12.1 Introduction

Mud volcanoes are surface structures formed by the eruption of mud and rock fragments from the subsurface. Colloquially, the term is used to describe a wide range of features, from decimeter-sized cones to features that are hundreds of meters high and create flows that extend laterally for a couple kilometers (Fig. 12.1). Submarine mud volcanoes, such as the serpentinite mud volcanoes in the Marianas forearc, can create edifices with diameters of 50 km that are 2.5 km high (e.g., Fryer 2012).

Fig. 12.1
figure 1

Example mud volcanoes, all form Azerbaijan, with people for scale in all photos. Azerbaijan is famous for the number and diversity of mud volcanoes (Aliyev et al. 2002). a Gryphons and fresh flows at the Dashgil mud volcano; b large Lokbatan mud volcano with vent in the background; c very viscous Koturdag mud volcano showing the mud breccia emerging at the vent in the foreground and flowing downslope. Photos by the authors

Full size image

The term “mud volcano” is generally reserved for large structures made from the eruption of mud breccias driven to the surface by buoyancy and overpressure (e.g. Mazzini and Etiope 2017). As the erupting materials are fluid-rich, a number of types of fluid seepage features are associated with mud volcanoes. Gryphons are small mud cones (e.g., Fig. 12.1a), up to a few meters high, can be numerous, and often are aligned with tectonic structures. Pools discharging water and gas and minor amounts of sediment are also common. All features can have episodic discharge.

The erupted materials are three-phase: solids, water, and gases, the latter being dominantly methane and CO2 (e.g., Kopf 2002). Fragments of country rock are also sometimes entrained. Individual blocks in erupted breccias can be greater than 1 m in diameter. Large mud volcanoes can be rooted deeper than 10 km, such as those near the Black and Caspian Seas (e.g. Mazzini and Etiope 2017).

Mud volcanism requires thick layers of unconsolidated sediment or brecciated sedimentary rocks with high pore pressures. Figure 12.2 illustrates schematically the source of materials and subsurface piercement structures associated with mud volcanoes. They are thus most common in areas with high sedimentation rates such as sedimentary basins and accretionary prisms. There are perhaps about ~103 mud volcanoes on land (Etiope and Milkov 2004; Etiope 2015). The number offshore is poorly known and their locations not well mapped. Milkov (2000) extrapolated mapped regions to estimate that there may be thousands to hundreds of thousands of submarine mud volcanoes. Their eruption is favored by compressional settings which act to increase pore pressure and they often form along the tops of anticlines with feeder dike orientations controlled by the regional stress field (e.g., Bonini 2012). Faults and fault intersections often help create pathways for fluid ascent (Mazzini et al. 2009). They are frequently related to pressurized hydrocarbon reservoirs and hence are sometimes targeted for hydrocarbon exploration.

Fig. 12.2
figure 2

(from Mazzini and Etiope 2017)

Schematic illustration of a mud volcano’s subsurface “piercement” structure and source of solids, fluids and gas

Full size image

12.2 Response of Mud Volcanoes to Earthquakes

A number of studies have documented eruptions of subaerial mud volcanoes within days of earthquakes (e.g., Snead 1964; Chigira and Tanaka 1997; Delisle 2005; Manga and Brodsky 2006; Mellors et al. 2007; Rukavickova and Hanzl 2008; Bonini 2009; Manga et al. 2009; Manga and Bonini 2012; Tsunogai et al. 2012; Bonini et al. 2016; Maestrelli et al. 2017). Tingay et al. (2018) provides a table summarizing the dates of the eruptions, and the earthquake magnitudes and their distance from the mud volcanoes. The number of documented triggered mud volcano eruptions is not large, 58 in this compilation. Of the total, 6 instances are triggered eruptions at the Niikappu mud volcano in Japan (Chigira and Tanaka 1997).

Figure 12.3 shows the relationship between earthquake magnitude and the distance between triggered eruptions of mud volcanoes and the earthquake epicenter. As with other hydrological responses to earthquakes, there appears to a clear upper bound on the distance over which eruptions might be triggered that increases as earthquake magnitude increases. This threshold is similar to that for streamflow responses and for the occurrence of liquefaction.

Fig. 12.3
figure 3

Relationship between earthquake magnitude and distance of mud volcanoes that erupted within days of the earthquake (filled blue triangle). Data from the compilation in Tingay et al. (2018). The small open red triangles are magmatic volcanoes that erupted within days of earthquakes from the lists in Manga and Brodsky (2006) and Lemarchand and Grasso (2007). The sloping black lines and lines of constant seismic energy density, as estimated in Wang and Manga (2010). The sloping red dashed line is a line showing one fault length (delineating the near-field)

Full size image

The repeated eruption of the Niikappu mud volcano, Japan, in response to earthquakes offers an excellent opportunity to better understand the conditions required for triggering. This is analogous to studying the response of a single water well to multiple earthquakes (Chaps. 66, 8) and response of a single stream to multiple earthquakes (Chap. 7). Manga et al. (2009) found that this mud volcano consistently obeyed the empirical threshold in Fig. 12.3 provided there was a repose time of at least 1–2 years between eruptions. Large, close earthquakes that occurred sooner did not trigger an eruption. This supports the arguments in Mellors et al. (2007) that a recharge period is needed before another eruption can be triggered.

Long-term triggering is more challenging to identify, but has been inferred for some mud volcanoes. Bonini et al. (2016) concluded that several mud eruptions were triggered in Azerbaijan within a year of earthquakes under conditions that the earthquakes were less than 10 fault length away and when coseismic static stress changes compressed the mud source and unclamped feeder dikes. Over longer time scales, Babyev et al. (2014) concluded that the Azerbaijan mud volcanoes have an increased eruption rate for several years after earthquakes and that dynamic rather that static strains play a dominant role in their delayed triggering.

12.3 Insights from Triggered Eruptions of Magmatic Volcanoes

The modest number of well-documented eruptions of mud volcanoes limits our ability to perform a meaningful analysis of the probability that they are triggered by earthquakes. This is less problematic for earthquake-triggered eruptions of magmatic (“real”) volcanoes since accounts and catalogues of eruptions are more accessible and have more data. We thus provide a brief overview of what is known about the triggered eruptions of magmatic volcanoes and implications for mud volcanoes. Early reviews of this topic are published by Hill et al. (2002) and Manga and Brodsky (2006).

There is a long record of inferring that earthquakes trigger the eruption of magmatic volcanoes (e.g., Darwin 1840; Yokoyama 1971; Nakamura 1975). Since the initiation of an eruption requires rocks to break, pathways for fluid flow to open, or pressures in magma bodies to change, it reasonable to expect that stress changes from earthquakes could initiate magma movement and then eruption.

The Smithsonian Institution maintains a catalog of volcanic eruptions that includes the date and magnitude of volcanic eruptions (Siebert and Simkin 2002; www.volcano.si.edu/world). This catalog is reasonably complete and meaningful for moderate to large explosive eruptions since about 1500 AD. It is thus possible to look for correlations between the occurrence of large (magnitude >8) earthquakes and volcanic eruptions over a period of at least 500 years, and with smaller earthquakes during the more recent past. In regions with a longer recorded history, e.g., Italy, a regional analysis may permit analysis of smaller earthquakes and eruptions extending further back in time.

Identifying a triggered eruption suffers from the complication that the surface manifestation of a triggered event may not occur for days to perhaps even years after the earthquake. The nature of any delay reflects the mechanism of triggering and the manner in which the magma erupts. The search for triggered eruptions is thus generally confined to a specific window in space and time. In a first global analysis, Linde and Sacks (1998) concluded that more eruptions occurred within a couple days of large earthquakes than could be expected by chance. Manga and Brodsky (2006) repeated the analysis and concluded that 0.4% of eruptions occur within 5 days of large earthquakes. This analysis (done with a more recent catalog of eruptions and earthquakes) is shown in Fig. 12.4 for eruptions with magnitude VEI ≥ 2 and within 800 km the earthquake, and subdivided between the nineteenth century and more recent times. VEI is the Volcanic Explosive Index, and a value of 2 corresponds to moderate explosive eruptions (Newhall and Self 1982). The number of events that appear to be triggered within days during the nineteenth century is lower than previous analyses as eruption dates have been refined (Watt et al. 2009). Lemarchand and Grasso (2007) performed a similar analysis that included both smaller earthquakes and eruptions for the period 1973–2005 and similarly found that 0.3% of eruptions interacted with earthquakes (though for these smaller events, the occurrence of eruptions is distributed approximately symmetrically in time around the earthquake). Intriguingly, the same type of global analysis since 1900 suggests that there is no short-term triggering of earthquakes (Fig. 12.4b).

Fig. 12.4
figure 4

(from Sawi and Manga 2018)

Histogram showing the relationship between the time of volcanic eruptions relative to the time of large (magnitude >8) earthquakes within 800 km. Bin size is 5 days. This figure repeats an analysis first done by Linde and Sacks (1998), but with expanded and updated earthquake and eruption catalogues

Full size image

Figure 12.4 shows the relationship between earthquakes and eruptions within a few days of the earthquakes, plotting examples listed in Manga and Brodsky (2006) and Lemarchand and Grasso (2007). If there is a threshold ground motion or stress for short-term triggering of magmatic volcanoes, it is similar to that for mud volcanoes (though the latter seem to be more sensitive to earthquakes on these short time scales).

The aforementioned studies focused on eruptions within days for which a statistical analysis is easier to perform (Linde and Sacks 1998). Delayed triggering is more difficult to establish and several studies have examined the space-time connections between earthquakes and eruptions (e.g., Marzocchi 2002; Watt et al. 2009). Proposed examples of delayed triggering include an increase in eruptions in the Cascade arc, USA in the 1800s following a large subduction earthquake in 1700 (Hill et al. 2002); volcanic eruptions following the M9.3 December 2004 Sumatra earthquake, the 1952 M9.0 Kamchatka earthquake, and the 1964 M9.2 Alaska earthquake (Walter and Amelung 2007); increased eruption rates after Chilean earthquakes, 1906 M8 and 1964 M9.5 (Watt et al. 2009); the 1991 eruption of Pinatubo 11 months after a M7.7 event (Bautista et al. 1996); increases eruption rates in the Lesser Antilles arc between the mid nineteenth century and early twentieth century following megathrust earthquakes in 1839 and 1843 (Feuillet et al. 2011). Sawi and Manga (2018) performed a global analysis considering the time period since 1964 for which earthquake catalogs are complete to magnitude 6 and the eruption record should be most reliable. Short term triggering, within days, was not apparent, regardless of tectonic setting, magma type, or eruption style. They did find a 5–12% increase in the number of eruptions within 2 months to 2 years after earthquakes—evidence for a modest increase in delayed triggering. For larger earthquakes (M7.5 or greater) and closer distances (within 200 km), the probability of eruption increases to 50% during the 5 years after the earthquake (Nishimura 2017). These changing probabilities of eruption can inform risk assessment.

It is worth thinking about the possible biases that enter the catalogues of events used to search for correlations between earthquakes and eruptions. For example, the report by Darwin (1840) turns out to be unreliable (Watt et al. 2009; Lara et al. 2020). There may be historical biases in recording events that are closely spaced in time. The analysis in Fig. 12.4 also only considers explosive eruptions with a VEI of 2 or greater. Smaller eruptions and effusive eruptions are more frequent. Yet, our record of VEI 2 eruptions is still probably only close to complete for about a century (with gaps during time periods of global disruption such as the world wars). Further, not all volcanoes are the same—some have open vents in which a persistently open path exists to the surface, and already-erupting systems do seem to be more sensitive to earthquakes than initiating new eruptions (Manga et al. 2009). Villarica, Chile, is one example, and it makes up 3 of the 11 triggered eruptions listed in Manga and Brodsky (2006).

12.4 Mechanisms

The mechanisms that trigger magmatic eruptions are likely to be more difficult to identify than the mechanisms that account for hydrological responses. This is because there are a greater number and complexity of processes that operate within magma reservoirs and that influence the ascent of magma. Here we review some of the mechanisms that have been proposed as triggers for both mud and magmatic volcanoes.

12.4.1 Static or Dynamic Stresses?

A central theme in studies of triggered eruptions is whether the triggering is controlled by static or dynamic stress changes. Manga and Brodsky (2006) argue that the static stress changes caused by earthquakes are in general too small to initiate eruption through any mechanism, and favor processes that are able to turn larger amplitude dynamic strains into some type of permanent or semi-permanent change. Bonini et al. (2016) analyzed the static stress changes on the feeder dikes beneath mud volcanoes from 9 settings on Earth. They identified a few settings where mud volcanoes erupted yet the static stress changes would have promoted clamping of dikes, thus favoring dynamic stresses for triggering the eruptions. In Azerbaijan, however, eruptions within a year of regional earthquakes are favored where dikes were unclamped—favoring a dominant role of static stress changes (Bonini et al. 2016). The strong tidal modulation of seismicity at Axial Volcano on the Juan de Fuca mid-ocean ridge provides important insights into the coupling of deformation and magma bodies. Scholz et al. (2019) show that magma bodies inflate and deflate in response to ocean tides, producing Coulomb stress changes larger than and opposite those from the tides themselves, thus controlling the seismicity. Scholz et al. (2019) further show that in this always seismically-active system, there is no stress triggering threshold.

The sequence of nine M > 5 events in central Italy from 2016–2017 triggered the eruption of 17 mud volcanoes and provides an excellent opportunity to assess the roles of static and dynamic stresses. Maestrelli et al. (2017) found a correlation with the amplitude of dynamic stresses whereas static stress changes were negligible or would have clamped feeder dykes. At least in this setting, dynamic stresses appear to dominate.

In support of an important role of static stress changes, Bonali et al. (2013) find a correlation between earthquakes that unclamp feeder dikes and whether an eruption follows an earthquake for a number of volcanoes in Chile. Fault geometry near the volcano thus matters, and for some already-active volcanoes, activity may be suppressed (Farias and Basualto 2020). These effects may be important to distances as great as 600 km from the epicenter of M8+ earthquakes (Bonali et al. 2015; Farias and Basualto 2020). Lupi and Miller (2014) argued that a reduction of compressive stresses after megathrust earthquakes may lead to pulses of volcanism. Walter and Amelung (2007) also document a systematic pattern of coseismic volumetric expansion at triggered volcanoes. It is not intuitive that the pressure decrease in magma bodies that would accompany volumetric expansion would promote eruption: eruption should require an overpressure to force magma out of the chamber, or to create new dikes.

Mud volcanoes are most numerous on the sea floor in the accretionary prisms above the subduction interfaces that produce the largest earthquakes on Earth. In these settings the stress changes from megathrust ruptures can produce very large stress changes, as large as 2–10 MPa close to the epicenter, and may increase the permeability of fault-controlled pathways supplying fluids and solids to the mud volcanoes (Bonini 2019). Figure 12.5a shows unclamping stresses from the 2004 M7.2 and M7.4 earthquakes in the Nankai trough, Japan. The mud volcano labelled MV#5 enhanced its methane release for several years after the earthquakes (Tsunogai et al. 2012). Large earthquakes may thus control the location and timing of fluid, gas and mud discharge in the accretionary wedge (Fig. 12.5b). Instrumenting discharge features may thus provide insights into the permeability evolution of conduits and the controls on fluid transport in these submarine seepage systems.

Fig. 12.5
figure 5

(adapted from in Bonini 2019)

a Unclamping stresses (changes in normal stress) on faults parallel to the subduction zone and at a depth of 7.5 km (colors). White rectangles outline the surfaces that slipped in 2004. Yellow circles show mud volcanoes and MV#5 responded to the earthquakes. b Schematic illustration processes that would enable fluids released from the subducted crust to move along faults and feed submarine mud volcanoes

Full size image

12.4.2 Mechanisms for Initiating Eruptions

Magmatic and mud volcanoes share in common that gases play a role in providing buoyancy, they erupt materials that are liquefied or fluidized, and the source is usually over-pressured. Mechanisms through which dynamic strains influence the nucleation or growth of bubbles, or liquefy sediment or crystal mushes, are in principle possible in both systems.

12.4.2.1 Mechanisms Involving Bubbles

Given the importance of bubbles in driving magma to the surface and powering volcanic eruptions, several triggering mechanisms have been proposed that invoke bubbles. One possibility is the nucleation of new bubbles in a supersaturated liquid by the periodic changes in pressure generated by seismic waves (e.g., Manga and Brodsky 2006; Crews and Cooper 2014). A second possibility, is that diffusion of gas from a supersaturated liquid into preexisting bubbles is enhanced by dynamic strains. When bubbles experience oscillatory strain, there is an asymmetric diffusion of gas into and away from the bubble owing to the change in shape – this process is called rectified diffusion (e.g., Sturtevant et al. 1996). Ichihara and Brodsky (2006) have shown that this process results in insignificant growth of bubbles. A third possibility, is that pore pressures rise as bubbles carry high pressures to shallower depths as they rise (e.g., Steinberg et al. 1989; Sahagian and Proussevitch 1992; Linde et al. 1994), a process called advective overpressure. This mechanism requires that both the bubbles and surrounding matrix/liquid are incompressible, and several studies have shown that these assumptions are not satisfied (Bagdassarov 1994; Pyle and Pyle 1995). A fourth possibility is that gas hydrates dissociate. Submarine mud volcanoes are often associated with gas hydrates (Milkov 2004) and enhanced methane emission has been attributed to earthquakes in both lakes (Rensbergen et al. 2002) and the ocean (Mau et al. 2007). However, triggered eruptions that have been identified so far are subaerial (this is very likely an observational bias), where gas hydrates should not exist. A final process is the mobilization of trapped bubbles by oscillating pressure gradients. Changing pressure gradients can dislodge bubbles trapped in pores (e.g., Beresnev et al. 2005; Deng and Cardenas 2013), increasing permeability and the ability of fluids to erupt. This process has been used to explain the triggered eruption of mud volcanoes (e.g., Rudolph and Manga 2012). At larger scales, sloshing of bubbly magma may cause magmatic foams to collapse, releasing gas and initiating eruptions (Namiki et al. 2016).

One intriguing possibility is that the volume expansion of a magma chamber can lead to a net increase in its overpressure owing to the growth of bubbles. Recall that Walter and Amelung (2007) found that triggered magmatic eruptions can occur in regions that experience volumetric expansion. Nishimura (2004) showed that the growth of bubbles that accompanies magma chamber expansion causes a decrease in the pressure difference between that inside and outside bubbles, and the surface tension energy liberated results in a net pressure increase in the magma. This effect is very small, except for very small bubbles (smaller than a few microns). Carr et al. (2018) suggested that the addition of CO2 to a magma reservoir, mobilized for example by an earthquake, can lead to water exsolution and help create the overpressure needed to change eruption rates over the time scales and magnitudes seen at Merapi, Indonesia.

12.4.2.2 Liquefaction

As mud volcanoes erupt liquefied or fluidized sediment, mechanisms that invoke liquefaction by dynamic strain seem reasonable. However, liquefaction is generally viewed as a shallow phenomenon because overburden stresses at greater depths requires that pore pressure changes become unrealistically large (e.g., Youd et al. 2004; Chap. 11). This should not be a limitation in the settings where mud volcanism occurs as the erupted materials initially had high pore pressures, and only modest increases in pore pressure may be necessary even if the overburden stresses are high. Liquefaction by dynamic strain has been invoked to explain mud eruptions (e.g. Lupi et al. 2013).

Liquefaction or weakening of magmatic suspensions has also been invoked to explain the seismic triggering of magmatic volcanoes (Hill et al. 2002; Sumita and Manga 2008). A reduction in strength would be manifest as a decrease in seismic velocity. Battaglia et al. (2012) documented a reduction in seismic velocity beneath Yasur volcano, Vanuatu, followed by a partial recovery, following a M7.3 earthquake 80 km from the volcano summit; in this instance, however, no eruption occurred.

12.4.2.3 Breaching Reservoirs

Water level changes in wells, as discussed in Chap. 8, can be explained in many instances by changes in permeability or the breaching of hydrological barriers that allow fluids and pore pressure to be redistributed. This is a viable mechanism to fluidize or liquefy unconsolidated sediments if there are reservoirs below the source layer with high enough pore pressure (e.g., Wang 2007; Cox et al. 2021). In some settings, the gases and fluids that erupt at mud volcanoes may be sourced much deeper than the erupted mud (e.g., Cooper 2001; Mazzini et al. 2009; Shirzaei et al. 2015), supporting the idea that fluid and gas migration play a role in initiating eruptions. This process may lead to, however, a time lag in the manifestation of the triggered eruption governed by the time scale for fluids and/or gas to migrate (e.g., Husen and Kissling 2001). Stress transfer following regional tectonic earthquakes can lead to accelerated seismic energy release, as documented at Tungurahua, Ecuador and Popocateptl, Mexico volcanoes, and enhanced seismic energy release may be documenting the beginning of the material failure that leads to eruption (De la Cruz-Reyna et al. 2010).

12.5 The Sidoarjo (Lusi) Mud Flow

On the morning of 29 May 2006, mud began erupting in the Porong subdistrict, Sidoarjo in East Java, Indonesia. The eruption continues to present. The eruption rate was and remains large, about 100,000–200,000 m3/day during the first few years, and about 80,000 m3/day in recent years (Miller and Mazzini 2018). The duration and erupted volume are unprecedented for a mud eruption on land. Several studies have attempted to forecast its expected longevity and all expect the eruption to continue for many more years (e.g., Davies et al. 2011; Rudolph et al. 2011, 2013). The eruption has led to a disaster with extensive human and environmental impacts. Villages were buried and more than 40,000 people were displaced; environmental pollution and poor conditions for those relocated create chronic health problems (Drake 2016; Fig. 12.6).

Fig. 12.6
figure 6

(from Bonini et al. 2016)

a Satellite image of the Sidoarjo (Lusi) mud eruption from August 2012 (extracted from Google Earth) that began eruption on 29 May 2006. The eruption continues to present. b Location of the vent (red circle) and the location and focal mechanism of the M6.3 Yogyakarta earthquake on 26 May 2006

Full size image

From the very beginning, the reason the mud erupted was the source of scientific controversy connected to the theme of this book. Early reports in the news, and then in the scientific literature (e.g., Mazzini et al. 2007; Sawolo et al. 2009), argued that the eruption was triggered by the M6.3 Yogyakarta earthquake 254 km away. Manga (2007), in contrast, used a compilation of previous examples of triggered mud eruptions to argue that the earthquake was too far away to trigger a new eruption and, moreover, that there were earthquakes that caused stronger ground motion or were even larger and closer and none of these earthquakes triggered an eruption. Tingay et al. (2008) showed that static stress changes were vanishingly small and Davies et al. (2008) showed that dozens of earthquakes caused stronger ground motions without causing eruptions. Rather than an earthquake-trigger, others proposed that ongoing drilling of the Banjarpanji-1 gas exploration well by PT Lapindo Brantas, about 100 m away from the vent where mud first erupted, initiated the eruption as a subsurface blowout (Davies et al. 2007, 2008; Tingay et al. 2008). These early studies initiated several more detailed analyses arguing for an earthquake trigger (e.g. Istadi et al. 2009; Lupi et al. 2013), against an earthquake trigger (Rudolph et al. 2015), or a drilling trigger (Tingay et al. 2015). A vigorous debate ensued in the scientific literature, including comments and replies (Davies et al. 2010; Sawolo et al. 2010). The debate continues to present, including two reviews (Miller and Mazzini 2018; Tingay et al. 2018).

For full disclosure, both authors of this book have written papers arguing why the earthquake did not cause the eruption and why drilling operations did. In brief, the strongest argument against an earthquake trigger is that there were other earthquakes that caused stronger ground motion but no eruption (Fig. 12.7). There is nothing special about the Yogyakarta earthquake. While it was a strike-slip event, and directivity effects can amplify ground motions and promote triggering (e.g., the Gwadar mud eruption in 2013 off the Makran coast responding to the M7.8 Balochistan earthquake with epicenter 383 km away), the orientation of the fault that slipped would not have enabled enhanced ground motion at the eruption site. The most compelling argument for a drilling trigger is the set of daily drilling reports themselves. Pressure data in the well record the birth of the eruption. Attempts to kill the eruption by pumping fluids into the well altered the eruption, providing evidence for a physical connection. The daily drilling reports are published in Tingay et al. (2018) and are annotated to help translate technical terms and to identify evidence and clarify what are interpretations and what are data.

Fig. 12.7
figure 7

(from Tingay et al. 2018)

A comparison of the Sidoarjo (Lusi) mud eruption and Yogyakata earthquake (green star) with other examples of triggered mud volcanoes shown with filled triangles (the same set shown in Fig. 12.3). Also shown with circles are all the earthquakes that occurred between 1976 and the eruption initiation (28 May 2006). Filled symbols are events deeper that 30 km

Full size image

In the scientific literature, the eruption is often called “Lusi”, a contraction for “Lumpur Sidoaarjo” with “lumpur” the Indonesian word for mud. Locals call it the “Lapindo” mudflow after the drilling company (Drake 2020). The name for the eruption itself is controversial because of the connotations and social context. There is a political dimension to the eruption connected to providing compensation to victims. Regardless of the trigger, the eruption clearly has devastated environments and communities and recovery will be very slow (Drake 2016).

12.6 Effect of Earthquakes on Already-Erupting Mud Volcanoes

Given the strong sensitivity of geysers—also already-erupting systems—to earthquakes, we might reasonably expect already-erupting mud volcanoes to be more sensitive to earthquakes than the triggering of new eruptions. Observations of magmatic volcanoes support this contention. For example, Harris and Ripepe (2007) report changes in eruption rate at Semeru volcano, Indonesia, in response to the 2006 M6.3 Yogyakarta earthquake based on satellite thermal imaging. At Mt Merapi, Indonesia, Walter et al. (2007) report increases in extrusion rate or fumarole temperature after regional earthquakes. The open vent volcano Stromboli, Italy, also responds to earthquakes (Speranza and Carniel 2008). Satellite thermal data show that there may be global increases in volcanic unrest following the largest earthquakes, M > 8.5 (Delle Donne et al. 2010). A more recent analysis of 14 M > 8 events found that 3 led to short-lived global thermal emission increases, and 2 decreased thermal emissions (Hill-Butler et al. 2020). At persistently active (open vent) basaltic volcanoes, there is an increase in SO2 emissions recorded by the Ozone Monitoring Instrument whereas decreases occur at andesitic volcanoes (Avouris et al. 2017).

The high level of earthquake activity in Indonesia and the longevity of the Sidoarjo eruption (Sect. 12.4) provide an opportunity to look for responses of the ongoing eruption to earthquakes. The main challenge is obtaining reliable quantitative information about the eruption. Responses to large distant and moderate regional earthquakes have been reported, though without data (Miller and Mazzini 2018). Figure 12.8 shows published eruptions rates and the magnitude of ground motion from earthquakes, computed from source mechanisms and using a local attenuation model (Davies et al. 2008). Despite anecdotal reports of responses, there is no obvious signature of responses in the reported discharge.

Fig. 12.8
figure 8

(from Bonini et al. 2016)

Eruption rate at the Sidoarjo (Lusi) mud eruption. Black curve is discharge from Mazzini et al. (2007). Red stars are computed ground motion for earthquakes in the USGS catalog using the attenuation relationship for east Java in Davies et al. (2008). Data provided by A. Mazzini

Full size image

The Salton Sea “mud volcanoes” in the Imperial Valley, California, technically gryphons and called hydrothermal features by Svensen et al. (2009), have responded multiple times to earthquakes and hence provide an opportunity to identify how frequency and amplitude of ground motion affect eruptions. Here, changes in eruptions are documented by increased gas flux and a greater number of fresh mud flows measured during discrete visits (Rudolph and Manga 2010). Using responses from 2 earthquakes and 4 no-responses, Rudolph and Manga (2012) conclude that for a given amplitude of ground motion, longer period waves are more effective at causing responses. The number of examples, however, is modest and a review of frequency-dependent triggering concluded that “the data supporting this conclusion are still extremely sparse” (Manga et al. 2012). This conclusion still holds several years later.

Menapace et al. (2017) installed pore pressure sensors near the conduit of a submarine volcano in the eastern Mediterranean Sea. They document pressure spikes after many earthquakes “but seemingly no triggered mud volcano eruptions”. They found that pressure changes are much more sensitive to earthquakes than are the eruptions of new mud volcanoes.

12.7 Concluding Remarks About Mud Volcanoes

It is important to recall that most eruptions at mud volcanoes and at magmatic volcanoes are not triggered by earthquakes. This implies that for triggered eruptions, the volcanic plumbing system must already be near failure, perhaps with stresses within less than a few percent of the failure stress (Manga et al. 2009). If we choose this failure stress to be the tensile strength of rock, say 10 MPa, extra stresses of only 0.5 MPa are needed to overcome a 5% deficit, and can be provided by static stress changes in the near-field or dynamic stresses in the intermediate-field. Understanding the relationship between stress changes and eruptions is important for revealing which volcanoes are poised to erupt and the mechanisms that initiate eruptions (National Academies 2017).

It is clear that the number of triggered events is small, and the amount and quality of data from erupting mud volcanoes is too limited, to conclusively answer the most interesting questions about triggered eruptions: Is triggering from static or dynamic stress changes? What is the mechanism of triggering? Does earthquake sensitivity increase once the eruption begins? Key to addressing questions about triggering are more examples, accurate timing, and ideally nearly co-located eruptions and seismometers or strainmeters to characterize the ground motion. Volcanic eruptions begin in the subsurface, and seismic and deformation signals that accompany the initiation of unrest prior to the surface expression of eruption may be critical for identifying the mechanisms that lead to eruption. For already-erupting mud volcanoes, continuous gas flux, pressure and temperature measurements or continuous GPS are promising approaches that offer high temporal resolution.