There is a hole in New Zealand. It is sitting midway between Auckland and Wellington, about as far from the sea as it is possible to get in New Zealand. Even a casual look at the map already shows the big lake right in the middle of the North Island, looking misplaced and oversized within the hilly region. It has an irregular shape, measuring at its largest 46 by 33 km. The average depth is just over 100 meters; it contains 60 km3 of water. That number is important, for this lake has not always been there. At one time all that water was rock. An errant volcano was to blame for the excavation.
Of course, volcanoes in New Zealand are not errant. They own the place. Auckland, like Naples, is build on an active volcanic field, and is the most likely place for the first volcanic eruption within a major city. The central region of the North Island is full of volcanoes. These do not erupt as frequently as say Iceland’s volcanoes, but when they do erupt they go for broke. Modern New Zealanders have never experienced their full fury. We have published about the Tarawera eruption of 1306, the largest New Zealand eruption of the millennium. The modern world has been lucky to avoid such eruptions.
Taupo is the location of the largest eruption of the previous millennium, around the year 230. It was a solid VEI-7 and among the largest eruptions of the holocene. But that wasn’t the eruption that formed Lake Taupo, although it did enlarge it. The lake formed some 25 thousand years ago in a much larger eruption. This was the most recent VEI 8 on Earth; it ejected 5 times as much volume as there is water in the current lake. Obviously the lake used to be deeper and also before the explosion the ground (or mountain) was much higher than modern ground level. It is hard to really imagine what New Zealand would have looked like after this explosion. Start with grey dust covering all of the North Island to a meter depth, with leafless and dying trees sticking out above it, and take it from there. Perhaps the kiwi is such a shy bird because it lives in fear of a repeat.
Following this earth-shattering eruption, there were a few small rhyolitic eruptions, until the system reactivated 12,000 years ago. Since that time there have been eruptions of a variety of sizes. The largest of these was the VEI-7 around 230 AD which caused a new caldera collapse.
Over the past 65,000 years, Taupo has had an average magma output rate of 0.2 m3 s−1. For a rhyolitic system which by necessity erupts infrequently (it takes time to properly prepare a batch of rhyolite in a volcanic bake-off) this is extreme. For comparison, Reykjanes managed to erupt 10 m3 s−1 for 6 months after a hiatus of 800 years, which averages out to 20 times less than Taupo – and Reykjanes is a proper basaltic system which should be far more productive than those slow rhyolites. Taupo wins hands down. This is how it does those VEI-7’s.
After the VEI-7 1800 years ago, there was a minor hick-up 30 years later which formed a rhyolitic reef and geothermal area in the lake. After that, Taupo went quiet. It remains reassuringly asleep, with no indication of any wake-up call. This is the volcanic rest of Charles Swinburne’s master piece, about the broken life of the goddess Persephone:
Then star nor sun shall waken,
Nor any change of light:
Nor sound of waters shaken,
Nor any sound or sight:
Nor wintry leaves nor vernal,
Nor days nor things diurnal;
Only the sleep eternal
In an eternal night.
One day Taupo will indeed move from its dormancy to the eternal sleep of extinction, as other New Zealand caldera volcanoes have done before it. But this is not yet that time. On average it erupts once every 1000 years, so the current dormancy is a normal part of its behaviour. Will we know when it is ready to wake up? There is already a restlessness to its sleep: perhaps the preparation for the next eruption has already begun. However, these stirrings do not indicate that it is yet ready to wake up.
New Zealand takes its volcanoes seriously and they are extensively monitored. A significant intrusion would find it difficult to hide. But a lake is not that easy to monitor. Water can hide what happens below, and a large lake can hide a lot. If the ground underneath the lake inflates, would we know? The shore line is a sensitive indicator, of course. Inflation around the shore will change the shore line and this shows up readily. That happened in 1922 when the ground deflated by several meters, a herald of good news as deflation does not normally presage an eruption. But there have been other times of unrest and the ground here moves both up and down.
Rhyolitic calderas such as Taupo often show signs of unrest even though actual eruptions are rare. These signs can include earthquakes, hydrothermal eruptions and gas emission. Taupo is no exception. There are frequent earthquakes. There are geothermal regions around Taupo. (Several of the five regions are used for geothermal energy: New Zealand does not let its volcanoes go to waste.) And sometimes there are bubbles in the lake.
A scale of unrest
How does one judge whether a volcanic stirring is a sign of danger or just a normal part of how a volcano dreams? Some volcanoes toss and turn more in their dreams than others. Ten quakes an hour may be a lot for one volcano but normal for another. Steam in an already active geothermal region is normal, but in a region that has always been quiet, it could be a sign to call the volcanic equivalent of 999. (No, I don’t know either which number to call to report an awakening volcano.) The baseline differs between volcanoes.
Sally Potter has proposed a classification scheme for volcanic unrest, on a scale reminiscent of the VEI, which she called the VUI or Volcanic Unrest Intensity. It aims to provide a pointer at how serious unrest is. The basic scale (or at least the accompanying work sheet) is reproduced below. It records the characteristics of the earthquake swarm, inflation, ground heat and gas flow. The activity can be minor or more significant but the precise definition will depend on the location. Even for identical volcanoes, the sensitivity of the measurements may vary from place to place.
The worksheet shows how the scale can be used. Each row of the worksheet is for one type of activity. The resulting number for that activity on the VUI scale is entered in the last column. There are nine such rows (actually there are ten but two of the rows are combined). Thee final column may have fewer numbers than nine: for instance if the gas flow isn’t measured, its slot is left empty in the final column. The adopted VUI number for the unrest episode is the average of all numbers in this column.
The run-up to the current Reykjanes eruption had some of the characteristics of a high VUI: the dense earthquake swarm which became shallow and the localized and large inflation. Geothermal activity may have been present, however the smoke or steam was also attributed to moss fires. The earthquake swarm was in itself not enough to give a high VUI: the inflation was the sign that the unrest was becoming heightened and that an eruption might be imminent. At the moment there is unrest at Ta’u in the American Samoa islands. There have now been three weeks of earthquake activity with the most recent report measuring 30 to 60 weak (magnitude 2 to 3) quakes an hour. The earthquakes are reported as sharp jolts which suggests they are shallow. But with only one micro-seismometer for the entire region, neither the location nor the depth is well known. There is some evidence of inflation from exposed coral reefs, however this may also be due to large sea tides. To assign a VUI level would require more data. As HVO has pointed out, not all earthquake swarms on volcanoes result in eruptions. But the rarity of such earthquake swarms in this location suggests that VUI 3 might be a reasonable expectation. This shows both the use of the scale and its limitations.
Previous reports of unrest at calderas have likely focussed on the strongest episodes, corresponding to VUI 4 on this scale. In those cases, the fraction of calderas that experienced an eruption after an episode of unrest ranged from 15% for silicic calderas to 50% for other calderas. For VUI 3 the fraction is likely much lower. The road to most eruptions will lead through a VUI 4. Assuming of course we are there to measure this.
Back to Taupo. You will see that the scale above has no specific numbers. How long is a ‘long’ duration? What earthquake magnitude should be considered as ‘high’? It depends on which volcano and how well it is monitored. For Taupo, the table below gives the numbers that the authors of the study used. For other volcanoes, the numbers would be different!
The magnitude and number of earthquakes are missing from this table. That is because the sensitivity has changed dramatically over time. Early on, only felt earthquakes would have been reported. After that there was a period with some basic seismographs. Modern seismographs have been installed since 1990. Each period needs its own numbers. Magnitude 4 or higher are considered as ‘high’ for Taupo. For the post-1990 period, a rate of 10-30 of such quakes per month would be considered as a ‘moderate rate’ and more than 30 as a ‘high rate’. For the older periods, lower rates qualify.
Defining unrest is especially hard for the period before 1940. Reports from this period are published in newspapers, correspondence and reports from travellers, mentioning earthquakes, seiches in the lake and changes in water level. But how do you assign numbers to a report that the ground was shaking for some time? How many quakes per hour does that correspond to, including ones too weak to be felt? And how reliable is the data? When something is written down years later, dates and even years are wrongly remembered, place names may be misspelled or just wrong, and quakes may be remembered stronger than they really were. Newspaper reports (to state the blindingly obvious) may dramatize reports, and may even include incorrect information especially if the information was sourced second or third hand. During periods with earthquake activity, people are more likely to notice other things such as geothermal activity, which they might have ignored at other times, making the unrest appear more widespread than it may have been in reality. It is hard to do science when the data is suspect. The New Zealand Seismological Reports have provided an annual catalogue of felt earthquakes in New Zealand since 1921: these data are considered more reliable.
To define a period of unrest, the area which is included needs to be defined. For example, the eruptions at Reykjanes came after earthquake swarms occurring over a larger area. Were the Thorbjorn swarms and inflation directly related to the Fagradalsfjall eruptions, or were they unrelated events that could have led to a separate eruption? The time period also needs to be considered. Did the Thorbjorn swarms, two years apart, form one unrest episode or two?
Taupo unrest episodes
16 episodes of unrest have been recognized at Taupo over the period since 1870. Minor unrest (VUI 2) occurred in 1877/8, 1890, 1899, 1961, 1964, 1967/8, 1974, February and December 1975, 1996–1999, 1999–2001, and 2008–2010. More significant unrest (VUI 3) happened in 1897,1922/3, 1964/5, 1983/4. There were no events corresponding to VUI 4.
1897: VUI 3
Earthquakes were felt from January to May 1897. In September, activity became much stronger. From 16 September, there was ‘nearly forty-eight hours of almost continuous quivering and shaking of terra firma’. On 30 September, at least 89 earthquakes were felt within a 2-h period. Further intense activity occurred in October. In early October, a newspaper reported that in Western Bay, ‘the water is warm and sulphurous in the small bay’. Rockfalls occurred in the same area following intense earthquakes on 17 October.
1922/23: VUI 3
In April 1922, earthquakes occurred 50 km northeast of Taupo. Over the next 5 months, the activity moved south to Taupo itself. By June, unusual low lake levels were reported, suggesting there was inflation at the northwest side of Lake Taupo. The Kaiapo fault now ruptured, which resulted in hundreds of 1-m tall water spouts, with a lake tsunami being reported. The rupture was followed by major subsidence (3.7 meters!) on the western side of the fault over a period of months. Earthquakes continued to be felt until at least January 1923.
1964/65: VUI 3
A swarm of earthquake lasted from December 1964 to January 1965: it produced 33 earthquakes of M4. It started in the Western Bay area of Lake Taupo, migrated to northern Lake Taupo and finally to Horomatangi Reef. It was attributed to a magmatic intrusion because of a report of volcanic tremor. This remains unconfirmed. There probably was 9 cm of uplift west of Horomatangi Reef: this uplift was found to have occurred sometime between 1956 and 1977, and this was the main unrest episode in that period.
1983/84: VUI 3
It started with 10 days of earthquakes during February 1983, centred near Kinloch, at a depth of 4-8 km, continuing at a lower rate until April. A hydrothermal eruption occurred on 9 April 1983 at Wairakei Geothermal Field. The northern caldera showed 5 cm of inflation: it has previously been subsiding. The source of inflation was at a dept of 4 km underneath the north edge of Lake Taupo. After a swarm under Kaiapo Bay in mid-June, the Kaiapo fault ruptured on 23 June over a length of 1.2 km in an magnitude 4 earthquake, with minor damage in the area. The inflation on the west side of the fault was fully reversed. The eastern side inflated by a further 1 cm. A volcanic earthquake and a hydrothermal explosion were reported. After October the activity decayed but small swarms were detected until March next year.
VUI 2 events
14 events of minor unrest were recorded. These show the same aspects as the moderate unrest episodes, just with lower intensity. They exhibited increased hydrothermal activity, earthquake swarms, low frequency quakes, and on some occasions inflation especially of the northern caldera. It often involved activity on the Kaiapo fault, which seems to be particularly affected by Taupo’s activity. Many of the minor episodes had as main event one or more hydrothermal explosions in one of the geothermal fields, excavating craters as large as 100 meters with ejecta destroying trees. Often the explosion itself was not observed and the exact date may be uncertain.
The area just north of Lake Taupo is dissected by several parallel faults. Kaiapo is the fault just west and north of Taupo. It is the only fault in New Zealand known to have ruptured twice in one century (1922 and 1983). The other active fault in the region is Whangamata (the orange-coloured fault on the west side), and there are two further fault systems in between, the Whakaipo and Ngangiho faults. Each of these has shown evidence for ground cracking during one or more episode of unrest, but only Kaiapo ruptured.
The north coast of Lake Taupo, showing the region of the faults of the Taupo rift. Kaiapo is on the far left, the Whaikapo fault in the centre and the Ngangigho fault on the right, each connecting to a small peninsula into the lake. The scarp of the Kaiapo fault is around 40 meters. Tongariro volcano is visible in the distance. Photo by D.B. Townsend. Source: Graham Leonard et al. Geology of the Rotorua area, published by GNS Science, 2010. https://www.researchgate.net/publication/285598682_Geology_of_the_Rotorua_area
Together, the faults delineate a rift system. New Zealand’s rift runs from the Bay of Plenty in the north southwards, but the amount of rifting decreases southwards to zero south of Taupo. In a rift, the extension causes subsidence: the faults accommodate this subsidence. North of Taupo, there is evidence for 40 meters of subsidence in the rift since the VEI-8 eruption 25,000 years ago. The Kapaio fault over the past 40 years has shown subsidence to the west and inflation to the east , consistent with rifting. But over the long term, Lake Taupo shows no subsidence. The subsidence must be compensated by the amount of magma entering the region. Here, tectonic subsidence and volcanic inflation give a zero net sum. Now throw in the frequent caldera collapse (and there are many calderas in the region, of all ages). Lake Taupo is both a collapsed caldera and a subsiding rift, located around an inflating region where magma appears to have moved up. For added complexity, you may also consider the geothermal activity which suggests the presence of circulating hot water which can also cause local inflation and subsidence.
Within Lake Taupo there is one region where the lake is only a few meters deep. It is shallow enough that a shipping warning light was deemed necessary, and this was conveniently placed on a platform installed to measure inflation of the lake bottom. This region is the Horomatangi reef. It is located next to the deepest part of the lake, and close to Motutaiko Island, the only island in Lake Taupo. The reef comes from volcanic eruptions. So does Motutaiko Island. (The name is in error, as ‘Motu’ already means ‘island’.) The reef was formed in the eruptions 30 years after the VEI-7 eruption, 1800 years ago. The island may be older: I have not found a date for it. The reef remains an active region: it is the location of geothermal activity, and magma may still be accumulating here. It is the most likely location of a future eruption.
Ups and Downs
Peter Otway and collaborators have been carrying out measurements of accurate levels around the lake since 1979, using 22 separate stations. They were the team who installed a platform on the Horomatangi Reef. The results show patterns where uplift and subsidence alternate in different areas. The plot below, taken from their paper, shows the location of the instruments and the change in levels over different time periods. It should be noted that the plots interpolate from the measurements: don’t look too hard for patterns where there were no instruments, such as in the centre of the lake!
The instruments first detected a pattern in March 1983, where Kinloch had started to rise, with lower rates of inflation on a line towards Taupo. At the same time, there was slight deflation on the southern side of the lake. By 13 June, Kinloch had risen by 47 mm. Three days later, the earthquake swarm started and unrest had begun. The swarm migrated southeastward. On 23 June, the Kaiapo fault ruptured. Now, Kinloch quickly deflated by 40 mm, followed by a slow further subsidence. By January, it has returned to its previous level. But on the east side of Kaiapo, the inflation continued and reached 55 mm by January. Plots b and c show the pattern over this time, before and after the rupture.
Clearly, over the 40 years there were a lot of ups and downs. By eye, it is already visible that most of the uplift is on the eastern side of the lake. That is confirmed when looking over the entire period. This is in the plot below, taken from the same paper. Kinloch is down, Horomatangi is up. The highest uplift is 135 mm, amounting to about 3 mm per year. The circles indicate where the measurements were taken and where the data is reliable.
The results clearly show that the area of the Horomatangi reef is inflating. This is on the eastern edge of the deep crater of the VEI-8 eruption and near the centre of the more recent VEI-7 eruption. The inflation is seen also on the eastern shore line.
The lines above show measurements along the north shore of Lake Taupo, crossing the various faults. Over the time, the west side of Kaiapo fault has remained at around the original level. On the east side there has been notable inflation.
This pattern agrees with the picture of average inflation across the area. It also explains why the Kaiapo fault is the most active here: it is located on the edge of the inflating region, so is subject to vertical strain. It is a tectonic response to a nearby magmatic intrusion.
The reef has been inflating for at least 50 years, but probably much longer. Around Lake Taupo are remnants of old shore lines. At Whakaipo Bay and Kinloch, these show that over the past 1800 years, there has been no net change in these regions. But at Rangatiro Point and Mine Point, near Horomatangi Reef, they show 5 to 8 meters uplift since the eruption 1800 years ago. That gives an average rate of inflation in this area of 2 to 3 mm/year – very similar to what was found by Peter Otway.
This inflation occurs in bursts lasting 1 to 4 years: these are the unrest periods. The inflation starts quite suddenly (within weeks) and also end suddenly. They are followed by a quiet period where there is a slight subsidence. In most places along the lake, the subsidence and inflation cancel out, but on the east side the inflation becomes permanent.
What is causing the inflation? The constant rate over a long period (with decade-long fluctuations) suggests magma intrusion is to blame. Assuming a 10 by 10 km region inflating at 3 mm/yr, about 0.5 km3 has accumulated since the eruption 1800 years ago. That is enough for a small eruption but not for anything on the scale of a typical major Taupo eruption. So where do those get their power from? The answer may lie in the Taupo rift. It is a subsiding graben, but around Taupo this subsidence is somehow being compensated. If we assume a 100-km long rift zone, 10 km deep and rifting at 1 cm/yr, then over 1000 years it requires an amount of 10 km3 of magma to fill the gap and compensate the rift subsidence. That is closer to what we need, although it probably overstates things since even Taupo does not do VEI-7’s every few thousand years.
Where does the magma come from? The caldera is rhyolitic, and so it is in storage and, like a fine wine, kept for ageing to bring out the best aroma. (Ok, volcanoes are not know for fine aromas.)
James Muirhead and collaborartors present the picture shown here. Underneath Taupo is a rift pillow where the mantle feeds the region. Above that is a zone with a partial melt of 10-20% which acts as a buffer. Above that is a mush zone which stores the (eventual) eruptible magma. The Horomatangi reef could be fed by a flow from this mush region to depths of a few km. The mush zone pulls in magma from a wider range of the Taupo rift. The volume of the mush zone is estimated at 250 km3.
What does this mean for us? The unrest episodes are driven by very small amounts of magma arriving at the shallow storage underneath the reef. The VUI scale is correct at not assigning these levels above VEI-3, and they are unlikely to lead to an eruption. (If they do, it would be a small event although probably with a spectacular view from Taupo.)
For a significant eruption, a much larger influx of magma would be needed. We have never seen the run-up a major eruption so have no idea how quickly that can happen. In the case of Krakatau (a very different system), there was a small eruption 200 years before the big one, and probably significant changes on the island in the century before the event, as we have discussed elsewhere. So we can suspect that we will have a century of warning. For now, it is important that studies such as those described here are continued. They may be our best early warning system.
To end this post, I reproduce a story from Lawrence Cussen, district surveyor in the 1880’s. It was published in 1897.
I am indebted to Major Scannell for the following interesting narrative of the sudden rise and fall of the water at the north end of the lake on or about 28th August, 1883. On the day mentioned, a little schooner, which used to ply across the lake, was lying afloat at Tapuaeharuru. Some men working close by noticed that the schooner was suddenly left high and dry. They went to shove her afloat again, and in doing so they noticed that the river had fallen about 2ft. In the course of fifteen or twenty minutes it rose again to its previous level. This phenomenon was noticed by four or five people. It occurred at half-past twelve o’clock in the day. On that same afternoon, between one and two o’clock, Sergeant-Major Smith and Sergeant Miles, of the Armed Constabulary Force, were bathing in a warm spring called Waiariki, situated on the bank of the Waikato River, about a mile from the lake; the bath was fenced round with stones on the side next the river, and it stood about 2ft. above the level of the water. They found their bath become suddenly cold, and were astonished to find that the river had risen to a level with it. It remained so for about five minutes, and then suddenly resumed its former condition. So far as I have been able to learn, this was the only occasion on which the phenomenon occurred.
The event that the major described was not associated with a period of Taupo unrest – Lake Taupo was sleeping at this time. It was instead due to a distant, volcanic event, not even in New Zealand. That event was the explosion of Krakatau on that day. Lake Taupo responded to the atmospheric pressure wave. I am not aware how Lake Taupo responded to the dramatic eruption of Hunga Tonga this year.
Albert, August 2022
- Taupō volcano’s restless nature revealed by 42 years of deformation surveys, 1979–2021. Peter M. Otway, Finnigan Illsley-Kemp & Eleanor R. H. Mestel, 2022, New Zealand Journal of Geology and Geophysics, DOI: 10.1080/00288306.2022.2089170. https://www.tandfonline.com/doi/full/10.1080/00288306.2022.2089170
- Introducing the Volcanic Unrest Index (VUI): a tool to quantify and communicate the intensity of volcanic unrest. Sarah Potter et al. (2015) Bull Volcanol. https://link.springer.com/article/10.1007/s00445-015-0957-4
- A catalogue of caldera unrest at Taupo Volcanic Centre, New Zealand, using the Volcanic Unrest Index (VUI). Sally H. Potter et al., Bull Volcanol (2015) 77: 78 https://link.springer.com/article/10.1007/s00445-015-0956-5
- Stretching, Shaking, Inflating: Volcanic-Tectonic Interactions at a Rifting Silicic Caldera. James Muirhead et al. (2022) Front. Earth Sci. https://doi.org/10.3389/feart.2022.83584