Frequency of eruptions at New Zealand volcanoes

This paper reviews the nature and history of activity and the extent of risk at 1 4 volcanoes and volcanic centres in New Zealand and the Kermadec Islands. Mean intervals between eruptions are calculated, or estimated by extrapolation, for eight classes of eruption, represented by order of magnitude volume increases from lO^m 3 to 1 0 l x m 3 (100 km ) Expected property losses in eruptions, divided by the approximate mean intervals, allow risk to be apportioned on an annual basis. In real terms the rhyolite volcanoes, between Kawerau/Lake Rotorua and the southern end of Lake Taupo, are easily the most destructive. Annually apportioned, however, the risk is highest for an eruption of about 10 m 3 at Mt Egmont. Cumulative volumes erupted with time are estimated for most of the volcanoes and, where possible, average rates of magma accumulation and subsequent eruption have been estimated. This enables any shortfall between the actual volumes erupted, and the expected volumes, to be estimated, thus giving a measure of eruption potential at the present time. This varies for different volcanoes, from about 0.04 km 3 up to several hundred cubic kilometres. The time elapsed since the last eruption, divided by the mean frequency for that class of eruption, gives an idea of the likelihood of further activity, although the usefulness of the results is limited by large standard deviations. In the short term, less than 100 years, an eruption of 1 0 7 m 3 at Mt Egmont again emerges as the most likely damaging event. In the medium term, of the order of a few hundred years, an eruption of c.1 km 3 in the OkatainaRotorua area, or in the district between Lake Taupo and Rotorua, becomes probable. The data on which the conclusions are based, together with the mean intervals accepted, and the times elapsed since the last eruptions, are given in Appendices, so that the nature of the facts, and hence a wide perspective on volcanic activity in New Zealand, can be the better appreciated. The picture is one of volcanoes dormant for long periods of time, with great destructive potential, any of which could awaken at any time. INTRODUCTION The frequency of eruptions, whether estimated on the basis of return period for a given size, or on the basis of annual frequency, is an essential parameter in calculating the risk for a given locality, for engineering or for any other purposes. Risk is exposure to hazard of something vulnerable and valuable. The "something" can be people, animals, property, or any asset of economic or aesthetic value and, in attempts to formulate risk mathematically, it is usually referred to . simply as value. Thus Risk has been Scientist, Geophysics Division, DSIR, Wellington. B U L L E T I N O F T H E NEW Z E A L A N D N A T I O N A L S O C I E T Y F O R defined (1) as Value x Vulnerability x Hazard. The use of different terminology can be confusing: for example, a slightly different definition (2) is Risk = Exposure x Location x Vulnerability x Hazards, where Exposure means Value, and Location, which is implicit in the term Vulnerability in (1 ) , is given prominence as a term on its own. At volcanoes, hazard is usually potential rather than actual because most volcanoes are dormant for 99 percent or so of their lifetimes, although many emit volcanic gases more or less continuously, and the resulting acidic and corrosive plumes can pose hazards downwind. The frequency of eruptions is therefore the parameter that defines, usually in a very approximate way, the transition time from potential to actual hazards at a volcanoe. Many attempts have been made to analyse periodicity of eruptions statistically. Landmark studies in this field are those by Wickman (3-8), and a partiE A R T H Q U A K E E N G I N E E R I N G , V O L . 18, N O . 1, M A R C H 1985


INTRODUCTION
The frequency of eruptions, whether estimated on the basis of return period for a given size, or on the basis of annual frequency, is an essential parameter in calculating the risk for a given locality, for engineering or for any other purposes.
Risk is exposure to hazard of something vulnerable and valuable. The "something" can be people, animals, property, or any asset of economic or aesthetic value and, in attempts to formulate risk mathematically, it is usually referred to . simply as value.
Thus Risk has been Scientist, Geophysics Division, DSIR, Wellington.
BULLETIN OF THE NEW ZEALAND NATIONAL SOCIETY FOR defined (1) as Value x Vulnerability x Hazard.
The use of different terminology can be confusing: for example, a slightly different definition (2) is Risk = Exposure x Location x Vulnerability x Hazards, where Exposure means Value, and Location, which is implicit in the term Vulnerability in (1 ) , is given prominence as a term on its own. At volcanoes, hazard is usually potential rather than actual because most volcanoes are dormant for 99 percent or so of their lifetimes, although many emit volcanic gases more or less continuously, and the resulting acidic and corrosive plumes can pose hazards downwind.
The frequency of eruptions is therefore the parameter that defines, usually in a very approximate way, the transition time from potential to actual hazards at a volcanoe.
Many attempts have been made to analyse periodicity of eruptions statistically.
Landmark studies in this field are those by Wickman (3-8), and a parti-EARTHQUAKE ENGINEERING, VOL. 18, NO. 1, MARCH 1985 cularly useful summary has recently been given by Scandone (9). The present paper does not attempt so thorough an analysis of the problem but reviews the data presently available on frequency of eruptions in New Zealand and its surrounding area, which is a first stage in the construction of meaningful risk maps.
Too little is yet known of the time dependence of some major types of activity for reliable risk maps to be drawn.
In particular, much more information is needed on the mechanism of the largest eruptions, and on the way in which these depend on tectonic events, and on each other.
At present it seems extremely likely that there is a strong interdependence in volcanic activity in a single tectonic zone, and that this is related to major tectonic episodes In the context of New Zealand, it is likely, but not yet proven, that major volcanic activity occurs in episodes related to changes in the pattern of tectonic stress due, perhaps, to variations in the rate at which the downgoing Pacific plate is underthrusting the overriding Indo-Australian plate.
At present this rate cannot be directly measured.
Major uncertainties in the volcanic history of New Zealand are due to burial of the older volcanic deposits by younger ones, to their removal by erosion, and to uncertainties in the dating process. Carbon-14 ( 11+ C) dates can only be obtained on material less than about 60,000 years old, and there are considerable inaccuracies associated with this and other methods of rock-dating.
Because of these limitations it is only feasible to study a comparatively small part of each volcano's history. Ruapehu, for example, has been active off and on for at least the last 250,000 years, but for eruptions as large as those of 1945, which were certainly significant in terms of destructive ability, it is only possible to go back 100 years or so.
Only for the very largest eruptions or collapses of the cone, represented by the thick lahar (mudflow) deposits around the mountain, is it possible that a complete sample has been preserved for the whole 250,000 years of activity, and even this is doubtful.

Contrary
to widely-held opinion, volcanoes that have not erupted for hundreds, thousands, or even tens of thousands of years, are unlikely to be extinct, Volcanoes typically erupt infrequently, sometimes at regular intervals, but more often apparently randomly.
These short episodes are separated by long periods during which molten rock (magma) rises once more into the upper crust and gas pressure builds up beneath the sealed-off vent.
A new eruption occurs when the gas pressure at the top of the magma column exceeds the strength of the overlying rocks.
This may happen when a fresh batch of magma is forced up to a shallow depth under the volcano, or when tectonic movements reduce the confining pressure on the magma allowing the dissolved gases that it contains to come rapidly out of solution, or when water comes into sudden contact with hot rock or gas causing steam explosions.
In all cases, the fundamental driving force of eruptions is gas, and the dramatic expansion that accompanies its exsolution process. Dormancy is therefore the normal state of a volcano, and no volcano can be considered extinct if the time that has elapsed since the last eruption is less than, or of the same order as the average interval between eruptions, even when this is measured in thousands or tens of thousands of years (10).
Only when the elapsed time is of the order of ten times the average interval between eruptions and, coupled with this, when there have been significant changes in the tectonic stress field which caused the volcano to form where it did in the first place, can it be reasonably considered that the volcano may indeed be extinct.
On this basis the following volcanoes and volcanic fields in the Kermadec Islands, New Zealand and the outlying islands to the south are dormant rather than extinct (see figure 1) Insufficient is known about activity in the Kermadec Islands, other than at Raoul, or at the Rumble submarine volcanoes, Timaru, Solander Island, and the Antipodes Islands, and therefore these are not considered further in this paper. The volcanic history at the remaining 14 volcanoes and volcanic fields is discussed below, and estimates of the mean frequency of eruptions are given for successive order of magnitude increases in the volume of rock ejected, from a lower limit of 10" m 3 up to 10 11 m 3 (100 km 3 ).

EFFECTS OF ERUPTIONS
The main factors that determine the effects of an eruption are the volume of material ejected and the rate at which this takes place.
These in turn depend partly on the chemical composition of the magma (magma is molten rock containing gases in solution), which in the New Zealand area ranges from basalt, through andesite and dacite, to rhyolite. Basalt has low viscosity and flows relatively easily: gases which come out of solution in the rising magma, as the confining pressure is decreased, are therefore able to escape readily and fail to build up the potential for large explosions. Increasing silica content in the magma leads to greatly increased viscosity. Andesite is much more viscous than basalt, and dacite than andesite, and consequently the explosive potential of dacite is very large.
The eruptions of Krakatoa in 1883 were of this type, as were those of Mont Pelee which destroyed the town of St Pierre in Martinique in 1902, Mt Lamington in Papua New Guinea in 1 951 , and Mount St Helens in 1980. Rhyolite, with greater silica content than dacite, and therefore still higher viscosity, has the highest explosive potential of all magmas.
The sudden interaction of water with magma, hot gas, or hot rocks surrounding a magmatic intrusion introduces a further complication.
Water may flash to steam, expanding many times in the process. This is the mechanism for hydrothermal, or socalled phreatic eruptions. Often these trigger underlying magmatic eruptions. Because of phreatic activity, even basaltic magmas can erupt explosively, as occurred in 1886 at Tarawera.
The explosive potential of magma, whether due to internal gas pressure or to rapid interaction with water, determines how the rock is erupted.
There are two principal types of erupted material, lava and pyroclastics. Lava is magma from which most of the gas has escaped: it flows relatively quietly or is extruded in the form of plugs or domes.
Pyroclastics, on the other hand, represent magma which has been blown apart explosively, forming a wide range of erupted debris, from large blocks and bombs, through pumice and lapilli, down to finegrained ash and dust.
Pyroclastic rock may be blasted to a great height above the volcano and be carried by the wind, finally dropping as airfall "tephra", or it may be blown sideways, often as a result of gravitational collapse of eruption columns, and be emplaced as dense ground-hugging pyroclastic flows.
These are turbulent, because of continual explosive gas release from the blocks and fragments of magma contained in the flow, and because of reactions between the gases and air trapped and entrained by its very rapid movement. Pyroclastic flows usually travel between 20 and 100 metres per second: when they are sufficiently hot and their volumes are large enough, the molten fragments of rock fuse together on coming to rest, forming welded ignimbites.
When these cool, they resemble dense concrete-like sheets which usually grade upwards into chaotic masses of unwelded soft pumice with scattered boulders Small pyroclastic flows can be formed by andesite, or even basalt eruptions.
The larger and more destructive ones are dacitic.
Ignimbrites, whether welded or unwelded, are nearly always rhyolite. Primary destructive effects of volcanic eruptions, apart from lava flows and gravity-controlled pyroclastic flows, and the far larger, totally devastating ignimbrites, are due mainly to airfall tephra, and at short distances from the volcano to the impact of rocks on ballistic trajectories.
Gas emission and acid rain may also prove very destructive. Table 1 shows approximate areas likely to be affected by different thicknesses of airfall tephra in eruptions of various sizes: windless conditions are assumed (that is, thickness contours are assumed to be circular about the volcano).

Secondary
destructive effects of eruptions are due, above all, to lahars (mudflows) and tsunamis.
Lahars may be generated in a variety of ways, by heavy rainfall washing unconsolidated tephra off steep slopes, by melting of snow and ice, by lava or pyroclastic flows entering streams, or by the direct ejection of water from crater lakes.
Tsunamis (socalled tidal waves) also form in many ways:± the largest, such as those at Krakatoa in 1883, by the impact of ignimbrites into the sea or large lakes (11). More detailed discussion of the effects of volcanic eruptions is outside the scope of this paper.
An excellent sourcebook on the subject has recently been published (12).

Kermadec Group -Raoul Island
Raoul is of interest from the point of view of volcanic risk only because there is a permanently manned meteorological station on the island.
The volcano is frequently active and last erupted in a very minor way, in 1964. The island, which has an area of only about 30 km 2 , is the summit of a large submerged volcano and includes at least two calderas* of about 3.5 kilometres in diameter.
The volcanic history of Raoul has been reconstructed back to about 3700 years ago (13). There are much older volcanic rocks exposed on the island, dating back to about 1.4 million years, but their relationships to individual eruptions are not clear. Offshore, about 240 kilometres to the southeast, volcanic ash has been found in deep-sea cores (14) showing that there were substantial eruptions (3-5 km 3 ?) about 30,000 years and 130,000 years ago. However, the ash is lacking in cores further north, closer to Raoul, and it is perhaps most likely that these eruptions Calderas are large, more or  less  circular areas of subsidence formed  after  large  explosive  eruptions,  when  the  roof  above  a  rapidlydrained  magma  chamber  collapses.  originated  from  the  Macauley  Island-Brimstone Island volcanic centre, further  south in the Kermadecs, about which very  little is known. During the last 3700 years or so, it is estimated that there have been 1 2 explosive eruptions at Raoul which have ejected more than about 10 8 m 3 (figure 2).
Volumes are very hard to estimate, and are rough approximates only, because the small size of the island means that most of the erupted debris fell offshore. Four of these eruptions probably ejected more than 10 9 m 3 (1 km 3 }, and one, the Fleetwood eruption about 2200 years ago, perhaps more than 10 km 3 .
Any eruption over about 10 8 m 3 would be destructive at the meteorological station on Raoul.
The eruptions are listed in Appendix 1, together with an estimate of their volumes.
The average interval between eruptions of > 10 8 m 3 works out as 327 ± 302 years, and that between eruptions of > 10 9 m 3 as 777 ± 182 years.
These means define a line which is shown extrapolated in figure 2a.
However, because the single large eruption about 2200 years ago dominates the volumes erupted during the period studied, a better estimate of the volume of eruptions versus time can be found from figure 2b.
The true rate probably lies between the upper and lower limits shown, that is, between about 138 and 670 years for a cumulative volume of 1 km 3 .
At present the shortfall between the expected and actual volumes erupted lies in the range 1.5 km 3 to 14 km 3 , and we are therefore in line, in a long-term sense, for an eruption in this range of magnitudes.
These figures, like all other average intervals and rates of eruption given in this paper, can only be taken as a rough guide to the frequency of eruptions. Where the standard deviations are approximately the same as the mean intervals, the eruptions are essentially random in their occurrence. Furthermore, the accuracy of rock-dating and estimation of volumes erupted is low.

Bay of Islands -Kaikohe Volcanic Field
This is a large area in Northland covering some 500 km 2 , in which eruptions, predominantly of basalt, have occurred sporadically for at least the last 2 million years. Very little is known in detail of the eruptions, and there are few reliable dates. The most recent activity took place at the Te Puke cones near Waitangi and may be about 1 7,000 years old, although this date is doubtful (15). A very rough estimate of 67,000 years for the mean interval between typical eruptions (of the order of 10 8 m 3 ) has been worked out from the fact that there have been about 19 such eruptions in 1.27 million years. Figure 3 shows an extrapolation of this interval by making the trend parallel to that derived for the very similar Auckland volcanic field.
Part of the area lies beneath the sea, and moderately explosive eruptions have occurred, and may do so again, as a result of interaction of water with the basaltic magma. Most eruptions, however, have been comparatively mild. Small volumes of rhyolite have been produced, notably at Putahi near Ngawha, by chemical differentiation of the basalt.
This process is capable of yielding only a small amount of rhyolite, quite different to the situation in the Taupo-Rotorua area where huge volumes of rhyolite are produced by large-scale melting of the upper crust.
Eruptions in this area are unlikely to exceed 10 9 m 3 (1 km 3 ). Some damage to buildings with flat roofs might occur when tephra thickness reaches 50 cm, especially when rain saturates the deposits. Table 1 shows that this thickness would occur within a radius of about 7 km from the vent, twice this distance perhaps with the effect of a strong wind. There are few large towns in the area, and the probability of an eruption close enough to damage Kaikohe, for example, is remote.
It is, however, unlikely that this volcanic field is extinct; further eruptions, but at very infrequent intervals, may be expected.
Small hydrothermal (phreatic) eruptions in the Ngawha geothermal field also take place occasionally and these could be destructive at short distances.

Whangarei Volcanic Field
Activity in this area, which includes the city of Whangarei and covers in all some 350 km 2 , is very similar to that in the Bay of Islands-Kaikohe district.
The most recent activity, however, seems to have occurred about 34,000 years ago, in the Kamo area.
This is twice as old as the supposed date for the latest Bay of Islands eruption, but is still a short time ago compared to the very long average interval between eruptions. This, deduced from the fact that there have been approximately 14 such eruptions (of c.10 8 m 3 ) in the past 2.3 million years, is about 166,000 years (15); see figure 3 where the trend centred on this interval is again drawn parallel to Auckland.
Like the Bay of Islands-Kaikohe field, the largest eruption to be expected will probably be about 1 km 3 .
The potential for damage is, however, greater than in the former area, since an eruption could well occur in or near Whangarei City.
The predominantly basaltic eruptions will be explosive, as in the past, if water is involved: most, however, will proceed comparatively quietly with the emission of lava flows.

Auckland Volcanic Field
It is unfortunate that the largest city in the country corresponds so closely to the extent of this volcanic field. Older volcanic cones lie to the south, in the area of the Bombay Hills, and are probably extinct. The Auckland field itself covers about 650 km 2 , and extends from north of Takapuna to near Manurewa (16, 17).
Like Bay of Islands-Kaikohe and Whangarei, activity is basaltic, more completely so in fact than either of the other two districts, and has produced more lava than tephra.
The presence of the sea over much of the field, however, means that many, of the past eruptions have been moderately explosive, and similar activity is expected in the future.
Although the largest single eruption ever likely to occur in the Auckland field will be of the order of 1 km 3 , as in the Bay of Islands and Whangarei districts, the potential hazards are very much greater than in either of these areas, because the field is built up and densely inhabited except for that part of it which lies beneath the sea.
Eruptions in shallow water will be explosive, although the comparatively small volumes of magma involved will limit the destructive effects (Table  1).
Small tsunamis may well be generated, and these are also likely to cause damage.
In assessing the frequency of eruptions in the Auckland field, a major difficulty has been the fact that many centres have erupted, each usually in a single cycle of activity, and usually with only a small volume of tephra.
Accordingly,, the products of different centres seldom overlap, and their relative ages are therefore hard to assess.
Also, few absolute ages have been obtained by lif C or other means, for the Auckland rocks. As in the Bay of Islands-Kaikohe and Whangarei fields, most cones in the Auckland field have been active only in a single eruptive cycle, which may last perhaps a few years to several tens of years, after which they become extinct. Rangitoto may be an exception.
This centre has produced by far the greatest volume of lava and ejecta in the Auckland area (Appendix 1 : figure  3) .
Although the evidence is sparse, most eruptions seem to take place in cycles of activity lasting for a few hundred to perhaps 1 500 years,-during which many centres may be active (17). Rangitoto last erupted only about 250 years ago, and began erupting some 500 years before that.
Although there has been no activity there in historic times, this short interlude may represent only a pause in the "present" or "Rangitoto" cycle of activity. It is clear from figure 3b that if a resurgence of activity were to take place today it would be grouped in with this "present" cycle which began about 750 years ago. As the volume of lava extruded to form Rangitoto has already been so much greater than in previous cycles of activity, any continuance of the present cycle is unlikely to involve eruptions bigger than about 10 8 m 3 9 In figure 3a the mean intervals between eruptions, determined from the groupings and estimated ages given in Appendix 1 , are given for events yielding >10 8 in 3 and >10 9 m 3 , and are shown extrapolated for smaller eruptions.
Note that in this context "eruptions" means "eruptive cycles": there is no evidence to indicate the periodicity of eruptions within a single cycle.
Hence little can be said of the probability of the next (comparatively minor?) eruption in the present "Rangitoto" cycle, beyond saying that an eruption up to the order of 10 8 m 3 is entirely likely, more so in fact than it would be between eruptive cycles, and that it will probably take place at or near Rangitoto rather than in some totally different part of the Auckland volcanic field.

Mayor Island
Mayor Island, with an area of about 15 km 2 , is the summit of a submerged rhyolitic volcano in the western Bay of Plenty. It is surmounted by a caldera about 3 km in diameter.
A long period of submarine eruptions had built up the cone to above sea level by about 42,000 years ago, when the products of a major eruption in the Okataina volcanic centre (see below) were deposited on the island.
Since then it is estimated that there have been 11 eruptive episodes of >c*10 7 m 3 , of which three have exceeded about 10 9 m 3 (19, 19): from this, approximate mean intervals between eruptions can be calculated, and are shown in figure 4a.
The eruptions are listed with their estimated dates in Appendix 1 , and are also shown in figure 4b, where the smaller eruptions (^10 8 m 3 ) are arrowed, It is clear that activity has been episodic, with a period of intermittent voluminous eruptions between about 10,000 and 6000 years ago.
How this period relates to previous episodes at the volcano, and hence to future ones, is not clear. More work is required, and it would be especially desirable to obtain some marine cores downwind of the island: from this it is likely that a reliable chronology of eruptions could be built up.
Although Mayor Island has no permanent inhabitants, it represents a high degree of volcanic risk.
Past eruptions have been entirely of rhyolite, and hence very explosive * Large pyroclastic flows impacting into the sea commonly give rise to tsunamis (11), and these could devastate a large part of the Bay of Plenty and Coromandel coastline. Furthermore, Auckland lies downwind of the island, insofar as the upper altitude winds are concerned (low altitude winds blow in the opposite direction, from west to east), and could receive a significant amount of ashfall in a very large eruption.
It is not known whether Mayor Island has the potential for eruptions as large as 100 km 3 , but any rhyolitic volcano should be suspected of having such a potential unless evidence can be found to the contrary.

White Island
White Island, in the eastern Bay of Plenty, is an andesitic to dacitic volcano almost entirely submerged.
A large crater open to the east lies only just above sea level, and smaller craters within this overall depression extend down well below sea level.
The island is small, no more than about 3 km 2 .
Its highest peak, 321 m, directly overlooks the crater and has a very steep headwall composed of rock much weakened by fumarolic alteration.
A system of arcuate faults, parallel to the crater tim, passes behind the peak on the side away from the crater.
This very unstable situation leads to massive rock-60 falls from the inner crater walls. One such, in 1914, blocked vents on the crater floor, causing gas pressure to build up. Resulting explosions mobilised debris and generated a lahar (mudflow) which overwhelmed the sulphur works and killed everyone on the island.
Similar events are likely in the future.
Nowadays there are no residents on White Island and, except in very large eruptions, risk is limited to occasional visits by scientists, tourists, fishermen or muttonbirders.
Large eruptions are, however, possible, up to a maximum of perhaps 1 km 3 . In such an event, subsidence or large explosions beneath the crater floor could conceivably allow seawater access to the present deep vents. If this were to happen suddenly, tsunamis might be generated which would be destructive on the Bay of Plenty coast.
However, calculations suggest that a large volume of water would need to come into very rapid contact with the near surface magma beneath the vents for this to happen, and that this would be most unlikely to occur (20). The risk of tsunamis from White Island is therefore considered to be much less than at Mayor Isla.id.
The volcanic history of White Island has been elucidated back to about 15,500 years before the present, by comparing the record of eruptions on the island (21) with marine cores obtained within a distance of about 70 km (22).
There have been seven eruptions during this period which have deposited a significant amount of ash in the marine cores: these, at a guess, have each ejected more than about 10 7 m 3 .
The latest eruption, which began in December 1976 and ended in January 1981, may rank with these seven previous events in terms of volume ejected.
Two eruptions about 4200 and 9000 years ago (22) were substantially larger (^10 8 m 3 ?), and are tentatively correlated with the eruptions which formed the west and east parts, respectively, of the main crater depression at White Island (21). It is likely that eruptions of this size gave rise to minor ashfall on the mainland, and small to moderate tsunamis may have been generated.
Eruptions are listed in Appendix 1 , and are shown on figure 5. From the relationship in figure 5b it is clear that there is a regularity in the volume produced with time.
This suggests, in spite of the latest eruption in 1 976, that the volcano currently has a potential for eruption of something like 4 x 10 7 m 3 . Estimates of volume are, however, very sketchy, and are certainly no better than rough order of magnitude approximations.

Okataina Volcanic Centre
The Okataina volcanic centre is a complex structure of overlapping calderas and associated rift zones, lying between Lakes Okataina, Rotoiti, Rotoma, Rerewhakaaitu, Rotomahana and Tarawera, in the area between Rotorua and Kawerau.
It includes the two volcanoes Tarawera and Haroharo, and is dominantly a rhyolitic centre, although small volumes of basalt have been erupted there also, as in 1 886 in the Tarawera eruption.
The volcanic history of the centre has been closely studied back to about 42,000 years ago, the time marked by a large eruption, the Rotoiti Breccia and associated Rotoehu Ash (see Appendix 1 ) in the north of the area. For the largest eruptions (£c.150 km 3 ), the chronology has been established back to about 230,000 years, the date of the first eruption which can be traced definitely to the Okataina centre (23).
Very large eruptions, with volumes of the order of 1 50-300 km 3 were erupted from the centre between 230,000 and 150,000 years ago, as welded ignimbrites.
These dominate the record of eruptions (figure 6b).
Since 150,000 y BP, activity has greatly declined, although these have been two large unwelded ignimbrites, at about 42,000 and 33 , 000 years ago.
All other eruptions during this period have been of no more than about 20 km 3 . This is still a large figure (the great Krakatoa eruption of 1883 ejected about 18 km 3 ), but it is insignificant compared to former eruptions. Note that the largest historic eruption in New Zealand, the Tarawera eruption of 1886, is barely visible on the scale of figure 6b. This event, which had a traumatic effect on the Rotorua district, killing over 150 people, was similar in size to the eruption of Mount St Helens, Washington, USA in 1980. This serves to highlight the tremendous potential for destruction that exists here and elsewhere in New Zealand.
A major difficulty in assessing the probability of future eruptions, which is common to all centres that erupt rhyolitic ignimbrites, arises from the very long intervals which typically separate the largest eruptions.
In order to project a reliable trend forward, a long history of past activity must be known. However, the older deposits are often eroded and are usually overlain by younger ones, so that realising this aim is almost always very difficult. Furthermore, the rate at which eruptions have taken place, the volume/time relationship illustrated in figure 6b, has not remained constant. Overall trend lines showing this relationship can be fitted to groups of eruptions, as has been done in figure 6b. From this, it is clear that the rates fluctuate considerably.
At present it seems likely that a shortfall of some 50 km 3 exists at Okataina between the volume actually erupted and the volume predicted by the extrapolation of the trend given by eruptions over the past 1 5,000 years or so.
Average intervals calculated from all known eruptions at Okataina of ^75 km 3 , and for all smaller eruptions over the past 42,000 years, are plotted in figure  6a.
This clearly shows the existence of two quite different trends, and there is the hint of a third, for the very largest eruptions.
This suggests that different mechanisms operate for the eruptions above and. below about 9 km 3 in volume.
Eruptions like that of Tarawera in 1 886, in the volume range 1-5 km 3 , would be destructive close to the volcano.
Those of 10 km 3 or so would cause damage at Kawerau and Rotorua: large lahars would probably form in the river valleys, and pyroclastic flows would travel considerable distances.
Eruptions of the order of 100 km 3 would be destructive over a large part of the North Island.
Volcanic risk is potentially, therefore, very great at Okataina, as it is at all the rhyolite volcanoes in the country.
Eruptions at Okataina tend to occur along two rift zones.
One is the wellknown Waimangu-Rotomahana-Tarawera rift in the south of the volcanic centre. This can be traced further northeast as a zone of weakness at least as far as Kawerau and Mt Edgecumbe. Mt Edgecumbe represents the extrusion of a large volume of degassed magma some 8000 to 1 0,000 years ago, probably related to one of the tephra eruptions in the Okataina centre proper (see Appendix 1). The other rift system parallels the Tarawera rift to the north, and includes the dormant Haroharo volcano and several subsidiary centres between Lake Okareka and Lake Rotoma.
In addition it has been suggested that there is a substantial magma body, of the order of 300 km 3 dense rock volume, near the eastern end of Lake Tarawera (24), between the two rift zones. If confirmed, it is likely that future large eruptions will take place in this area. Figure 7 shows how volcanic activity in the Okataina centre has been split between the two rift systems. The volumes erupted on the Tarawera rift have been far greater than at Haroharo because of the three large ignimbrites erupted before about 150,000 years ago.
The rate of eruptions (figure 7b) has fluctuated greatly, but from the average intervals (figure 7a) there is no evidence to suggest anything but an overall straight line relationship. Volumes erupted at Haroharo have been greater during the past 50,000 years than at Tarawera and the rates have also fluctuated greatly. The average intervals between eruptions on the Haroharo rift system (figure 7a) show the same two trends that can be seen for Okataina as a whole (figure 6a). At present the shortfall at Tarawera seems to be of the order of 40 km 3 , and at Haroharo about 80 km 3 , although there is some doubt about how the trend lines should be drawn.
Numerous very large ignimbrites were erupted from the central volcanic region before 250,000 years ago.
Many have not yet been traced to their origin, and it is possible that some were erupted from the Okataina volcanic centre. These ignimbrites of unknown origin are discussed below under the Maroa volcanic centre and are illustrated on figure 9 (see also Appendix 1 ).

Rotorua Caldera
Rotorua caldera was probably formed by subsidence after the eruption of the Mamaku ignimbrite about 140,000 years ago (the youngest of the major welded ignimbrites).
Previous activity, if any, is unknown and subsequent eruptions (see figure  6b) have been insignificant in volumetric terms.
The lava dome of Ngongotaha (c.10 km 3 ?) was extruded (23) at some unknown date between the Mamaku ignimbrite eruption and the Rotoiti eruption from Okataina about 42,000 years ago. There was at least one lesser eruption (of the order of 1 km 3 ) between 42,000 and 13,800 years ago.
These rough dates allow a crude approximation of average intervals to be made, and suggest a return period of the order of 70,000 years for eruptions £ 10 km 3 , and about 47,000 years for eruptions of £1 km 3 . This trend is shown in figure 6a.
Rotorua caldera may be a "one shot" caldera in which no further ignimbrite eruption of the scale of the Mamaku ignimbrite will occur again.
However it is the youngest of all the rhyolite centres except Mayor Island, and it is equally possible that a series of major rhyolite eruptions will take place there in the future.
Volcanic risk is high because of the presence of Rotorua City, and damage would probably be severe in any eruption of 1 km 3 or above.

Maroa Volcanic Centre
The Maroa volcanic centre as defined (25) lies south of the Waikato River between Orakeikorako and Atiamuri* It is a cluster of rhyolite domes within an area about 15 km in diameter, lying within a larger, roughly circular" area, slightly elongated in the east-west direction, which is about 45 km in maximum diameter. This larger area, or outer ring structure, is defined by rhyolite domes near Whakamaru and Mokai in the west, -Oruanui and Kaimanawa in the south and Horohoro and the Waiotapu area in the north; to the east it is bounded by the Kaingaroa Fault. Dacite domes, as well as rhyolite ones, are found near Waiotapu, and minor basalts are exposed near Whakamaru.
The whole area is one of subsidence, cut by northeast trending faults which separate uplifted blocks such as Haroharo and Paeroa from sinking grabens such as the Guthrie, Ngakuru and Waikite valleys. For the purposes of the present discussion the Maroa volcanic centre is taken to include the outer ring structure.
It is bounded by the Rotorua caldera and the Okataina volcanic centre to the north and northeast, and by the Taupo volcanic centre to the south.
Very large welded ignimbrites have been erupted intermittently from the Maroa volcanic centre, possibly from near Mangakino, over a period of at least 900,000 years, from c.1.05 million y to 150,000 y BP. These have been extensively studied (2 6-32) and correlated to some extent with tephra found in marine cores at distances of up to 2000 km from the North Island (14, 33).
There is still considerable doubt both as to the source and the number of individual ignimbrite eruptions, and there are anomalies in the dates which have been found.
The volumes are huge and can only be approximately estimated. In spite of all these uncertainties, however, a broad picture of the volcanic activity at Maroa can be obtained (figure 8b).
Uncertainties, marked by the cumulative volumes shown as maximum and minimum in the figure, make it impossible to estimate whether the rate has remained constant up to about 1 50,000 years ago, or declined during the last 280,000 years. Either way, unless the centre is extinct, which is possible but unlikely, it would seem that a substantial shortfall exists, of at least 100 km 3 and possibly of as much as 700-900 km 3 , between the volume actually erupted and the volume expected on the basis of the time elapsed.
These estimates, however, are very approximate.
Very little is known of the chronology of smaller eruptions at Maroa.
A rough estimate of about 10,000 years for the average interval between eruptions of the order of 10 s m 3 (34) has been used in figure 8a to define the average volume versus time relationship. The trend marked A was used in a previous study (35). That marked B has been adjusted to give a better fit to the data presented here for the largest eruptions.
It is clear from the record preserved in the deep sea cores (33) that large eruptions have occurred intermittently in the North Island for at least the last 3.7 million years. Many of the older ones, before about 1 million years ago, probably originated in the Coromandel district, which was active before the Taupo volcanic zone.
The exact source of these ignimbrites is unknown, and it is only those that have taken place since c.1.05 million years ago that can be traced definitely to the Maroa or Mangakino centres or their surroundings (figure 8b and Appendix 1). In order to examine the overall way in which the rate of eruptions has changed with time, an estimate of volumes and dates has been plotted in figure 9 (see also Appendix 1 ) . All rhyolite eruptions are shown, including those from the Okataina, Taupo and Rotorua centres, as well as those which can be definitely traced to Maroa/Mangakino and those whose source is unknown.
The average intervals versus volumes are shown in figure 9a, and on figure 9b the cumulative volumes erupted, from which a steady increase in the rate of eruption can be clearly seen. This began about 1.5 million years ago and climaxed about 300,000 years ago. It has declined quite markedly over the last 200,000 years. These changes in rate are plotted separately on figure  10. Extrapolation of the trends into the future is dificult, but it Is clear that the rate remains high, probably of the order of 1 km 3 /300-500 years.
Any eruption of more than 1 km 3 or so in the Maroa area would cause damage to forests and farmland. The larger eruptions would be highly destructive. Lahars would probably be generated in the Waikato River and would cause extensive damage to dams, power plants and towns downstream. Taupo Volcanic Centre   The  Taupo  volcanic  centre,  which  abuts the Maroa centre to the north and  the andesitic  Tongariro  centre to the south, consists of a large area of subsidence, 30 to 35 km in diameter now largely covered by Lake Taupo, and probably includes several overlapping calderas.
It is cut by a number of prominent northeastsouthwest trending faults, along which many of the rhyolite intrusions are aligned.
There are minor basalt extrusions close to several of the faults and one andesite extrusion on the fault-bounded eastern margin.
Tauhara, east of Taupo borough, and a small dome further southwest, are composed of dacite. It is likely that at least three major welded ignimbrite eruptions originated in the Taupo volcanic centre between about 320,000 and 21 5,000 years ago.
If this interpretation is correct, the Taupo centre is somewhat older than Okataina (230,000 years?), but much younger than Mangakino/Maroa (1.1 million years?). Eruptions are listed with their dates and estimated volumes in Appendix 1, and average intervals and cumulative volumes are shown on figure 11. There is little evidence to suggest the current rate of eruption (figure 11b) because the record is entirely dominated by the gigantic eruptions about 320,000 years ago.
On the basis of activity during the last 40,000 years it has been suggested (40) that no large eruption will take place at Taupo for 1 0,000 years or so, although there could well be small eruptions of less than about 5 km 3 .
A similar calculation for Okataina suggested a shortfall at present of about 16 km 3 , compared with my estimate (see above) of about 50 km 3 .
Uncertainties in the dates and source vents of the earlier eruptions, and especially in the volumes emitted (see 40 for a thorough study of this problem) mean that estimates of future eruptions are inevitably impre-Tongariro Volcanic Centre The Tongariro volcanic centre includes the two large andesitic massifs of Tongariro and Ruapehu, and the smaller cone Ngauruhoe.
It also includes older andesite volcanoes which are now probably extinct, such as Tihia, Kakaramea and Pihanga.
The centre lies south of Lake Taupo and marks the southern extension of the Taupo volcanic zone.
The present active vents are aligned along a north northeast trending zone which continues beyond Tongariro, to the northeast of Lake Rotoaira where a number of small explosion craters have formed. Lake Rotopounamu fills the largest of these.
Eruptions are possible in the future anywhere along this zone, from a short distance southwest of Turangi to just south of Ohakune, where there are also explosion craters which formed comparatively recently. A description of the three presently active or potentially active volcanoes -Tongariro, Ngauruhoe and Ruapehu -follows.

Tongariro:
The oldest of the three is Tongariro.
This is a complex andesite volcano with many craters, some containing lakes, and some still fumarolic. Activity probably started here half a million years or so ago, and during its long life there have been several moderately large explo-. sive eruptions.
The largest which have been recognised, for example the Mangamate eruptions about 8000 years ago (41) probably amounted to 3-4 km 3 or so, and somewhat resembled the 1980 Mount St Helens eruption in scale.
The largest which is considered ever likely to occur at Tongariro would be of the order of 1 0 km 3 . Such an eruption would generate lahars and spread ash over a wide area, but damage would be limited as the area is not densely populated, and only Turangi, among the larger towns in the district, would be affected.
Power lines and communications would be cut, and widespread contamination of rivers and lakes by lahars and airfall tephra would cause damage and disrupt the Tongariro power scheme.
Little detailed work has been done on the past eruptions of Tongariro and in particular, little is known of the sequence of lahars that built up the extensive ring-plain around the volcano. In considering eruptions of ^ 1 km 3 , it is necessary to include Ngauruhoe with Tongariro as Ngauruhoe is really only a parasitic cone on the Tongariro structure. The marked activity associated with the birth of the present Ngauruhoe cone about 2500 years ago, and the earlier Mangamate eruption about 8000 years ago, suggests that a typical interval between eruptions of c.1 km 3 is about 5500 years. Small historic eruptions in the 19th century, between 1855 and 1896, and a possible small eruption in 1 926, define a rough periodicity of about 24 years for eruption of the order of 10 s m 3 (see Appendix 1). The trend given by these values is marked on figure 12a.
The cumulative volumes erupted over the past 12,000 years at Tongariro and Ngauruhoe are shown on figure 12b.
There is insufficient information to project a reliable trend forward as a basis for estimating future activity, but a shortfall of the order of 4,5 km 3 seems likely* Ngauruhoe: As mentioned above, Ngauruhoe is really only a parasitic feature on the flank of Tongariro. The present cone began to form only about 2500 years ago. A previous, much larger cone on or' near the same site is inferred from the remnants of glacial valleys to have existed at the end of the Ice Age, from perhaps 30,000 to around 10,000 years ago.
It is likely that this cone was destroyed in the Mangamate series of eruptions about 8000 years ago (see Tongariro above).
The 1  Apart from this r the risk arising from eruptions at Ngauruhoe is slight.

Ruapehu:
Like Tongariro, Ruapehu is a large complex andesitic volcano. 11 s summit plateau is made up of several extinct craters, and the present active crater contains a lake which is kept heated by constant fumarolic activity. The volume of the lake fluctuates somewhat but is generally close to 10 7 m 3 .
Because it lies at an altitude of about 2530 m and several major rivers rise on Ruapehu, the lake-filled crater dominates much of the surrounding district and represents a high degree of volcanic risk.
Eruptions have been going on intermittently at Ruapehu for at least the last 250,000 years, but the bulk of the mountain is less than 1 00,000 years old r and all the upper part formed less than 50,000 years ago (42 Only a very rough chronology can be pieced together for these major lahars (43) , which probably accompanied largescale collapses of parts of the cone. It is unclear whether all of them were accompanied by eruptions. What information is available is given in Appendix 1 and is plotted on figure 13b. Smaller lahars, which are still an order of magnitude larger than any that have happened in historic times, are recorded by deposits laid down by the Whangaehu River, which drains Crater Lake.
There are only two such deposits above the marker bed laid down by the Taupo pumice eruption about 1800 years ago, and they have been dated (44) at 756 ± 56 and 407 ± 70 years BP.
These are shown on figure 13c.
It is likely, if such lahars were to occur again, that they would represent the explosive ejection of all the water at present in Crater Lake, or the sudden catastrophic collapse of the southeast wall impounding the lake, with rapid release of all the water.
In the former case there would be devastating lahars on all sides of the mountain: these would affect not only the upper Whakapapa skifield but Iwikau and Whakapapa villages also (Top O 1 The Bruce and the Chateau). In the latter case all the water would be channelled down the Whangaehu River, and the resulting lahars would cut the main electricity transmission lines and the Desert Road, and would inundate Waiouru.
During historic times, smaller lahars (c.10 6 m 3 ), amounting to a^ tenth or so of the volume of Crater Lake, have several times caused damage on and around the mountain (45). Such events took place in 1 861, 1 895, 1 925, 1 969 and 1 975 and are shown on a reduced scale in figure 1 3d.
The mean interval between these is about 23 ± 11 years.
Numerous smaller lahars occur in the Whangaehu River at intervals of a few years.
These are not destructive.

Little
is known of the dates of tephra and lava eruptions at Ruapehu. There was a major eruption or series of eruptions which produced the Rangataua lava flows (;> c.10 9 m 3 :42) and the Okupata Tephra ( 1 0 8 -1 0 9 m 3 :41) between about 8000 and 13,000 years ago.
This eruptive episode almost certainly corresponds to the latest large lahar (figure 13b), the Murimotu lahar which laid down the large mounds near the bottom of the Chateau Road. During historic times there have been two occasions, in 1861 and 1 945, when lava was present at the surface in Crater Lake and notable eruptions took place. Many phreatic eruptions have been recorded throughout the historic period (45: see Appendix 1 ) . These have a mean periodicity of about 2.5 years.
Estimates of the mean intervals for eruptions are shown on figure 13a.
The largest likely eruption at Ruapehu would be of the order of 10 km 3 (about twice the 1 980 Mount St Helens eruption) and, as can be seen from the figure, the frequency of such an event is very low. The present shortfall, based on the record of past lahars, seems to be about 0.06 km 3 .

Dominating
Taranaki, the isolated andesitic volcano Mt Egmont is thought by many to be extinct but is certainly only dormant.
It last erupted in a comparatively minor way about 1755 AD. Somewhat larger eruptions took place about 100 years earlier.
Previous eruptions back to about 25,000 years have been intensively studied, and for the largest lahars a rough chronology has been worked out for the last 50,000 years (46, 47).
The mountain is surrounded by laharic ring-plains more extensive than those at Ruapehu or Tongariro.
These have also formed by large scale collapses (sector collapses) of the cone. The largest, which was probably triggered by a major explosive eruption, occurred about 23,000 years ago and attained a volume of 12 to 15 km 3 (see Pungarehu lahars. Appendix 1 ) , covering large areas at the western foot of the volcano.
Such lahars, although hopefully on a smaller scale, are likely to occur again. The present summit, formed by a lava dome intruded 300 years or so ago, is greatly oversteepened and liable to collapse.
Future eruptions would be very likely to cause this to happen.
The largest likely eruption at Mt Egmont would be of the order of 10 km 3 and would devastate a large part of Taranaki.
However, the periodicity inferred by extrapolating the data in figure 14a suggests that such an event occurs only once in 140,000 years or so. This is probably longer than the time elapsed since Mt Egmont began erupting (50-70,000 years ago?), and as one such event (the Pungarehu lahars) has already occurred it is unlikely that eruptions in the near future will attain this volume.
Smaller eruptions, of 1 km 3 or so, are estimated .to have a typical return period of the order of 12,000 years. Such events, and still smaller ones also, would generate lahars and cause damage over a substantial area. Towns such as Inglewood would be affected (48), and ashfall would affect grazing and water supplies over most of the region. Because of the steepness of the cone and the way in which it looms over the landscape, even quite small eruptions, of the order of 10 7 m 3 , would be liable to generate lahars which would be destructive in river valleys around the mountain.
Estimated cumulative volumes are plotted against time in figure 14b.
The volume emission rate has remained approximately constant, at about 1 km 3 /5900 years, since the Pungarehu eruption c.23,000 years ago.
A shortfall of the order of 0.35 km 3 appears to exist at present. On the basis of average intervals between eruptions (figure 14a) a smaller eruption, of the order of 1 0 7 -1 0 8 m 3 would seem more likely.
At all events it is abundantly clear that the volcano is dormant rather than extinct.

QUANTITATIVE VOLCANIC RISK AND FREQUENCY OF ERUPTIONS
In a previous study (35) estimates were made of the total losses expected in eruptions at all the volcanoes discussed above. Naturally these are dominated by the largest eruption expected for each volcano.
Neglecting Raoul Island, figures range from $10 million to $3000 million for the andesites, $35 million to $3100 million for the basalts, and $14,000 million to $25,000 million for the rhyolites.
The upper limit for a single rhyolite eruption is therefore broadly equivalent to the total external and internal deficit which at present burdens the nation's economy! Clearly, however, it does not make much sense to regard the rhyolite volcanoes as the source of the greatest shortterm threat, short-term in this sense being anything less than 500 or 1000 years. Some allowance must be made for the mean frequency of eruptions.
A simple way of combining the two sets of data, the risk on the one hand, and the mean frequency on the other, is to apportion the risk on an annual basis, by dividing the amounts at risk in each magnitude range at each volcano by the mean interval between eruptions. The mean intervals accepted and used for this purpose are listed in Appendix II.
Risk calculated on an annual basis in this way ranges from about $700 to $3 million for the andesites, $400 to $400,000 for the basalts, and $200,000 to $1.25 million for the rhyolites (neglecting Raoul Island, as before, where the values at risk are very low). Apportioned annually, the greatest risk for the andesites stems from Mt Egmont, for the basalts, from Auckland, and for the rhyolites, from Okataina.
In the country as a whole, by far the greatest volcanic risk, on an annually apportioned basis, is due to Mt Egmont, with Okataina a poor second. This is all the more significant when it turns out that the greatest annually apportioned risk for Okataina is for the largest eruption class (100 km 3 ) while for Mt Egmont it is for an eruption of only 10 7 m 3 , a size that has a high degree of probability of occurrence.
Larger and more damaging eruptions at Mt Egmont, when considered on an annual risk basis, amount to much less because of the proportionately much longer intervals that separate the large andesite eruptions compared to those at the rhyolite volcanoes.
Ruapehu, which is second to Mt Egmont among the andesites in terms of annual risk, similarly reaches its highest risk value for a comparatively small eruption, in this case 10 6 m 3 only, an order of magnitude smaller than Mt Egmont.
All other volcanoes have both their actual and annually apportioned risk at a maximum for their largest eruptions, except Raoul Island where the values at risk are so close to the volcano that peak loss, apportioned annually, is reached for 10 8 m 3 eruptions.
A measure of "likelihood" of eruptions is given in figure 15, assuming that eruptions are not a random process and that a volcano carries the "memory" of past events.
This assumption is likely to be correct, since refilling of magma chambers, accumulation of gas pressure, and similar processes must to a large extent determine the time of eruptions. For each volcano the time elapsed since the last eruption in each magnitude class is divided by the mean interval between eruptions (Appendix II) and the results are plotted.
Values below 1 indicate that the time elapsed since the last eruption is les 3 than the mean interval for that magnitude range. This serves to highlight those volcanoes and eruption sizes which are, in terms of the mean interval, "overdue". It is important to realise, however, that the large standard deviations on most intervals mean eruptions may still occur below the level 1 , and equally that those volcanoes showing high values of "likelihood" may not necessarily be the first to erupt.

FUTURE ERUPTIONS: EVIDENCE FOR MAGMA BODIES
Although basaltic magmas can rise rapidly to the surface from beneath the Earth's crust, andesitic-dacitic, and rhyolitic magmas accumulate in reservoirs within the crust and remain there for long periods before eruption.
These reservoirs are not totally, or even mostly, filled with molten rock, but with partial melts which typically amount to 20-30 percent of the total rock volume (40).

In
New Zealand where the source region for large rhyolite eruptions is large and ill-defined, including all the central North Island from the north of Lake Rotorua to the south of Lake Taupo, it is a matter of urgency to locate underground sources of magma which might give rise to future large eruptions.
In this respect, a knowledge of the history of past eruptions is very helpful. Eruptions have occurred repeatedly along the Haroharo and Tarawera rift systems at Okataina and from the Horomatangi Reef (see figure 1)  This has been done on a small scale in the Tongariro National Park (49).
An analysis of gravity and magnetic measurements suggests that a large magma body, with about 300 km 3 of partial melt, exists at shallow depth beneath the eastern end of Lake Tarawera (24). Another substantial magma body may underlie the Guthrie-Ngakuru area since a swarm of apparently volcanic earthquakes took place there in May 1983. This is the area previously identified as the Kapenga volcanic centre (24, 32).
A third source of magma is inferred, on seismological grounds, to lie along the line of the Ngangiho Fault (see figure 1) which cuts both the Taupo and Maroa volcanic centres. These are the only places within the area liable to rhyolite eruptions where evidence has been found up to the present time of the existence of magma bodies.

CONCLUSIONS
In assessing volcanic risk, a major practical difficulty lies in the very long intervals, often thousands or tens of thousands of years, between the largest, most damaging eruptions. This is accentuated by the fact that in New Zealand, which is famous for its rhyolite ignimbrite eruptions, activity has been on such a gigantic scale that much of the evidence concerning eruptions in the present cycle of activity, which is it vital to assess in order to forecast future activity, lies buried and inaccessible beneath later deposits.
In the short term, 100 years or less, the greatest volcanic risk in New Zealand lies in moderate eruptions at Mt Egmont of the scale that have occurred many times in the past. In the medium term, say 100-200 years, a destructive eruption of 1 km 3 or thereabouts at Raoul Island becomes a likely event.
In the long term, very large rhyolite eruptions at Okataina or Maroa/Kapenga are increasingly likely, and these may well take place from magma bodies inferred to underlie the east end of Lake Tarawera, and the Guthrie graben (Kapenga volcanic centre).
No systematic search for magma bodies has yet been undertaken, and there are likely to be other sources of underground magma in the central volcanic district.
One such probably lies along the line of the Ngangiho Fault.
By plotting cumulative volumes erupted against time, it is possible to estimate the current potential for eruptions at many of the volcanoes. This is expressed on the diagrams as a shortfall between the volumes actually erupted and those estimated by extrapolating past rates of eruption to the present time. The current potentials are roughly as follows: Raoul Island 1.5-14 km 3 ; Auckland 0.1 km 3 likely in the present (unfinished?) episode of activity; White Island c.0.04 km 3 ; Okataina c.50 km 3 ; Maroa 1 00-900 km 3 unless the centre is extinctm, which is possible but unlikely; Tongariro (including Ngauruhoe) c.4.5 km 3 ; Ruapehu 0.06 km 3 ; and Mt Egmont c.0.3 5 km 3 There is insufficient information to project a reliable trend for Mayor Island, Rotoroa Caldera, Taupo, or for Ngauruhoe on its own in spite of the large volume of data on small eruptions. However, the potential at Taupo is probably less than about 5 km 3 .
At Ruapehu it is uncertain whether the large lahars are all eruptionrelated and the estimate may therefore be unreliable.

Quantitative
estimates of volcanic risk expressed as property losses, excluding casualties, amount to about $3000 million for the largest basalt and ande-site eruptions, and to about $25,000 million for the largest rhyolite eruption. These have been normalised to correct for differences in the mean frequency of eruptions, and are expressed as annually apportioned risks. For an individual volcano expected losses, annually apportioned reach $3 million for the andesites (Mt Egmont), $1.25 million for the rhyolites (Okataina), and $400,000 for the basalts (Auckland).
A measure of likelihood of eruptions is obtained by dividing the time since the last eruption by the mean frequency of eruptions.
For those that could be significant in terms of damage (^10 7 m 3 ), the greatest short-term likelihood exists for an eruption of about 10 7 at Mt Egmont, which coincidentally is also the eruption for which apportioned annual risk is highest.
The mean frequency of such an event is less than 100 years.
In the medium term an eruption of about 1 km 3 at Raoul Island, with a mean frequency of about 800 years, is considered probable.
In the longer term, eruptions of the order of 100 km 3 appear likely in the Maroa area and at Okataina.
These have mean frequencies of about 100,000 years and 25,000 years respectively.
There is, however, a possibility that the Maroa centre may be extinct.
Further work is needed to obtain a better estimate of the dates of past lahars at Tongariro and Ruapehu. This is crucial in assessing volcanic risk. So too is the need for a better understanding of the timing, sources, volumes and degree of interdependence of the large rhyolite eruptions. More data are urgently needed on the history of eruptions at Mayor Island, and could best be obtained .by marine coring downwind of the island. Finally, a major integrated effort is required to locate any large magma bodies that may exist within the crust from southern Lake Taupo to north of Rotorua.

ACKNOWLEDGEMENTS
Among his many colleagues who have contributed ideas towards this paper, the author wishes to thank especially J W Cole, P C Froggatt, B F Houghton, E F Lloyd, V E Neall and C J N Wilson. The publication of this paper is subject to the approval of the Director, Geophysics Division, DSIR.
Thanks are also due to S R Hicks for help with some of the diagrams .

22.
Kohn B P; Glasby G P, 1 978 : Tephra distribution and sedimentation rates in the Bay of Plenty, New Zealand.

Nairn I A, pers comm, 1981:
Unpublished report for National Civil Defence Planning Committee on Volcanic Hazards, entitled "Potential eruption hazards at Okataina Volcanic Centre".

24.
Rogan A M, 1980: Geophysical studies of the Okataina Volcanic Centre.
Unpublished PhD Thesis held in the library, Auckland University, 272 pp.

25.
Healy J, 1 962: Structure and volcanism in the Taupo Volcanic Zone, New Zealand.

26.
Martin R C, 1961: Stratigraphy and structural outline of the Taupo Volcanic Zone.

27.
Blank  These indicate a current shortfall in the volume erupted (that is, an eruption potential) of between about 1.5 and 14 km 3 .

Figure 3
Auckland volcanic field a) Log-log plot of mean intervals between eruptions. Filled circles are calculated values, and asterisks their corresponding standard deviations. Unfilled circles are rough estimates for the Whangarei and Bay of Island-Kaikohe volcanic fields, for which the trends have been drawn arbitrarily parallel to that for Auckland.

b)
Cumulative volume erupted versus time (Auckland volcanic field only), to which an average rate of eruption trend has been fitted. Note that there is no evidence for the frequency of eruptions within an eruption cycle, and that it is unlikely that the current cycle has ended.

Figure 4
Mayor Island a) Log-log plot of mean intervals between eruptions. Unfilled circles are rough estimates only.

b)
Cumulative volume erupted versus time. Activity has been episodic, and there is insufficient evidence for an average rate of eruption to be determined.
Heavy arrows mark eruptions of about 10 7 m 3 .
These are too small to show as steps on the diagram.

Figure 5
White Island a) Log-log plot of mean intervals between eruptions. Filled circles are calculated values, and asterisks their corresponding standard deviations. The unfilled circle is a rough estimate. The line marked A was used in the quantitative risk survey referred to (35). That marked B is a better fit to the data and yields the values given in Appendix II.

b)
Cumulative volume erupted versus time.
The average rate of eruption is shown by the line which suggests that there is a current shortfall in the volume erupted (that is, an eruption potential) of about 5 x 10 7 m 3 .

Figure 6
Okataina volcanic centre and Rotorua caldera a) Log-log plot of mean intervals between eruptions. Filled circles are calculated values, and asterisks their corresponding standard deviations. The values seem to lie on three linear segments with slope changes at about 9.1 x 10 9 m 3 and 1.1 x 10 11 m 3 .
The unfilled circles are rough estimates for the Rotorua caldera.

b)
Cumulative volumes erupted with time for the Okataina volcanic centre and Rotorua caldera. The Okataina data suggest several changes in the average rate of eruption.
The latest trend, beginning about 14,000 years ago, suggests a current shortfall (that is, an eruption potential) of about 50 km 3 .
The Rotorua data are insufficient to estimate the average rate of eruption. The values for Haroharo define two trends, with a change in slope at about 10 10 m 3 .
The Tarawera data are less complete, but appear to fit a single trend (the lowest value has large standard deviations).

b)
Cumulative volume erupted versus time.
The data for both Tarawera and Haroharo suggest several changes in the average rate of eruption. Current shortfalls (that is, eruption potentials) may be about 80 km 3 at Haroharo and about 40 km 3 at Tarawera.
However the lines are not well defined and it would be possible to draw a trend line for Haroharo which would give a shortfall of about 20 km 3 .

Figure 8
Maroa volcanic centre The Maroa volcanic centre is taken to include the Mangakino and Kapenga volcanic centres. On this figure only those ignimbrites that can be definitely traced to sources within this area are included.
Others are shown on figure 9. a) Log-log plot of mean intervals between eruptions. Filled circles are calculated values and asterisks their corresponding standard deviations. Unfilled circles are rough estimates.
Line A was used in the quantitative risk survey referred to (35). Line B gives a better fit to the recalculated values at the high end and yields the values given in Appendix II.

b)
Cumulative volume erupted versus time. Uncertainty over which ignimbrites originated from the Maroa/Mangakino/Kapenga area is the reason for the two cumulative totals shown.
Average rates of eruption fitted to these range from 1 km 3 /182 years to 1 km 3 /1333 years, and suggest current shortfalls in the volume erupted (that is, an eruption potential) of between 900 km 3 and 125 km 3 respectively.
All known rhyolite eruptions (all sources, including Okataina, Rotorua, Maroa/ Mangakino/Kapenga, and Taupo, and unknown sources in the North Island).

a)
Log-log plot of mean intervals between eruptions. Filled circles are calculated values, and asterisks their standard deviations (envelope outlined) Unfilled circle is a single value. Wavy lines mark discontinuities in the data. Above the wavy line marked 1, mean intervals were calculated over the past 3.7 million years; between those marked 1 and 2, the period considered was the past 1.3 million years; and below 2, it was the past 42,000 years.
The vertical discontinuity marked A may therefore be suspect, whereas that marked B is probably real. This implies a different mechanism for eruptions between about 20 km 3 and 575 km 3 , the point which marks the abrupt change in slope at the upper end of the diagram, from both smaller and larger eruptions.

b)
Cumulative volume erupted versus time.
Many changes have taken place in the average rate of eruption. An interpretation of these is given by the straight lines fitted to the step function.
The figures are years needed to accumulate 1 km 3 , the eruptions taking place intermittently and with large volumes.
Rate of eruption of rhyolite versus time during the past 4 million years. The data plotted are the rates of accumulation of 1 km 3 shown on figure 9b. The rate reached a peak about 325,000 years ago of about 1 km 3 /125 years.
It has since declined, but remains high at approximately 1 km 3 /300-500 years.
An overall period of about 2.5 million years is suggested by the data.

Taupo volcanic centre a)
Log-log plot of mean intervals between eruptions. Filled circles are calculated values, and asterisks their corresponding standard deviations. The unfilled circle is a rough estimate based on a single observation. Two trends are indicated by the data with a change in slope at about 1.55 x 10 l0 m 3 .

b)
Cumulative volume erupted versus time.
The large volumes erupted before 200,000 years ago make it hard to assess a reliable rate of eruption for this centre.
The long period of quiescence c.215,000 years ago covers a period which has not been well studied at Taupo, and may be suspect. The steps represent only large lahars (sector collapses): there are no data available on lava or tephra eruptions apart from the Rangataua lavas and Okupata tephra, shown arrowed, which probably occurred at about the same time as the Murimotu lahars.
The rate of production of lahars at Ruapehu is approximately 1 km 3 /600,000 years: on this basis there is a current shortfall (lahar potential) of about 6 x 10 7 m 3 . There may be no physical reason for any regularity in the production of lahars since some may not be eruptionrelated.
Hence this estimate is very doubtful. c) Represents cumulative volume versus time of lahars in the Whangaehu valley since the Taupo pumice (arrowed) at about 1800 y BP.  Events define two trends, which intersect at 8.3 x 10 6 m 3 .

b)
Cumulative volume of lahars and eruptions versus time.
A steady rate of 1 km 3 /c.5840 years is apparent since the major cone collapse represented by the Pungarehu lahars about 23,000 years ago.
The figure shows the time since the last eruption or eruption-related event for a given magnitude class at a given volcano, divided by the mean frequency (see Appendix II) for the same magnitude class and volcano.
Filled circles are well-determined values, and open ones are approximate values. Values below 1 represent elapsed times since the last eruption which are shorter than the mean intervals.
The higher above the line representing 1, the more likely eruptions are to occur. For significant eruptions in terms of risk (those of ^10 6 m 3 ) Maroa, Mt Egmont (^10 7 m 3 ), Taupo (^10 7 rn 3 ), Raoul Island (^10 9 m 3 ) and Okataina and Rotorua (both ^lO 11 m 3 ) emerge as the most likely events. Note that the mean frequencies of these eruptions are all greater than 500 years, except for Mt Egmont for which the mean frequencies of 10 6 m 3 and 10 7 m 3 eruptions are both less than 100 years (see Appendix II).     Eruptions designated "Breccia" were largely or entirely phreatic, i.e. the result of interaction with water. Those designated "Tephra" were predominantly magmatic. ** Volumes given are "fresh-fallen", i.e. as they actually appear in an eruption. These volumes are expanded in the process of gas release, by a factor which is usually close to 2, from dense rock equivalent volumes, which are approximately the volumes of the magma before eruption. A similar result is obtained for > 10 9 m 3 by including Stratford lahars (£50 000 y BP: c.l.5xl0 9 m 3 ): 5 episodes in c.50 000 y = 1:10 000 y.
For smaller (eruption-related?) events, an idea of the mean frequency can be obtained i) from the fact that there are 6 large Maero debris flow units (lahars), >c.l0 7 m 3 , above the Newall tephra (404 y BP); i.e. to 1984. A.D., 438 y, mean frequency 1:73 years, and ii) from the fact that estimates for the number of small debris flows (>5xl0^m 3 ) during the last 500 years vary between 14 and 23; this gives a mean frequency of 1:29 (28.7) years.