Bull Volcanol DOI 10.1007/s00445-010-0393-4

RESEARCH ARTICLE

Volcanic : global observations and constraints on source mechanisms

Stephen R. McNutt & Earle R. Williams

Received: 2 June 2008 /Accepted: 4 June 2010 # Springer-Verlag 2010

Abstract Lightning and electrification at volcanoes are is more common at night (56%) and less common in important because they represent a hazard in their own daylight (44%). Reporting also varied substantially from right, they are a component of the global electrical circuit, year to year, suggesting that a more systematic observa- and because they contribute to ash particle aggregation and tional strategy is needed. Several weak trends in lightning modification within ash plumes. The role of water occurrence based on magma composition were found. The substance (water in all forms) in particular has not been bimodal ash plume heights are obvious only for andesite to well studied. Here data are presented from a comprehensive dacite; and basaltic-andesite evenly span the range of global database of volcanic lightning. Lightning has been heights; and rhyolites are poorly represented. The distribu- documented at 80 volcanoes in association with 212 tions of the latitudes of volcanoes with lightning and eruptions. The Volcanic Explosivity Index (VEI) could be eruptions with lightning roughly mimic the distribution of determined for 177 eruptions. Eight percent of VEI=3–5 all volcanoes, which is generally flat with latitude. eruptions have reported lightning, and 10% of VEI=6, but Meteorological lightning, on the other hand, is common in less than 2% of those with VEI=1–2. These findings the and decreases markedly with increasing latitude suggest consistent reporting for larger eruptions but either as the ability of the to hold water decreases less lightning or possible under-reporting for small erup- poleward. This finding supports the idea that if lightning in tions. Ash plume heights (142 observations) show a large (deep) eruptions depends on water substance, then the bimodal distribution with main peaks at 7–12 km and 1– origin of the water is primarily magma and not entrainment 4 km. The former are similar to heights of typical thunder- from the surrounding atmosphere. Seasonal effects show and suggest involvement of water substance, that more eruptions with lightning were reported in whereas the latter suggest other factors contributing to (bounded by the respective autumnal and vernal equinoxes) electrical behavior closer to the vent. Reporting of lightning than in . This result also runs counter to the expectations based on entrainment of local .

Editorial responsibility: JC Phillips Keywords Volcanoes . Lightning . Electrification . Electronic supplementary material The online version of this article Ash plumes . Water (doi:10.1007/s00445-010-0393-4) contains supplementary material, which is available to authorized users. S. R. McNutt (*) Introduction Geophysical Institute, University of Alaska Fairbanks, 903 Koyukuk Drive, P.O. Box 757320, Fairbanks, AK 99775, USA Electrical discharge in volcanic eruptions (i.e., volcanic e-mail: [email protected] lightning) is fairly common yet relatively understudied. Lightning is reported at only a few of the 55–70 volcanoes E. R. Williams that erupt each year (Simkin 1993), so it appears that either Massachusetts Institute of Technology 48-211, Parsons Laboratory, there are systematic problems with reporting, or some Cambridge, MA 02139, USA special set of circumstances must prevail to favor the Bull Volcanol production of lightning, or both. Lightning and electrifica- videos, was also compiled. The bulk of this work was tion at volcanoes are important because they represent a completed before modern computer search engines, such as hazard in their own right (people were killed by volcanic Google™, were in common use. lightning at Paricutin and Rabaul; see McNutt and Davis Key references included the Bulletin of Volcanology, the 2000,p.45–47), they are a component of the global Journal of Volcanology and Geothermal Research, the electrical circuit, and because they contribute to ash particle Smithsonian Institution’s Scientific Event Alert Network aggregation and modification within ash columns (Lane and Global Volcanism Network Bulletins, and the books by and Gilbert 1992; Gilbert and Lane 1994a, b; James et al. McClelland et al. (1989), Taylor (1958), and Newhall and 2003; Textor et al. 2005a, b; Mather and Harrison 2006; Punongbayan (1996), amongst others. We note that books James et al. 2008). Conventional thinking has been that published after about 1984 had better indexes than older interactions between dry silicate ash particles, such as works, probably a result of the increased use of electronic word collisions and fractures, have been dominant processes processing and the ease of making searches by keywords. causing electrification, while the role of water is secondary. Information that was collected included the name of the This is in contrast to early studies of Surtsey, Iceland, in volcano, the date (and time, where noted), the Volcanic which seawater played a prominent role (Blanchard 1964; Explosivity Index (VEI; Newhall and Self 1982; Simkin Anderson et al. 1965; Blanchard and Björnsson 1967). and Siebert 1994) of the eruption, the plume height Further, the existence and role of particles in ash maximum, the magma composition, whether it was day or plumes, now recognized as a fundamental aspect of night, any significant ancillary observations, and references thundercloud electrification, has been noted in only a few (Appendix). Latitudes and longitudes for volcanoes are studies at volcanoes (e.g. Thorarinsson 1966; Rose et al. available from Volcanoes of the World (Simkin and Siebert 1995a, b, 2000; Hoblitt 2000; Thompson 2000). In this 1994). VEI values were generally taken from Simkin and report, the expansion of steam and its subsequent conden- Siebert (1994) after checking that the date of the lightning sation and freezing are emphasized as primary processes observations coincided with the date of an eruption that was whose magnitude and effects have been understudied with assigned a VEI. We also estimated many VEI values respect to volcanic lightning. ourselves based mainly on reports of tephra volumes and Despite the frequent occurrence of volcanic lightning, plume heights. and many spectacular photographs, only two systematic compilations of basic lightning facts have been published (McNutt and Davis 2000; Mather and Harrison 2006). Here Analyses and results we present results of a more comprehensive literature search on the occurrence of volcanic lightning and attempt Reporting problems to summarize the effects of observed parameters such as height of plume, volume of tephra and gases (mainly water We first discuss reporting problems because these must be vapor), and magma composition. We also discuss reporting understood before other features of the data can be parameters because these affect the historical record of interpreted. Considerations of the detectability of volcanic lightning occurrence. Few instrumental studies of volcanic lightning over the full range of sunlight conditions must be lightning have been undertaken (Hoblitt 1994; McNutt and determined. Of the 131 eruptions for which the time of day Davis 2000; Paskievitch et al. 1995; Thomas et al. 2007, was known, we found that 44% occurred during daylight 2010), but this study is complementary to all of them. hours and 56% at night (Appendix). This suggests that Accordingly, some recommendations are included regard- lightning is more easily seen against a dark background and ing desirable electrical studies that will be necessarily may be missed during bright daytime conditions. As a aimed at constraining mechanisms for volcanic lightning. specific example of this effect, the eruption of Mount Spurr, Alaska, on September 17, 1992, occurred at night and had dramatic displays of lightning readily noted by people on Data and methods the ground. However, the amount of lightning for this eruption (7 flashes recorded instrumentally) was nearly an The main information source for this study is a compre- order of magnitude less than for the other two 1992 hensive literature compilation that was carried out over a eruptions (61 and 57 flashes, also recorded instrumentally), period of about 10 years. A standard bibliography keyword which were of similar size but occurred during the day, search yielded only a handful of studies, so instead the when ground observations of lightning were lacking literature was searched systematically, colleagues were (McNutt and Davis 2000). All three Spurr eruptions used polled, photographs were collected, and information from the same instrumental data to determine the number of newspapers and other sources, such as poster displays and flashes; details are given in McNutt and Davis (2000). Bull Volcanol

An additional reporting problem with regard to daytime level for eruptions with VEI=3 and larger following the observations has to do with the fact that many volcanoes standard conventions (Simkin and Siebert 1994). For are located in tropical areas where lightning is common comparison, a histogram of all known VEI is shown in from . Thus volcanic lightning may not be Fig. 1 (top). There are many more small eruptions than noticed if an observer is not aware of an eruption, because large ones, so the numbers increase from right to left. The the lightning would be dismissed as being due to normal drop off for VEI=1 is likely due to under reporting (Simkin patterns. Conversely, lightning that and Siebert 1994). happens to occur near an erupting volcano might be The number of occurrences of volcanic lightning at associated with the eruption even if it is not caused by the various VEI values is also shown in Table 2 along with all eruption, depending on the interpretation of the observer. known VEI and percent of cases showing lightning versus For some regions, such as the Alaska/Aleutian arc, all VEI. The percent of eruptions with lightning is nearly lightning is not common (J. Painter, National Weather the same for VEI=3, 4, and 5. This suggests a standard Service, pers. comm., 1997; D. Dissing, pers. comm., reporting efficiency. The percentage of VEI=6 eruptions is 2003). During a rare thunderstorm near Augustine Volcano somewhat higher. Such large eruptions attract attention and in 1998, several local citizens called the Alaska Volcano are generally well reported. We also note the common Observatory to ask if Augustine was erupting, because they observation that large eruptions create local darkness remembered that the 1976 and 1986 eruptions had been because of the dense ash plumes, a condition under which accompanied by vigorous lightning (Kienle and Swanson lightning can be more easily seen. Further, the heights of 1985; J. Kienle pers. comm. 1992). The volcano was not the ash plumes are >15 km for this range of VEI, so these erupting at the time of the 1998 thunderstorm. In contrast, are taller than most thunderclouds (also referred to as deep Augustine’s 2006 eruptions were accompanied by abundant convection by atmospheric scientists) and we infer that lightning (Thomas et al. 2007, 2010). similar water based charge generation or separation We determined the number of eruptions with reported mechanisms may be acting (see Discussion section below). lightning versus year (Table 1). Considerable variation is The percentage of eruptions with recorded lightning with noted, with values ranging from 0 to 19. The peak year was VEI=2 and smaller drops off sharply. This behavior 1979. Given that between 55 and 70 volcanoes erupt per suggests either a systematic reporting problem, or that year (Simkin and Siebert 1994) this suggests that up to 27– lightning is simply less common in these smaller eruptions, 35% of erupting volcanoes can produce lightning. Another or both. way to view this is that volcanic lightning is generally under-reported. It is not clear why the reporting varies so Bimodality of eruption plume heights much from year to year. There is a rough tendency for the number of reports to be higher following famous eruptions. Eruptions with VEI=3 have ash plume heights of 3–15 km For example, 1951 is the only year from 1950 to 1978 to (Simkin and Siebert 1994); these straddle typical thunder- have more than 10 eruptions reported with lightning. Ten of heights, and we find many cases with lightning and 11 reports were from Lamington volcano, which had a large also many without. For VEI=1 and 2 eruptions there are explosive eruption that year. In other years with high relatively fewer cases, but these are important because the numbers of cases generally one volcano contributed most of ash plume heights are <5 km, less than summertime the cases. The large number of reports for 1979 (19) thunderclouds (tops at 7 to 20 km a.s.l.; Byers and Braham includes many reports for Aso and Sakurajima; apparently 1949; Williams 1985), and they suggest that some other some sort of systematic study of lightning/electrical activity lightning producing mechanisms may be acting. A histo- was under way in Japan. gram of ash plume heights is shown in Fig. 2a for the 142 cases for which we have data. The histogram shows a Lightning occurrence and volcanic explosivity index bimodal distribution with one peak in the range of 7 to 12 km and another peak in the range of 1 to 4 km. The We found that lightning has occurred in association with higher altitude peak represents the typical heights of 212 eruptions at 80 volcanoes (Appendix). These are ordinary thunderstorms. The low altitude peak, however, significantly higher numbers than the 55 volcanoes cited is significantly lower than thunderstorm values. in McNutt and Davis (2000) or 58 cited in Mather and No comprehensive or extensive compilation of plume Harrison (2006) and reflect our additional work since then. heights is known to us, based on numerous searches and VEI values were known or could be estimated for 177 discussions with colleagues. We expect plume heights, eruptions, and a histogram of these events is shown in which are a component of VEI, to have a similar Fig. 1 (bottom). Note that heights are estimated above the distribution to the VEI data in Fig. 1 (top). Restated, a vent for eruptions with VEI=2 and smaller, but above sea histogram of all plume heights would be expected to have Bull Volcanol

Table 1 Number of eruptions with lightning per year Year No. Eruptions Year No. Eruptions

2009 6 1972 0 2008 6 1971 1 2007 2 1970 0 2006 15 Augustine (14) 1969 0 2005 13 Colima (11) 1968 2 2004 5 1967 0 2003 0 1966 2002 0 1965 1 2001 0 1964 2 2000 3 1963 1 1999 1 1962 1 1998 1 1961 0 1997 1 1960 0 1996 3 1995 2 1956 1 1994 2 1953 1 1993 1 1952 1 1992 4 1951 11 Lamington (10) 1991 9 1950 1 1990 12 Redoubt (12) 1949 1 1989 1 1948 1 1988 2 1947 1 1987 1 1946 1 1986 4 1945 1 1985 1 1944 2 1984 1 1943 1 1983 1 1937 1 1982 13 Galunggung (6) 1933 1 1981 2 1931 1 1980 3 1929 1 1979 19 Aso (7), Sakurajima (8) 1924 1 1978 7 1914 2 1977 1 1912 1 1976 5 1911 1 Prior to 1960 only years with 1975 1 1906 1 eruptions are listed 1974 1 1902 3 Prior to 1900, the year 1707 is 1973 2 the only year with more than 1707 2 one report of lightning smoothly increasing numbers moving to smaller heights. linear relation to estimate the number of plume heights for Because no compilation exists, we created a synthetic each km. Data were then renormalized so the total number distribution of heights as follows. We used VEI data to of eruptions was the same as the starting values. This generate a plot of number of plumes at each km of altitude procedure gives us estimates of the total number of by assuming all heights for a given VEI were at the mid- eruptions for each km of plume height. We then can plot point of the altitude range. For example, the height range the number of eruption plumes with lightning for each for VEI=2 is 1–5 km so we assumed 3 km as the mean height as a percentage of all eruptions, as shown in Fig. 2b. height. For VEI=5 we used 25 km and for VEI=6 we used Here we observe that all the values for heights of 1–6km 31 km. We then computed a linear regression for the VEI are 0.3% or less (mean 0.018%), whereas those between 7 values from 2 to 6 (heights of 3 to 31 km) and used the and 13 km are 0.3% or greater (mean 0.07%). This Bull Volcanol

Fig. 1 Histogram of Volcanic Explosivity Index (VEI) for all eruptions in Volcanoes of the World, by Simkin and Siebert, 1994 (a) and for eruptions accompanied by lightning (b) using data from the Appendix. Note that the eruptions accom- panied by lightning are skewed towards higher VEI values. The plume heights for VEI 3 and larger are similar to the heights of thunderclouds

difference of more than a factor of 3 suggests that eruptions 3.8 km high and this was accompanied by only a single with ash plume heights of 7 to 13 km, the same range as lightning flash. The latter flash showed a duration of about thunderstorms, are more efficient at producing lightning 10 ms, substantially shorter than typical thunderstorm than smaller eruptions. Additional reasons for this system- lightning (Thomas et al. 2010). Thus the available instru- atic difference are given in the discussion section. mental data, although sparse, do suggest a systematic Recent instrumental data from Augustine volcano in difference in lightning between large plumes and small ones. Alaska may also shed some light on this issue. The eruption The study of photographs of volcanic lightning also on January 28, 2006 at 05:31 UT had an ash plume 10.5 km supports a systematic difference. Photos of small plumes high (a.s.l.), similar to thunderstorms. It had abundant (1–3 km), typically time exposures, show lightning with lightning in the plume, with over 300 lightning flashes few branches and typical lengths of a few tens to hundreds starting about 5 min after the onset of the main phase of the of meters (e.g. National Geographic Sept. 2007, p. 14–15). eruption (Thomas et al. 2007, 2010). Most of these events Some photos from larger eruptions, on the other hand, show were intracloud flashes with durations of 30–600 ms, lightning with many branches and lengths of several km typical of values for thunderstorms. Measurements of (e.g. Galunggung; Katili and Sudrajat 1984). We note that duration were made using Lightning Mapping Array the dozen or so photos we examined are suggestive but not (LMA) data (Thomas et al. 2004). In contrast, a smaller definitive. A more comprehensive study of this topic is eruption at 08:37 UT January 28 had an ash plume only warranted.

Magma composition Table 2 Number and percent of occurrences of volcanic lightning at various VEI values The chemical composition of erupting magma is of interest VEI No. cases No. cases percent in studies of volcanic lightning because this characteristic is with Lightning all eruptions known to influence the dissolved water content in the 1 7 845 0.83 magma (with important effects on electrification in its own 2 61 3477 1.75 right), with important consequences for the style of the 3 81 869 9.32 eruption (i.e., Strombolian, Plinean, etc.). The broad 4 22 278 7.91 categorization for chemical composition in eruptions ranges – 5 7 84 8.33 from basalt to andesite to dacite to rhyolite (48 77% SiO2), 6 4 39 10.26 with a systematic decline in equilibrium magma tempera- ture from about 1,200°C to 800°C over this range. Water 7 0 4 0.00 contents range from about 0.1 to 6.5 weight percent, and 80 0– are systematically higher with increasing silica content Bull Volcanol

Fig. 2 a Histogram of ash plume heights for eruptions accompanied by volcanic light- ning. The broad peak >7 km includes heights of similar dimension and larger than typi- cal thunderclouds. This suggests that similar mechanisms may be acting. Also note the second peak from 1 to 4 km. These heights are significantly smaller than thunderclouds and suggest a possible second mechanism or mechanisms. b Histogram of ash plume heights for eruptions accompanied by volcanic light- ning plotted as percentages of the numbers of all eruptions for given ash column heights

(Wallace and Anderson 2000). In an ideal case we would high and supports the general conclusions given here, have eruptions of identical size (plume height and volume, however, we suspect that the sample is not uniform. This etc.) with only the composition (SiO2 percent) varying. topic will benefit from better data from future eruptions. Unfortunately the tests enabled here with real-world Ideally, future cases of instrumental recording of lightning observations fall short of this mark. To address this issue will also include the supporting information on magma we compiled information concerning the magma composi- composition for each individual eruption. tion for as many of the cases in the Appendix as possible. Some of the cases were compiled previously (Mather and Latitude effects Harrison 2006) and we added 30 or so new cases to that list. One problem is that information on a volcano is often The current data set on volcanic lightning (Appendix)is given based on study of geological samples over a long now sufficiently extensive to explore the variation of events time span, rather than for a specific eruption that produced with latitude, with results that serve to constrain source lightning. Hence we were constrained to choose represen- mechanisms for the lightning. Figure 3 (top) shows the tative values for magma composition based on published number of volcanoes with at least one eruption with data. lightning, binned in 5° increments of latitude and normal- All identified eruptions with lightning as functions of ized for surface area within these increments, and summed plume height and magma composition are listed in Table 3. for northern and southern latitudes. Figure 3 (bottom), Basaltic eruptions span the range from 1 to 21 km, with no constructed in similar fashion for sake of contrast, shows clear groupings as a function of plume height. The same is the latitudinal distribution of all volcanoes in the Holocene true for the basaltic-andesite cases, which range from 0 to record (Smithsonian Institution 2009; Simkin and Siebert 19 km. Andesite cases, which were the most numerous, 1994). Figure 3 (middle) shows the latitudinal variation of give the first indication of two groups of plume heights as all eruptions known to produce lightning. also seen in Fig. 2. They also span the largest range from 0 Despite a notable gap in all the distributions in Fig. 3 in to 21 km. Andesitic-dacite cases form a single group the 20°–30° range, attributable to a deficit in island arc between 7 and 13 km plume height. Dacite cases range length in this region, the overall distributions are decidedly from 4 to 33 km (Pinatubo at the maximum) with no flat with latitude. Indeed, the percentage of all historical obvious grouping. Four of the six rhyolite cases were volcanoes with documented lightning (Fig. 4, top) is largely associated with small eruptions from 0 to 2 km plume independent of latitude. If the entrainment of meteorolog- height; the high value is for Chaiten, May 2008. This is too ical water was significant, then the tropical values should small a sample to be able to generalize. Thus the overall be significantly higher. As a check, the percentage of data do not appear to display any simple composition- eruptions with lightning is also plotted versus latitude dependent trends. The number of cases in Table 3 is fairly (Fig. 4, bottom). Here several volcanoes contribute heavily Bull Volcanol

Table 3 Volcanic Lightning occurrence, magma composi- Plume Height km B BA A AD D R tion, and ash plume height 0–1 BA A,A R,R 1–2 B,B BA A,A,A,A R,R 2–3 B A,A,A,A,A 3–4 B,B A,A 4–5BADR 5–6B A 6–7 BA,BA 7–8 BA A,A,A AD D 8–9 B BA AD 9–10 A,A,A AD,AD,AD 10–11 B A,A,A,A AD,AD,AD,AD D 11–12 B A,A 12–13 AD,AD D,D 13–14 B BA,BA A 14–15 BA 15–16 A 16–17 B 17–18 18–19 BA,BA 19–20 R – Repeated letters indicate multi- 20 21 B A D,D ple observations 21–22 B basalt; BA basaltic-andesite; A 22–23 D andesite; AD andesitic dacite; D 33 D dacite; R rhyolite to specific latitude bins, but overall the distribution is still dependence in the available water, because geochemistry— flat with no significant increase for the tropics. These and water content—is independent of latitude. If, on the observations suggest that magmatic water already brings other hand, the water is derived by entrainment of water volcanic plumes above the threshold needed to produce vapor from the local atmosphere (e.g., Rose et al. 1995a, b; lightning, hence the role of meteoric water is secondary. Sparks et al. 1997; Carey and Bursik 2000; Textor et al. In marked contrast with the latitudinal behavior of 2003), then a pronounced latitudinal dependence in volcanic lightning, Fig. 5 shows the latitudinal distribution volcanic lightning is expected. The latter expectation is of natural (meteorological) lightning after Williams (1992). not supported by the available observations in Figs. 3 A pronounced decline with increasing latitude is evident. and 4. Some water vapor is added to ash columns via Similar findings are evident in the classical study of thunder entrainment, but our data suggest this is not in sufficient days for the globe (Brooks 1925). These trends are amounts to alter the electrical activity. In other words, there generally attributable to the temperature dependence of may be a threshold of water substance concentration saturation water vapor embodied in the Clausius-Clapeyron required in a plume for lightning to occur, but this appears relation (Williams 1995, 2005). For a rough doubling of to be a function of the large amount of water in the magma saturation water vapor mixing ratio for every 10°C of and not the smaller amount in the entrained air. temperature change in the latter relation, this can amount to an order-of-magnitude effect between equatorial and polar Seasonal effects regions. For thunderstorms the water substance is respon- sible for generating and moving charges between water A second test of these ideas pertaining to water substance droplets, ice, and . Hence more water in tropical air involves the comparison of volcanic lightning in the winter means more lightning there. and in the summer, on account of the pronounced differ- Water substance is fundamentally important for volcanic ences in surface temperature between these two . lightning, according to at least one idea (Williams and For purposes of this comparison, only extra-tropical McNutt 2005). If this water is derived from the magma volcanoes (with latitudes greater than 23° in both hemi- rising in the eruption, one does not expect a latitudinal spheres) were considered. ‘Winter’ eruptions were consid- Bull Volcanol

Fig. 3 a Number of volcanoes with lightning versus latitude; b number of eruptions with lightning versus latitude; and c number of volcanoes versus latitude. The three plots show basically the same trend: the distribution of volcanoes, erup- tions, and volcanoes at which lightning is observed occur rather uniformly from 5 to 20 and from 35 to 65° latitude, with a conspicuous gap from 25 to 30°. This plot is very different from the plot of thundercloud lightning versus latitude shown in Fig. 5. The numbers for volcanoes and eruptions are normalized for the total area within each 5° interval of latitude

ered bounded by the autumnal and vernal equinoxes, and does indeed occur. The available evidence, however, ‘summer’ eruptions, vice versa. With these definitions, 44 suggests that it is not an important effect in terms of eruptions from the Appendix are wintertime, in contrast increasing the water content of volcanic plumes. Previous with only 36 in summer. This result runs counter to the calculations (Williams and McNutt 2005) show that the expectations based on the entrainment of local water vapor, magma-derived water content in volcanic plumes is already whereas geochemistry is seasonally independent. The considerably higher than that of thunderstorms. greater number of eruptions in winter may find an An additional effect must be noted here. We had explanation in the temperature structure of the local assumed in the previous paragraph that eruptions were atmosphere, a circumstance that assures that more of the evenly distributed by . However, Mason et al. (2004) magma-derived water substance is transformed to ice demonstrate seasonality of eruptions with higher numbers during the eruption. Note that the results found here do occurring in boreal winter. Thus it is possible that the not mean that entrainment of water vapor does not occur; it greater number of eruptions with lightning in winter may be

Fig. 4 a Percent of volcanoes with lightning versus latitude. b Percent of eruptions with lightning versus latitude. High values in (b) between 30–35 and 60–65° represent observations from Japan and Alaska. These distributions are essentially flat, suggesting that entrainment of meteorological water does not play a major role in volcanic lightning occurrence Bull Volcanol

observations from the constellation of US DoD satellites (Moore and Rice 1984; Pack et al. 2000).

Ash plume heights

The bimodal ash plume height distributions of Fig. 2 (top; and normalized version in Fig. 2 (bottom)) suggest that two broad classes of charging mechanisms may be acting. It is important to distinguish what we actually measured— plume heights—from the lengths of lightning flashes which can be much smaller. First, note that the ash plume heights are for the tops of the plumes. Lightning occurs within the ash plume at various azimuths (not necessarily vertical) and Fig. 5 Latitudinal distribution of lightning from space (i.e. global ISS-b satellite observations) showing a dominant contribution from flashes typically have smaller dimensions. For the higher the tropics (+23°). This strong decline with latitude is attributable to ash plume peak from 7 to 12 km, and including all reports the distribution of water in the atmosphere. Dominant signals are from greater than 7 km, the mechanism is plausibly the same as thunderstorms, not volcanoes. Note the differences between this figure for ordinary thunderstorms. The concentration of water in and volcanic lightning data in Figs. 3 and 4, particularly between 25– 30° and >50°. Figure from Williams (1992) based on data from eruption plumes is likely greater than for typical thunder- Orville and Henderson (1986) storms (Williams and McNutt 2005), so the stage is set for efficient charge separation via the mixed phase microphys- ics of water substance (Williams 1995). Indeed, the a function of more eruptions in that season, and the relative expected abundance of water may lead to an ice coating ease of forming ice at colder winter conditions. of the ash particles that interferes with a charge separation mechanism based on the collisions of dry silicate particles, but it is suggested here that tall (>7 km) Discussion plumes are essentially dirty thunderstorms. Further evi- dence for this point of view may be found in Williams and Lightning occurrence McNutt (2005). A recent study by Durant et al. (2008) concluded that volcanic plumes are “overseeded” with ice Our compilation shows that volcanic lightning is quite particles compared with thunderstorms, and have higher common, having occurred at 212 eruptions from 80 concentrations of ice particles of smaller average size. This volcanoes (Appendix). However, Simkin (1993) estimates lab based finding agrees with our inferences based on that 55–70 volcanoes erupt worldwide each year. From lightning distribution. Modeling studies by Textor et al. Table 1 we find the year with the greatest number of (2005a, b) also discuss formation of ice and water lightning reports is 1979 with 19 cases. This suggests that droplets. Durant et al. (2008) noted that ice formation only 27–35% of eruptions are accompanied by lightning, occurs on ash particles at temperatures of 250 to 260 K assuming one eruption per volcano per year. These must be (−10 to −20 C); these temperatures are found at heights of over-estimates because most volcanoes erupt more than several km depending on the atmospheric model, latitude, once per episode. There are several reasons that such low and time of year. Plumes may be somewhat warmer than percentages of eruptions are accompanied by lightning. the surrounding ambient air, so heights of ice formation First, our table may be incomplete; we did our best, but we would be a bit higher. We suggest that the formation of ice cannot guarantee completeness. Second, specialized con- particles within plumes larger than 7 km is the reason for ditions may be necessary, and these may occur during only the systematic increase in lightning percent shown in a small percentage of eruptions. Third are reporting Fig. 2 (bottom). Once ice particles exist, the processes problems mentioned above, such as daylight or tropical separating charges operate efficiently as in ordinary environments and the need for alert observers in areas thunderstorms. The charge distribution in thunderstorms where the general population may be very sparse. To as a function of height and temperature is shown by improve reporting and to remove possible bias, we Krehbiel (1986) and lightning initiation is common at recommend that a representative suite of volcanoes be heights of 5–6 km where temperatures are −10 to −20 C instrumented and a systematic reporting scheme be set up. and ice particles form. This is the same range of temper- A more comprehensive documentation of eruptions world- atures found by Durant et al. (2008) for ice formation on wide (with and without lightning) could also be achieved ash particles. The data of Fig. 2 (bottom) suggest that this through greater exploitation of continuously available process—involving water and ice—is about 3 times more Bull Volcanol efficient than the process or processes dominant at subsequent eruptions at 11:04 UT and 16:42 UT, which had altitudes of 6 km and below. ash plumes 7.2 and 7.0 km high. Each of these produced a The secondary peak of Fig. 2 (top), from 1 to 4 km, weaker (but longer lasting) vent discharge signal at the time suggests that a second process or suite of processes is of the eruption, and correspondingly weaker pressure acting. The ash plume heights for these eruptions are amplitudes on the infrasound sensor (Petersen et al. smaller than those for thunderstorms and the near-vent 2006), but each had more near vent lightning flashes, 28 temperatures are likely quite elevated, both effects casting for the 11:04 UT eruption and 6 for the 16:42 UT eruption. doubt on any significant ice phase. Accordingly, a To summarize, the continuous electrical signals at the time mechanism similar to thunderstorms seems unlikely. We of the eruptions (vent discharges) were positively correlated suggest that several mechanisms are operating simulta- with pressure amplitudes, but negatively correlated with the neously. The first is ash particle collisions, which have been number of lightning flashes. We can reconcile these shown to generate reasonable charges in laboratory experi- observations by noting that the eruptions with longer ments (James et al. 2000). A second related mechanism durations produced more lightning flashes (McNutt et al. consists of fracturing particles, so called fracto-emission 2010). This suggests that efficient charge producing (James et al. 2000; Gilbert and Lane 1994b). A possible mechanisms (high pressure amplitudes at the vent) alone third mechanism involves moving insulating bodies relative are not enough to produce lightning; there must also be a to ionizing fluids (also called streaming potential; e.g. sufficiently large amount of tephra to carry the electric Morgan et al. 1989). A familiar example of this is gasoline charge and redistribute it in space to produce conditions moving through hoses. (This is why gas stations have signs favorable for lightning generation. requiring containers to be placed on the ground for filling.) A later phase of sustained eruptions at Augustine (in the This effect requires an electrical double layer at the contact interval January 29-February 2, 2006) showed lightning between the fluid and wall or pipe. This effect, if it occurs, only in association with four stronger pulses (Thomas et al. may be most likely for basalt because basalt is the most 2010), even though the ash columns were persistently 5 km fluid . Alternatively, gases separating from the magma or more in height. The ash plume height was 3.8–7.2 km or groundwater may be involved. With the cases we have for each of the four pulses (Thomas et al. 2010). However, for study, however, we cannot verify or quantify this the pressure amplitudes were low (all less than 13 Pa at mechanism. It is mentioned for the sake of completeness. 3 km distance), so even though abundant tephra was Boiling of seawater, as discussed by Blanchard (1964), available, the charging mechanisms were only sufficiently Blanchard and Björnsson (1967) and James et al. (2008)is vigorous for lightning production when the stronger pulses a very efficient charge generating mechanism. However, the occurred. We infer that virtually all the visual observations number of cases in the Appendix that involve seawater is of lightning in the Appendix are either near vent lightning very small, so we infer this to be a special case rather than or plume lightning; the vent discharges (continuous signals the more ubiquitous general case involving ash. during eruption) have only been observed instrumentally and there are not yet any corroborating visual observations. Recent instrumental observations From a theoretical point of view, the maximum electric field at the surface of a uniformly charged spherical volume The 2006 eruptions of Augustine (Power et al. 2006) of radius R is provide some clues about charging mechanisms (Thomas et E ¼ rR=3" al. 2007; Thomas et al. 2010). The explosive eruption on o 3 January 28, 2006 at Augustine 08:37 UT produced where ρ is the space charge density (C/m ) and εo is the significant nearly continuous radio frequency electrical permittivity of free space. The maximum field between two discharge activity (vent discharges; 1–2 ms, 10 s of m uniformly charged spherical volumes of opposite polarity is scale length) for the 20 s during which the tephra and gases twice this value, by superposition. These simple calcula- were exiting the vent (Thomas et al. 2010; these were tions show that dielectric breakdown can be achieved in similar to the continuous charges as shown in Fig. 1a of of small size only if the space charge density is very Thomas et al. 2007). The 08:37 UT eruption had the highest large. Typical values of ρ for thunderstorms are 1 nC/m3 pressure amplitude (105 Pa) of the 13 explosive eruptions which requires R of several km to produce lightning. For as measured on an infrasound sensor 3 km from the vent some observed volcanic lightning, the scale length is a few (Petersen et al. 2006). From this we infer relatively higher hundred meters, which suggests space charge densities gas content and high ejection rates. However, the ash plume about one order of magnitude larger. Dielectric breakdown was only about 3.8 km high, and this eruption produced initiated in the high field region is sufficient to produce only a single lightning flash (near vent lightning; 10–30 ms, VHF radiation, but subsequent thermalization of the 100 m to km scale length). This behavior contrasts with the streamer channel is required to form lightning flashes. Bull Volcanol

(Thermalization is movement of the channel towards and electric field intensities as functions of time. Are thermal equilibrium, a transition in the state from high particles already charged and simply hoisted into the air by electron energy alone to high energy for all constituents eruptions? Or does the eruption process itself produce the (Gallimberti et al. 2002)). Small volumes of space charge charges? Previous works by Gilbert and Lane (1994b) and may not provide sufficient charge (and current) for James et al. (2000) investigated charging mechanisms and thermalization, if laboratory studies are a guide (Boschung rates but not with respect to lightning per se. If particle et al. 1977). collisions are needed, then we would expect greater These considerations suggest that the nearly continuous concentration of fines in ash columns with greater charge, electrical activity—vent discharges—in the high pressure and textural evidence of collisions in the particles that fall emission at Augustine on January 28, 2006 at 08:37 UT to the ground. We are not aware of studies that looked for was caused by a mechanism generating large space charge these features in particular with respect to volcanic density in a relatively compact (3.8 km height) that lightning production, and we suggest that such studies are produced discharges with durations much shorter than needed. Information is also needed on the distribution of typical lightning flashes in thunderclouds. The larger, electric charge carried by tephra particles over the full range deeper clouds (7.0–7.2 km) in the 11:04 UT and 16:42 of particle sizes. Some work on charge versus size has been UT eruptions plausibly involved an in situ mechanism, documented by Miura et al. (2002) but the eruptions possibly involving ice particle collisions, which allowed for studied did not produce lightning. In ordinary thunder- both dielectric breakdown and sufficient charge to thermal- clouds, the ability to access such information is difficult. In ize channels and provide for discharges with durations volcanic eruptions, the access is extraordinarily difficult. more typical of those in thunderclouds. In the later January 29-February 2, 2006 Augustine eruption phase, involving Practical considerations smaller clouds (<~5km) and also low ejection rates, again, only four episodes of near vent lightning activity were A certain concentration or density of charge is needed to present. The behavior during most of the continuous phase cause lightning, so there ought to be a relation between ash is plausibly explained by the absence of high space charge particle concentration and lightning occurrence if indeed density and by a cloud radius R insufficient for either the ash particles are the charge carriers. This suggests that dielectric breakdown or channel thermalization. lightning may be used as a real-time predictor of charge density, hence ash particle density. The occurrence of Other meteorological phenomena lightning would then add information of value to determine ash hazards to aviation. Lightning could thus improve the Electrical charge separation is known to occur in two other ability to verify when significant ash clouds are being meteorological phenomena that are distinct from thunder- produced, and by identifying the different types or regimes clouds: dusty gust fronts (Williams et al. 2009) and dust of lightning, such as near-vent or plume lightning, also devils (Freier 1960;Crozier1964; Ette 1971). These improve estimates of the sizes of eruptions. This would be involve lofting of silicate minerals into the boundary layer especially true for volcanoes where lightning detection by strong winds by an entirely dry process. Though the systems are in place, which could provide information in detailed physical process for charge separation is still poor weather or darkness. poorly understood at the particle scale, the observations Better instrumental data are needed, including specific support a selective charging of the smaller dust particles experiments for individual eruptions as well as better with negative electricity. This polarity is opposite to the utilization of existing resources. For example, there is a need behavior of ordinary thunderclouds and the few volcano for a systematic search for volcanic lightning in the Lightning clouds that have been documented electrically (Cobb 1980; Imaging Sensor (LIS) and Optical Transient Detector (OTD) Hobbs and Lyons 1993; Lane and Gilbert 1992; Gilbert and satellite data (Christian et al. 2003). Volcanic lightning Lane 1994b; Hoblitt 1994; McNutt and Davis 2000). Miura presents many interesting research opportunities, which are et al. (2002) showed cases with both positive and negative complementary to both atmospheric sciences and volcanol- charge in fine particles from eruptions of Sakurajima. ogy. It is hoped that future research with better data will Furthermore, dusty gust fronts and dust devils occur in further illuminate the trends identified in this work. dry environments, whereas even ‘dry’ volcanic eruptions contain a significant quantity of water. These considerations and comparisons cast doubt on the gust front/ Conclusions process alone in accounting for volcano electricity. To better constrain these ideas, instrumental data are Data from a comprehensive global database show that needed. For example we need to know particle velocities lightning has been documented at 80 volcanoes in Bull Volcanol association with 212 eruptions. Eight percent of VEI=3–5 Anderson T, Flett JS (1903) Report on the eruptions of the Soufriere in St. Vincent, and on a visit to Montagne Pelee in Martinique. eruptions have reported lightning, and 10% of VEI=6, but – – Philos Trans R Soc Lond A 200:353 553 less than 2% of those with VEI=1 2. These findings Anderson R, Bjornsson S, Blanchard DC, Gathman S, Hughes J, suggest consistent reporting for larger eruptions but either Jonasson S, Moore CB, Survilas HJ, Vonnegut B (1965) less lightning, a different mechanism, or possible under- Electricity in volcanic clouds. Science 148:1179–1189 reporting for small eruptions. Ash plume heights show a Anonymous (1863) The reason why geography and geology. Houlston – – and Wright, London, 88 bimodal distribution with main peaks at 7 12 km and 1 Arason P (2005) Lightning during volcanic eruptions in Iceland. Geophys 4 km. The former are similar to heights of typical thunder- Res Abstr 7:05369, SRef-ID: 1607-7962/gra/EGU05-A-05369 storms and suggest involvement of water substance, Arason P, Sigurdsson E, Johannsdottir H, Juliusson G, Kristmundsson whereas the latter suggest other factors contributing to G (2000) Volcanogenic during a sub-glacial eruption in Iceland (abstract) International Conference on Lightning electrical behavior closer to the vent. Reporting of lightning Protection, ICLP 2000. Rhodes, Greece, p 100 is more common at night (56%) and less common in AVO (Alaska Volcano Observatory) Bi-monthly report (1996) Double daylight (44%). Reporting also varied substantially from issue, September-December 1996, Alaska Volcano Observatory, year to year. Several weak trends in lightning occurrence Fairbanks, Alaska, 8, 5, and 6, 56 Benediktsson P (1996) Eruption at the ice cap: volcanic unrest at based on magma composition were found. The bimodal ash Vatnajokull, 1996 (video). Sjonvarpid, Reykjavik, Iceland plume heights are obvious only for andesite to dacite; basalt Blanchard DC (1964) Charge separation from saline drops on hot and basaltic-andesite evenly span the range of heights; and surfaces. Nature 201:1164–1166 rhyolites are poorly represented. The distributions of the Blanchard DC, Björnsson S (1967) Water and the generation of volcanic electricity. Mon Weather Rev 95:895–898 latitudes of volcanoes with lightning and eruptions with Blong RJ (1984) Volcanic hazards: a sourcebook on the effects of lightning roughly mimic the distribution of all volcanoes, eruptions. Academic, Sydney which is generally flat with latitude. Meteorological Bluth GJS, Scott CJ, Sprod IE, Schnetzler CC, Krueger AJ, Walter LS lightning, on the other hand, is common in the tropics and (1995) Explosive emissions of sulfur dioxide from the 1992 Crater Peak eruptions, Mount Spurr volcano, Alaska. In: Keith decreases markedly with increasing latitude. This finding TEC (ed) The 1992 eruptions of Crater Peak, Mount Spurr, supports the idea that lightning in large (deep) eruptions Alaska. US Geol Surv Bull 2139:37–45 depends on water substance, and the origin of the water is Boschung P, Hilgner W, Luttgens G, Maurer B, Widmer A (1977) An primarily magma and not entrainment from the surrounding experimental contribution to the question of the existence of lightning-like discharges in dust clouds. J Electrostat 3:303–310 atmosphere. Seasonal effects show that more eruptions with Brook M, Moore CB, Sigurgeirsson T (1973) Lightning in volcanic lightning were reported in winter (bounded by the respec- clouds. EOS Trans Am Geophys Union 54:701 tive autumnal and vernal equinoxes) than in summer, again Brooks CEP (1925) The distribution of thunderstorms over the globe. – suggesting that the water is primarily from magma. The Geophys Mem London 24:147 164 BVE (Bulletin of Volcanic Eruptions) (1989) Volcanol Soc Japan, Tokyo small number of modern instrumental observations from BVE (Bulletin of Volcanic Eruptions) (1990) Volcanol Soc Japan, Augustine Volcano shows features that are complementary Tokyo to the data from the global compilation. Byers HR, Braham RR (1949) The thunderstorm project. U.S. Weather Bureau, US Department of Commerce, Washington, DC Carey S, Bursik M (2000) Volcanic plumes. In: Sigurdsson H, Acknowledgements Discussions on this subject with RV Anderson, Houghton BF, McNutt SR, Rymer H, Stix J (eds) Encyclopedia P Arason, M Baker, D Blanchard, CGJ Ernst, J Gilbert, T Grove, of volcanoes. Academic, San Diego, pp 527–544 P Herzegh, R Hoblitt, P Hobbs, C Kessinger, P Krehbiel, T Mather, Carroll JJ, Parco SA (1966) Social organization in a crisis situation: J Murray, C Newhall, A Oswalt, D Pack, C Rice, W Rison, W Rose, the Taal disaster. Philippine Sociological Society, Manila D Schneider, T Simkin, C Textor, R Thomas, A Tupper, J Ewert, and Christian HJ, Blakeslee RJ, Boccippio DJ, Boeck WL, Buechler DE, R Wunderman are gratefully acknowledged. We thank two anonymous Driscoll KT, Goodman SJ, Hall JM, Koshak WJ, Mach DM, reviewers for their comments which helped to improve the manuscript. Stewart MF (2003) Global frequency and distribution of lightning ’ ERW s work on this problem has been assisted by the NASA ASAP as observed from space by the Optical Transient Detector. J (Advanced Satellite Aviation Assets) program under J Murray. This Geophys Res 108:4005. doi:10.1029/2002JD002347 work was partially supported by the National Science Foundation Clarke ED (1821) Lectures to the Royal Society of London under contract ATM-0538319. Cobb WE (1980) Electric fields and lightning in the Mt. St. Helens volcanic cloud. EOS Trans Am Geophys Union 61:987 WD (1964) The electric field of a New Mexico dust devil. J – References Geophys Res 69:5427 5429 Davis CM, McNutt SR (1993) Lightning associated with the 1992 eruptions of Mt. Spurr Volcano, Alaska. EOS Trans Am Geophys Abe K (1979) Seismicity of the caldera-making eruption of Mount Union 74(43):649 Katmai, Alaska in 1912. Bull Seismol Soc Am 82:175–191 Durant AJ, Shaw RA, Rose WI, Mi Y, Ernst GGJ (2008) Ice Alaska Volcano Observatory (1993) Mt. Spurr’s 1992 eruptions. EOS nucleation and overseeding of ice in volcanic clouds. J Geophys Trans Am Geophys Union 74:217, and 221–222 Res 113:D09206. doi:10.1029/2007JD009064 Alcaraz AP (1989) Some notes on false alarms of volcanic activity Erskine WF (1962) Katmai. Abelard-Schuman, London and mudflows. In: Latter JH (ed) Volcanic hazards: assessment Ette A (1971) The effect of dust on atmospheric electric and monitoring. IAVCEI Proc Volcanol 1:163–168 parameters. J Atmos Terr Phys 33:295–300 Bull Volcanol

Fedotov SA, Masurenkov YuP (1991) Active volcanoes of Kamchatka. James MR, Lane SJ, Gilbert JS (2003) Density, construction, and drag Nauka, Moscow coefficient of electrostatic volcanic ash aggregates. J Geophys Fisher RV, Heiken G, Hulen JB (1997) Volcanoes: crucibles of Res 108:2435. doi:10.1029/2002JB002011 change. Princeton University Press, Princeton James MR, Wilson L, Lane SJ, Gilbert JS, Mather TA, Harrison RG, Francis P (1976) Volcanoes. Penguin, New York Martin RS (2008) Electrical charging of volcanic plumes. Space Freier GD (1960) The electric field of a large dust devil. J Geophys Sci Rev 137:399–418. doi:10.1007/s11214-008-9362-z Res 65:3504 Johnson RW, Threlfall NA (1985) Volcano town: the 1937–43 Rabaul Gallimberti I, Bacchiega G, Bondiou-Clergerie A, Lalande P (2002) eruptions. Robert Brown & Associates, Bathurst, NSW Fundamental processes in long air gap discharges. CR Phys Judd JW (1888) In: Symons GJ (ed) The eruption of Krakatoa, and 3:1335–1339 Subsequent Phenomena. Report of the Krakatoa Committee of Gilbert JS, Lane SJ (1994a) The origin of accretionary lapilli. Bull the Royal Society. Trubner and Co, London, pp 1–56 Volcanol 56:398–411 Juhle W, Coulter H (1955) The Mt. Spurr eruption, July 9, 1953. EOS Gilbert JS, Lane SJ (1994b) Electrical phenomena in volcanic plumes. Trans Am Geophys Union 36:199 In: Casadevall, TJ (ed) Volcanic ash and aviation safety: Katili JA, Sudrajat A (1984) Galunggung: the 1982–1983 eruption. Proceedings of the first international symposium on volcanic Volcanol Survey , , p 3 ash and aviation safety. US Geol Surv Bull 2047:31–38 Katsui Y, Kawachi S, Kondo Y, Ikeda Y, Nakagawa M, Gotoh Y, Goldsmith O (1852) A history of the Earth and animated nature. Yamagishi H, Yamazaki T, Sumita M (1990) The 1988–1989 Blackie, Glasgow explosive eruption of Tokachi-dake, Central Hokkaido, its Gorshkov GS (1959) Gigantic eruption of the volcano Bezymianny. sequence and mode. Bull Volcanol Soc Japan 35:111–129 Bull Volcanol 20:77–109 Kienle J, Swanson SE (1985) Volcanic hazards from future eruptions Gorshkov GS, Dubik YM (1970) Gigantic directed blast at Shiveluch of Augustine volcano, Alaska. Report UAG R-275, University of Volcano (Kamchatka). Bull Volcanol 34:261–288 Alaska, Fairbanks, Alaska Gourgaud A, Camus G, Gerbe M-C, Morel J-M, Sudradjat A, Vincent Kikuchi K, Endoh T (1982) Atmospheric electrical properties of PM (1989) The 1982–83 eruption of Galunggung (Indonesia): a volcanic ash particles in the eruption of Mt. Usu volcano, 1977. J case study of volcanic hazards with particular relevance to air Meteorol Soc Jpn 60:548–561 navigation. In: Latter JH (ed) Volcanic hazards: assessment and Krafft M (1993a) Vulkane, Feuer der Erde. Otto Maier, Ravensberg, monitoring. IAVCEI Proc Volcanol 1:151–162 Germany, pp 56–57 Green JA (1944) Paricutin, the cornfield that grew a volcano. Natl Krafft M (1993b) Vulkane, Feuer der Erde. Otto Maier, Ravensberg, Geogr Mag 85(155):129–156 Germany, p 103 Gutierrez C (1972) A narrative of human response to natural Krehbiel P (1986) The electrical structure of thunderstorms. In: Krider disaster: the eruption of Paricutin. In: Nolan ML (ed) San Juan EP, Roble RG (eds) The Earth’s electrical environment. National Nuevo Parangaricutiro: Memories of Past Years. Environmen- Academy Press, Washington DC, pp 90–113 tal Quality Note No. 07. Texas A&M University, College Lane FW (1965) The elements rage. Chilton, Philadelphia Station Lane SJ, Gilbert JS (1992) Electric potential gradient changes during GVN (Global Volcanism Network) Bulletin (1991; 1992; 1993; 1994; explosive activity at Sakurajima Volcano, Japan. Bull Volcanol 1995) Smithsonian Institution, Washington, DC 54:590–594 Hamilton W (1772) Observations on , Luhr JF, Simkin T (eds) (1993) Paricutin: the volcano born in a and other volcanoes, in a series of lectures addressed to the Royal Mexican cornfield. Geoscience Press, Phoenix Society of London MacDonald GA (1972) Volcanoes. Prentice-Hall, Englewood Cliffs, Havskov J, De la Cruz-Reyna S, Singh SK, Medina F, Gutierrez C New Jersey (1983) Seismic activity related to the March-April, 1982 Mason BG, Pyle DM, Wade WB, Jupp T (2004) Seasonality of volcanic eruptions of El Chichon volcano, Chiapas, Mexico. Geophys eruptions. J Geophys Res 109:B04206. doi:10.1029/2002JB002293 Res Lett 10:293–296 Mather TA, Harrison RG (2006) Electrification of volcanic plumes. Hobbs PV, Lyons JH (1993) Electrical activity associated with the Surv Geophys 27:387–432. doi:10.1007/s10712-006-9007-2 May 18, 1980 eruption of Mt St Helens, Unpublished report, McClelland L, Simkin T, M, Nielsen E, Stein TC (1989) Global September, 1993 volcanism, 1975–1985. Prentice-Hall, Englewood Cliffs, New Jersey Hoblitt RP (1994) An experiment to detect and locate lightning McKee CO, Johnson RW, Lowenstein PL, Riley SJ, Blong RJ, de associated with eruptions of Redoubt volcano. J Volcanol saint Ours P, Talai B (1985) Rabaul caldera, Papua New Guinea: Geotherm Res 62:499–517 Volcanic Hazards, surveillance, and eruption contingency plan- Hoblitt RB (2000) Was the 18 May lateral blast of Mt St Helens the product ning. J Volcanol Geotherm Res 23:195–237 of two explosions? Philos Trans R Soc Lond A 358:1639–1661 McNutt SR (1994) Volcanic tremor amplitude correlated with eruption Hoblitt RP, Murray TL (1990) Lightning detection and location as a explosivity and its potential use in determining ash hazards to remote eruptions monitor at Redoubt Volcano, Alaska. EOS aviation. US Geol Surv Bull 2047:377–385 Trans Am Geophys Union 71:146 McNutt SR, Davis CM (2000) Lightning associated with the 1992 Hoblitt RP, Wolfe EW, Scott WE, Couchman MR (1996) The eruptions of Crater Peak, Mount Spurr Volcano, Alaska. J preclimactic eruptions of Mount Pinatubo, June 1991. In: Newhall Volcanol Geotherm Res 102:45–65 CG, Punongbayan RS (eds) Fire and mud: eruptions and of McNutt SR, Tytgat G, Estes S, Stihler S (2010) A parametric study of Mount Pinatubo, Philippines. University of Washington Press, the January 2006 explosive eruptions of Augustine volcano, Seattle, pp 457–511 Alaska. US Geol Surv Prof Paper 1769 (in press) Illingworth AJ (1985) Charge separation in thunderstorms: small-scale Melson W (1994) The eruption of 1968 and tephra stratigraphy of processes. J Geophys Res 90:6026–6032 Arenal volcano. In: Sheets PD, McKee BR (eds) Archaeology, Jaggar TA (1906) The volcano Vesuvius in 1906. Tech Quart Proc 19 volcanism, and remote sensing in the Arenal region, Costa Rica. (2):104–115 Univ Texas Press, Austin, pp 24–47 James MR, Lane SJ, Gilbert JS (2000) Volcanic plume electrification: Mercalli G (1907) I vulcani attivi della terra. U Hoepli Press, Milano experimental investigation of a fracture-charging mechanism. J Miller TP, Chouet BA (1994) The 1989–1990 eruptions of Redoubt Geophys Res 105:16641–16649 Volcano: an introduction. J Volcanol Geotherm Res 62:1–10 Bull Volcanol

Miura T, Koyaguchi T, Tanaka Y (2002) Measurements of electric Pratt WE (1911) The eruption of , January 30, 1911. charge distribution in volcanic plumes at Sakurajima Volcano, Philipp J Sci 6A(2):63–83 Japan. Bull Volcanol 64:75–93 Rogers JA, Maslak S, Lahr JC (1980) A seismic electronic system Moore JG, Rice CJ (1984) Chronology and character of the May 18, with automatic calibration and crystal reference. US Geol Surv 1980 explosive eruptions of Mount St Helens. Explosive Open-File Rep 80–324:1–130 volcanism: inception, evolution and hazards. National Academy Rose WI, Delene D, Schneider D, Bluth G, Kreuger A, Sprod I, Press, Washington DC, pp 133–142 McKee C, Davies H, Ernst GGJ (1995a) Ice in the 1994 Rabaul Morgan FD, Williams ER, Madden TR (1989) Streaming potential eruption cloud: implications for volcano hazard and atmospheric properties of Westerly Granite with applications. J Geophys Res effects. Nature 375:477–479 94:12449–12461 Rose WI, Kostinski AB, Kelly L (1995b) Real-time C-band radar Nairn IA, Hewson CAY, Latter JH, Wood CP (1976) Pyroclastic observations of 1992 eruption clouds from Crater Peak, Mount eruptions of Ngauruhoe Volcano, central north island, New Zealand, Spurr Volcano, Alaska. In: Keith TEC (ed) The 1992 eruptions of 1974, January and March. In: Johnson RW (ed) Volcanism in Crater Peak, Mount Spurr, Alaska. US Geol Surv Bull 2139:19–26 Australasia. Elsevier, Amsterdam, pp 385–405 Rose WI, Bluth GJS, Ernst GGJ (2000) Integrating retrievals of Natl Geogr (1982) (photo) Nov issue, National Geographic magazine, volcanic cloud characteristics from satellite remote sensors: a 162, 656–657 summary. Philos Trans R Soc Lond S A 358:1538–1606 Natl Geogr (1993) (photo) July issue, National Geographic magazine, Rosenbaum JG, Waitt RB Jr (1981) Summary of eyewitness accounts 183, 83–84 of the May 18 eruption. US Geol Surv Prof Pap 1250:315–320 Natl Geogr (2007) (photo) Sept issue, National Geographic magazine, Rulenko OP (1981) Electrical processes in the vapor and gas clouds of 212, 14–15 the Karymskaya Sopka Volcano. Dokl Akad Nauk SSSR 245:33– Natl Geogr (2008) (photo) Nov issue, National Geographic magazine, 35 214, 16–17 Ryan MP (ed) (1994) Magmatic systems. Academic, San Diego Newcott WR, Menzel P (1993) Lightning: nature’s high-voltage Saavedra VA, Ramis RO (1991) Volcanoes. CEPA, Gallery, S.A., spectacle. Natl Geogr 184:83–103 Madrid, Spain Newhall CG, Punongbayan RS (1996) Fire and mud: eruptions and Sabit JP, Pigtain RC, de la Cruz EG (1996) The west-side story: lahars of Mount Pinatubo, Philippines. University of Washington observations of the 1991 Mount Pinatubo eruptions from the Press, Seattle west. In: Newhall CG, Punongbayan RS (eds) Fire and mud: Newhall CG, Self S (1982) Volcanic Explosivity Index (VEI): an eruptions and lahars of Mount Pinatubo, Philippines. University estimate of explosive magnitude for historical volcanism. J of Washington Press, Seattle, pp 445–455 Geophys Res 87:1231–1238 Sapper K (1905) den vulkangebieten Mittelamerikas and Westindiens. Niida K, Katsui Y, Suzuki T, Kondo Y (1980) The 1977–1978 eruption Verlag der E. Schweizerbartschen Verlagsbuchhandlung, Stuttgart, of Usu Volcano. J Fac Sci Hokkaido Univ Ser IV 19:357– 394 pp 101–153 Ondoh T (1990) Unusual intensity enhancements of low frequency Schneider M (1995) Eruption! Mt. Ruapehu, N.Z. (postcard) Kiwi atmospherics associated with great eruptions of Izu-Oshima Vista collection, New Zealand volcano in November 1986. J Geomagn Geoelectr 42:237–256 SEAN (Scientific Event Alert Network) Bulletin (1986) Smithsonian Orville RE, Henderson RW (1986) Global distribution of midnight Institution, Washington, DC lightning: September 1977 to August 1978. Mon Weather Rev Shepherd JB, Aspinall WP, Rowley KC, Pereire J, Sigurdsson H, 114:2640–2653 Fiske RS, Tomblin JF (1979) The eruption of Soufriere Volcano, Pack DW, Rice CJ, Tressel BJ, Lee-Wagner CJ, Oshika EM (2000) St. Vincent, April-June 1979. Nature 282:24–28 Civilian uses of surveillance satellites, Crosslink, The Aerospace Shore D (1975) The man who fought Vesuvius. Weekend Mag, July Corporation, December issue 30:6–7 Palmieri L (1873) Indagini spettroscopiche sulle sublimazioni Simkin T (1993) Terrestrial volcanism in space and time. Annu Rev vesuviane. Rend R Acc Sc Fis Mat Serie I vol. XII Earth Planet Sci 21:427–452 Paskievitch JF, Murray TL, Hoblitt RP, Neal CA (1995) Lightning Simkin T, Fiske RS (1983) Krakatau 1883: the volcanic eruption and associated with the August 18, 1992, eruption of Crater Peak its effects. Smithsonian Institution Press, Washington DC vent, Mount Spurr Volcano, Alaska. In: Keith TEC (ed) The 1992 Simkin T, Howard KA (1970) Caldera collapse in the Galapagos eruptions of Crater Peak, Mount Spurr, Alaska. US Geol Surv Islands. Science 169:429–437 Bull 2139:179–182 Simkin T, Siebert L (1994) Volcanoes of the world, 2nd edn. Perret FA (1924) The Vesuvius eruption of 1906: study of a Geoscience Press, Tucson volcanic cycle. The Carnegie Institution of Washington, Smithsonian Institution (National Museum of Natural History) (2009) Washington DC Global Volcanism Program, Volcanoes of the World, http://www. Petersen T, De Angelis S, Tytgat G, McNutt SR (2006) Local infrasound volcano.si.edu/ Cited 1 Oct 2008 observations of large ash explosions at Augustine volcano, Alaska, Sonafrank GHC, Power J, March G, Davies JN (1991) Acquisition during January 11–28, 2006. Geophys Res Lett 33:L12303. and automatic processing of seismic data at the Geophysical doi:10.1029/2006GL026491 Institute, University of Alaska. Seismol Res Lett 62:23 Pond JA, Smith SP (1886) Observations on the eruption of Mount Sparks RSJ, Bursik MI, Carey SN, Gilbert JS, Glaze L, Sigurdsson H, Tarawera, Bay of Plenty, New Zealand, 10th June, 1886. New Woods AW (1997) Volcanic plumes. Wiley, New York Zealand Inst Trans Proc 19:342–371 Stearns HT (1925) The explosive phase of Kilauea Volcano, Hawaii, Poster Display (1995) The 1995 International Workshop on Volcanoes in 1924. Bull Volcanol 1:193–208 Commemorating the 50th Anniversary of Mt. Showa-Shinzan, Taylor GAM (1958) The 1951 eruption of Mount Lamington, Papua. Hokkaido, Japan Aust Bur Miner Resour Geol Geophys Bull 38:1–117 Pounder C (1980) Volcanic lightning. Weather 35:357–360 Textor C, Graf HF, Herzog M, Oberhuber JM (2003) Injection of Power JA, Nye CJ, Coombs ML, Wessels RL, Cervelli PF, Dehn J, gases into the stratosphere by explosive volcanic eruptions. J Wallace KL, Freymueller JT, Doukas MP (2006) The reawaken- Geophys Res 108:4606. doi:10.1029/2002JD002987 ing of Alaska’s Augustine volcano. Eos Trans Am Geophys Textor C, Graf HF, Herzog M, Oberhuber JM, Rose WI, Ernst GGJ Union 87(37). doi:10.1029/2006EO370002 (2005a) Volcanic particle aggregation in explosive eruption Bull Volcanol

columns. Part I: Parameterization of the microphysics of hydro- Volcanes de Chile (1991) (tourist book with photos) meteors and ash. J Volcanol Geotherm Res 150:359–377 Volcano Quarterly, Tanaka J (ed) (1995) The village square of Textor C, Graf HF, Herzog M, Oberhuber JM, Rose WI, Ernst GGJ volcanodom. Issaquah, Washington (2005b) Volcanic particle aggregation in explosive eruption Wallace P, Anderson A (2000) Volatiles in magmas. In: Sigurdsson H, columns. Part II: Numerical experiments. J Volcanol Geotherm Houghton B, McNutt SR, Rymer H, Stix J (eds) Encyclopedia of Res 150:378–394 volcanoes. Academic, San Diego, pp 149–170 Thomas RJ, Krehbiel PR, Rison W, Hunyady SJ, Winn WP, Hamlin T, Ward PL, Endo ET, Harlow DH, Allen R, Marquez D, Eaton JP Harlin J (2004) Accuracy of the lightning mapping array. J (1974) Development and evaluation of a prototype global Geophys Res 109. doi:10.1029/2004JD004549 volcano surveillance system utilizing the ERTS-1 satellite data Thomas RJ, Krehbiel PR, Rison W, Aulich G, Edens H, McNutt SR, collection system. US Geol Surv Open file Rep 74–124:154 Tytgat G, E (2007) Electrical activity during the 2006 Wilcox RE (1959) Some effects of recent volcanic ash falls with Mount St. Augustine volcanic eruptions. Science 315:1097 especial reference to Alaska. US Geol Surv Bull 1028-N:409– Thomas RJ, McNutt SR, Krehbiel P, Rison W, Aulich G, Edens H, 476 Tytgat G, Clark E (2010) Lightning and electrical activity during Williams ER (1985) Large-scale charge separation in thunderclouds. J the eruptions of Augustine volcano. US Geol Surv Prof Pap 1769 Geophys Res 90:6013–6025 Thompson D (2000) Volcano cowboys—the rocky evolution of a dangerous Williams ER (1992) The Schumann resonance: a global tropical science. Thomas Dunne Books, St Martin’s Press, New York thermometer. Science 256:1184–1187 Thompson G, McNutt SR, Tytgat G (2002) Three distinct regimes of Williams ER (1995) Meteorological aspects of thunderstorms, Chapter volcanic tremor associated with eruptions of Shishaldin volcano, 3 in CRC Handbook of Atmospheric Electrodynamics, 27–60. Alaska, April 1999. Bull Volcanol. doi:10.1007/s00445-002-0228-z CRC Press, Boca Raton Thorarinsson S (1966) Surtsey: the new island in the North Atlantic. Williams ER (2005) Lightning and : a review. Atmos Res Almenna Bokateloged, Reykjavik 76:272–287 Tillard S (1812) A narrative of the eruption of a volcano in the sea off Williams ER, McNutt SR (2005) Total water contents in volcanic the island of St. Michael. Philos Trans R Soc Lond 102:152–158 eruption clouds and implications for electrification and lightning. Uman MA (1987) The lightning discharge. International Geophysical In: Pontikis C (ed) Recent progress in lightning physics, Chapter Series, 39. Academic Press, Orlando 6. Research Signpost Publishing, Kerala, India, pp 81–93 Viemeister PE (1972) The lightning book. MIT Press, Cambridge Williams E, Nathou N, Hicks E, Pontikis C, Russell B, Miller M, Viramonte JG, Ubeda E, Martinez M (1971) The 1971 eruption of Bartholomew MJ (2009) The electrification of dust-lofting gust Cerro Negro. Geol Service of Nicaragua, Managua, Nicaragua fronts (‘’) in the Sahel. Atmos Res 91:292–298