctic Ocean
Philippines and SE Asia
Africa and Red Sea
Melanesia and Australia
Philippines and SE Asia
N. America inc. Alaska
Philippines and SE Asia
Middle East and Indian Ocean
Middle East and Indian Ocean
Middle East and Indian Ocean
Africa and Red Sea
Hawaii and Pacific Ocean
Hawaii and Pacific Ocean
It is clear from both the number of volcanoes and eruptions that the majority are part of the circum-Pacific ‘Ring of Fire’. Table 2 also lists the number of eruptions per QEAR volcano in each region. The New Zealand to Fiji region has the highest number of eruptions per QEAR volcano, although this region only has 16 QEAR volcanoes out of 74 recognised volcanoes. This may be a reflection that few of the region’s volcanoes have experienced explosive activity in the Quaternary, or that a small number have been studied in detail. JTM and the Atlantic Ocean are both second with a mean of 6.8 eruptions per QEAR volcano. To test whether these differences are meaningful we have applied the Pearson’s chi-squared test to the data in Table 2.
Pearson’s chi-squared is used to measure whether the difference between variables (in this case, regions or eruptions) is one that could have occurred on the basis of chance alone. Firstly, we tested the number of QEAR volcanoes per region compared to the total number of identified volcanoes. The test assumes that if a particular region has a certain share of the total number of volcanoes (e.g. 10% in the case of JTM) then, if there was no statistical difference between the numbers of QEAR volcanoes in each of the regions, it would be expected to have a roughly equal share of the QEAR volcanoes. However, the test shows that there is a significant (p < <0.01) difference between these values. In particular, JTM, Mexico and Central America, Kuril Islands and Europe all have significantly greater proportions of QEAR volcanoes than expected given the overall distribution of all identified volcanoes; Kamchatka and Mainland Asia, Indonesia, and the Middle East and Indian Ocean all have much lower proportions in comparison. This does not change when the influence of the JTM region is removed from the analysis. Although some of the difference here could be due to some regions having more volcanoes prone to explosive eruptions than others, the result remains statistically significant.
When considering the expected number of eruptions given the proportions of QEAR volcanoes, chi-squared results also show eruptions in some regions are significantly better recorded than others. Excluding the JTM data from the analysis, those regions with the most complete record are New Zealand to Fiji, Kamchatka and Mainland Asia, and Iceland and the Arctic Ocean; regions with the most under-recording are Africa and the Red Sea, Melanesia and Australia, and Indonesia. Unlike the results discussed in the previous paragraph, the tendency for some regions to have more explosive volcanoes would not influence these results as this has already been accounted for by using QEAR values (each QEAR volcano has had at least one M ≥ 4 eruption). Thus these results indicate that recording levels differ significantly between various regions of the world.
With much of the LaMEVE data being from the JTM region, characteristics particular to that region are likely to skew analysis of global data. As such, some of the following analyses investigate the whole global dataset, just the JTM dataset, and the rest of world (ROW) dataset (global excluding JTM).
Dating of eruptions
Percentage of eruptions in LaMEVE dated using each technique group
Dating technique group
Radiocarbon (corrected), Radiocarbon (uncorrected), U-Th series, Ar-Ar, K-Ar, Isothermal Plateau Fission Track, Fission track
Stratigraphy, Sediment accumulation, Magnetic stratigraphy, Oxygen isotopes, Ice core, Relative dating, Radiocarbon and Stratigraphy, Tephrochronology
Thermoluminescence, Dendrochronology, Varve count, Electron Spin Resonance, Anthropology
Prior to 2 ka radiocarbon dating dominates with between 20% and 60% of eruptions dated by 14C analysis from 2 to 50 ka (Figure 2b). Radiocarbon dating is only viable to about 50 ka (Fairbanks et al. 2005; red line, Figure 2b) and is only possible if organic material has been preserved. Uncorrected 14C dates from the literature were calibrated and entered in LaMEVE using IntCal 09 (northern hemisphere) and ShCal04 (southern hemisphere) in CALIB Radiocarbon Calibration Program 6.1.0 (Stuiver and Reimer 1993) for dates up to 26 ka. Fairbanks et al. (2005) was used for calibration of dates of 26 to 50 ka. It is not always clear in the literature whether reported dates are calibrated or not; these dates are entered without calibration as ‘Unknown’ dating technique. Calib 7.0, released in 2013, now calibrates radiocarbon ages to 50 ka and will be used for calibration of dates in Version 3 of the LaMEVE database.
Eruptions older than 50 ka are predominantly dated using 40Ar-39Ar and K-Ar radiometric techniques. Over 50% of eruptions between 50 and 500 ka are dated using radiometric techniques; this increases to >75% for eruptions older than 1.5 Ma.
Eruption size and volume
Magnitude is the eruption size measure used in all quantitative analysis in this paper. It is frequently sourced from published works where it is based on volume estimates for tephra fall deposits, pyroclastic flow deposits (ignimbrites) and intra-caldera deposits. Where unavailable in the literature magnitudes were calculated specifically for LaMEVE from the published volume data or VEI values. There are numerous volume estimation methods, but unfortunately this is commonly not provided or clearly described in the source literature. The most widely used method for calculation of volume of tephra fall deposits is based on the methodologies of Pyle (1989;2000). LaMEVE contains magnitudes derived in a variety of ways. Papers may report anything from one to up to six pertinent parameters, namely VEI, bulk tephra volume, DRE, magnitude, tephra density and magma composition (from which magma density can be calculated). 46% of the LaMEVE magnitude entries are derived from literature sources, which provide either one or more of VEI, bulk tephra volume and DRE. Another 46% are derived from directly reported magnitude values. Most literature does not directly report magnitudes; however the two Japanese databases include this metric. 8% of magnitudes in LaMEVE are derived from sources describing the occurrence of large magnitude explosive eruptions without further information: these are judged to have been M ≥ 4. For calculated magnitudes the tephra density is taken as 1000 kg/m3 if there is no information on deposit density in the original source and no DRE cited. Magma density is assumed to be 2500 kg/m3 (andesite) if the magma type is not known. The bulk tephra and DRE volumes are closely related. There is some overlap in the DRE volume range for each magnitude band due to variations in magma density values (from 2300 to 2700 kg/m3).
Recent studies to assess uncertainties in volume estimates from tephra fall deposits (Connor and Connor 2006; Burden et al. 2013; Engwell et al. 2013) indicate that even the best-documented tephra fall deposits have volume uncertainties greater than 10%. Uncertainties in volumes are not commonly reported, but where they are estimates are typically below half an order of magnitude. Problems also arise when attempting to distinguish between bulk and DRE volumes in literature sources, a matter also highlighted by Mason et al. (2004). Furthermore, substantially diverse volumes can be reported for a single eruption in different literature sources, with occasionally incompatible bulk and DRE volumes between sources and even within the same source. In 64 cases, mostly from Japan, reported magnitudes are either greater or less than the magnitude implied by the reported DRE volume. All data have been included in LaMEVE but labelled as either ‘preferred’ or ‘alternate’ values. Our analysis in this paper is based on the preferred magnitude values only. The identification of ‘preferred’ is a subjective judgement based on the assessed quality of the various sources of information. Given these quite large uncertainties analysis of LaMEVE at higher resolution than order of magnitude bins is not justified. Thus here we analyse LaMEVE data collectively in bins of M4-4.9, M5-5.9, M6-6.9, M7-7.9 and M8-8.9.
The number of Quaternary eruptions in LaMEVE in each magnitude band
Number of eruptions
Temporal variations in recorded magnitude
The median age, T50, of eruptions in each magnitude band
Median time (T50)
50% of time covered by LaMEVE database
M ≥ 4 (entire database)
M ≥ 5
M ≥ 6
M ≥ 7
M ≥ 8
The improvement in the completeness of the geological record of events from about 35 ka to the present day is evident in eruptions of all magnitudes except M ≥ 7, which is more consistent through time. Although M4-4.9 events are most common within the overall dataset (Table 4) these are poorly represented in the geological record (>1 ka) and only exceed the recording of M5-5.9 eruptions after approximately 10.5 ka (Figure 4b). From 1 ka to the present eruptions follow the expected order of decreasing eruption frequency with increasing magnitude. The data indicate that pre-Holocene geological preservation is very poor for M < 5 eruptions.
When separated from the rest of world (ROW) data the Japan, Taiwan and Marianas (JTM) dataset shows a more complete record of small magnitude events back into the Pleistocene (see Additional file 1: Figure S1). Here, eruptions of M4-4.9 become the most frequent eruption type at about 18 ka in contrast to around 6 ka in the ROW dataset. The JTM data also show a more steady level of recording of events of M6 and above, whilst only M7 and above eruptions show this feature in the ROW data. This difference demonstrates a more comprehensive eruption record in the JTM region, including the previously published and independent volume calculations of Hayakawa (2010), and that records of small magnitude events are more incomplete elsewhere. The difference may reflect a contrasting geological history as well as admirably thorough studies of Quaternary Japanese tephras.
The increasing under-recording of eruptions with decreasing magnitude is also shown through the calculation of the b-value from the data in Table 4. The Gutenberg-Richter law in seismology states that the frequency of earthquakes of different magnitudes normally generates a b-value of 1.0 with the logarithmic earthquake magnitude scale (Gutenberg and Richter 1956). The b-value generated from the LaMEVE dataset, 0.46 (Additional file 1: Figure S3), indicates a much higher frequency of large eruptions in comparison with smaller events. The JTM and ROW datasets analysed separately produce a similar b-value.
We dismiss the possibility that global trends in the event rate are caused by a dramatic global increase in volcanic activity. Rates of volcanism are principally controlled by rates of plate motion and mantle convection related to hot spots, which change slowly over many millions of years. Thus, global changes related to plate and hot spot processes over the time scales being considered are not credible. However, these arguments are less compelling on a regional scale where, for example, local changes in plate motions and changes from upper plate extension to compression in arcs or vice versa might lead to changes in rates of explosive volcanism. There is evidence for an increase in global volcanism related to deglaciation (e.g. Huybers and Langmuir 2009), however this would not cause a decrease back in time but rather fluctuations related to glacial cycles. In a later section we look at the evidence for event rate variations related to glaciations. Any real fluctuations in rates of volcanism with time (non-stationarity) should affect all magnitudes approximately equally, were it a significant factor. However, the decrease back in time is a strong function of magnitude (Figure 5), a relationship easily explained by under-recording, but essentially inexplicable by a physical process. We do not exclude the possibility of non-stationarity in Quaternary global explosive volcanism but claim that the evidence shows that it is masked by under-recording.
Completeness of the eruption record as a function of time
The ‘expected’ (red) lines in Figure 5 are derived from the total cumulative number of eruptions from 50 to 1 ka. From 1 ka to the present day there is an increasing number of recorded M < 7 eruptions (not shown in Figure 5). The expected frequency of eruptions can be calculated from the historical period (1 ka to 2013) and from this a comparison can be made with the recorded number to examine recording levels as a function of time. Here we adopt a simplified analysis of the available data to provide a first order estimation of the percentage of eruptions recorded at particular times.
The mean number of eruptions per century during different historical time periods
1900 to 2013 AD: Mean eruptions per century
1500 to 1900 AD: Mean eruptions per century
450 to 1500 AD: Mean eruptions per century
450 to 1500 AD rate as percentage of 1900 to 2013 AD rate
Eruption preservation potential
The median age, T 50 , for data in each magnitude bin for time periods ending at 1 ka and 2013
T50for whole dataset
T50for geological data (≥ 1 ka)
T50 increases with increasing magnitude, however even for M8-8.9 eruptions at 935 kyr, T50 is 359 kyr less than the central point, suggesting under-recording for even the largest events. All T50 values for lower magnitudes are considerably smaller than the 1.3 Ma central point of time for the database, which is simply explained by under-recording. The relationship between magnitude and T50 reflects the probability of preservation of the eruption deposits, which is primarily a function of volume and time. Large volume eruptions (with consequently larger magnitudes) have thicker deposits and cover larger areas than smaller volume eruptions, resulting in surface area to volume ratios commensurate with an increased length of time required for erosion to remove in situ physical evidence of the eruption. Factors such as the depositional environment (e.g. marine deposition) and climate therefore strongly influence preservation. Anthropogenic activities are less likely to remove the largest and thickest deposits, whilst smaller deposits in populated zones may well be quarried or otherwise removed without detailed study.
The same relationship is found for the data when divided into the JTM and ROW datasets, with increased divergence at lower magnitudes due to the superior record in the JTM region (Additional file 1: Figure S4).
There is regional variability (Figure 8; Additional file 1: Table S1), with the Kuril Islands, Indonesia, and the Philippines and South East Asia all indicating the greatest levels of under-recording with T50 ages of <250 years BP for M4-4.9 eruptions. The Middle East and Indian Ocean and Antarctica have the highest T50 ages (around 150 ka) for M4-4.9 eruptions, although we note only six eruptions of this size are recorded between them. The Mediterranean and West Asia (40 M4-4.9 eruptions), Africa and the Red Sea (17 M4-4.9 eruptions) and Japan, Taiwan and the Marianas (312 M4-4.9 eruptions) could therefore be interpreted to have the least under-recording in M4-4.9 eruptions, with T50 ages of ~15 ka to 38 ka. A similar degree of recording is suggested in Canada and the Western USA, Hawaii and Pacific Ocean, Mexico and Central America, South America and the West Indies by their comparable T50 ages of around 2 ka. These regions lie within a narrow longitudinal range and this, along with those regions showing greatest under-recording, indicates an anthropogenic recording bias based on longitudinal location.
Indonesia and the Philippines and SE Asia show relative under-recording of M5-6.9 events, and Iceland and Arctic Ocean display consistent under-recording of M4-6.9 eruptions. JTM shows thorough recording with the most populous dataset of eruptions and consistently high T50 ages. The variability between recorded events from present to 20 ka in a selection of regions can also be seen in Figure S5 (Additional file 1), which highlights the reduction in recording of eruptions in Indonesia, the southwest Pacific and Iceland before 1 ka.
Cumulative erupted volume
Glacial control on eruption incidence
During the Last Glacial Maximum (LGM) ice sheets and high altitude glaciers reached their maximum extent in the northern hemisphere. LGM glaciation peaked between 33 ka and 26.5 ka and onset of ice retreat initiated at about 20–19 ka (Clark et al. 2009). Major ice retreat occurred at the beginning of the Holocene and retreat was largely complete by 7 ka (Peltier 1994).
Huybers and Langmuir (2009) investigated a link between volcanic activity and deglaciation through analysis of a selection of regions from the GVP and Bryson et al. (2006) databases, both of which are included and expanded upon in LaMEVE. Kutterolf et al. (2013) identified cyclicity in activity in marine tephras across 1.2 Ma, with increased volcanism following periods of deglaciation. They also detected an increase in eruptions in non-glaciated regions and suggested that stress field changes affected regions beyond the limits of glaciations, perhaps related to sea level rise. Kutterolf et al. (2013) did not investigate a global dataset but instead the activity in the Central American Volcanic Arc and locations around the Pacific rim, resulting in a relatively small dataset for each glacial cycle. Watt et al. (2013) proposed that the volcanic record for regions identifies localised cycles and that correlations with deglaciation may be coincidental rather than demonstrating causation. Watt et al. (2013) examined the published eruption record on a regional scale and found varying responses to deglaciation, with an estimated increase in eruption rate in post-glacial activity by a factor of about 2, much lower than the factor of 4 to 5 estimated by Huybers and Langmuir (2009).
Here we compare LaMEVE data in glaciated and non-glaciated regions around the world, with the ice extent derived from Huybers and Langmuir (2009) and Watt et al. (2013) (Additional file 1: Table S2). Areas which underwent significant deglaciation were the Southern Andes, Alaska, the Cascades, Iceland, Western Europe and Eastern Russia; the tropical Americas, Africa and Southeast Asia were not glaciated (Huybers and Langmuir 2009 and references therein). Figure 12b and c highlight the 15 to 5 ka period, distinguishing between high latitude (glaciated) and low latitude (non-glaciated) regions. The deglaciated regions show a peak in volcanic activity between about 11 and 9 ka (Figure 12b) with a much lower number of eruptions prior to this, reflecting temporal improvements in eruption records. The peak in glaciated regions falls within the 12 to 7 ka peak identified by Huybers and Langmuir (2009). Watt et al. (2013) also found a peak in volume output between about 11 and 8 ka. The increased peak in glaciated regions in the LaMEVE data (Figure 12b) is thus consistent with the hypothesis of increased volcanism through ice-unloading. Although no peak is evident in non-glaciated regions there is a noticeable increase at the beginning of the Holocene (Figure 12c). A difference between our study and others arises due to the inclusion of M < 4 events in the analysis of Huybers and Langmuir (2009), Kutterolf et al. (2013) and Watt et al. (2013). However, large explosive events dominate the record further back in time and so this difference may have a limited effect. A peak occurs earlier in New Zealand data (12 to 11 ka) and may reflect the earlier onset of deglaciation in the southern hemisphere (Sikes et al. 2013).
Huybers and Langmuir (2009) developed an eruption factor, Ef, comparing eruption frequency in the glaciated and non-glaciated regions. Applied to a particular selection of glaciated and non-glaciated regions they found the peak in volcanism in glaciated regions. However Watt et al. (2013) demonstrate that a small change to assumptions on glacial extent and the chosen regions significantly changes the size and occurrence of this peak and suggested that analysis of the Ef is only valid if the eruption rates in glaciated and non-glaciated regions are equal, which, due to considerable spatial variation in the eruption rates, we know is not the case. Analysis of LaMEVE supports this, illustrating the regional variability in the quality of the eruption record, with considerably different numbers of events between time periods, regions and between glaciated and non-glaciated areas in the same time period (Additional file 1: Figure S5). Comparison of a selection of regions indicates a higher proportion of eruptions were recorded in glacial regions. Correction of the data as suggested by Watt et al. (2013) and calculation of the Ef (not shown) nonetheless still indicates a post-glacial peak in the LaMEVE data. Thus the finding of a peak during deglaciation seems robust.
Analysis of the LaMEVE data indicates increased numbers of eruptions in regions undergoing deglaciation at the end of the LGM. A difficulty arises from the analysis of dating techniques (Table 3, Figure 2), which indicates the predominance of radiocarbon dating from 50 to 2 ka, with approximately 50% of eruptions dated by radiocarbon analysis during the Holocene. Extensive ice sheets reduce both the likelihood of tephra preservation and the material required for carbon dating through the absence of soils, lakes and vegetation. This decreases the probability of the recording of eruptions in glacial stages. Thus, the apparent decline in eruption numbers prior to about 11 ka may be caused by a sampling bias. However, such a bias would not explain the large peak in Figure 12b with a decline in recording after about 9 ka. It is the comparison of this early post-glacial activity with that later in the Holocene that provides evidence for real variation in eruption frequency over time, with deglacial pulses of volcanism observed lasting a few thousand years (Watt et al. 2013). Our peak is dominantly controlled by a few regions: Kamchatka and Mainland Asia, South America, Alaska and the Aleutian Islands and New Zealand. A pulse of increased effusive volcanism in Iceland is discussed by Watt et al. (2013) and references therein, which is much less well constrained in the explosive record, with data from LaMEVE indicating 5 eruptions between about 12 and 9 ka with increased frequency later in the Holocene.
Column height and intensity
Maximum column height is provided for 11% of eruptions in the LaMEVE database, of which 65% are Holocene in age. The column height is derived from literature sources, where the column height is either determined from direct observations of historical eruptions, or estimated through application of the Carey and Sparks (1986) maximum clast size dispersal model. Uncertainties in plume heights from the maximum clast method have been quantified by Burden et al. (2011) and the uncertainty in inferred intensities can also be assessed empirically from the correlation of column height data and independently determined intensity estimates from historic eruptions (Sparks et al. 1997). Uncertainties in column heights derived by the maximum clast method are typically less than 5 km (equivalent approximately to a factor of 2.5 in intensity). Given a known column height, uncertainties in inferred intensities are typically less than a factor of 3. Thus the proxy intensity estimates have uncertainties of about half an order of magnitude or less.
Baines and Sparks (2005) demonstrated that M ≥ 6.5 eruptions are capable of producing horizontal spreads which greatly exceed the vertical height. New calculations of column height may therefore be required for M ≥ 6.5 eruptions, which account for about 10% of the eruptions with associated column heights in the LaMEVE database. The LaMEVE data show column height increases with magnitude, but there is considerable scatter, with, for example, column heights of 30 km recorded for eruptions ranging from M4-8 (Additional file 1: Figure S6). There are clusters at a column height of 30 km and M4.0 and 5.0, indicating that there is some estimation of size values with rounding of results.
Here we have analysed the LaMEVE data to produce a synopsis of the record of global explosive volcanism (M ≥ 4). We have assessed understanding of the global eruption record, identified knowledge gaps and areas for improvement.
Our analysis of the LaMEVE database has highlighted that the record of global volcanism has both spatial and temporal biases and is incomplete. Under-recording is a well-established attribute of global volcanic datasets (Simkin 2003) and also strongly affects the historical records, which dominate the last few centuries (Simkin and Siebert 2000). Recent analyses of global Holocene datasets (Coles and Sparks 2006; Deligne et al. 2010; Furlan 2010) demonstrate a rapid decrease in historical recording back to 1500 AD, with possible change points at 1500 and 1900 (Furlan 2010). The analysis of Holocene data (M ≥ 4) using an under-recording model by Deligne et al. (2010) suggests that recording was steady from 10 to 2 ka. For example, Deligne et al. (2010) estimated that recording of M ≥ 6 eruptions reflects about 15-20% of the true record in this period. Deligne et al. (2010) concluded that the Holocene was too short a period to sample M ≥ 7 eruptions. The analysis of Mason et al. (2004) considers a longer time-scale, but this study is limited to M ≥ 8 eruptions. LaMEVE contains 26 Quaternary events with M ≥ 8; in contrast, Mason et al. (2004) considered only eight M ≥ 8 Quaternary events out of 36 spanning a 38 Myr period. LaMEVE considerably improves the compiled record of M ≥ 7 eruptions, enabling an assessment of geological recording over a much longer period of time than the Holocene. Although M ≥ 7 events are also affected by under-recording, this is much less severe than for M4-M6 events as indicated by the T50 being much closer to the median time of the Quaternary (Figure 7). The T50 value for M8-8.9 can be used to make a conservative estimate of under-recording: 9 missing events prior to 1.255 Ma would move T50 to the median time for the Quaternary; this would suggest 25% under-recording of M ≥ 8 eruptions in the first half of the Quaternary. However, given that the most recent half of the Quaternary is likely also to be missing some events, assuming uniformity, there is likely >25% under-reporting of M ≥ 8 eruptions.
where To is the maximum deposit thickness and bt is effectively a measure of deposit area or hazard footprint (see Pyle (1989) for further details). Pyroclastic flow volumes are typically calculated by determining deposit area and average thickness. Thus, a simple explanation for the relationship found in Figure 7 is that preservation potential is linearly proportional to deposit thickness and to the square root of the deposit area, resulting in the observed 3/2 exponent for volume and 1/2 for mass.
Cumulative curves of the number of eruptions normalised to the number of recorded events prior to 1 ka (Figure 5) enables quantification of the decline in under-recording by comparison with recording rate. These results show the strong dependence of geological under-recording on magnitude. This supports the underlying approximation of stationarity since all magnitudes would be affected to the same extent by any real fluctuations in rates of volcanism. However, there is a marked increase in events in glaciated regions in late glacial times (notably 9–11 ka). While this peak may be affected by sampling biases, it supports previous suggestions (Huybers and Langmuir 2009; Watt et al. 2013) that there was a global increase of volcanism related to deglaciation.
There are clear regional biases in LaMEVE, which highlights the very uneven distribution of knowledge about volcanism around the world. Japanese data are strongly represented in LaMEVE and so could bias global analysis of the data. Japan’s record for M4-M6 eruptions extends back much further in time than for other regions, reflecting commendable scope of geological investigations, and a favourable environment for preserving tephra deposits.
We can use LaMEVE to identify major knowledge gaps and therefore areas which could benefit from focussed research efforts. LaMEVE also provides a resource for producing regional magnitude-frequency relationships and for developing these relationships for individual morphologic types of volcano. However, breaking up the LaMEVE data into subsets based on region, country or volcano type will exacerbate the problem of uneven quality and coverage. This would therefore require development of statistical models that account for under-recording and develop the concept of exchangeability. Exchangeability is based on the premise that if one takes data from a population of objects that are thought to be similar (in this case volcanic sub-regions or individual volcanoes), one can characterise the statistical properties of a sub-set of objects that are relatively well characterised and assume that they can represent those that are poorly characterised. We suggest that such an approach could be applied using Japan as the well-characterised sub-set to develop a statistical model of stratovolcanoes, island arcs or other types exemplified in the Japanese region.
A final issue highlighted by the population and analysis of the LaMEVE database is the uneven quality of the data sources. Many studies do not state the method used to estimate volumes of tephra deposits in sufficient detail and there is no standardisation of methodologies. Uncertainties are commonly not assessed or stated. Looking towards future research, there needs to be significant improvements in literature descriptions of age and volume data to ensure that unambiguous data are available for consistently named eruptions with appropriate derivation methods cited. The introduction of guidelines and adoption of standardised internationally agreed methods is needed. For example, when providing unit dates through radiocarbon analysis it should be made clear whether the ages are calibrated and which algorithm was used in the calibration. The methods and assumptions behind volume estimates and conversions to DRE values should also be reported, and estimates of uncertainties in ages and volumes should be given. Standardisation of data reporting and methodologies would improve clarity and transferability throughout volcanology research.
Our analysis of the LaMEVE database, the most comprehensive inventory of known large (M ≥ 4) explosive eruptions for the Quaternary, identifies major biases in time and space in the scientific community’s record of large explosive eruptions. In particular, under-recording is a dominant feature of the data which worsens with increasing time from present. However, recording improves markedly with magnitude and the database likely includes 70% or more of M ≥ 7 eruptions. Regional biases include better recording in the northern hemisphere and, notably, eruptions from the Japan, Taiwan and Marianas region constitute over 40% of all LaMEVE data.
Since historical influences on eruption recording have been widely studied our analysis of LaMEVE has focussed on geological recording, defined as the time prior to 1 ka. Consideration of volume data through time indicates that there are very strong temporal magnitude-related biases in the data. The record is shown to be incomplete and we suggest that this is partially controlled by and quantifiable through the eruption age and size, with the probability of recording an eruption in the geological record, as defined by a median preservation time scale, being proportional to the square root of the magnitude. This suggests a simple explanation of preservation being proportional to deposit thickness and square root of the deposit area.
The major influence of under-recording prevents much interpretation of temporal variations in volcanism. However, we find that there is an increase in explosive volcanism at the end of the last ice age in glaciated areas; this trend is not observed in areas unaffected by glaciation. These observations support the hypothesis of glacial unloading triggering enhanced volcanism (Huybers and Langmuir, 2009; Watt et al. 2013). The strength of this signal may be exaggerated by biases related to the (lack of) preservation of syn-glacial pyroclastic deposits and carbon within them. The eruptions of the largest magnitudes, M8-8.9, show only modest under-recording, with approximately 25% missing events. The M ≥ 8 data do not indicate any major temporal variability during the Quaternary.
LaMEVE can identify knowledge gaps and constrain return periods of explosive eruptions of different magnitudes on global, regional and local (individual volcano) scales. However, the fact that major spatial and temporal biases together with under-recording are strongly dependent on magnitude means that the database should not be used in its raw form for such estimates. Statistical analyses to correct for biases and under-recording are essential. We recommend that the principle of exchangeability or equivalent correction techniques be applied to enable proper usage of LaMEVE data for assessment of volume production rates, magnitude-frequency estimates and hazard applications.
aPublically accessible at http://www.bgs.ac.uk/vogripa;
bPublically accessible at http://www.volcano.si.edu;
cThroughout this article, a distinction in the notation between geohistorical dates in years before present (e.g. ka, Ma) and geohistorical durations in years (e.g. kyr, Myr) has been made, as per the article by Aubry et al. (2009).
Dense Rock Equivalent
Global Volcano Model
Global Volcanism Program
Japan, Taiwan and the Marianas
Thousands of years ago
Thousands of years
Large Magnitude Explosive Volcanic Eruptions
Last Glacial Maximum
Millions of years ago
Millions of years
Quaternary Explosive Activity Recorded
Rest of World
Volcanic Explosivity Index
Volcano Global Risk Identification and Analysis Project
Volcanoes of the World.
The authors would like to acknowledge the funding bodies for this project: the European Research Council (VOLDIES grant), the Natural Environment Research Council (Global Volcano Model grant), the British Geological Survey and also Munich Re in the initial stages. Thanks to Susanna Jenkins and Henry Odbert for helpful discussions at the drafting stage.
- Active Volcanoes of Japan Database: Geological Survey of Japan. Geological Survey of Japan 2008. . Accessed regularly https://gbank.gsj.jp/volcano-AV/cgi-bin/volcano-list.cgi . Accessed regularlyGoogle Scholar
- Aubry M-P, Van Couvering JA, Christie-Blick N, Landing E, Pratt BR, Owen DE, Ferrusquía-Villafranca I: Terminology of geological time: Establishment of a community standard. Stratigraphy 2009, 6(2):100–105.Google Scholar
- Auker MR, Sparks RSJ, Siebert L, Crosweller HS, Ewert J: A statistical analysis of the global historical volcanic fatalities record. J Appl Volcanol 2013, 2(2):1–24. 10.1186/2191–5040–2-2 10.1186/2191-5040-2-2Google Scholar
- Baines PG, Sparks RSJ: Dynamics of giant volcanic ash clouds from supervolcanic eruptions. Geophys Res Lett 2005, 32(24):L24808.View ArticleGoogle Scholar
- Bryan SE, Ukstins Peate I, Peate DW, Self S, Jerram DA, Mawby MR, Marsh JS, Miller JA: The largest volcanic eruptions on Earth. Earth Sci Rev 2010, 102: 207–229. 10.1016/j.earscirev.2010.07.001View ArticleGoogle Scholar
- Bryson RU, Bryson RA, Ruter A: A calibrated radiocarbon database of late Quaternary volcanic eruptions. Earth Discussions 2006, 1(2):123–134. 10.5194/eed-1-123-2006View ArticleGoogle Scholar
- Burden RE, Philips JC, Hincks TK: Estimating volcanic plume heights from depositional clast size. J Geophys Res 2011, 116: B11206.View ArticleGoogle Scholar
- Burden RE, Chen L, Phillips JC: A statistical method for determining the volume of volcanic fall deposits. Bull Volcanol 2013, 75: 707.View ArticleGoogle Scholar
- Carey S, Sigurdsson H: The intensity of plinian eruptions. Bull Volcanol 1989, 51(1):28–40. 10.1007/BF01086759View ArticleGoogle Scholar
- Carey S, Sparks RSJ: Quantitative models of the fallout and dispersal of tephra from volcanic eruption columns. Bull Volcanol 1986, 48(2):109–125.View ArticleGoogle Scholar
- Chesner CA, Rose WI, Deino A, Drake R, Westgate JA: Eruptive history of Earth’s largest Quaternary caldera (Toba, Indonesia) clarified. Geology 1991, 19: 200–203. 10.1130/0091-7613(1991)019<0200:EHOESL>2.3.CO;2View ArticleGoogle Scholar
- Clark PU, Dyke AS, Shakun JD, Carlson AE, Clark J, Wohlfarth B, Mitrovica JX, Hostetler SW, McCabe AM: The Last Glacial Maximum. Science 2009, 325(5941):710–714. 10.1126/science.1172873View ArticleGoogle Scholar
- Coles S, Sparks RSJ: Extreme value methods for modelling historical series of large volcanic magnitudes. In Statistics in Volcanology, vol 1. Edited by: Mader HM, Coles SG, Connor CB, Connor LJ. Geological Society of London: Special Publication of IAVCEI; 2006:47–56.Google Scholar
- Connor L, Connor C: Inversion is the key to dispersion: understanding eruption dynamics by inverting tephra fallout. In Statistics in volcanology of special publications of IAVCEI, 1. Edited by: Mader HM, Connor CB, Coles SG, Connor LJ. London: Geological Society; 2006:231–242.Google Scholar
- Crosweller HS, Arora B, Brown SK, Cottrell E, Deligne NI, Guerrero NO, Hobbs L, Kiyosugi K, Loughlin SC, Lowndes J, Nayembil M, Siebert L, Sparks RSJ, Takarada S, Venzke E: Global database on large magnitude explosive volcanic eruptions (LaMEVE). J Appl Volcanol 2012, 1(4):1–13. 10.1186/2191–5040–1-4Google Scholar
- Decker RW: How often does a Minoan eruption occur?. London: The Thera Foundation; 1990.Google Scholar
- Deligne NI, Coles SG, Sparks RSJ: Recurrence rates of large explosive volcanic eruptions. J Geophys Res 2010, 115: B06203.View ArticleGoogle Scholar
- Ellis BS, Mark DF, Pritchard CJ, Wolff JA: Temporal dissection of the Huckleberry Ridge Tuff using the 40Ar/39Ar dating technique. Quat Geochronol 2012, 9: 34–41.View ArticleGoogle Scholar
- Engwell SL, Sparks RSJ, Aspinall WP: Quantifying uncertainties in the measurement of tephra fall thickness. Journal of Applied Volcanology 2013. in press in pressGoogle Scholar
- Fairbanks RG, Mortlock RA, Chiu T-C, Cao L, Kaplan A, Guilderson TP, Fairbanks TW, Bloom AL, Grootes PM, Nadeau M-J: Radiocarbon calibration curve spanning 0 to 50,000 years BP based on paired 230Th/234U/238U and 14C dates on pristine corals. Quat Sci Rev 2005, 24(16–17):1781–1796.View ArticleGoogle Scholar
- Furlan C: Extreme value methods for modelling historical series of large volcanic magnitudes. Stat Model 2010, 10(2):113–132. 10.1177/1471082X0801000201View ArticleGoogle Scholar
- Geyer A, Martí J: The new worldwide collapse caldera database (CCDB): A tool for studying and understanding caldera processes. J Volcanol Geotherm Res 2008, 175(3):334–354. 10.1016/j.jvolgeores.2008.03.017View ArticleGoogle Scholar
- Gibbard PL, Head MJ, Walker MJC, Subcommission T: The Subcommission on Quaternary Stratigraphy (2010) Formal ratification of the Quaternary System/Period and the Pleistocene Series/Epoch with a base at 2.58 Ma. J Quat Sci 2010, 25: 96–102. 10.1002/jqs.1338View ArticleGoogle Scholar
- Gutenberg B, Richter CF: Earthquake magnitude, intensity, energy, and acceleration: (Second paper). Bull Seismol Soc Am 1956, 46(2):105–145.Google Scholar
- Hayakawa Y: Tephra of the World database. 1996.http://www.hayakawayukio.jp/catalog/eworld . Accessed regularlyGoogle Scholar
- Hayakawa Y: Hayakawa’s 2000-year and one million-year database. 2010.http://www.hayakawayukio.jp/database/index.php?kind=1m . Accessed regularlyGoogle Scholar
- Huybers P, Langmuir C: Feedback between deglaciation, volcanism, and atmospheric CO2. Earth Planet Sci Lett 2009, 286(3–4):479–491.View ArticleGoogle Scholar
- Kutterolf S, Jegen M, Mitrovica JX, Kwasnitschka T, Freundt A, Huybers PJ: A detection of Milankovitch frequencies in global volcanic activity. Geology 2013, 41: 227–230. 10.1130/G33419.1View ArticleGoogle Scholar
- Machida H: Quaternary Volcanoes and Widespread Tephras of the World. Glob Environ Res 2002, 6: 3–17.Google Scholar
- Mason BG, Pyle DM, Oppenheimer C: The size and frequency of the largest explosive eruptions on Earth. Bull Volcanol 2004, 66(8):735–748. 10.1007/s00445-004-0355-9View ArticleGoogle Scholar
- Mastin LG, Guffanti M, Servranckx R, Webley P, Barsotti S, Dean K, Durant A, Ewert JW, Neri A, Rose WI, Schneider D, Siebert L, Stunder B, Swanson G, Tupper A, Volentik A, Waythomas CF: A multidisciplinary effort to assign realistic source parameters to models of volcanic ash-cloud transport and dispersion during eruptions. J Volcanol Geotherm Res 2009, 186(1–2):10–21.View ArticleGoogle Scholar
- Németh K, Cronin SJ, White JDL: Kuwae Caldera and Climate Confusion. Open Geology J 2007, 1: 7–11. 10.2174/1874262900701010007View ArticleGoogle Scholar
- Newhall CG, Self S: The Volcanic Explosivity Index (VEI) - An estimate of explosive magnitude for historical volcanism. J Geophys Res Oceans Atmos 1982, 87(C2):1231–1238. 10.1029/JC087iC02p01231View ArticleGoogle Scholar
- Ninkovich D, Sparks RSJ, Ledbetter MT: The exceptional magnitude and intensity of the Toba eruption, Sumatra: An example of the use of deep-sea tephra layers as a geological tool. Bull Volcanol 1978, 41(3):286–298. 10.1007/BF02597228View ArticleGoogle Scholar
- Oppenheimer C: Limited global change due to the largest known Quaternary eruption, Toba 74 kyr BP. Quat Sci Rev 2002, 21: 1593–1609. 10.1016/S0277-3791(01)00154-8View ArticleGoogle Scholar
- Peltier WR: Ice Age Paleotopography. Science 1994, 265(5169):195–201. 10.1126/science.265.5169.195View ArticleGoogle Scholar
- Petraglia MD, Korisettar R, Pal JN: The Toba volcanic super-eruption of 74,000 years ago: Climate change, environments, and evolving humans. Quat Int 2012, 258: 1–4.View ArticleGoogle Scholar
- Pyle DM: The thickness, volume and grainsize of tephra fall deposits. Bull Volcanol 1989, 51: 1–15.View ArticleGoogle Scholar
- Pyle DM: Sizes of volcanic eruptions. In Encyclopedia of Volcanoes. Edited by: Sigurdsson H, Houghton BF, McNutt SR, Rymer H, Stix J. London: Academic Press; 2000.Google Scholar
- Pyle DM, Beattie PD, Bluth GJS: Sulphur emissions to the stratosphere from explosive volcanic eruptions. Bull Volcanol 1996, 57: 663–671. 10.1007/s004450050119View ArticleGoogle Scholar
- Robock A: Volcanic eruptions and climate. Rev Geophys 2000, 38: 191–219. 10.1029/1998RG000054View ArticleGoogle Scholar
- Rose WI, Conway FM, Pullinger CR, Deino A, McIntosh WC: An improved age framework for late Quaternary silicic eruptions in northern Central America. Bull Volcanol 1999, 61: 106–120. 10.1007/s004450050266View ArticleGoogle Scholar
- Siebert L, Simkin T, Kimberly P: Volcanoes of the World. 3rd edition. Smithsonian Institution: University of California Press; 2010.Google Scholar
- Sikes EL, Medeiros PM, Augustinus P, Wilmshurst JM, Freeman KR: Seasonal variations in aridity and temperature characterize changing climate during the last deglaciation in New Zealand. Quat Sci Rev 2013. 10.1016/j.quascirev.2013.01.031 10.1016/j.quascirev.2013.01.031Google Scholar
- Simkin T: Terrestrial volcanism in space and time. Annu Rev Earth Planet Sci 1993, 21: 427–452. 10.1146/annurev.ea.21.050193.002235View ArticleGoogle Scholar
- Simkin T, Siebert L: Earth's volcanoes and eruptions: an overview. In Encyclopedia of Volcanoes. Edited by: Sigurdsson H, Houghton BF, McNutt SR, Rymer H, Stix J. London: Academic Press; 2000.Google Scholar
- Small C, Naumann T: The global distribution of human population and recent volcanism. Environmental Hazards 2001, 3: 93–109.View ArticleGoogle Scholar
- Sparks RSJ, Bursik MI, Carey SN, Gilbert JS, Glaze L, Sigurdsson H, Woods AW: Volcanic Plumes. Chichester: Wiley; 1997.Google Scholar
- Storey M, Roberts RG, Saidin M: Astronomically calibrated 40Ar/39Ar age for the Toba supereruption and global synchronization of late Quaternary record. Proc Natl Acad Sci U S A 2012, 109: 18684–18688. 10.1073/pnas.1208178109View ArticleGoogle Scholar
- Stuiver M, Reimer PJ: Extended 14C data base and revised CALIB 3.0 14C age calibration program. Radiocarbon 1993, 35(1):215–230.Google Scholar
- Volcanoes of the World 4.0: Smithsonian Institution, Washington, DC. 2013.http://www.volcano.si.edu . Accessed regularlyGoogle Scholar
- Watt SFL, Pyle DM, Mather TA: The volcanic response to deglaciation: Evidence from glaciated arcs and a reassessment of global eruption records. Earth Sci Rev 2013, 122: 77–102.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.