El Chichón volcanic activity before and after the M_w8.2, 2017, Chiapas earthquake, México. Is El Chichón ready to erupt?

Abstract

El Chichón volcano is the most active volcano in the state of Chiapas, México, and experienced its last Plinian eruption (VEI = 5) in 1982. To better assess its volcanic hazard, we studied its readiness to erupt by estimating changes in its internal stress state. These stress changes are difficult to calculate accurately, for example in the absence of focal mechanisms, but their existence can be indirectly revealed by the presence of volcano-tectonic earthquakes, for example following a large tectonic earthquake. We show that the seismic rate recorded at El Chichón volcano increased slightly after the large M_w8.2 Tehuantepec earthquake of 8 September 2017, Chiapas. However, this rate quickly returned to its background level after only 2 months, without any external volcanic manifestations, suggesting that the volcano is not ready to erupt in the near future. Previous observations of slight increases in the volcanic seismicity rate following large earthquakes have been explained by the presence of active hydrothermal systems in the vicinity of the volcano. We propose a similar explanation for El Chichón volcano which is known for its large hydrothermal system. Furthermore, the characteristics of the 2017 seismicity (spatial and magnitude distributions), and the horizontal-to-vertical spectral ratio also confirm the presence of high amounts of water near the volcano. We show that the 2017 volcano-tectonic seismicity is of hydrothermal rather than magmatic origin, in agreement with recent independent geochemical and aeromagnetic studies.

Introduction

It has long been suggested that an eruption of a ready-to-erupt volcano (i.e., in a metastable equilibrium) may be triggered by an earthquake with specific magnitude and distance (Latter 1971; Tokarev 1971; Yokoyama 1971; Nakamura 1975; Carr 1977; Sharp et al. 1981; Barrientos 1994; Linde and Sacks 1998; Hill et al. 2002; Marzocchi et al. 2002; Manga and Brodsky 2006; Walter and Amelung 2007; Eggert and Walter 2009; Watt et al. 2009; Bonali et al. 2013; Kennedy 2017; Nishimura 2017, 2021; Sawi and Manga 2018; Boulesteix et al. 2022; Legrand 2022). However, of course, not all earthquakes induce volcanic eruptions, and not all volcanoes are affected equally by a large-magnitude earthquake (Bretón et al. 2022). It is not yet well understood why some volcanoes are more efficient to have a triggered eruption after an earthquake than others. However, it seems that it depends on stress changes beneath the volcano that local or regional earthquakes can modify (Sharp et al. 1981) or it depends on the geometry and the number of the magma chambers/reservoirs feeding the volcano (Martí et al. 2013). These two effects control the magma ascent (Bretón et al. 2022). Seropian et al. (2021) also mentioned that the presence of active hydrothermal systems beneath volcanoes is a favorable condition for triggering eruptions after a large earthquake. However, when a volcano is far from erupting, an earthquake, even large and near, will not affect the activity of such a “dormant” volcano. In fact, many volcanoes do not erupt after a large earthquake, suggesting that many of them are not ready to erupt at the moment of the occurrence of the earthquake. For example, after the M_w9.5 Chilean earthquake of 1960, although Cordón Caulle, Puyehue volcano reactivated 48 h after the earthquake, many other Chilean volcanoes did not erupt (Barrientos 1994).

Assessing the internal stress state of a volcano is difficult, but one way to achieve this objective is by quantifying any rate-change of the volcano-tectonic earthquakes (VTs) activity. Such changes have been observed prior to most volcanic eruptions and used as an efficient forecasting tool (McNutt 1996; White and McCausland 2016, 2019). They can also be detected beneath a volcano after the occurrence of a large earthquake in the region. The Yellowstone National Park region is an example of how the 2002 M_w7.9 Denali (Alaska) earthquake 3100 km away has affected its volcanic seismicity. An abrupt increase in seismicity was recorded in this volcanic region, particularly nearby the hydrothermal systems, coincident with the earthquake and during the following 30 days, due to dynamic triggering (Husen et al. 2004; Pankow et al. 2004). The Denali earthquake triggered other volcanic seismic swarms at Katmai Volcano, Mount Rainier Volcano, the Long Valley Caldera, and the Geysers and Coso geothermal fields (Moran et al. 2004; Prejean et al. 2004). Another example of dynamic triggering is the occurrence of a volcanic seismic swarm triggered at Pavlof volcano by the 2011 M_w9.1, Tohoku, Japan earthquake, 4497 km away from the volcano (Prejean and Hill 2018). Between 2006 and 2013, nine out of 12 Alaskan volcanoes increased their seismic rate following regional or teleseismic earthquakes of magnitude greater than 7 (Prejean and Hill 2018).

Here, we studied the changes in seismic rate at El Chichón volcano in response to the M_w8.2 Tehuantepec earthquake of 8 September 2017, Chiapas, with an epicentral distance of 271 km (Fig. 1). Using a local velocity model constrained by the Vp/Vs ratio and the horizontal-to-vertical spectral ratio (hereafter H/V ratio) obtained from the P- and S-arrival times of these 2017 earthquakes, we detected and located the local seismicity during 2017. We then showed that the seismic rate increased slightly after this large earthquake, and quickly returned to its pre-earthquake level within just 2 months. This suggests that El Chichón is not prepared for an imminent eruption.

Fig. 1

Modified from López-Loera et al. (2020)

Geodynamic context. a The three Cocos, Caribbean, and North American plates. Red triangles are active volcanoes, and white triangles are scattered volcanic structures of the modern Chiapanecan Volcanic Arc (MCVA). CAVA is the Central American Volcanic Arc, and LTVF is Los Tuxtlas Volcanic Field. White lines are the main faults. Cities: Coatzacoalcos (Co), Villahermosa (Vi), Tuxtla Gutiérrez (TG), Salina Cruz (SC), Tonalá (To), Tapachula (Ta), Chetumal (Ch). b Zoom of the red rectangle in Fig. 1a. The modern Chiapanecan volcanic arc (MCVA). Cities: Tuxtla Gutiérrez (TG), San Cristóbal de las Casas (SCC).

Geological setting

El Chichón volcano is part of the trench-trench-transform triple junction area involving the North American continental plate, the Cocos oceanic plate, and the Caribbean plate (Guzmán-Speziale et al. 1989; DeMets 2001; Ratschbacher et al. 2009; Franco et al. 2012; Garduño-Monroy et al. 2015). The Cocos (CO) plate subducts beneath the Caribbean (CA) and the North American (NA) plates along the Middle/Meso American Trench (MAT). The onshore extension of the Tehuantepec Ridge located on the CO plate cuts off the El Chichón volcano (Manea and Manea 2006) and coincides with a change in the subduction angle. To the west of the Tehuantepec Ridge, the slab dips at low angles of 25–35°. To the east of the ridge, the slab dips at higher angles, up to 40–45°, and the depth of the seismicity reaches 270 km (Stoiber and Carr 1973; Burbach et al. 1984; Ponce et al. 1992; Pardo and Suárez 1995; Rebollar et al. 1999; Manea et al. 2005). El Chichón volcano is located above this drastic change in plate shape (Manea and Manea 2006). It has also been proposed that a tear of the CO subducting plate allows the presence of an anomalously hot region of the mantle and the migration of the magma to the surface (De Ignacio et al. 2003; Calò 2021; Arce et al. 2024).

The active andesitic stratovolcano El Chichón (1100 m a.s.l.) is the youngest (~ 0.21 Ma) volcano of the relatively young (~ 2–3 Ma) modern Chiapanecan volcanic arc (MCVA, Damon and Montesinos 1978; Duffield et al. 1984; Mora et al. 2012; Fig. 1). The MCVA is small (~ 150 km long) and consists of a few (~ 10) scattered volcanic structures (white triangles in Fig. 1). El Chichón is located at a horizontal distance of ~ 325–330 km from the MAT, and at a vertical distance of ~ 200–220 km above the subducting CO slab (Syracuse and Abers 2006; Manea and Manea 2006, 2008).

The tectonics around El Chichón

El Chichón volcano is part of a complex system composed of several pull-apart basins generated by ~ nine distant left-lateral strike-slip faults (~ 120–170 km long), which explain the rectangular shape of El Chichón structure (Guzmán-Speziale and Meneses-Rocha 2000; Meneses-Rocha 2001; Fig. 1). There are three main fault systems in its vicinity, some of which cut through the volcano (Duffield et al. 1984; García-Palomo et al. 2004; Macías et al. 2008; Garduño-Monroy et al. 2015):

• 1) A system composed of several parallel E-W left-lateral strike-slip faults. One of these, the San Juan left-lateral strike-slip fault, cuts the Somma crater at the summit of the volcano. It is well identified in the eastern part of the volcano and appears to continue west of the crater (Mazot and Taran 2009; Mazot et al. 2011). Another one is the Chichón-Catedral fault, which cuts both the San Juan fault and the Somma crater, but strikes in a NNW-SSE direction in the northern part of the crater and in a NW–SE direction in the southern part of the crater (Garduño-Monroy et al. 2015).

• 2) A conjugate fault system, perpendicular to the one described in 1), consists of N-S right-lateral strike-slip faults.

• 3) A system composed of normal faults oriented in the N45° E direction, called the Chapultenango fault system, produces a half-graben geometry on which El Chichón volcano is located (Macías et al. 1997a; García-Palomo et al. 2004). It is a local extensional regime with en échelon faults within a pull-apart basin, associated with earthquakes of normal focal mechanisms (García-Palomo et al. 2004).

Past volcanic activity

El Chichón volcano was not recognized as an active volcano until 1928 by Müllerried (1932; 1933), who conducted fieldwork in the region due to the occurrence of local earthquakes felt by humans. Recent stratigraphic studies show that El Chichón volcano had frequent and violent activities during the Holocene, with a response interval of ~ 300 years, ranging from a minimum of ~ 100 years to a maximum of ~ 600 years, making El Chichón a highly dangerous volcano (Tilling et al. 1984; Espíndola et al. 2000; Macías 2007; Macías et al. 2008; Mendoza-Rosas and De la Cruz-Reyna 2010; De la Cruz-Reyna and Tilling 2015; Scolamacchia and Capra 2015). At least 12 episodes of explosive volcanic activity occurred during the Holocene (Rose et al. 1984; Tilling et al. 1984; Espíndola et al. 2000; Macías 2007; Macías et al. 2003, 2008; Scolamacchia and Capra 2015; Table 1).

Table 1 List of past eruptions of El Chichón volcano (Macías et al. 2008)

The last Plinian (VEI = 5) eruption of El Chichón in 1982 killed ~ 2000 people. It consisted of three explosions that ejected a total of ~ 1.5 km³ of pyroclastic surge, -flow and -fall deposits and ~ 1.1 km³ of magma (dense rock equivalent = DRE; Carey and Sigurdsson 1986). They blasted the two long pre-existing andesitic summit lava domes (covered by trees) nested within the 2 × 1.5 km² old Somma rim created a new 1-km wide and > 200-m deep summit crater within this old Somma crater (De la Cruz-Reyna and Tilling 2015) and caused almost total devastation in a radius of 10 km around the crater (De la Cruz-Reyna and Martin Del Pozzo 2009).

On 25 April 1982, three small lakes were observed inside the new crater formed by the eruption, and in November of the same year, after the rainy season between June and November (with high rainfall of ~ 4 m/year), they merged into a single hot (> 50 °C) crater lake (Casadevall et al. 1984; Peiffer et al. 2015). In January 1983, the lake was ultra-acidic (pH of 0.56) and hot (T = 52–58 °C; Casadevall et al. 1984). This lake never disappeared after its formation, although its volume changed considerably over time (Taran et al. 1998; Tassi et al. 2003; Rouwet et al. 2004, 2008, 2009). Three years after the appearance of the lake, its pH and temperature stabilized around 2.5 and 30 °C, respectively (Armienta et al. 2000; Taran and Rouwet 2008), indicating a reduced input of magmatic gases to the hydrothermal system. In the following years, numerous studies focused on the chemical and isotopic composition of the lake, confirming that the volcano had undergone a rapid transition from magmatic to hydrothermal conditions (Armienta and De la Cruz-Reyna 1995; Taran et al. 1998; Rouwet et al. 2008; Taran and Rouwet 2008; Mazot and Taran 2009; Taran and Peiffer 2009; Mazot et al. 2011). Prior to the 1982 eruption, a vigorous hydrothermal system with intense fumarolic activity existed within the Somma crater, close to the two central domes (Müllerried 1933; Damon and Montesinos 1978; Cañul and Rocha 1981). After the explosion, the high precipitation of the region (~ 4 m/year) mainly fed the hydrothermal system (Taran et al. 1998; Taran and Rouwet 2008). Since the last eruption in 1982 and the associated seismic activity (Havskov et al. 1983; Medina et al. 1988, 1990, 1992; Yokoyama et al. 1992; Jiménez et al. 1999; Espíndola et al. 2006; Legrand et al. 2015), El Chichón volcano has not shown clear signs of reactivation. In particular, no new lava dome has yet appeared in the Somma crater. A new lava dome is expected to appear a few years after a major eruption, such as at Mount Pinatubo, where a new lava dome appeared between July and October 1992, almost a year after the major eruption of June 1991 (Stimac et al. 2004), and before a new eruption.

Seismic data

We processed the 2017 seismic activity recorded at El Chichón (López-Landa 2020) using a local seismic network consisting of three broadband three-component seismometers. NPCH and FLCH were QA-120 s nanometrics broadband seismometers with a bandwidth between 120 s and 145 Hz connected to a Centaur acquisition system, installed in 2016, 1 year before the M_w8.2 Tehuantepec earthquake on 08 September 2017 (07 September 2017 in local time). The VVCH was a Guralp 40 T seismometer with a bandwidth between 60 s and 100 Hz. The acquisition is performed at a sampling rate of 100 Hz. The VVCH is the best station (the closest to the volcano, with a higher signal-to-noise ratio than the other two stations). After the M_w8.2 Tehuantepec earthquake, the VVCH seismometer was out of service for 15 days from 16 September 2017 to 01 October 2017, due to a power supply problem. The NPCH seismometer was also out of service from 07 June 2017 to 19 April 2018, and the FLCH from 24 June 2017 to 18 October 2018. No data are available for 2018 as VVCH was out of service, and NPCH and FLCH stations did not record clear seismicity. In 2017, no earthquake was recorded simultaneously by the three seismometers. Only 71 earthquakes were recorded and located by two seismometers (VVCH, NPCH) or (VVCH, FLCH), and 322 earthquakes were recorded and located by one seismometer (VVCH).

Method

Locating earthquakes in volcanic environments is always challenging due to the highly heterogeneous nature of the medium. The determination of an adequate velocity model is then essential to reduce hypocenter errors, especially at volcanoes where only a few seismometers are installed, as at El Chichón. For the 2017 El Chichón earthquakes, when picking the P- and S-wave arrival times tp and ts (respectively), we noticed that the difference ts-tp is small (~ 0.38 s on average), so the seismicity is shallow. Therefore, we need to pay attention to the first few kilometers of our velocity model.

To determine the velocity model, we proceeded in three steps. In the first step, we calculated a unique Vp/Vs ratio for the whole area, where Vp and Vs are the P- and S-wave velocities, respectively, using the generalized Wadati diagram. In the second step, we determined a relative (i.e., the velocity and thickness of the layers are not calibrated) velocity model determined by the H/V technique. In the third step, we determined an absolute velocity model by calibrating the relative depths of the velocity model with geological information. Each step is described in more detail below.

In general, the Wadati diagram (t_s-t_p versus t_p, where t_p and t_s are the arrival times of the P- and S-waves) is plotted for each earthquake, giving the original time t₀ of the earthquake and the average Vp/Vs ratio between the earthquake hypocenter and the seismometers. However, by making a double difference between the S- and P- arrival times at all the stations for a given earthquake, the t₀ of each earthquake disappears, and the diagram is a line passing through the origin with a slope equal to Vp/Vs. This diagram is called the generalized Wadati diagram (see details in Pinares 2006; Legrand et al. 2021; Montenegro et al. 2021, and the Annex). We found a high value of Vp/Vs = 1.91 (Fig. 2) compared to the classical value of 1.73 value obtained for most deeper earthquakes.

Fig. 2

The generalized Wadati diagram derived from the P- and S- arrival times using the (VVCH, NPCH) and (VVCH, FLCH) station pairs

Site effects and subsurface properties can be determined using the horizontal-to-vertical spectral ratio H/V technique (Sánchez-Sesma et al. 2011; Piña-Flores et al. 2017). We used the ambient seismic noise recorded at the three seismometers (Fig. 3a) over 4 days. The H/V ratios have similar shapes at the FLCH and NPCH stations (blue and red lines in Fig. 3a respectively), showing little amplification and suggesting similar media. In contrast, the H/V ratio at VVCH is very different (black line in Fig. 3a), showing a strong amplification effect at 2 Hz. This H/V curve suggests that the medium beneath VVCH is very different from that beneath NPCH and FLCH stations. The modeled H/V ratio at VVCH (black dashed line in Fig. 3a) is associated with the velocity profile shown in Fig. 3b and Table 2. This velocity model results from the H/V inversion in the 1–15-Hz frequency range, and considers an initial mean velocity model under volcanoes (Lesage et al. 2018; Perton et al. 2022). However, the inversion of H/V ratios does not provide a unique solution. Without knowledge of a precise geological profile, it is difficult to constrain the absolute value of the thickness of these layers. Therefore, additional information has to be considered. Here, we constrained the layer thicknesses from the geological knowledge of the region (Garduño-Monroy et al. 2015). The first ~ 45–50 m are composed of pyroclastic deposits (fall and pyroclastic density currents) and volcanoclastic Holocene deposits. The layers from ~ 45–50 m to ~ 250 m are consolidated pyroclastic and volcaniclastic deposits of the Somma volcano (ages range from ~ 300 to 55 ka). The layers from ~ 250 m are Miocene sedimentary rocks (eMscu and P-IMscu formations, Garduño-Monroy et al. 2015). Only small Vs values (< 300 m/s) in the first 50 m allow the reproduction of the H/V peak amplitude at the VVCH station (Fig. 3a, Table 2), suggesting the presence of a fluid-saturated medium, confirmed by the presence of shallow ash and soft soil filled with water (De la Cruz-Reyna et al. 1989; Peiffer et al. 2022).

Fig. 3

a H/V ratios for the three seismic stations: VVCH (black line), NPCH (red line), and FLCH (blue line). The black dashed line is the H/V modeled curve found at the VVCH station, which was used to determine the velocity model shown in b

Table 2 Velocity model, with Vs = 1.91 Vp calculated from Fig. 2. The eMScu unit (sandstones and claystones), PIMuscu unit (sandstones and calcareous claystones), and IK-evls unit (evaporites and limestones) discontinuities are discussed in the text and Garduño-Monroy et al. (2015)

All the earthquakes are located at least near the VVCH station, since the ts-tp differences are smallest at this station. Therefore, we used the 1D local velocity profile below VVCH (Table 2 and Fig. 3b) for the whole area.

In 2017, no earthquake was recorded simultaneously by the three seismometers. Therefore, we located the earthquakes with one or two station(s). The method of location with one seismometer (VVCH) is performed classically by determining the P-wave polarization direction, and the distance with the difference ts-tp of the P- and S-wave arrival times (Havskov and Ottemoller 2010) and the velocity model of Table 2. The two-seismometer method is performed by using the combined information of the two P-wave polarization directions and the differences ts-tp (Havskov and Ottemoller 2010). We have only located the earthquakes for which the P-wave was clear enough to calculate the polarization direction (in practice this corresponds to magnitudes ≥ 1.8). We calculated the duration magnitude using the formula M_d =  − 0.87 + 2 × log10 (duration) + 0.0035 × distance (Lee et al. 1972).

Results

The earthquake-locations

In Fig. 4, we show the 393 earthquake hypocenters of magnitude ≥ 1.8 obtained with two seismometers (the 71 red dots), and with the single VVCH seismometer (the 322 black dots), from 01 January 2017 to 31 December 2017. The earthquakes located with two stations mainly correspond to the period before the large M_w8.2 Tehuantepec earthquake of 8 September 2017, because after this earthquake, only one station (VVCH) was partially operational. Therefore, we cannot see any spatial change in these hypocenters before and after the M_w8.2 earthquake. Most of the earthquakes are shallow (depths less than ~ 2 km), have a NW–SE orientation (Fig. 4), and are clustered within the Somma crater. The epicenters have a median azimuth of ~ 286° from the VVCH station when determined by two stations (Fig. 5a) or by one station (Fig. 5b). As expected for single station locations, the hypocenters obtained with a single seismometer (black dots in Fig. 4) are more spatially diffuse than those obtained with two seismometers (red dots in Fig. 4). Nevertheless, they are also oriented in the NW–SE direction and towards the western part of the volcano, as for the hypocenters located with two stations.

Fig. 4

Locations of the three seismometers (yellow triangles, VVCH, Viejo Volcán Chichón, NPCH, Nicapa, FLCH, Francisco León) of El Chichón volcano in 2017. Locations of the 393 earthquakes of magnitude ≥ 1.8 in 2017 (red dots: 71 locations using two seismometers, black dots: 322 locations using the VVCH seismometer)

Fig. 5

Back-azimuths of the earthquakes at the seismic VVCH station for a events located with two stations, b best-located events using one station (VVCH), and magnitude ≥ 1.8

The Catedral-Chichón fault and the NW–SE well-located earthquakes

The earthquake epicenters (Fig. 4) are mainly oriented in a NW–SE direction, with a strike of ~ 286°, similar to the azimuth of the Catedral-Chichón fault that crosses the crater lake (Mazot and Taran 2009; Mazot et al. 2011; Garduño-Monroy et al. 2015). The vertical distribution of the best-located earthquakes (red dots in Fig. 4) is consistent with the dip of a sub-vertical strike-slip fault mechanism.

Seismic characteristics

Figure 6 a shows the monthly distribution of the maximum duration magnitude M_d (Lee et al. 1972) calculated at the VVCH station. Figure 6 b shows the monthly number of earthquakes. Figure 6 c shows the monthly energy, and Fig. 6d shows the daily number of earthquakes. The two largest local earthquakes (M_d = 2.89 on 15 September 2017 and M_d = 2.81 on 27 November 2017) occurred 7 and 80 days after the M_w8.2 Tehuantepec earthquake, respectively (Fig. 6a). Prior to the M_w8.2 September earthquake, the VT seismic rate was ~ 58 + / − 19 earthquakes/day (with a maximum of 87) between January and June, and increased to 110 + / − 3 between July and August. The highest rate was in September (during the month of the earthquake) with a value of up to 158 (an underestimate due to the 15-day data gap), and 142 in October, the month after the earthquake. By November, the number of VT earthquakes had already returned to the average before the M_w8.2 Tehuantepec earthquake. Two months after the earthquake, from November to December, it decreased to 39 + / − 1, a value close to the pre-earthquake background level of ~ 58 + / − 19 (Fig. 6). Thus, the VT seismic rate increased slightly after the M_w8.2 earthquake, and only for 2 months, suggesting that this earthquake did not significantly affect the internal stress state of the volcano.

Fig. 6

Monthly distribution of a the maximum magnitude, b the number of earthquakes, and c the energy. The energy (in Joule) is calculated from the magnitudes using the relationship: log₁₀(E) = 5.24 + 1.44 M. d Daily number of earthquakes. The vertical red line corresponds to the day of the 08 September 2017 M_w8.2 Tehuantepec earthquake

Discussion

Large amount of fluids beneath El Chichón

The seismic data reveal the presence of a large amount of fluid beneath the volcano using the following: (1) the H/V inversion, (2) the $\frac{{V}_{P}}{{V}_{S}}$ ratio, and (3) the magnitude distribution of the earthquakes.

• 1) The H/V analysis (Fig. 3) shows the presence of small Vs values in the first 50 m below the VVCH station. This thickness may correspond to pyroclastic deposits (fall and pyroclastic density currents) and volcanoclastic Holocene deposits, where meteoric water can easily circulate and accumulate. The VVCH station is located near the “barranca” (ravine), in a place where thick deposits could have accumulated more easily during the 1982 eruption. The circulation of fluids in the shallow part of the medium could also be favored by the presence of the Catedral-Chichón fault, which almost crosses the VVCH station.

• 2) The value of $\frac{{V}_{P}}{{V}_{S}}$ also indicates the presence of fluids. A typical value of the ratio of the P- and S-wave velocities $\frac{{V}_{P}}{{V}_{S}}= \sqrt{\frac{\lambda +2\mu }{\mu }}$, where $\lambda$ and $\mu$ are the Lamé parameters, is obtained for $\lambda =\mu$ (i.e., $\frac{{V}_{P}}{{V}_{S}}=\sqrt{3}\sim 1.73$). The lowest value is obtained for $\lambda =0$ (i.e., $\frac{{V}_{P}}{{V}_{S}}=\sqrt{2}\sim 1.41$). For soft soils, especially on volcanoes, V_s is small, so $\frac{{V}_{P}}{{V}_{S}}$ can be high, up to 1.8–2.0 (Aki and Richards 2002). Therefore, this ratio typically varies from ~ 1.4 to ~ 2.0. The El Chichón ratio V_p/V_s = 1.91 ± 0.10 (Fig. 2) is therefore a high value and may be due to a large amount of fluid, as it is the case in a highly porous medium. As this ratio is calculated from earthquake depths between 0 and 2 km, the presence of fluids is confirmed, at least in the first 2 km beneath El Chichón.

• 3) The high b value of the Gutenberg-Richter law may indicate the presence of fluids. The Gutenberg-Richter law takes into account the distribution of earthquake magnitudes (Ishimoto and Iida 1939; Gutenberg and Richter 1944). It describes the number N(t) of earthquakes with a magnitude greater than or equal to the magnitude M during a time window (t) as: ${\text{log}}_{10}\left[N\left(t\right)\ge M\right]=a\left(t\right)-bM$, where a(t) is the number of earthquakes of magnitude ≥ 0 during the time window (t), and (b) is a constant over the time interval (t). The b value describes the number of small earthquakes relative to the number of large earthquakes (i.e., the larger b, the greater the number of small earthquakes). The b value of the Gutenberg-Richter law is typically equal to 1 or less, for tectonic earthquakes (Frohlich 1993), whereas for volcanic earthquake sequences, b values are much larger than 1 (typically between 1.2 and ~ 2) (e.gWiemer and McNutt 1997; Wyss et al. 1997, 2001). Here, we were unable to calculate the b value after the 2017 M_w8.2 Tehuantepec earthquake because we did not have 15 days of seismic data after the earthquake. We were only able to calculate the b value before this earthquake, which corresponds to 396 earthquakes for all magnitudes. We used the maximum likelihood method (Aki 1965) and Mc calculated at the rupture of the slope of the Gutenberg-Richter law. We found a b value of 1.41 ± 0.15 (Fig. 7).

Fig. 7

Gutenberg-Richter law of El Chichón volcano, calculated from 01 January 2017 to 08 September 2017 (396 earthquakes of all magnitudes), corresponding to the seismicity before the M_w8.2 Tehuantepec earthquake

The b value of 1.41 ± 0.15 (Fig. 7) is greater than 1.2. This is a high value and confirms the existence of a highly heterogeneous and porous medium that may contain large quantities of fluids. As this value is obtained for earthquake depths between 0 and 2 km, it is consistent with the existence of fluids within the first 2 km below the volcano, as previously inferred from the Vp/Vs ratio. It is more likely that these are hydrothermal fluids rather than magma, since there are no other manifestations that should occur in the presence of shallow magma, such as high deformation, temperature increase, drying of the crater lake, an increase in the temperature of the spring water, or, more tellingly, the emplacement of a new lava dome in the crater.

The presence of hydrothermal fluids is consistent with previous studies and past volcanic activity in several respects. A well-developed hydrothermal system within the first few kilometers of the crust beneath El Chichón has already been identified (Casadevall et al. 1984; Taran et al. 1998; Taran and Rouwet 2008). Surface manifestations of this system include an acid lake, boiling springs, and steam vents within the crater at the summit of the volcano, as well as several groups of thermal springs located on the north-western and southern outer slopes of the volcano (Rouwet et al. 2008; Taran et al. 2008).

Peiffer et al. (2011, 2015) proposed the existence of two thermal aquifers located at two different depths: a shallow volcanic aquifer located within the upper volcanic deposits (up to a few hundred meters below the crater), and a deeper one within the limestone basement (> 2 km b.s.l.). Both aquifers are probably connected by the fracture zone associated with the volcanic conduit and by the local fault systems. The low permeability of the Tertiary rock sequence (separating the volcanic deposits and the limestones) probably “channelizes” the upflow of thermal fluids through the fractured volcanic conduit (Fuentes-Arreazola et al. 2023). Thermal fluids are further mixed with meteoric water as they flow to the surface. Precipitation is intense in the area, averaging ~ 4 m/year (characteristic of humid tropical regions), which has carved many deep ravines (“barrancas”) on the volcano slopes. Fuentes-Arreazola et al. (2023) have also shown the existence of additional water reservoirs using aeromagnetic data. One would be an unconfined aquifer, located at the base of the volcanic edifice in contact with the Miocene sedimentary rocks (eMscu and P-IMscu units described by Garduño-Monroy et al. 2015). The contact between the volcanic deposits and the Cretaceous limestones and evaporites IK-evls units of these authors) would also hold some water.

Another important aspect is the presence of an almost permanent crater lake at the summit of the volcano. As the San Juan fault crosses the crater of El Chichón volcano in an EW direction, and the Cadedral-Chichón fault in a NW direction, a large amount of water may be using these faults to circulate in the subsurface. Water circulation below and around the crater would explain the high observed Vp/Vs ratio, the high b value, the shape of the H/V ratio at the VVCH seismometer, and the concentration of VT seismicity below the crater.

The hydromagmatic eruptions documented in the stratigraphic records and the water-saturated surge deposits (Macías et al. 1997b) also indicate the presence of a high amount of fluids. In particular, the 1982 eruption experienced both hydromagmatic and magmatic explosions, with the presence of fluids indicated by the associated VT seismicity patterns, such as the Gutenberg-Richter and Omori laws before and during the 1982 eruption (Legrand et al. 2015). The seismicity indicated the absence of fluids after the eruption (Legrand et al. 2015), which effectively disappeared almost completely during the eruption.

As discussed in the “Past volcanic activity” section, El Chichón has produced many hydro-magmatic eruptions. This confirms the high amount of hydrothermal fluids during the eruptive process and in the inter-eruptive periods, which continuously feed the hydrothermal reservoirs. The existence of a hydrothermal reservoir beneath El Chichón and its partial destruction during the 1982 eruption is confirmed by the abundance of illite and smectite minerals found in the wet surge deposits of the 1982 eruption (Macías et al. 1997b).

Is the induced seismicity due to magma motions or hydrothermal systems?

The increase of seismic activity beneath an active volcano could also be due to the effect of magma movement. However, the absence of deeper seismicity (at depths of ~ 10–30 km) in 2017, where the magma reservoir is thought to be located (Macías et al. 2003; López-Loera et al. 2020; Fuentes-Arreazola et al. 2023), shows that there is no evidence for magma intrusion, in contrast to what happened in 2020 at the SW of the Paricutin volcano, México (Legrand et al. 2023). The absence of a large positive magnetic anomaly beneath El Chichón (Fuentes-Arreazola et al. 2023) also suggests the absence of a recent magmatic intrusion.

The 2017 VT seismicity, which is shallow (depth <  ~ 2 km), is located on the W and NW flanks of El Chichón volcano. This location corresponds to the well-established hydrothermal system (Taran et al. 1998; Capaccioni et al. 2004; Peiffer et al. 2011), and the Catedral-Chichón fault (Fig. 1). On some occasions in 2017, the VT seismicity was clustered into small swarms of several earthquakes within a few minutes (Fig. 8), as observed at other volcanoes with high hydrothermal activity and discussed in the Introduction.

Fig. 8

Examples of two swarms (time series and spectrograms) on a 28 March 2017 at 15h01 and b 02/10/2017 at 15h50

We suggest that the 2017 shallow VT seismicity, concentrated below the crater, is partly due to the large amount of water beneath El Chichón volcano, confirmed by the H/V ratio study, the high Vp/Vs ratio, and the high b value, and the fact that the crater is centered above the upwelling, boiling, and condensation zone. The 2017 seismicity is therefore likely to be of hydrothermal rather than magmatic origin.

Has the M_w8.2 Tehuantepec earthquake of 08 September 2017 increased the volcanic seismicity rate at El Chichón volcano?

The M_w8.2 Tehuantepec earthquake was located at ~ 270 km from El Chichón, with a back-azimuth to the centroid of 195°. This earthquake clipped out the broadband seismograms at El Chichón. The closest accelerometer to El Chichón that recorded the earthquake is in the city of Tuxtla Gutierrez, 70.4 km away, with a back-azimuth of − 10.6°. The maximum peak ground acceleration (PGA) at Tuxtla Gutierrez is 70.3 cm/s², and the maximum peak ground velocity (obtained by integrating the acceleration and filtering between 0.1 and 10 Hz) is 5.1 cm/s. An interpolated map of the PGA gives a value of about 40–50 cm/s² at El Chichón (internal report of the Instituto de Ingeniería, UNAM). The earthquake-volcano distance corresponds to a near-field distance (i.e., at a distance “$d$” of the same order of magnitude as the characteristic length $\sqrt{S}$ of the earthquake, where S is its rupture surface, (see Legrand 2022 for details), with a triggering index = $d/\sqrt{S}$= 2.3. This index value indicates that the volcano is located at a near-field distance from the earthquake. Thus, the large M_w8.2 Chiapas earthquake had the potential to reactivate El Chichón through static and/or dynamic stress changes. However, this did not happen, confirming that El Chichón is in a quiescent state, i.e., far from erupting in the near future (a few years).

Conclusion

We propose that the shallow (depths < 2 km) VT seismicity recorded in 2017 is of hydrothermal rather than magmatic origin. The high amount of water beneath El Chichón volcano is confirmed by the H/V ratio study, the high Vp/Vs ratio, and the high b value. The VT-seismicity rate of El Chichón volcano increased after the large M_w8.2 Tehuantepec earthquake on 8 September 2017, but only slightly and briefly (during only 2 months). This earthquake was strong enough and close enough to El Chichón to potentially reactivate the volcano if it was ready to erupt, but no eruption occurred. Therefore, we infer that El Chichón volcano is not ready for a major eruption in the next few years. A volcano ready to erupt would have shown a greater change in VT rate after such a large earthquake, over several months, and a deeper (at depths between ~ 10 and ~ 30 km) seismicity, which is not the case, indicating the absence of a deep magma intrusion. Furthermore, a new magma intrusion would have produced other manifestations (higher temperatures of the crater lake, its eventual complete evaporation, high deformations, changes in the chemistry of the water springs and fumaroles or, more revealing, the emplacement of a pre-eruptive lava dome in the crater). Nevertheless, El Chichón must be continuously monitored in case of reactivation of this very active volcano in Chiapas. The VVCH seismometer is crucial in such a continuous and real-time monitoring system, as well as other broadband seismic stations to improve the earthquake locations.

Annex. The generalized Wadati law

In the simple case of a homogeneous half-space, the arrival time ${t}_{P}$ of the P-wave of velocity ${V}_{P}$ from the hypocenter of one earthquake occurring at time ${t}_{0}$ to a station located at a distance $\Delta$ from the earthquake is as follows:

$${t}_{S}= {t}_{0}+ \frac{\Delta }{{V}_{s}}$$

(1)

For a S wave:

$${t}_{S}= {t}_{0}+ \frac{\Delta }{{V}_{s}}$$

(2)

We deduce:

$${t}_{S}\hspace{0.33em}-{t}_{P}=\hspace{0.33em}\Delta \hspace{0.33em}\left(\frac{1}{{V}_{S}}-\frac{1}{{V}_{P}}\right)$$

(3)

Using Eq. (1) into Eq. (3), we have the classical Wadati law: (${t}_{S}\hspace{0.33em}-{t}_{P}$ versus ${t}_{P}$):

$${t}_{S}\hspace{0.33em}-{t}_{P}=\hspace{0.33em}\left({t}_{P}\hspace{0.33em}-{t}_{0}\right)\hspace{0.33em}{V}_{P}\left(\frac{1}{{V}_{S}}-\frac{1}{{V}_{P}}\right)=\left(\frac{{V}_{P}}{{V}_{S}}-1\right){t}_{P}-\left(\frac{{V}_{P}}{{V}_{S}}{t}_{0}-{t}_{0}\right)$$

where the $\frac{{V}_{P}}{{V}_{S}}$ ratio can be deduced from the slope of the line, and ${t}_{0}$ can be deduced from the value at the zero-time ${t}_{P}$.

When the same earthquake “k” is recorded at two seismometers “i” and “j”, its occurrence time ${t}_{0}^{k}$ can be canceled by doing the double difference of S-wave arrival times versus the difference of P-wave arrival times. For each kth earthquake occurring at time ${t}_{0}^{k}$ recorded at each pair ith and jth of stations of distance ${\Delta }_{i,k}$ and ${\Delta }_{j,k}$ respectively, we have:

$${t}_{S}^{i,k}\hspace{0.33em}-{t}_{S}^{j,k}=\hspace{0.33em}\left({t}_{0}^{k}+\frac{{\Delta }_{i,k}}{{V}_{S}}-\hspace{0.33em}{t}_{0}^{k}-\frac{{\Delta }_{j,k}}{{V}_{S}}\right)=\hspace{0.33em}\frac{1}{{V}_{S}}\left({\Delta }_{i,k}-{\Delta }_{j,k}\right)$$

$${t}_{S}^{i,k}\hspace{0.33em}-{t}_{S}^{j,k}\hspace{0.33em}=\hspace{0.33em}\frac{{V}_{P}}{{V}_{S}}\left({t}_{p}^{i,k}\hspace{0.33em}-\hspace{0.33em}{t}_{0}^{k}-{t}_{p}^{j,k}\hspace{0.33em}+\hspace{0.33em}{t}_{0}^{k}\right)$$

$${t}_{S}^{i,k}\hspace{0.33em}-{t}_{S}^{j,k}\hspace{0.33em}=\hspace{0.33em}\frac{{V}_{P}}{{V}_{S}}\left({t}_{p}^{i,k}\hspace{0.33em}-{t}_{p}^{j,k}\right)$$

which is the generalized Wadati law, where the ${t}_{0}^{k}$ disappeared. Hence, if we plot ${t}_{S}^{i,k}\hspace{0.33em}-{t}_{S}^{j,k}$ versus ${t}_{P}^{i,k}\hspace{0.33em}-{t}_{P}^{j,k}$ for each kth earthquake, we have a straight line of slope $\frac{{V}_{P}}{{V}_{S}}$ passing through the origin.

We can also, in a same graph, plot different earthquakes “k” at all the stations, where all the ${t}_{0}^{k}$ will disappear. The main implicit hypothesis is that the $\frac{{V}_{P}}{{V}_{S}}$ is the same in all the region including the earthquakes and the stations. The generalized Wadati law gives a better estimation of $\frac{{V}_{P}}{{V}_{S}}$ than the classical Wadati law.