The most recent eruption of Kīlauea volcano on the island of Hawaii began in 1983. For 35 years, most of its magma emerged from a set of fissures in the volcano called the upper east rift zone. But on 3 May 2018, Kīlauea’s lower east rift zone opened up, giving way to a massive outpouring of lava that devastated the southeastern part of the island1 (Fig. 1). An important question is why this change occurred in May 2018, rather than earlier or later in the course of the eruption. Writing in Nature, Farquharson and Amelung2 propose that record-breaking levels of rainfall in early 2018 increased groundwater pressures which, in turn, made it easier for rock to break and hence magma to rise to the surface at new locations.
The creation of a pathway that brings magma to Earth’s surface begins with the mechanical failure of rocks. This failure can occur in two ways: new cracks can open, or existing faults can slip. Both processes can be promoted by pressure changes in groundwater. For the former, increases in fluid pressure decrease the amount of stress needed to open new cracks. For the latter, faults can slip when the stresses acting parallel to the fault (shear stresses) overcome those perpendicular to the fault (normal stresses). These normal stresses act to clamp the fault shut. Increasing fluid pressure in rocks lowers normal stresses without changing shear stresses, thus promoting fault failure.
Heavy rainfall increases water levels underground and thus pressure in groundwater. The volcanic rocks in Hawaii are very permeable, which allows water to infiltrate and pressure changes to propagate to a depth of several kilometres, close to where magma is stored. Fluid-pressure changes take time to propagate from the surface to those depths. Thus, downward migration of rock failure over time, along with a time lag between the accumulation of water at the surface and failure at depth3, would be key indicators that rainfall was the cause of rock failure at Kīlauea.
Farquharson and Amelung modelled pressure changes at Kīlauea caused by rainfall in the months leading up to the eruption on 3 May 2018. Their model showed an increase in pressure of tens to hundreds of pascals at depths of several kilometres. On the basis of these changes, along with four sets of observations indicating that eruptions at Kīlauea are associated with patterns of substantial rainfall, the authors propose that heavy rainfall promoted the rock failure that enabled magma to flow into the lower east rift zone.
Is their hypothesis plausible? The pressure changes computed by their models are small — smaller than stresses from tides. However, if rocks are already close to breaking, such changes might be sufficient to initiate failure. The 2018 eruption was accompanied by a magnitude-6.9 earthquake, and examples of earthquakes caused by pressure changes on this scale are abundant4. For example, the widespread increase in earthquake frequency in the central and eastern United States in the past decade results from wastewater injection into permeable rocks that increases water pressure and changes stresses5.
The geological record also confirms that changes in stresses at Earth’s surface can modulate volcanic activity. On land, volcanism is promoted by the retreat of glaciers6. Sea-level changes between glacial and interglacial periods can modulate eruption rates at mid-ocean ridges7. Stresses from large earthquakes increase the probability of volcanic eruptions8 and can change activity at volcanoes that are already active9.
Although it is well established that changes in water pressure promote earthquakes, they are not necessarily a direct cause of magma eruption. To begin moving through Earth’s crust, magma must create large enough stresses in the surrounding rocks to open a pathway. Earthquakes triggered in the crust around that stored magma, however, can actually relieve stress — as such, they might make it more difficult for magma to erupt10.
Ultimately, whether fault failure from water-pressure changes can occur close to stored magma, as hypothesized by Farquharson and Amelung, remains uncertain. The first magma to erupt from the lower east rift zone in 2018 was old, perhaps left over from an earlier, 1955 eruption11, implying that the rift zone was already hot. As a result, groundwater in the rift zone might have been vapour at shallow depths12, and at greater depths it could have been a supercritical fluid (a substance that is not in a distinct liquid or gas phase, but has properties of both). The high compressibility of both vapours and supercritical fluids would dampen the magnitude of pressure changes in the authors’ model, making failure less probable.
How, then, can we test the hypothesis that rainfall initiated the lower east rift zone eruption? Unfortunately, subsurface pressure measurements — and hydrogeological data more generally — are rarely part of volcano monitoring. Instead, as with many geoscience and Earth-history questions, we have to look back in time using the geological and historical record of eruptions. In support of their hypothesis, Farquharson and Amelung analysed all reported eruptions at Kīlauea since 1790, and showed that the volcano tends to erupt at the wettest time of year.
Should we increase alert levels at volcanoes after heavy rainfall? We could ask the same question about other stress changes, such as those from regional earthquakes. This is an open question. These stress changes are small, and hence, if anything, modulate the exact timing of the surface eruption. At Kīlauea, there were other sources of stress — in fact, a change in eruption behaviour had been anticipated on the basis of ground-deformation measurements and inferred magma movement. The Hawaiian Volcano Observatory issued a warning on 17 April that a new vent might open1.
The possibility that external processes initiate volcanic eruptions is a reminder that volcanoes are part of a dynamic Earth system. Volcanic eruptions influence all surface environments, including climate and weather13. Changes in those surface environments, such as heavy rainfall, might also influence eruptions. We are only just beginning to understand these interactions.
Nature 580, 457-458 (2020)