Scientists stand at the crater rim at Kīlauea Volcano on 26 October 2019
U.S. Geological Survey scientists stand at the crater rim at Kīlauea Volcano on 26 October 2019, preparing to use UAS to sample the water lake below. The appearance of this lake after Kīlauea’s 2018 eruption could presage explosive eruptions to come. Credit: Photograph by J. Adams, U.S. Geological Survey

On a misty, gray morning, scientists watched with anticipation as a water sampler dunked beneath the surface of a small lake. Once water was retrieved, the researchers made preliminary measurements of pH and conductivity in the field before preparing the remaining sample for the lab.

Such a scene might be common for hydrologists, ecologists, or water chemists, but this was no ordinary sampling expedition. Last October, it was a first for volcanologists at the U.S. Geological Survey’s (USGS) Hawaiian Volcano Observatory (HVO). The water they collected was not from a typical mountain stream or tropical wetland, but instead from the recently formed and historically unprecedented crater lake at the summit of Kīlauea Volcano. Further, the sample was not obtained by merely walking to the lakeshore and filling a bottle; its retrieval required the use of unoccupied aircraft systems (UAS)—drones in everyday language—operated by a pilot a kilometer away, high above on the crater rim.

The lava lake in Kīlauea’s Halema’uma’u Crater On 3 May 2018 (left) compared with a water lake in the crater on 3 May 2020 (right)
Fig. 1. On 3 May 2018, the lava lake in Kīlauea’s Halema‘uma‘u Crater was approximately 200 meters wide (left). Two years later, Kīlauea’s water lake, shown here on 3 May 2020, is approximately 130 meters wide (right). Lake widths are given for the dimensions from left to right in both photos. Credits: Photographs by C. Parcheta and M. Patrick, U.S. Geological Survey

The 2018 eruptive sequence drained the lava lake and shallow magma reservoir, feeding fissures along the volcano’s lower East Rift Zone that covered 35 square kilometers with lava.

The circumstances were even more remarkable considering that just 18 months earlier, in April 2018, the crater held not a water lake but an active lava lake—the manifestation of a decade-long summit eruption at Kīlauea (Figure 1). What followed from May to September 2018 was one of the largest eruptions, and the most destructive, at Kīlauea since 1790.

The 2018 eruptive sequence drained the lava lake and shallow magma reservoir, feeding fissures along the volcano’s lower East Rift Zone that covered 35 square kilometers with lava [Neal et al., 2019]. That partial evacuation of Kīlauea’s summit magma resulted in a collapsed crater more than 500 meters deeper and 825 million cubic meters larger than it was before the eruption [Anderson et al., 2019], the first water lake in written history at Kīlauea, and a new era of science and monitoring for HVO.

A New Explosive Period?

Since the start of the written record of Kīlauea about 200 years ago, the volcano has repeatedly produced lava flows and lava fountains. This style of effusive eruptive activity is what most people, including many geologists, picture when they think of Kīlauea. Such eruptions, although sometimes destructive, are not often life-threatening. What many people may not realize, however, is that Kīlauea has a long history of large, hazardous explosions.

Cross sectional view of rock and soil deposits left by a past volcanic eruption
Fig. 2. These deposits from 18th century explosive eruptions on the rim of Kīlauea caldera are 1 kilometer northwest of the present-day water lake. The scale shows centimeters and inches. Credit: Photograph by D. Swanson, U.S. Geological Survey

In fact, a period of predominantly explosive activity lasting some 300 years, including an eruption that killed hundreds of people in 1790 (Figure 2), came to an end in the early 1800s. An even longer explosive period took place between 200 BCE and 1000 CE [Swanson et al., 2014].

Why Kīlauea’s eruptive style cycles between effusive and explosive is not entirely understood, but water originating from sources other than magma probably plays a critical role. Textural evidence—such as the dominance of older, recycled rock fragments and the quenched (rapidly cooled) nature of some particles—in Kīlauea’s ash layers indicates that many, perhaps most, explosive eruptions have occurred when groundwater or lake water was vaporized, either by heat given off by the magma (resulting in phreatic eruptions) or by the magma itself (resulting in phreatomagmatic eruptions). Although groundwater alone could have been involved in many explosions, ash from some specific past eruptions suggests that a lake was present at times. That ash contains more dissolved water and sulfur compared with ash that did not interact with lake water and was erupted directly to the atmosphere [Mastin et al., 2004]. Such a lake could have formed after a collapse of the caldera floor and after sufficient cooling of the magma conduit [Hsieh and Ingebritsen, 2019].

Determining whether Kīlauea is likely to erupt explosively—and issuing warnings before it does—is a formidable task for Hawaiian Volcano Observatory (HVO) scientists and their colleagues.

Now there is water in a collapsed caldera once again. Are we entering a new cycle of explosive, rather than effusive, activity at Kīlauea? Will we see dangerous explosions next? Hazards associated with such explosions, including ballistic ejection of rocks, ashfall, and surges of searing ash and gas, could endanger nearby communities as well as visitors to the popular Hawai‘i Volcanoes National Park.

Unfortunately, phreatic activity often begins suddenly and is difficult to forecast. The world saw the havoc that such eruptions can wreak in September 2014, when a small phreatic eruption at Japan’s Mount Ontake killed 63 hikers. A December 2019 phreatic eruption at Whakaari (White Island), New Zealand, resulted in 21 fatalities. Many of Kīlauea’s past explosions have been much larger than those recent events at Ontake and Whakaari.

Determining whether Kīlauea is likely to erupt explosively—and issuing warnings before it does—is a formidable task for HVO scientists and their colleagues.

Unfamiliar Scientific Territory

HVO was founded in 1912 by Thomas Jaggar, an American volcanologist whose pursuits involved measuring lava temperature, collecting volcanic gases, and installing seismometers to detect earthquakes. Over the years, HVO improved on his methods and added other monitoring techniques. In the 1960s, scientists started measuring ground tilt, which can indicate inflation and deflation of subsurface magma reservoirs, electronically. They began using sunlight to measure sulfur dioxide (SO2) emissions in the 1970s, and they employed GPS sensors to detect ground motion beginning in the 1990s [Tilling et al., 2014]. Computational and instrumental advances since then have further revolutionized data collection and analysis.

All the measurements at Kīlauea in HVO’s 108-year history have been made during the volcano’s modern period dominated by effusive activity. More than 50 explosive eruptions did take place at Kīlauea’s Halema‘uma‘u summit crater in 1924, but none were as large as those in 1790 and earlier. Despite indications of water possibly influencing the 1924 explosions [Jaggar and Finch, 1924; Stearns, 1925], no lake was present.

Even without the elevated explosive hazard, the presence of a water lake today warrants adding water sampling to HVO’s repertoire, as water interacts with volcanic gases, particularly SO2. Because magma progressively releases gases during its ascent, increases in SO2 emissions can herald coming volcanic unrest and eruptive activity. However, if those gases dissolve into water rather than escaping to the atmosphere, that geochemical signal could be missed.

By late 2018, Kīlauea’s SO2 emission rate was the lowest measured since the advent of such monitoring in the 1970s, and those low levels persist today. The low rate could result from magma being too deep to release much sulfur, from reduced magma supply to the summit region, or from SO2 released from shallower magma dissolving into water. Thus, the chemistry of the lake water and its dissolved constituents can provide clues about magma beneath Kīlauea’s summit.

The association of surface water with enhanced explosive potential only heightens the need to sample and study the water regularly. But the lake is located at the base of a steep, unstable crater, hundreds of meters below the crater rim, which makes sampling on foot impossible and deployment of geophysical and geochemical sensors for continuous monitoring very difficult. Those steep sides, a lack of emergency landing sites, and the potential for dangerous volcanic gases to pool in the crater also make sampling via helicopter inadvisable.

Yet the need to sample the water remains. So in the tradition of continually pursuing advances in volcano monitoring technology, HVO turned to UAS to draw water from the volcano’s new lake.

Send in the Drones

When Kīlauea’s activity began to escalate in 2018, HVO tapped unoccupied aircraft systems (UAS) to support monitoring efforts, particularly in areas that were inaccessible or too hazardous for ground crews or traditional aircraft.

Unoccupied aircraft systems have become more common in volcanology in recent years, but prior uses of UAS by the USGS for volcano monitoring had occurred primarily through university partnerships or through the USGS Volcano Disaster Assistance Program‘s work with international colleagues. However, when Kīlauea’s activity began to escalate in 2018, HVO tapped UAS to support monitoring efforts, particularly in areas that were inaccessible or too hazardous for ground crews or traditional aircraft. The ensuing 4-month-long federal government UAS response was the longest and largest of its kind—involving more pilots, flights, and data collection than previous efforts—and it marked the first time the USGS deployed its own UAS fleet in response to a volcanic crisis.

UAS were used extensively throughout the eruption to provide quick-turnaround data to scientists and on-demand, 24/7 support to emergency managers. Aboard various UAS, including small and large multirotor and fixed-wing aircraft, were gas sensors as well as cameras that provided livestream video to emergency operations centers. Videos taken over the lava channel enabled volcanologists to calculate the rates at which lava was erupting. Photogrammetry surveys provided data for high-resolution digital elevation models (DEMs), which yielded estimates of erupted lava volume, flow advance rates, and flow path forecasts. At Kīlauea’s summit, a time series of DEMs captured the caldera collapse in unprecedented detail.

After the 2018 eruption, it was clear that the need for UAS at Kīlauea would persist. For years, scientists had made gas measurements at Kīlauea’s summit both on foot and by vehicle traverse. But the road used for these measurements had collapsed into the crater. Gas vents (fumaroles) that HVO sampled to determine gas chemistry were also destroyed by the collapse. Although new fumaroles have since appeared, they are all inaccessible, located on the steep walls of the deepened crater.

Since 2018, the use of UAS has supplanted the now unfeasible former methods. A specialized sensor package that samples emitted gases to measure concentrations of carbon dioxide (CO2), SO2, hydrogen sulfide, and water can be mounted on UAS. Data collected are then used to assess the spatial distribution of outgassing, to determine the chemistry of fumarolic gases, and to make preliminary assessments of the volcano’s CO2 emission rate, which can provide information about deep magma supply to the volcano.

Close study of the chemistry and textures of older deposits at Kīlauea is crucial for assessing the volcano’s past, and UAS provide the safest way to begin investigating rocks newly exposed by the crater collapse that have never before been studied [Anderson et al., 2019]. But as with the fumaroles, many sites of interest are on steep crater walls or even on cliff faces. HVO will use the imaging capabilities of UAS to assess which deposits will be most valuable for study and which might be safely accessible for sampling.

A scientist kneels on a tarp watching a multirotor unoccupied aircraft system (UAS) take off (top); the UAS hovers over the surface of Kīlauea’s water lake in October 2019 (bottom)
Fig. 3. A scientist from the Hawaiian Volcano Observatory readies the water sampler, attached to the UAS, for takeoff (top). The UAS (small dark object near center of photo) hovers over the lake surface on 26 October 2019 (bottom). Credits: Photographs by (top) M. Warren and (bottom) C. Parcheta, U.S. Geological Survey

With UAS becoming a critical tool at HVO, turning to them to sample the new water lake, first spotted in July 2019, was a logical step. Over the past decade, UAS have been used for sampling nonvolcanic water bodies [Lally et al., 2019]. More recently, volcanic crater lakes have been successfully sampled using UAS, including Kusatsu-Shirane Volcano, Japan [Terada et al., 2018], and Whakaari [Kilgour and Scott, 2019].

However, preparations were not as straightforward as simply devising a sampler and heading into the field. Kīlauea’s summit, and Halema‘uma‘u Crater in particular, is regarded as sacred by Native Hawaiians and other groups, and it is located within Hawai‘i Volcanoes National Park, under the purview of the National Park Service. As part of obtaining formal permission from the National Park Service for UAS water sampling, the park’s Kūpuna consultation group, which includes Native Hawaiian organizations and individuals as well as other interested parties, was asked to identify and address concerns about this work occurring on a sacred landscape. The group was supportive of the scientific objectives of sampling, which allowed HVO to move forward with utilizing UAS as the least invasive and safest tool to accomplish the task.

On 26 October 2019, HVO scientists obtained the first sample from the growing lake (Figure 3). A second effort in January 2020 also succeeded in retrieving water samples and temperature data.

A Watery Window to Activity Below

Prior to the appearance of the water lake, the only window into the state of groundwater at Kīlauea’s summit was through periodic water level and chemistry monitoring of a research well drilled in the 1970s about 1 kilometer from Halema‘uma‘u Crater. Sulfate (a product of SO2 dissolving in water) and chloride ion concentrations in the well water increased ahead of the 2008 summit lava lake eruption, indicating that groundwater and magmatic gases were interacting [Hurwitz and Anderson, 2019].

However, the concentration of sulfate in the water lake is more than 7 times the highest concentrations ever measured in the well. A concentration that high suggests that significant amounts of SO2, which would be emitted into the atmosphere in the absence of water, may instead be dissolving in the lake (and/or the surrounding groundwater), thereby decreasing the SO2 emission rate to the atmosphere. Alternatively, rain percolating through the ground before reaching the lake could be leaching sulfur-rich deposits that accumulated in the summit area before the former lava lake drained.

Either scenario would affect the chemistry of the new lake, which differs from that of crater lakes at other volcanoes. Hyperacidic volcanic lakes, including those at Poás in Costa Rica, Kawah Ijen in Indonesia, and Whakaari, typically have a pH below 1 (highly acidic), whereas the pH of Kīlauea’s lake is roughly 4 (moderately acidic) and the pH of the well water is about 7 (neutral).

Laser rangefinder measurements made from the crater rim show that the lake has been deepening by nearly a meter per week since it appeared, although this rate seems to be slowing in recent months. Further sampling missions will be necessary to identify any fluctuations in water chemistry and to determine their causes as the lake continues to grow and change.

The chance to monitor an incipient volcanic lake is not unprecedented, but it is rare. Kīlauea’s crater lake provides an opportunity to improve the scientific community’s understanding of how such lakes evolve and interact with magmatic systems below.

Deciphering Kīlauea’s Signals

Phreatic eruptions are observed at other volcanoes, but we do not yet know the precursors that might presage such an explosion at Kīlauea specifically.

It is undoubtedly a new era for Kīlauea Volcano and HVO. Even as we at the observatory use modern tools like UAS, we continue to work to understand the new paradigm of this water lake and an increased potential for explosive eruptions. Given the history of explosive eruptions preserved in the geologic record and in Hawaiian oral tradition, similar lakes may once have been common at Kīlauea; we just have not seen one in modern times.

Phreatic eruptions are observed at other volcanoes, but we do not yet know the precursors that might presage such an explosion at Kīlauea specifically. Will these precursors be the same as those at other volcanoes or unique to Kīlauea? What signals might we detect in the water chemistry or the seismicity at the volcano that could clue us in on—and allow us to sound alerts about—potential hazards on the horizon?

Addressing these questions requires that HVO adapt its monitoring and mitigation strategies. At the same time, we must work with local communities, local government, and Hawai‘i Volcanoes National Park to prepare best for whatever happens next at Kīlauea.


Sampling of the lake was conducted under research permit HAVO-2019-SCI-0046 from the National Park Service, and we thank Hawai‘i Volcanoes National Park for assistance and support. We also thank colleagues who made the UAS sampling possible, especially Frank Younger, Sara Peek, Tamar Elias, Rich Thurau, Joe Adams, Todd Burton, Tina Neal, and the Department of the Interior’s Office of Aviation Services. Wendy Stovall and Larry Mastin are thanked for reviews that improved this article.


Anderson, K. R., et al. (2019), Magma reservoir failure and the onset of caldera collapse at Kīlauea Volcano in 2018, Science, 366(6470), eaaz1822,

Hsieh, P. A., and S. E. Ingebritsen (2019), Groundwater inflow toward a preheated volcanic conduit: Application to the 2018 eruption at Kīlauea Volcano, Hawai‘i, J. Geophys. Res. Solid Earth, 124(2), 1,498–1,506,

Hurwitz, S., and K. R. Anderson (2019), Temporal variations in scrubbing of magmatic gases at the summit of Kīlauea Volcano, Hawai‘i, Geophys. Res. Lett., 46(24), 14,469–14,476,

Jaggar, T.A., and R. H. Finch (1924), The explosive eruption of Kilauea in Hawaii, 1924, Am. J. Sci., ser. 5, 8(47), 353–374,

Kilgour, G., and B. Scott (2019), Sampling the inaccessible: A drone at work on Whakaari/White Island, GeoNet,

Lally, H. T., et al. (2019), Can drones be used to conduct water sampling in aquatic environments? A review, Sci. Total Environ., 670, 569–575,

Mastin, L. G., et al. (2004), What makes hydromagmatic eruptions violent? Some insights from the Keanakāko‘i Ash, Kı̄lauea Volcano, Hawai‘i, J. Volcanol. Geotherm. Res., 137(1), 15–31,

Neal, C. A., et al. (2019), The 2018 rift eruption and summit collapse of Kīlauea Volcano, Science, 363(6425), 367–374,

Stearns, H. T. (1925), The explosive phase of Kilauea Volcano, Hawaii, in 1924, Bull Volcanol., 2(2), 193–208,

Swanson, D. A., et al. (2014), Cycles of explosive and effusive eruptions at Kīlauea Volcano, Hawai‘i, Geology, 42(7), 631–634,

Terada, A., et al. (2018), Water sampling using a drone at Yugama crater lake, Kusatsu-Shirane volcano, Japan, Earth Planets Space, 70(1), 64,

Tilling, R. I., et al. (2014), The Hawaiian Volcano Observatory: A natural laboratory for studying basaltic volcanism, in Characteristics of Hawaiian Volcanoes, U.S. Geol. Surv. Prof. Pap., 1801-1, 1–64,

Author Information

Patricia A. Nadeau (, Hawaiian Volcano Observatory, USGS, Hilo; Angela K. Diefenbach, Cascades Volcano Observatory, USGS, Vancouver, Wash.; Shaul Hurwitz, California Volcano Observatory, USGS, Moffett Field; and Donald A. Swanson, Hawaiian Volcano Observatory, USGS, Hilo


Nadeau, P. A.,Diefenbach, A. K.,Hurwitz, S., and Swanson, D. A. (2020), From lava to water: A new era at Kīlauea, Eos, 101, Published on 25 September 2020.

Text not subject to copyright.
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.