As promised in my last post, I’m here to talk about the 1790 Keonehelelei eruption of Kilauea. “The falling sands.” This is an article that has been developing for quite some time now. The 1790 eruption is, after all, a lot of things at the same time. The most obvious and no less important is that this was a deadly and destructive eruption of Kilauea that may have killed up to a few hundred people. But beyond that, it is also a scientifically fascinating event on how to make fluid gas-poor magma, which is normally harmless, produce an extremely violent blast. And even more so, the 1790 eruption marks a drastic turning point in Kilauea’s historical activity, a shift of activity from the East Rift Zone to the summit, an event that lies at the heart of Kilauea’s most dynamic period, surrounded by catastrophic lava lake drain-outs, and fascinating rift zone action involving satellite shields, recurrent rifting, fissure eruptions, and pit crater collapses. To top it off, this event happened within a fascinating cultural and historical context, where it was perhaps the last Kilauea eruption to be truly seen through the lens of Pele’s wrath by the local inhabitants, and it possibly intervened in the course of a civil war over the control of the island when part of the army of Keoua was wiped out by the explosion, to the advantage of his opponent Kamehameha, who would eventually emerge victorious over the island and would later go on to unify the archipelago and take major decisions that would shape their history. Perhaps this can give some idea of why the 1790 eruption and the volcanic events around it seem so fascinating to me, and why they are a major part of why I’m obsessed over Kilauea volcano.

The recent intrusive activity in the northern part of Kilauea’s caldera, along the ring fault, made me worried about the possibility that processes similar to those in action during explosive eruptions of Kilauea could be developing (the possibility of a ring dike). But now, having seen InSAR data of the intrusion event, and having refreshed my memory on Kilauea’s past explosive activity, I see little to no indication that the current situation bears any resemblance. For one, the intrusion has stopped and seems to be making no further attempts at intruding northward. Most importantly, deformation data doesn’t show any obvious subsidence of the caldera floor, and seismicity has been very limited around the caldera’s faults, so overall the caldera is not looking unstable and doesn’t seem to be making any moves towards subsidence or collapse at this moment. In the end, lateral draining is the main or only process through which volcanoes like Kilauea collapse, and at least that scenario can probably be ruled out for now. But with no further ado, let’s begin our story.

Before the 1790 eruption

Thanks to a few radiocarbon dates, some information from natives, and the lacking vegetation cover of certain flows, we know that, during the 18th century, activity along the East Rift Zone of Kilauea was quite extensive. USGS geologist Robin T. Holcomb mapped these 18th-century flows to have erupted along the entire rift zone from Pauahi Crater to the Puna Geothermal Venture. These eruptions were mostly small but intense and flooded the pit craters of the East Rift Zone, filling some with sizable ponded lava lakes.

The British missionary and writer William Ellis recorded information from the natives about some of these eruptions when he journeyed the island in 1823. According to Ellis’s information, the Kaimu district (in the Lower East Rift Zone) was flooded by lava during the reign of King Alapaʻinui, which lasted about 1725-1754. The lava in the Kaimu district corresponds to the shield volcano Heiheiahulu, which is known from other evidence to date to the 18th century. The volume of this long-lasting eruption, including the minimum lava delta volume, is around 0.99 km3. Another flow that is from around this time erupted from a fissure to the south of Leilani and paves much of the ground of the Malama Ki Forest Reserve. This south of Leilani flow has a volume of 0.13 km3 (deltas included), and was bare lava with no vegetation at the time Ellis walked over, though he didn’t give an age. Paleomagnetism, recorded in the magnetic minerals of the lava, lies in between that of Heiheiahulu and the Lava Tree State flow, so maybe took place in between the two, around the mid-18th century. Lastly, another eruption (the one that formed Lava Tree State) took place in the reign of Kalaniʻōpuʻu, in the 1755-1782 period, when they told Ellis that there was a violent lava flood contained to the north of the hills of Kaliu, Malama, and Puulena (north of Leilani). This flow is small, around 0.03 km3, but extraordinary; it erupted out of a collapsed trench-like structure that is similar to the Great Crack of the Keaiwa 1823 flow, and this trench can be traced up to the wall of Napau Crater. This same flood created the tree molds of Lava Tree State Monument by plastering tree trunks in lava to a height of 4.5 m above the final flow surface. I gather that this event was a smaller version of the later Keaiwa eruption of 1823. Fissure activity in the East Rift Zone would have filled Napau Crater with lava and led to a catastrophic outburst when the lake grew into a shallow dike down-rift all the way to the Leilani area.

I mention this 18th-century activity because it may be important to understand what happened in 1790. The eruption of 1790 marked a turning point in Kilauea’s activity. After the frequent East Rift Zone activity of the preceding century, Kilauea entered over a century of activity limited to the summit, which consisted of filling of the summit caldera. A key event around this time was also the formation of 4 or more new pit craters along the ERZ: West Makaopuhi, West Pauahi, Inner Alae, and Puhimau. West Makaopuhi collapse cuts and thus postdate the 18th century lava, mapped as such by Holcomb, filling the older eastern pit. West Pauahi also seems to postdate the 18th-century flows that flooded the Central Pauahi pit. The four craters I mentioned were all almost entirely unvegetated in aerial photos of 1954, so likely date to the same event. Other UERZ pit craters have a thicker forest cover, including some that postdate the Aila’au lavas of around 1450 AD, which are very evidently older on account of their more abundant vegetation. As such, there may have been some sort of deep drainage event at some point towards the end of the 18th century, maybe 1790 itself, that created the four craters and possibly switched Kilauea’s mode of activity. However, unlike many maps suggest, no lava erupted in the Lower East Rift Zone in 1790 as per any historical records, if such a lateral eruption and collapse happened, then it must have taken place in the submarine continuation of the East Rift Zone, the Puna Ridge, as a large underwater lava flood, which is not unlikely at all given the Puna Ridge makes up over half of the ERZ’s total length.

 

1790 Keonehelelei eruption

To native Hawaiians, the 1790 blast is known as Keonehelelei, meaning “the falling sands”. This was the last (or near-last) of a sequence of explosive events that created the Keanakakoi Tephra of Kilauea, between about 1500 and 1800 AD. This eruption is shrouded in mystery, given it caused around hundreds of casualties, killed by pyroclastic density currents. The deadly effects of the Keonehelelei eruption that would be expected from, say, an arc stratovolcano are instead a complete standout in Kilauea’s last 200 years of regular observations. Let’s keep in mind that Kilauea erupts tholeiite basalt, a magma type that is very fluid and has the lowest gas content of all magmas, it’s a type of magma that very rarely results in substantial hazards, so 1790 and other explosive eruptions of Kilauea are highly anomalous in a basaltic ocean island volcano.

There are two different lines of information on the 1790 eruption. First, the limited but “direct” narrations recalled by Hawaiians upon questioning by Ellis and other authors. Second, the exhaustively studied stratigraphy of the summit explosive activity, which, while very well studied, can be misinterpreted, being only indirect evidence of the eruption.

Fred Klein and Thomas L. Wright’s Catalog of Hawaiian Earthquakes opens with an interesting entry of seismic activity at the time of the 1790 eruption. It goes: “The natives say that in the days of Kamehameha, 80 shocks were felt in one day, some of them so hard as to throw down houses. This was soon followed by an eruption of the volcano which destroyed some lives and did much damage.” It’s listed in the date of the Keonehelelei eruption (November 1790). The authors themselves gave a M 6.4 estimate of the seismic activity, which itself is probably quite conservative. It would be hard for earthquakes to have reached this size if the decollement fault had not shifted (the fault that runs between the former seafloor and the volcanic pile under the southern side of the island), and decollement earthquakes in relation to eruptive activity likely imply some sort of rifting event going on (like it happened in 1868 or 2018), particularly considering the large number of earthquakes. Even so, it’s not entirely clear the relation of this swarm to the eruption, but in my mind it stands as evidence that a lateral eruption in the rift zone (underwater), may have triggered collapse of the summit.

There are two main narrations of the eruption. The eruption was impactful because Hawaii was in the midst of a years-long civil war, which at the time was down to two relatives fighting over control, Keoua and Kamehameha. Keoua was marching near Kilauea when the eruption happened and suffered substantial losses; 80 warriors died, according to Ellis. They were travelling with their families, which were also killed, and some other travellers also seem to have died, so the death toll likely stands in the few hundreds or so.

The main narration is given by US missionary Sheldon Dibble:

“But Keoua was at Hilo. He heard of the invasion of the enemy and hastened to the scene of action. His path led by the great volcano of Kilauea. There they encamped. In the night a terrific eruption took place, throwing out flame, cinders and even heavy stones to a great distance, and accompanied from above with intense lightning and heavy thunder. In the morning Keoua and his company were afraid to proceed and spent the day in trying to appease the goddess of the volcano, whom they supposed they had offended the day before by rolling stones into the crater. But, on the second night and on the third night also there were similar eruptions. On the third day they ventured to proceed on their way, but had not advanced far, before a more terrible and destructive eruption than any before, took place. (…) The army of Keoua set out on their way, in three different companies. The company in advance had not proceeded far, before the ground began to shake and rock beneath their feet, and it became quite impossible to stand. Soon a dense cloud of darkness was seen to rise out of the crater, and almost at the same instant the electrical effect upon the air was so great, that the thunder began to roar in the heavens and the lightning to flash. It continued to ascend and spread abroad till the whole region was enveloped, and the light of day was entirely excluded. (…) Soon followed an immense volume of sand and cinders which were thrown in high heaven and came down in a destructive shower for many miles around. Some few persons of the forward company were burned to death by the sand and cinders and others were seriously injured. All experienced a suffocating sensation upon the lungs and hastened on with all possible speed. The rear body, which was the nearest the volcano at the time of the eruption (probably encamped in the north rim of the caldera, where the 1790 deposits are thinnest), seemed to suffer the least injury, and after the earthquake and shower of sand had passed over, hastened forward to escape the dangers which threatened them and rejoicing in mutual congratulations that they had been preserved in the midst of such imminent peril. But what was their surprise and consternation, when on coming up with their comrades of the centre party, they discovered them all to have become corpses. (…) So much like life they looked, that they at first supposed them merely at rest, and it was not until they had come up to them and handled them, that they could detect their mistake. The whole party, including women and children, not one of them survived to relate the catastrophe that had befallen their comrades. The only living being they found was a solitary hog, in company with one of the families which had been so suddenly bereft of life.”

Another narration is from a translated moʻolelo, from Kamakau, a Hawaiian writer:

“When Keoua Kuahuula had finished living in Hilo, his chiefs and soldiers with him decided to return to Kau. They went up Olaa, passing by way of the volcano and on to Kalanihale at Kapapala. Those who came behind the procession of the chief Keoua Kuahuula were all killed by Pele in the following manner: The sand and stones sprang up and became a pillar, standing higher than the mountains Mauna Loa and Mauna Kea. Those who lived at Kawaihae shore (northwest coast of Hawaii, Kohala area) saw this wonderful fire pillar standing in the sky, and at the top of this pillar a flame of fire was flashing (“lapalapa”, a flashing as of lightning). Another wonderful thing seen was, the rising of a small hill lifting up a large hill (this could come from sort of way of explaining the pine tree shape of the plinian column, narrower at the base then wider on top, or could be some sort of description of a co-ignimbrite cloud producing ground-hugging conical plumes next to the main column), the column that was belching forth sand and big stones, and they are the stones lying over the banks of Kilauea and round about. This was continued for several days, and those who passed through there were killed; the remains of men, women, and children lying dead in good form (…). Some of these sand hills are now seen at Kilauea and some are found down at Ahua (Apua?). In less than two weeks after the event the sand reached Punaluu, where Keoua was living at the time in the kapu. The sand came from Ahua to Punaluu; Wailau and Ninole (bays) were closed up , and Punaluu became a fish pond until this day (this last part seems to refer to ash of the 1790 eruption falling over the ocean, or being reworked by erosion into the ocean, and then washed ashore by sea currents, the ash/lapilli fell towards Apua coastline and then casted ashore to the southwest along the shore, clogging some bays with ash)”

On the 12th of Feb, of 1794, while climbing Mauna Loa, Archibald Menzies makes the following remarks regarding Kilauea, showing some minor explosive eruptions were probably continuing:

“We were now within a few miles of the Volcano of which there seemed to be, this day, a considerable eruption, and as the wind blew from that direction, the smoke dust and ashes arising from it proved very troublesome to our eyes in travelling with our faces towards it”

To this historical evidence adds the geologic evidence of the eruption stratigraphy. There are several layers that could, and I think do, belong, to the 1790 eruption and the following years of activity. The first is unit H, this eruption happened after a long interval following the previous explosive activity, since there is incipient soil that separates it from the earlier unit G of the Keanakakoi Tephra. These are the geologic units described in “Products, processes, and implications of Keanakāko‘i volcanism, Kīlauea Volcano, Hawai‘i” by Donald A. Swanson and Bruce F. Houghton and the events of 1790 I think they may belong to listed in stratigraphic order:

Unit H: Unit H consists of a series of 12 subunits of varied character, each of them seemingly separate volcanic blasts which changing conditions, from wet to dry, sometimes mainly fall deposits or at other times PDCs. This unit is mainly wall-rock, also called lithics, which is to say pyroclastic material made of blown-out bedrock rather than fragmented magma. Though a bit of juvenile material (fragmented magma) does occur. Many units have cross-bedding from pyroclastic density currents near the caldera rim, and certain subunits have many ballistic blocks, some reaching a diameter of over 20 cm. The coarse layers with the ballistic blocks seem to have erupted from the rough location of Halema’uma’u, and they would have been deadly in areas close to it (maybe a km to a few). Some units have ash pellets, also called accretionary lapilli, a sort of volcanic hail typical of wet volcanic plumes, often co-ignimbrite plumes, so there must have been water interaction. Some lithic ash up to 17 km E-NE of the caldera is thought to have fallen around this time (implying tall columns into the jet stream), though the timing is not confirmed. Even though in recent interpretations Unit H is not included in the 1790 eruption, I think there is strong evidence that this unit was erupted in 1790. First, there is no evidence of a hiatus with the next units (erosion, soil, and such), but mainly that this unit is the only one that matches the description of the two days of explosions before the deadly blast that took a toll on Keoua’s troops. According to the story, there would have been several different explosions, and these ejected ballistic blocks. Only Unit H matches this description.

Unit I (Footprints Ash): Unit I is an ash and ash pellets (accretionary lapilli) unit that is most notable for having thousands of indented footprints of men, women, children, and also a few from pigs, mainly in the area around Mauna Iki. From the moment they were discovered in 1920, they were proposed to have been left behind by Keoua’s army during the 1790 eruption (as the narrations mention the army included the families of the warriors as well as livestock they were using as food supply). These footprints deformed the ash in a way that it shows it was wet when it was stepped on, and some footprints also have mud-cracks. Abundant accretionary lapilli occur in this layer, of up to 3.5 cm diameter (the size of a nut), and indicates a wet eruption cloud. It was deposited in two main lobes running to the S and SW of the summit, respectively. Like the previous Unit H, I is mostly lithic (rock blown out from the bedrock). I will discuss this layer more below and why I think this is a dilute PDC that was the start of the lethal blast responsible for the many casualties.

Subunit J1 (Mystery Unit): J1 is a fall deposit of lithic ash, lapilli, and blocks blown from the bedrock, with some rare bombs and lapilli of fragmented magma. Some ejected blocks are over 1 m across. It was dispersed to the SE towards the Apua area, being 1.5 m thick in the caldera rim and found as far as 30 km, but must continue over the ocean to a substantial distance. It may have originally been a few to several cms thick as far as the Puna coast, though having been quite reworked. Glass samples of the Mystery Unit include a major range in MgO of 3.6–9.4 weight%, with the more evolved melts being more common, some of them low-silica basaltic andesites, which I think is the most evolved melt known to have erupted from the summit of Kilauea. This unit is most evidently the tall volcanic cloud visible from the NW part of the island that was filled with lightning, also the destructive shower of ash that came down over a wide area, Keonehelelei, at about the time the lethal blast happened, it must have also been responsible for clogging the bays along the south coast of Hawaii with drifted ash that would have fallen over the ocean, or washed into it by rain.

Subunit J2: This subunit is entirely made of wall-rock lithics. There are common ballistic blocks and common cross-bedding from pyroclastic density currents. It was deposited within about 3 km of the west caldera rim, and both it and Unit I look to have erupted from the NW sector of the caldera, about a km north of Halema’uma’u or so. This layer has been recently interpreted as the lethal blast, but I think it may have been from the later parts of the main explosion, given it overlies J1, which is the deposit that, judging from the stories, may have been the co-ignimbrite cloud (rising above the PDCs that killed the second and part of the first parties).

Subunit K1 (Golden Pumice): This is a high fountain pumice-fall deposit that erupted from the SW corner of the caldera in Halema’uma’u, so probably quite deep below the rim, and was carried SW,  probably by the low-altitude trade winds. It’s up to 1 m thick at the SW rim of the caldera. It’s subdivided in U1 and U2, each of which starts with a higher abundance of outgassed and less vesicular tephra followed by more vesicular tephra, so possibly it included a pair of high fountain episodes from the same vent. Melt (glass) from the Golden Pumice was remarkably primitive, mostly 9- 10.5 weight% MgO. Regarding its timing, it’s clearly not in the part of the 1790 eruption narrated in detail by the stories and could have happened well after, though it could also maybe be part of the several days of continuing activity mentioned by Kamakau, and overall I think it is unlikely to have happened much more than a few years after 1790.

Subunit K2 (Eastern Pumice): This is a thin pumiceous fall deposit that erupted from a vent some hundreds of meters north of Keanakakoi Crater, around the E end of the South Sulphur Banks that existed in the early 18th century. It was dispersed SW, probably by the low-altitude trade winds. It was a brief high fountain that deposited Pele’s hairs-tears overlain by pumice. According to maps like those of Wilkes in 1841 the area of the caldera where the Eastern Pumice happened was only slightly subsided, no more than 100 m below the caldera rim so for no spatter to have landed in the caldera rim and also given how oddly thin (only up to 10 cm) but far-dispersed this unit is (probably much more than the 10 km where it’s still barely thinner than at the caldera rim) this event must have been very unusual probably similar to the brief jetting events observed at the end of some fountain episodes of the ongoing 2024-2026 eruption, but sustained. Likely there was a jet of mostly gas, carrying some pumice and lava drops, shooting in an incandescent column many hundreds of meters high, enveloped in abundant black tephra. The Eastern Pumice chemistry is also enigmatic, the glass (molten portion of the magma) is less primitive than the Golden Pumice, mainly 7-9 weight% MgO, but is notable for the high potassium compared to previous and usual Kilauea eruptions, being among the most K2O-rich/alkaline Kilauea eruptions documented.

Unit L: Series of five subunits with various characteristics, all of them thin fall deposits, some of lithic ash or blocks, and others of Pele’s hair and tears. They seem to have erupted from multiple locations on the southern side of the caldera in the general area of the sulphur banks. They were generally dispersed south, so probably had plumes at least some kms into the air to catch the southward winds. It’s possible that the activity Menzies saw in 1794 climbing Mauna Loa was related to Unit L. Unit L Pele’s hair, which may have come from a similar area of the caldera (the south sulphur banks) as the Eastern Pumice, carries the same unusual potassium-rich composition as the Eastern Pumice.

As I’ve mentioned, I suspect that it’s the Footprints Ash, the Unit I rich in large accretionary lapilli and containing the footprints, that may have killed part of Keoua’s army and comprises the start of the main explosion. The latest interpretations put the small local PDCs of J2 (2-3 km west-southwest of the caldera) as the ones responsible for the death of the army, but I think there are several inconsistencies. First, that Unit I, I think, should be interpreted as a PDC deposit. I’ve always been drawn to view this unit as akin to the 1911 PDC event of Taal, given the similarities in deposit, both left a muddy deposit rich in accretionary lapilli and context: a mafic volcano in the midst of being drained laterally. The accretionary lapilli of 1790 would have dropped out of steam-rich coignimbrite clouds rising above the PDCs. Looking at the deposit itself, the thinning rate is very low over distance, which led to considering this a phreatoplinian fall deposit, but if it truly was a phreatoplinian intensity eruption then it should have been transported to the east or southeast, which is the direction the jet stream almost always blows, and the dispersal of the subsequent Mystery Unit. The way it has a few different lobes that run in downslope directions from the caldera seems consistent with PDCs. At the same time the ash would have been carried by the co-ignimbrite plumes several kms up into the jet stream and rained to the SE as unit J1 (the Mystery Unit).

A PDC interpretation as part of the main blast I think, makes more sense with the stories narrated, particularly that of the forward company of Keoua, and in multiple ways. First, considering that the path of the footprints, and in general the path of travellers, follows the caldera around the W before behind SW, then, regardless of Unit J2 or I being the ones responsible, the rear company must have still been near the northern rim of the caldera, outside their reach. And yet it’s stated the rear company was actually the closest to the volcano, so it seems unlikely the center company was still within the restricted area of the J2 flows, and even more unlikely the forward company that had left even earlier was in that same area too. The forward one must have been some, or more like several kilometers downslope at that point, and yet according to the descriptions, the forward company was impacted by PDCs that burned them and clogged their airways. The area around the caldera rim was also heavily bombarded by lithic blocks that would have been deadly on impact, yet the descriptions of the central company don’t mention injuries from lithic block impacts, and in the forward company only sand and cinders are mentioned as being experienced, also suggesting they were farther away than the reach of unit J2. Instead, I think the SW lobe of the Footprints Ash that reached about 16 km from the vent is likely the pyroclastic density current deposit responsible for the deaths in the forward company and the annihilation of the central one. The timing of J2 seems less consistent with the story, given it overlies Unit J1, the Mystery Unit ashfall, while in the story the explosion seems to start very violently and immediately envelop the forward company in darkness, seemingly more consistent with Unit I at the base of the deposit being the ash cloud that rapidly overtook them. In this interpretation, it’s also easier to understand how the thousands of footprints were made in the area SW of the caldera, since the rear company would have walked over the freshly emplaced mud and engraved their passage into it. That said, this is just an interpretation, and it doesn’t necessarily imply all of the unit was emplaced as PDCs either.

The hidden processes

Having described the event, the main question now is, of course, what caused the eruption itself? I think the answer may be quite complex and that as many as three different elements probably came into play: The rifting/lateral eruption and collapse, the intrusion of a ring dike, and the presence of large amounts of water or shallow groundwater in the caldera. Regarding the lateral draining and collapse of the caldera, I’ve already mentioned a few lines of evidence, namely the collapse of about four pit craters in the East Rift Zone somewhere around this time, and earthquakes that threw down houses, which may have come from the decollement fault as the flank shifted southward, pushed by intrusion of magma in the East Rift Zone. But it’s also worth noting the unusual chemistry erupted in the Eastern Pumice and Unit L Pele’s Hair.

The Eastern Pumice (K2) and Unit L erupted from the southeast sector of the caldera, an area that historically has often erupted the most primitive magmas at the summit (as opposed to Halema’uma’u being more evolved). Its location close to the edge of the caldera complex, and over the ERZ connector, is probably key in this primitive character. The unusual K2O rich (also enriched in certain trace elements, titanium, others) chemistry erupted in these two units could have come from a batch of magma that had already entered Kilauea’s deep system and influenced its activity leading to 1790 but had not yet invaded the shallow Halema’uma’u chamber (as has been observed in the past decades at times, namely that Halema’uma’u chemically lags behind the southern caldera area).

And here I have to go back to the recent historical period to explain this. I’ve always found it curious that the lavas erupted at the time of the 1955 and 1960 eruptions, and to a lesser degree the 1960s, were more K2O rich and often also picritic (carrying large amounts of olivine), and also that during this time Kilauea behaved differently, in a more “rifty” manner. In particular, in the past, I noticed the chemical composition of Kilauea eruptions from 1959 to 2018 show a consistent and oddly geometrical gradual change in their composition. Lavas became more potassium-rich over time at a given magnesium content, most times erupting at typical values of around 8-6 wt% MgO, but occasionally eruptions produced higher MgO lavas. These more primitive eruptions followed the odd pattern that, over time, the maximum MgO content dropped, and the highest MgO always stood at the same K2O content. There are two ways that lavas could have more K2O at a given MgO content: you raise the K2O without changing the other, or you increase the MgO without changing K2O. This is because, with a higher MgO, the magma will crystallize away olivine to go back down, and doing so the K2O which stays in the melt will have increased. The odd pattern observed in 1959-2018 eruptions suggests it may be the later case: that the magmas richer in K2O were simply originally much higher in MgO, that the magma in 1959 had originally about 16 wt% MgO, close to the lower bound of komatiite, but in its way up rapidly developed olivine crystals and concentrated potassium in the melt.

The chemistry of the Eastern Pumice is relevant to the volcano in the way that I think 1959 is also relevant, which has to do with the deep spreading of the volcano. I’ve always found it curious how the volcano behaved in the 1950s and 1960s, the intrusions of 1955 and 1960 started all the way down the magmatically active rift, under the Puna Geothermal Venture, which is much further downrift than in 2018 (the intrusion initiated from Pu’u’o’o). In the 1960s and, to a lesser degree, the early 1970s, Kilauea was an intrusive volcano; spreading rates were of 18 cm/yr, faster than the East Pacific Rise, compared to 11 cm/yr in the years before Pu’u’o’o or 8 cm/yr in the early Pu’u’o’o eruption (which I think fell even further). The volcano was also in a regime of frequent sizable dike intrusions, sometimes multiple events per year, like the 5 dikes of 1963. But why would I think to connect this to the chemistry? That is because the spreading of Kilauea volcano is known to be accommodated by a structure known as the “deep rift” which drives the southward motion of Hawaii’s south flank and opens up space for shallower dike intrusions. The deep rift leaves dense olivine cumulates behind that form a high-gravity, high-seismic-velocity core under the rift axis. And the reason the chemistry would be relevant is that a high Mg composition like that of Kilauea Iki in 1959 would be the perfect fuel for the deep rift to grow and shove away the flank, crystallizing dense olivine that acts as a wedge, opening up the edifice.

Now, if the above is correct, this immediately underscores the relevance of the magma composition of the Eastern Pumice, which, in terms of potassium enrichment and trace element patterns, is about similar to the 1960s magma (see the table below). It’s possible a batch of high-MgO magma rose into the deep system before 1790 and started crystallizing large amounts of olivine (high density) in the deep rift; this dense crystal cargo, dropped inside the volcanic edifice, literally forced Kilauea open and triggered a dike intrusion, maybe erupting in the Puna Ridge and pushing the south flank away, while collapsing both the summit and portions of the East Rift Zone. A similar thing, but with less dramatic results, would have happened in 1959 when gassy picrite basalt erupted from Kilauea Iki, two months later, a dike intrusion started all the way down in the present-day Puna Geothermal Venture location, which is the tip of Kilauea’s deep rift. The 1960 Kapoho eruption followed, which erupted more of the picrite and did a very small collapse of Halema’uma’u. So what was the initial culprit in both 1790 and 1960? I’d go for the magnesium and the olivine crystals it goes into, though a lot more played into the 1790 outcome.

The table below lists eruptions by their “original” MgO, working under the assumption that they came from an initial K2O content of 0.38 wt%. As can be seen, Unit L and the Eastern Pumice plot quite high in the list. Considering that, volumetrically the shield eruptions in blue dominate, eruptions that, most of them, erupted multiple cubic kilometres, then the explosive eruptions, which mostly fall in higher values, can be considered to be on the high MgO side. Even so, explosive eruptions tap into the Halema’uma’u reservoir and stored magma that may not be representative of the “fresh” magma arriving into the deep rift, and unlike the summit overflows, they are not voluminous enough to flush the system of this old magma, so there could have been more magnesian melts in action, which I think the Eastern Pumice may have shown, bypassing the Halema’uma’u Chamber in the way the south caldera rim usually does. More relevant perhaps is that the summit overflows and larger ERZ shields (Kane Nui O Hamo, and Pu’u’o’o) all concentrate around the lower-end estimates of about 9 wt% MgO, suggesting that chemistry plays a more important role in shaping Kilauea’s eruption style than I had previously thought, making Kilauea’s long-term character highly unpredictable since it’s unclear if there’s any way of anticipating these chemical changes.

The chemistry of the incoming magma would have likely set the stage, creating a favourable situation for spreading of the volcano and rifting, but was not the main mechanism of the eruption itself. I strongly suspect the main mechanism was a so-called ring dike intrusion.

In general, explosive eruptions at ocean island volcanoes involve caldera collapses, for example Kilauea in 1924 and 2018, and Piton de la Fournaise in 2007. Fernandina (Galapagos) collapsed in 1968, likely due to an underwater flank eruption, while blowing up in a particularly violent VEI 5. It may well have involved a ring dike given the explosive vent was on the ring fault, though it can’t really be proven. Tholeiite volcanoes like Kilauea, Piton, or Fernandina haven’t really been seen to ever behave so explosively outside collapse situations. A volcano that is not an ocean island volcano but kinda behaves as such and has clear evidence for a ring dike producing explosive activity is Miyakejima in the Izu Islands of Japan, which collapsed due to magma removal during a lateral dike intrusion in 2000. During the later stages of the 2000 caldera collapse, some substantial explosive phases happened, reaching subplinian intensity. Now, the term ring dike has never really been used alongside Miyakejima, as far as I know, but the lines of evidence I find converge pretty straightforward at this type of intrusion. The explosive craters of Miyakejima formed a line along the southern caldera ring fault, in the eruption aftermath, the craters degassed a gigantic amount of magma, producing so much SO2 that it forced to evacuate the island for several years; and lastly, satellite-based monitoring has shown a large intrusion under the south rim of the caldera that continues to deflate slowly (most likely the same one that produced all the SO2).

Taking into account the above considerations, I think it’s not unreasonable to consider 1790 to be a direct result of Kilauea’s summit caldera collapsing, and maybe even a ring dike. Kilauea provides additional evidence that its explosive eruptions and the 1790 eruption involved ring dike intrusions. The most clear comes from a different explosive event, the ~1500 AD Basal Reticulite. This outburst was sourced from multiple vents in a semicircle around the northern half of the caldera, in a series of very powerful lava fountains. The power of fountains can be gauged by how most of the deposit is reticulite, a golden-colored type of tephra, made of bubbles with polyhedral shapes, that can reach vesicularities as high as 98%. Reticulite is normally only a minor product in the most powerful Hawaiian fountains, but in the Basal Reticulite it dominates. Another sign of its power is that the tephra was scattered concentrically in every direction around the caldera for several kilometers. Whatever the wind direction was at the time of the eruption, it doesn’t matter, the umbrella would have dropped reticulite several kilometers upwind. The arrangement of vents into a semi-ring along the northern caldera side stands as a fairly simple sign of ring dike involvement.

In 1790, the main reason to suspect a ring dike is the presence, when Ellis climbed to the summit, of massive sulphur deposits along the south side of the caldera known as the South Sulphur Bank. These sulphur deposits covered the entire southwest slope of the caldera along a length of 2-3 km. This is an area that was buried in over 2 meters of ash by the 1790 eruption, so the deposits must have been formed in only 30 years before Ellis’ visit. A sulfur deposit like this, so big, has not again formed at Kilauea, which implies some exceptionally thick magma intrusion took place along the ring fault and continued to degas spectacularly for years or decades after the event. I find it likely this magma was intruded in 1790 and fed the explosive eruptions of the Eastern Pumice and Unit L. The sulfur banks formed in this area because there was a downdropped block which, at the time, was above the lava fill, but the ring dike likely also extended around the west side of the caldera and did not have the equivalent expression there due to the lava fill that was rising and burying that side. However, in 1838, there is a description of streams of sulphur cascading down the northwest wall of Kilauea’s caldera, roughly the side of the caldera that sourced the main explosion of the 1790 eruption. I think the ring dike was also present along this area and degassing under the lava lake, some of the sulfur deposited under the vast ponded lava lake of those decades may have melted and poured out of walls on the caldera above the lake surface.

The third major ingredient in the 1790 eruption was water. Be it groundwater or a water lake, the mud pellets (accretionary ash) and overall muddy character of Unit I and also Unit H imply that water was involved. Given how liquid water expands to nearly 2000 times the volume when turning into steam, water flashing into steam definitely seems like a rather easy way to get a big blast. But the first question is if a lake was present in 1790.

There is a statement of William Ellis in his 1823 journal regarding Kilauea that may be relevant: “in earlier ages it used to boil up, overflow its banks, and inundate the adjacent country; but that, for many kings reigns past, it had kept below the level of the surrounding plain, continually extending its surface and increasing its depth, and occasionally throwing up, with violent explosion, huge rocks or red-hot stones.” This information refers to the evolution of Kilauea. As it has become clear afterward, the first part of the text mentions the summit shield that existed around 1100 to 1500 AD and which produced frequent overflows all the way to both the south and north coasts of the island. This would have been followed by a series of collapses and explosive activity starting around 1500. The information by Ellis suggests the overall trend from ~1500 to 1823 was that of the caldera growing larger in area and deeper, as subsequent caldera collapses took place, so it seems possible the caldera was already quite deep by the time of the 1790 eruption, but was it deep enough to hold a lake?

The water table at Kilauea sits quite deep, at an elevation of 600 m above sea level. Added to the absence of direct mentions of a lake, this makes it very unlikely a water lake existed inside Kilauea in 1790, particularly when considering that the last thing we know the volcano to have done before 1790 is filling Napau Crater to the brim with lava, to an elevation of 800 m. Even if lava transport was perfectly efficient into the East Rift Zone, I doubt the summit crater could have been anywhere under 800 m elevation to sustain a long eruption inside Napau (else it would have filled the summit crater first). The only situation in which there might have been an actual lake is if Halema’uma’u had collapsed shortly before 1790, which I overall think to be unlikely. In the decades before 1790, if there was a deep crater,  it would have filled with lava to some elevation. As such, the eruption of 1790 likely occurred through groundwater interaction alone. But this raises another question. 1790 showed much more marked signs of water interaction than the other strong explosive eruptions in 1500 and around 1650, and without this added water interaction it’d probably not have been so deadly. So why was it more phreatomagmatic?

One possibility is that the successive collapses had created vast amounts of breccia underground that were holding a growing volume of groundwater. That said, Kilauea caldera has seen many collapses before 1500, so this explanation may not really work. Another possibility, which I now lean towards, without being very confident, is that the caldera was indeed deeper before 1790 so that a thinner rock and lava lid over the aquifers could be blown away more easily. At the time of the 1500 Basal Reticulite, Kilauea stood at the “literal” height of the Observatory Shield with about 700 meters of rock standing above the water table; by 1790, the amount of lava fill over the water table had probably decreased, maybe to only 200 m in the more extreme scenario of very efficient lava transport to the Napau area. With only these 200 meters of lava over it, the superheated groundwater would have easily popped away its lid. The growing amount of water interaction that the explosive events around 1500, 1650, and that of 1790 gradually display possibly reflects a caldera complex that overall kept deepening over time, as suggested by Ellis. Not because of the formation of a water lake, but rather by bringing the water closer to the surface.

Taking all the above together, I can now make a full account of how I think the 1790 eruption proceeded. How the deadliest historic US eruption, the most important turning point of Kilauea’s activity in the last few centuries, and one of the most fascinating events in Hawaiian volcanism, if not volcanism as a whole, came to be:

It would have been prepared by centuries of a caldera being present at the summit, which had collapsed more than once, and would have brought the floor down towards the water table. It was also preceded by several decades of intense fissure activity along the East Rift Zone of Kilauea, with a particularly voluminous expression in the Lower East Rift Zone. Shortly before the 1790 eruption a change in the composition of the magma supply would have taken place, with the arrival of melts that were enriched in K2O and probably originally in MgO. Large amounts of olivine would have immediately crystallized from it and sunk into the deep rift of the volcano, forcing it open. The spreading would have triggered an eventual failure. The location of this failure would have been the only place it could have gone unobserved, the tip of the deep rift above Kapoho/below the Puna Geothermal Venture. From here, the magma intrusion would have likely reached into the Puna Ridge (the underwater ERZ) and erupting massively deep below the ocean surface, initiating the collapse of the summit and of four or so pit craters along the East Rift zone (Puhimau, West Pauahi, Inner Alae, and West Makaopuhi).

As is typical of Kilauea collapses, it’d have first started at Halema’uma’u, the heart of Kilauea volcano. Initial collapse of the central conduit of Kilauea would have been akin to that of the start of the 2018 collapse or to the 1924 eruption, when the Halema’uma’u lava lake drained and produced explosions as the lowering magma column explosively and episodically released magmatic gas through the crumbling rock funnel. If the ring dike had already emerged by the time of the Unit H explosions, I do not know, but I do not think it to be necessary. These explosions can be explained by the collapse of the Halema’uma’u conduit and the (likely) shallower groundwater of 1790 flooding it and interacting with the hot rock and magma. The explosions of Unit H sent pyroclastic density currents and ballistic rocks to distances of up to a few kilometers, and lasted about 2 days, corresponding to the explosions that halted Keoua’s advance.

In only 2 days, the situation would have escalated spectacularly from a collapsing pit at Halema’uma’u to decollement earthquakes throwing down houses, the entire caldera dropping, and full-blown ring dike intrusion developing around it. The intrusion of the ring dike around the southern and western sides of the caldera, intrusion that would have been rapidly widening in the context of a caldera collapsing, would have suddenly brought enormous volumes of magma close to the surface, superheating shallow groundwater under the caldera ring fault, creating a ticking bomb. Either a large decollement or less likely a collapse event earthquake seems to have been the detonating factor of the main blast. Shaking threw people out of balance just before a black cloud illuminated by flashes of lightning rose from the caldera. If my interpretation of Unit I is correct, then this event would have fed pyroclastic density currents, maybe as far as 16 km across the Kau desert, downslope of the caldera, killing part of Keoua’s forward company, and annihilating the middle one. Convective clouds would have risen above the PDCs dropping ash pellets and joining into a plume that drifted to the SE, carrying ash over the flank of the volcano and the ocean (Unit J1), giving the eruption its name, Keonehelelei, “the falling sands”,

Both the initial explosions, as well as the main explosion, would have owed their lithic composition to the steam-driven explosions that ejected the overlying bedrock into ballistic blocks and ash. Chemistry also indicates that at the time of the 1790 eruption the magma of the summit magma storage would have been exceptionally evolved compared to summit activity of the last 200 years, with MgO going as low as 3.6 wt% and to the primitive edge of basaltic andesite composition, Though given the little involvement of fresh magma in the eruptive activity, this evolved basalt would probably not have played an important role in its explosiveness. Instead, after the main explosion, activity tapped into more primitive melt and changed to “magmatic” mode, producing the high lava fountains of the Golden and Eastern Pumice. I do not know if the occurrence of the lava fountains was part of the collapse or if it happened up to years afterwards, but the eruptions would have been sourced from the same ring dike that triggered the main explosion of 1790.

The Golden Pumice also provides a potential source of the lava lake setup that marked the first century of regularly recorded activity at Kilauea. In 1823, when Ellis visited, and over the subsequent visits by various observers, the activity in the caldera seems to have consisted of a convecting lava lake in the SW corner supplting from below a vast ponded lake covering much of the total surface of the caldera. The ponded lake drained multiple times over the same area in 1823, 1832, 1840 and the last time in 1868, when it was finally destroyed. The convecting lava lake at the SW end, the Halema’uma’u lava lake persisted semi-continuously until 1924, when a deep draining of the volcanic system, during a dike intrusion at Kapoho, collapsed the conduit for good. While lakes could have formed at later times, through other processes, the occurrence of the Golden Pumice shows the actual moment they probably formed. The Golden Pumice, likely erupted where the ring dike reached the lowest elevation in the caldera walls and was producing high-end lava fountain activity in at least two separate fountain episodes. These would have flooded the bottom of the newborn caldera with a ponded lava lake, thus potentially creating the broader caldera lake that existed until 1868. At the same time, the wide magmatic shaft of the Golden Pumice eruption would have been able to immediately develop into a convecting lava lake conduit, maybe creating the convecting Halema’uma’u lava lake that lasted until 1924.

Lastly, over the years or decades following 1790, the thick ring dike intrusion under the southern side of the caldera would have continued to produce small but powerful explosions and vigorous degassing that created the south sulfur bank. The 1790 event also mostly ended East Rift Zone activity for over a century, something that I suspect has more to do with the damage to the ERZ’s plumbing by the collapsing pit craters along the rift zone than with the events at the summit.

Final remarks

And so the story of 1790 comes to a close. This is a “short” overview of an event that could easily have enough material for a small book. Some aspects are evidently speculative, but based on evidence. Water clearly played a role; a caldera collapse and ring dike intrusion seem also almost certain. The exact chemical changes or the exact role that they played in the event may be a bit more unclear at this point, and definitely a field where Kilauea could use some more science into its old lava flows and perhaps some drilling under the surface to reveal its past chemical variations.

This eruption, what came before and what came after, like the Keaiwa eruption that was a direct consequence of the stage set in 1790, was a demonstration of Kilauea’s dynamic activity at its best. 1790, and the surrounding events in time, I think make the core of what Kilauea is and of its most eccentric aspects as a volcano, a story to be kept in mind whenever Kilauea is the focus of our thoughts.

It missed regular observation of Kilauea’s activity by only 30 years, which shrouds the eruption in mystery. But the way of volcanoes is the way of mystery. Even if it had been observed, the volcano would have kept things hidden. Eruptions lurking below the ocean beyond the tip of Cape Kumukahi, intrusions following underground passages, magma changes in the deep rift system of Kilauea hidden below the level of the summit’s shallow storage, water flowing through cracks and breccia layers in amounts that one can only but guess, island-scale faults reaching delicate thresholds… All while bigger processes across time and space, below and above, shape the happenings of its magma system.

 

References

Swanson, D. A., Weaver, S. J., & Houghton, B. F. (2014). Reconstructing the deadly eruptive events of 1790 CE at K lauea Volcano, Hawai’i. Geological Society of America Bulletin, 127(3–4), 503–515. https://doi.org/10.1130/b31116.1

Swanson, D.A., and Houghton, B., 2018, Products, processes, and implications of Keanakāko‘i volcanism, Kīlauea Volcano, Hawai‘i, in Poland, M.P., Garcia, M.O., Camp, V.E., and Grunder, A., eds., Field Volcanology—A Tribute to the Distinguished Career of Don Swanson: Geological Society of America Special Paper 538, p. 159–190, https://doi.org/10.1130/2018.2538(07).

Ellis, W. (1827). Narrative of a tour through Hawaii, or Owhyhee with observations on the natural history of the Sandwich Islands, and remarks on the manners, customs, traditions, history, and language of their inhabitants. https://doi.org/10.5962/bhl.title.163283

Holcomb RT. 1987.. Eruptive history and long-term behavior of Kilauea Volcano. . In Volcanism in Hawaii, ed. RW Decker, TL Wright, PH Stauffer , pp. 261–350 Washington, DC:: US Geol. Surv.

Thomas L. Wright and Fred W. Klein: Two hundred years of magma transport and storage at Kilauea Volcano, Hawai’i, 1790–2008 March 2015 Bulletin of Volcanology 77(4) DOI:10.1007/s00445-015-0915-1

Wolfe, E.W., and Morris, J., 1996, Sample data for the geologic map of the island of Hawaii: U.S. Geological Survey IMAP 2524, 1 map on 3 sheets; 157 x 140 cm., sheets 104 x 81 cm. or smaller, folded in envelope 30 x 24 cm. +1 pamphlet (51 p. : map ; 28 cm.), https://doi.org/10.3133/i2524B.

Lynn, K.J., Downs, D.T., Chang, J.M., Lundblad, S.P., Mills, P.R., McDade, B., Deligne, N.I., Gansecki, C.A., Schmith, J., Decker, M.F.I., Zoeller, M.H., Trusdell, F.A., Carr, B.B., Patrick, M.R., Parcheta, C., Nalesnik, A., Dietterich, H., Hazlett, R., Johnson, P.A., Gallant, E., Mulliken, K.M., Nadeau, P.A., Cappos, M.J., DeSmither, L., Peek, S., Damby, D., Dotray, P.J., van Helden, K.M., Shea, T., Hammer, J.E., Mourey, A.J., and Loewen, M., 2024, Sample details and near-real-time ED-XRF data collected during the 2020-2023 Halema‘uma‘u eruptions of Kīlauea volcano, Island of Hawaiʻi: U.S. Geological Survey data release, https://doi.org/10.5066/P1XFKKYH.

Lynn, K.J., Downs, D.T., Decker, M.F.I., Ruth, D.C.S., Forster, L., Poepoe, K.K., Lundblad, S.P., Winslow, H., DeSmither, L., Barnett, J., Deligne, N.I., Zoeller, M.H., Patrick, M.R., Gallant, E., Sealing, C., Mulliken, K.M., Dotray, P.J., Ellis, A.P., Chiaro, G., Robbins, M., Johnson, E.R., Lerner, A., Lopez, B., Bennington, N., Nadeau, P.A., Phillips, D.A., Neuman, H., Gansecki, C., Haleamau Rubio, K., Johanson, I., Warren, S.M., Cappos, M., Adams, R.G., Wall, K., Thomas, N., Benage, M., Wright, H., Dechert, A.E., Gauer Pasqualon, N., Flinders, A., Bagnardi, M., Weiss-Racine, H., Corsa, B., Hayes, Q., Kohagura, N., Cauley, C., and Bustos, A., 2026, Sample details and compositional data collected during episodes 1–39 (2024-2025) of the 2024-ongoing episodic Halema‘uma‘u eruption of Kīlauea, Island of Hawaiʻi: U.S. Geological Survey data release, https://doi.org/10.5066/P1GFWGX9

Lynn, K.J., Downs, D.T., Decker, M.F.I., Lundblad, S.P., Winslow, H., McDade, B.R., Mills, P.R., DeSmither, L., Barnett, J., Poepoe, K.K., Deligne, N.I., Zoeller, M.H., Patrick, M.R., Mulliken, K.M., Dotray, P.J., Ellis, A.P., and Forster, L., 2025, Sample details and compositional data collected during the 2024 Southwest Rift Zone and middle East Rift Zone eruptions of Kīlauea, Island of Hawaiʻi: U.S. Geological Survey data release, https://doi.org/10.5066/P13Z5F2T.

M.O. Garcia, A.J. Pietruzka, M.D. Norman, J.M. Rhodes Kīlauea’s Puʻuʻōʻō Eruption (1983–2018): a synthesis of magmatic processes during a prolonged basaltic event. Chem. Geol, 581 (2021), Article 120391

Garcia MO, Mucek* AE, Lynn† KJ, Swanson DA, Norman MD (2018) Geochemical evolution of Keanakāko‘i Tephra, Kīlauea Volcano, Hawai‘i. In M. P. Poland, M. O. Garcia, V. E. Camp, & A. Grunder, Field Volcanology: A Tribute to the Distinguished Career of Don Swanson. Geological Society of America. https://doi.org/10.1130/2018.2538(09)

I. R. Sides, M. Edmonds, J. Maclennan, D. A. Swanson, B. F. Houghton. Eruption style at Kīlauea Volcano in Hawai‘i linked to primary melt composition. Nature Geoscience, 2014; DOI: 10.1038/ngeo2140