4

How Do Earth Systems Interact with Eruptions?

Implicit in the goals of eruption forecasting is the assumption that improved forecasts will help to mitigate the immediate impacts of volcanic eruptions (see Chapter 3). Also critical, however, are long-term forecasts of very large eruptions and their potential for both global and long-lived impacts to Earth’s environment. Volcanoes affect a host of Earth systems and vice versa. Thus, two central questions about the spatial and temporal impacts of large volcanic eruptions are (1) How do landscapes, the hydrosphere, and the atmosphere respond to volcanic eruptions? and (2) How do volcanoes respond to tectonic and climate forcing?

4.1 HOW DO LANDSCAPES, THE HYDROSPHERE, AND THE ATMOSPHERE RESPOND TO VOLCANIC ERUPTIONS?

The products of volcanic eruptions change landscapes and introduce particles and gases into the atmosphere and oceans. The immediate impacts of small to large (Volcano Explosivity Index [VEI] ≤6) volcanic eruptions on Earth systems are generally well known (Section 2.3) through observations of historical eruptions. However, the impacts of larger eruptions, such as the last super-eruption 26,000 years ago (Oruanui, New Zealand), are less well understood. Important unanswered questions are whether the impacts of very large eruptions can be anticipated by scaling up the impacts of smaller eruptions (e.g., Self, 2006) or whether the impacts of very large eruptions may be self-limiting (e.g., Oppenheimer, 2002; Timmreck, 2012; Timmreck et al., 2009). That is, will very large eruptions have unanticipated consequences for the environment and hence for human populations?

Effect on Landscapes

Volcanic eruptions can profoundly change the landscape, initially through both destructive (flank failure and caldera formation) and constructive (lava flows, domes, and pyroclastic deposits) processes, which destroy vegetation and change the physical nature of the surface (e.g., porosity, permeability, and chemistry). After explosive activity ends, secondary hazards may continue to affect local and global environments for months, years, or decades. These hazards include explosions within pyroclastic flows that occur within a few months of pyroclastic density current emplacement (Torres et al., 1996), catastrophic breakouts of lakes dammed by volcaniclastic material years after the damming event (Manville and Cronin, 2007), rainfall-generated lahars that mobilize loose pyroclastic debris for years to decades after a large eruption (Major et al., 2000; Rodolfo et al., 1996), phreatic eruptions from hydrothermal systems (e.g., Barberi et al., 1992), and sudden releases of CO₂ from volcanic lakes (e.g., Funiciello et al., 2003; Zhang, 1996).

More generally, changes in the infiltration capacity

of disturbed landscapes can greatly increase flooding and sediment transport (Pierson and Major, 2014) or, conversely, enhance remobilization of volcanic ash by wind for decades, centuries, or even millennia after a large eruption. Volcanic dust, in particular, is easily remobilized from the surface of pyroclastic deposits, as illustrated by frequent dust storms downwind of historically active volcanic regions (e.g., Liu et al., 2014; Wilson et al., 2011). Studies on the adverse effects of remobilized ash on ecosystems are few, but are increasingly recognized as an important component of ecosystem response and recovery. On even longer time scales, the landscape continues to respond by erosion and redeposition of loose surface material, rearrangement of drainage systems, regrowth of often different vegetation, and reintroduction of fauna. There are no comprehensive studies of the nature and time scales of landscape and ecosystem response, although detailed studies have traced recovery after individual volcanic eruptions (e.g., Dale et al., 2005; Del Moral and Bliss, 1993; Dull et al., 2001; Egan et al., 2016; Gunnarsson et al., 2017; Long et al., 2014; Walker et al., 2013).

Effect on the Subsurface Hydrosphere

The effects of eruptions on Earth surface processes are easy to observe and thus are fairly well quantified. Less apparent are the effects of reawakening magmatic systems on subsurface processes, particularly hydrothermal systems important for generation of energy and, over longer time spans, formation of ore deposits. Observable interactions of magmatic and groundwater systems include geophysical and geochemical signals that can be difficult to distinguish from signals of magmatic unrest. Although volcanic eruptions are commonly preceded and followed by phreatic eruptions from hydrothermal systems (e.g., Barberi et al., 1992), phreatic eruptions may also occur without warning during periods of repose and so pose a substantial forecasting challenge. Similarly, magmatic CO₂ leaked slowly into volcanic lakes can suddenly destabilize and release lethal dense gas plumes (e.g., Funiciello et al., 2003; Zhang, 1996).

Beneath the surface, magmatic–geothermal systems can generate geothermal energy and create ore deposits. Porphyry deposits in volcanic arcs provide about 75 percent of the world’s copper, 50 percent of its molybdenum, 20 percent of its gold, and many metals that underpin emerging low carbon technologies (Sillitoe, 2010). It had generally been assumed that voluminous explosive volcanism is incompatible with porphyry formation. Active magmatic systems, however, are able to provide the requisite metal-bearing brines (e.g., Chelle-Michou et al., 2017), and copper ore precipitates when this brine interacts with sulfur-rich gases released from the underlying magmatic system (Blundy et al., 2015). This newly emerging understanding posits an active role for magmatism, and raises new questions about the timing of magmatism and ore formation.

Effect on the Atmosphere and Climate

Large volcanic eruptions can inject enough H₂O, CO₂, SO₂, and other volatiles (e.g., halogen species) into the upper troposphere and stratosphere to influence atmospheric chemistry and climate (Robock, 2000; Figure 4.1). Although CO₂ emitted from erupting and passively degassing volcanoes is the major pathway for mantle-derived CO₂ to enter the atmosphere (Kelemen and Manning, 2015), it is a minor component of the global mass of atmospheric CO₂ (Burton et al., 2013). For this reason, CO₂ release from all but the very largest eruptions is unlikely to change climate significantly (Self et al., 2014), although methane and CO₂ release from igneous intrusions in carbon-rich sediment can greatly increase gas emissions (e.g., Aarnes et al., 2010; Svensen et al., 2007).

The short-term effects of explosive volcanic eruptions on climate arise from the injection of volcanic SO₂ into the stratosphere where it transforms to sulfate aerosols that can persist for years, backscattering sunlight and cooling Earth’s lower atmosphere and surface (Robock, 2000; see Section 2.3). Emissions of SO₂ from human activities and volcanoes, including diffuse emissions from nonerupting volcanoes, are shown in Figure 4.2. Volcano location plays an important role, with tropical eruptions being more capable of producing global impacts because seasonal variations in the Intertropical Convergence Zone facilitate transfer of aerosols between hemispheres (e.g., Kravitz and Robock, 2011; Oman et al., 2006). For this reason, even relatively small, but frequent, injections of SO₂ into the stratosphere by moderate tropical eruptions (VEI ≤4)

may sustain the background stratospheric sulfate layer and affect climate (e.g., Santer et al., 2014; Solomon et al., 2011; Vernier et al., 2011). Less well understood are the impacts of major volcanic injections of halogen gases (Cl, Br) into the stratosphere, which could cause significant ozone depletion and generate localized ozone holes (e.g., Cadoux et al., 2015; Kutterolf et al., 2013).

The best documented global climate impact of large explosive eruptions is cooling, typically followed by winter warming of Northern Hemisphere continents, as illustrated by the 1991 eruption of Pinatubo (McCormick et al., 1995; Robock, 2000). In that event, ~10⁴ teragrams of erupted magma injected 30 teragrams of aerosols into the stratosphere, the largest stratospheric loading of the past century (Figure 4.1). The negative radiative forcing caused largely by stratospheric sulfate aerosols resulted in a global tropospheric cooling of 0.2°C relative to the baseline from 1958–1991. Adjusted for the warming effect of the El Niño–Southern Oscillation (ENSO), the overall temperature decrease was 0.7°C. This temperature decrease is similar to those estimated for other sulfur-rich eruptions, such as Krakatau (1883) and Tambora (1815) in Indonesia and El Chichon (1982) in Mexico. Such temperature anomalies are short lived, so that by 1993 the tem-

perature anomaly caused by the Pinatubo eruption had already decreased to –0.1°C (McCormick et al., 1995).

The relationship between cooling and large explosive eruptions is complex and includes not only the effect of SO₂ gas but also the effects of other emitted material (particularly H₂O, halogens, and ash), as well as the details of atmospheric chemistry that control the production and size of volcanic aerosols (e.g., LeGrande et al., 2016; Timmreck, 2012; Timmreck et al., 2009). For example, SO₂ is a greenhouse gas that could counteract the cooling effect of sulfate aerosols (Schmidt et al., 2016). Thus, the balance between SO₂ and aerosols in different parts of the atmosphere is complicated, as is the resulting climate response.

Large explosive eruptions can also affect global circulation patterns such as the North Atlantic Oscillation and ENSO (Robock, 2000), although the mechanism(s) by which this happens are not well understood (LeGrande et al., 2016). Finally, eruptions have been linked to substantial but temporary decreases

in rainfall and river discharge (e.g., Oman et al., 2006; Trenberth and Dai, 2007) and the occurrence of tropical cyclones in the North Atlantic (Guevara-Murua et al., 2015). Documentation of the atmospheric impact of recent explosive eruptions provides important constraints for testing short-term climate model predictions and for exploring the effects of proposed geoengineering solutions to global warming (e.g., Robock et al., 2008, 2009).

Large effusive eruptions have a somewhat different effect on the atmosphere because of their long durations (e.g., Schmidt et al., 2016; Thordarson and Self, 2003). Basaltic eruptions, in particular, can be both voluminous and long lived, and can therefore affect local, regional, and possibly global climate. Historical examples from Iceland, such as the Laki eruption of 1783–1784 and the Bárðarbunga eruption of 2014–2015, provide an interesting contrast. The former had a regional (Northern Hemisphere) impact in the form of dry fogs of sulfuric acid (H₂SO₄), while the latter produced dangerously high local levels of SO₂. The difference reflects not only the larger volume of the Laki eruption, but also the season (summer versus winter) because sunlight plays an important role in the oxidation of SO₂ to H₂SO₄ (Gislason et al., 2015; Schmidt et al., 2010). In the extreme, the large volume and long duration of ancient flood basalts may have perturbed the atmosphere over time scales of decades to centuries to even millennia (Figure 4.1).

The effects of injecting large amounts of water by volcanic eruptions into the dry stratosphere could affect climate by accelerating the formation of sulfate aerosol by OH radicals or by decreasing the ozone formation potential of the system (Glaze et al., 1997; LeGrande et al., 2016). Studies of very large flood basalt eruptions suggest that both the formation of sulfate aerosols and the depletion of ozone played a significant role on climate over Earth’s history (Black et al., 2014). These examples emphasize the need to better characterize plume gas and aerosol chemistry as well as coupling of gas-phase chemistry with aerosol microphysics in climate models. Because satellite-based remote sensing observations of volcanic gases are heavily biased toward SO₂ (e.g., Carn et al., 2016), obtaining a complete volatile inventory for explosive eruptions required for a full chemistry simulation of volcanic plumes is still a major challenge.

Effect on the Oceans

Large eruptions affect Earth’s oceans in a variety of ways. Volcanic ash may be a key source of nutrients such as iron and thus capable of stimulating biogeochemical responses (Duggen et al., 2010; Langmann et al., 2010). During the week following the 2003 VEI 4 eruption of Anatahan, Northern Mariana Islands, for example, satellite-based remote sensing detected a 2–5-fold increase in biological productivity in the ocean area affected by the volcanic ash plume (Lin et al., 2011). These impacts can be particularly pronounced in low-nutrient regions of the oceans. A more indirect and longer-term impact of very large volcanic eruptions is caused by the rapid addition of CO₂ and SO₂ to the atmosphere, which affects seawater pH and carbonate saturation. Carbon-cycle model calculations (Berner and Beerling, 2007) have shown that CO₂ and SO₂ degassed from the 201-million-year-old basalt eruptions of the Central Atlantic Magmatic Province could have affected the surface ocean for 20,000–40,000 years if total degassing took place in less than 50,000–100,000 years. Ocean acidification from the increased atmospheric CO₂ may have caused near-total collapse of coral reefs (Rampino and Self, 2015). Rapid injection of large amounts of CO₂ into the atmosphere by volcanic eruptions also provides the best analog for studying the long-term effects of 20th-century CO₂ increases on ocean chemistry. Targeted investigations of these large eruptions have the potential to establish quantitative estimates of the volatile release and residence in the atmosphere as well as the effects on ocean acidification, carbon saturation, coral mortality, and biodiversity.

Over the long term, large eruptions can release thousands of gigatons of methane from organic-rich sediments. Light δ¹³C signatures interpreted to represent such a release (Svensen et al., 2009) have been recognized in carbon isotope stratigraphic records at the Permian–Triassic (252 Ma) and Triassic–Jurassic (201 Ma) boundaries, as well as in the Paleogene (56 Ma; Saltzman and Thomas, 2012). The latter represents a well-documented thermal maximum associated with extensive volcanism that accompanied the opening of the North Atlantic Ocean. Reconstructing the volcanic carbon emission record through geologic time and assessing the potential for large releases of reduced carbon from organic sediments is challenging and requires

a firm understanding of the processes that currently degas carbon and other volatiles to the atmosphere and how those signatures may be preserved in the geologic and ice core records.

Finally, some secondary volcanic hazards are generated in the ocean. Tsunamis can be generated directly by explosive submarine eruptions (e.g., Fiske et al., 1998), or indirectly by volcanic flows (pyroclastic, lahar) or debris avalanches produced by volcano flank collapses (e.g., Paris, 2015). Even small volcano-triggered tsunamis can produce significant waves (e.g., Day, 2015).

Key Questions and Research Priorities on the Response of Landscapes, the Hydrosphere, and the Atmosphere to Volcanic Eruptions

4.2 HOW DO VOLCANOES RESPOND TO TECTONICS AND CHANGES IN CLIMATE?

Volcanic eruptions can be triggered when the pressure in a subsurface magma body exceeds the confining pressure in the surrounding crust, or when underpressure initiates collapse. The latter includes a contribution from surface loading (e.g., ice sheets). Active volcanoes are therefore sensitive to changes in stress, particularly those systems that are “primed” for eruption (Bebbington and Marzocchi, 2011). An external forcing mechanism that either increases magmatic overpressure or reduces the confining pressure can potentially trigger an eruption. The sources of such perturbations operate on time scales that range from near-instantaneous stress changes associated with tectonic processes such as earthquakes, to longer-term variations due to climate change such as changes in sea level and melting of ice sheets. A deeper understanding of external stimuli (tectonics, earthquakes, changes in sea level or glaciers) provides an important test of mechanisms for melt accumulation and triggering thresholds (Figure 4.3) and is necessary for improved hazard mitigation.

Tectonics

Tectonics influences volcanism by controlling the composition and amount of magma generated in the mantle and the thickness of the crust and the stresses that hinder or promote magma intrusion and ascent. Quantifying these connections would benefit from a better understanding of the properties of the crust that host magma bodies as well as the conditions that enable the propagation of dikes (Section 2.1). For example, large, silicic magma bodies that can produce caldera-

forming eruptions are more likely to develop in thicker crust, whereas more frequent eruptions of less evolved magmas are more likely to develop in thinner, extended crust (e.g., Cembrano and Lara, 2009). There are many exceptions, however. For example, one of Earth’s most frequently active silicic volcanic systems, the Taupo volcanic zone (New Zealand), is located in an extensional area. Tectonic stresses also affect magma storage and the size of eruptions (e.g., Robertson et al., 2016).

Tectonics also influences the morphology and stability of volcanoes. Volcanoes may develop on large tectonic faults (e.g., Socompa; Wadge et al., 1995) or generate faults around their base by gravitational and magmatic deformation (e.g., Etna; Acocella and Neri, 2005). Movement on tectonic faults intersecting volcanic edifices may increase the risk of flank collapse and the generation of debris avalanches, but at the same time may inhibit magmatic processes by relieving stress (e.g., Ebmeier et al., 2016). Regional stresses and faults may control the alignment of dikes, but the extent to which ambient stresses are modified by the development of magma reservoirs (e.g., Andrew and Gudmundsson, 2008; Karlstrom et al., 2009) and loading by volcanic edifices (e.g., Pinel and Jaupart, 2003) remains an open question.

Earthquakes

On a global scale, volcanism and large earthquakes are strongly spatially correlated. Most of Earth’s explosive volcanoes are adjacent to subduction zones, which also generate the largest earthquakes. Temporal coincidences between earthquakes and eruptive activity have been documented since at least the writings of Pliny (his encyclopedia published in the 1st century AD). Analysis of recent earthquake and eruption catalogs shows a spike in volcanic eruptions within a few days after major (M >8) earthquakes, hinting at short-term eruption triggering at distances of many hundreds of kilometers from the epicenter (e.g., Linde and Sacks, 1998; Manga and Brodsky, 2006; Walter and Amelung, 2007). Eruption rates in the southern Andes may have increased for up to 12 months following some large earthquakes (Watt et al., 2009). However, large earthquakes do not always trigger volcanic eruptions. For example, neither the 2010 Maule nor the 2011 Tohoku earthquakes, which were of large magnitude and occurred in active and well-instrumented volcanic arcs, have been linked to triggered eruptions, perhaps because few volcanoes are “critically poised” and susceptible to triggering at any given time. The possibility of delayed triggering (e.g., the 1991 Pinatubo eruption 11 months after the M 7.8 1990 Luzon earthquake) becomes increasingly difficult to establish with time after an earthquake (Hill et al., 2002).

Persistently active volcanoes such as Merapi, Indonesia, may be particularly prone to triggered responses (e.g., Walter et al., 2007). The orientation

of the earthquake focal mechanism with respect to distal volcanoes may also determine whether a triggered response occurs (e.g., Delle Donne et al., 2010). Eruptions have been attributed to earthquake-induced compression (e.g., Bonali et al., 2013; Feuillet et al., 2011; Nostro et al., 1998) or expansion of the crust (e.g., Fujita et al., 2013; La Femina et al., 2004; Walter and Amelung, 2007), nucleation or growth of bubbles (e.g., Crews and Cooper, 2014), mobilization of crystal-rich magmas by dynamic strains (e.g., Sumita and Manga, 2008), initiation of convection (e.g., Hill et al., 2002), and resonance phenomena (e.g., Namiki et al., 2016) in magma chambers. On longer time scales, earthquake-triggered ascent of deeper magmas or gases may play a role. Despite decades of study, however, the mechanisms through which seismic waves and static stress changes initiate eruptions and influence ongoing eruptions, even on short time scales, remain unknown.

Earthquakes can also trigger noneruptive unrest (seismicity, gas emissions, and changes in hydrothermal systems) at volcanoes (e.g., West et al., 2005). Indeed, hydrothermal systems are particularly sensitive to earthquakes (e.g., Ingebritsen et al., 2015). The availability of decadal or longer time series of satellite observations have facilitated investigation of links between volcanic unrest and earthquakes, especially for volcanoes without ground-based instruments. These observations reveal a range of noneruptive volcanic responses to earthquakes, including ground deformation, changes in surface heat flux, induced volcanic seismicity, and hydrologic changes (e.g., Delle Donne et al., 2010; Harris and Ripepe, 2007). Some responses suggest that eruption is less likely. Subsidence recorded at several Chilean and Japanese volcanoes following the 2010 Mw 8.8 Maule, Chile (Pritchard et al., 2013) and the 2011 Mw 9 Tohoku, Japan (Takada and Fukushima, 2013), earthquakes was attributed to coseismic release of hydrothermal fluids and enhanced subsidence of a hot, weak plutonic body, respectively. Deep long-period seismicity also decreased at Mauna Loa after the 2004 Mw 9.3 Sumatra earthquake (Okubo and Wolfe, 2008).

Volcanoes can also influence other volcanoes nearby (e.g., Linde and Sacks, 1998). Coupled eruptions have been documented, with pairs occurring within 50 km of each other (e.g., Biggs et al., 2016; Figure 4.3). The ability to predict and explain volcano responses to earthquakes and other volcanoes would be a significant advance that would aid in the interpretation of persistent unrest, such as Long Valley, California.

Climate

Although it is well understood that volcanic eruptions can impact climate (Section 4.1), relatively little attention has been paid to the potential impacts of future climate change on volcanic activity and hazards (Tuffen, 2010). On various time scales (annual to millennial), volcanoes and volcanic regions may respond to the slow surface deformation associated with seasonal and climatic cycles, such as the growth and melting of glaciers and ice sheets, and changes in sea level (e.g., Jellinek et al., 2004; Maclennan et al., 2002; Mason et al., 2004; Mather, 2015; McGuire et al., 1997; Rawson et al., 2016; Tuffen, 2010; Watt et al., 2013). Surface pressure changes induced by these processes can affect rates of decompression melting in the mantle, drive magma ascent through deformation of the crust, or lead to volatile exsolution and eruption.

Identifying correlations between volcanic activity and climate cycles relies on accurate and complete catalogs of eruptions and intrusions. Major eruptions (VEI >5) are infrequent, but their occurrence is usually, although not always, well preserved in geologic or proxy records (e.g., Rougier et al., 2016). Smaller eruptions (VEI 0–3) are more frequent and hence provide better statistics, but catalogs of such events are incomplete (e.g., Watt et al., 2013). Seasonal fluctuations of up to 50 percent of average eruption rates occur in some regions for small (VEI 0–2) eruptions (Mason et al., 2004). This fluctuation is attributed to surface deformation associated with the seasonal transfer of water between the oceans and landmasses, with volcanic eruptions more likely during periods of surface pressure change.

Large-scale melting of ice can affect the timing of eruptions. Increases in volcanic activity lag ice retreat by several thousand years at stratovolcanoes in California and Chile (Jellinek et al., 2004; Rawson et al., 2016), whereas volcanic activity in Iceland accelerated more quickly following the last deglaciation (e.g., Maclennan et al., 2002). Although glacial unloading is effectively instantaneous on geologic time scales, the lag times probably reflect the variable depth of magma supply and the transit time through the crust. At some

arc volcanoes, observed lag times are shorter for eruptions of silicic magmas, which reside in shallow crustal magma chambers, than for less evolved magmas that are replenished by decompression melting in the mantle (e.g., Jellinek et al., 2004; Rawson et al., 2016).

Melting of ice leads to rising sea levels, but the volcanic response to sea-level change may promote or suppress eruptions depending on volcano type and location (McGuire et al., 1997). At mid-ocean ridges, changes in magma production may be recorded in seafloor topography (Crowley et al., 2015) and may provide CO₂-driven feedbacks with 10⁵-year time lags (Burley and Katz, 2015). Hence, the feedbacks between volcanism, ice removal, and sea-level rise may be global (e.g., Huybers and Langmuir, 2017) but may also be highly variable on local and regional scales.

Changing sea level may indirectly affect eruptions by affecting flank collapse or other mass wasting events (e.g., Coussens et al., 2016). In addition, unloading the volcano may initiate eruptions (e.g., Cassidy et al., 2015). The interrelationship between flank collapse, climate, and volcanic eruptions is best deciphered from the marine sediment archive, accessible by deep sea drilling.

Although volcanic responses to glacial cycles and sea-level changes are likely the dominant climatic influence on volcanism, weather and climate can impact volcanism in other ways. Volcanic activity can be triggered by rainfall (e.g., Matthews et al., 2009; Violette et al., 2001), and there is evidence that the likelihood of volcanic flank collapse may increase in a wetter climate (e.g., Deeming et al., 2010). Future climate change may also shift the extent and/or location of the tropical rain belt, potentially decreasing eruption column heights and the ability of plumes to cross the tropopause and deliver materials to the stratosphere (e.g., Aubry et al., 2016). Our ability to forecast volcanic eruptions and their impacts in the context of a changing climate is therefore contingent on an improved understanding of the feedbacks between volcanic activity and other Earth systems.

Key Questions and Research Priorities on the Response of Volcanoes to Tectonics and Changes in Climate