Implications for volcano monitoring

Extrapolating petrological data of ancient or historic eruptions can be useful to anticipate the potential future behavior of volcanoes (Weber and Castro, 2017; Stock et al., 2017; Muir et al., 2015; Singer et al., 2014). Geochemical patterns in plagioclase and orthopyroxene crystals record recurrent cycles of fresh magma injection and differentiation prior to each of the last three Plinian eruptions of Nevado de Toluca. Inter-Plinian Block-and-Ash flows are compositionally very similar to the Plinian eruptions and are likely sourced from the same plumbing system as the Plinian events (Arce et al., 2006). The consistency in pre-eruptive processes and timescales between these events, separated in time by millennia (Fig. 4.1), would suggest that similar magmatic events will precede a future Plinian eruption of Nevado de Toluca. The surprisingly mafic orthopyroxene compositions found in all studied eruptions stress the important role of deep magmatic processes on the reactivation of the subvolcanic magma reservoir. Magma injection at deep crustal level could be indicated by deep seismicity and broad deformation pattern surrounding a volcano and registered by satellite or ground based geodetic techniques such as InSAR, levelling or GPS (e.g. Magee et al., 2018). Numerical modelling shows that such broad deformation pattern are mainly controlled by chamber depth, crustal heterogeneity and rheology (Hautmann et al., 2010; Gottsmann and Odbert, 2014). While such pattern have been associated with the eruption of mafic magmas (Sturkell et al., 2013; Fedotov et al., 2010), we emphasize that at Nevado de Toluca signs of deep crustal unrest might herald potential Plinian eruptions. Given the volcano’s current long dormancy of over three ka, which is comparable to the order of magnitude of time scales recorded in plagioclase core diffusion, it is likely that renewed magma input into the shallow magma reservoir is required for Plinian or other eruptive activity to resume. Extended periods of dormancy lasting many centuries to millennia are frequently observed at Arc volcanoes (Hildreth, 2007), which can hamper recognition of the volcanic hazard potential by authorities and the local population. Our data shows that Nevado de Toluca is a long-lived mushy system that can transition into a critical state on sufficiently short time scales relevant for volcanic hazard assessment.

Figure 4.11. Comparison of overpressure generated due to magma injection and the depth range of the silicic Nevado de Toluca reservoir. Critical pressures (black lines) for the rupture of a magma chamber and pressure inside a magma chamber due to injection of more magma were calculated as function of depth, using the analytical equations of Canon-Tapia (2014). Dashed curves show the pressure due to magma injection for two different tensile strength of the crust (red: 0.5 MPa; tan: 6 MPa) and are labeled for the ratio of volume change to magma chamber volume (ΔV/V). The grey shaded area marks the depth range of the silicic magma reservoir that fed Plinian eruptions from Nevado de Toluca (Arce et al., 2013; 2006).

Considering the potential effect of magma input on magma reservoir pressurization and failure, we calculated the approximate volume of magma to be injected to generate sufficient overpressure to trigger a Plinian eruption of Nevado de Toluca (Fig. 4.11; Cañon-Tapia, 2014).

The critical pressure at which failure of a reservoir occurs can be calculated as a function of material properties such as tensile strength and Poisson’s ratio of the rock, which was varied between 0.5-6 MPa and 0.25-0.5, respectively (Gudmundsson, 2011; Perras and Diederichs, 2014). Overpressure generated by magma injection was modeled for several volume ratios between input (ΔV) and reservoir volume (V), using a bulk modulus of magma between

1.15∙10¹⁰ and 2 ∙10¹⁰ Pa and the pressure derivative between 4 and 7 (Canon-Tapia, 2014). ΔV/V ratios between 0.002 and 0.01 generate sufficient overpressure to induce failure of a magma reservoir at depth corresponding to the feeding reservoir of the Plinian eruptions of Nevado de Toluca. Combining these volume fractions with the erupted magma volumes for the three eruptions (Fig. 4.1), recharge volumes of 0.0017-0.0085 km³ for the LTP, 0.0036-0.018 km³ for the MTP and 0.016-0.08 km³ are required to trigger the UTP eruption by magma injection. Small volcanic eruptions, occurring with recurrence intervals of days to weeks on Earth, produce bulk volumes very similar to the calculated recharge trigger threshold between 0.001 and 0.01 km³ (Pyle, 2015), indicating that even a relatively small injection of magma in the shallow reservoir may be sufficient to trigger Plinian events at Nevado de Toluca. Using median opx diffusion timescales for the UTP and LTP of 18 years, and 44 years for the MTP yield magma injection rates between 9.44x10^-5 km³ yr^-1 and 8.89x10^-4 km³ yr^-1 required which are likely episodic rather than continuous injections, as such low rates might be not sustainable due to conduit freezing. to reach the critical volumes and trigger the eruptions. The estimates volumes and injection rates are minima, as our calculations do not consider inelastic effects or magma compressibility (e.g.

Kilbride et al., 2016). Nevertheless, volume changes of this magnitude measured during unrest are resolvable from modeling of volcano deformation data derived by InSAR (Jay et al., 2014) or GPS (Sigmundsson et al., 2010) and should be considered as potential indicator of a future Plinian eruption, especially if following the signs of deep crustal magmatic activity.

4.5 Conclusions

Detailed chemical and textural analyses of plagioclase, amphibole and orthopyroxene crystals for three major Plinian eruptions of Nevado de Toluca volcano in Central Mexico reveal a recurring pattern of magmatic processes prior to these events. The major and trace element zoning in plagioclase is consistent with cycles of magma differentiation separated by injection of magma from depth. Recharge magmas are chemically heterogeneous basalts or basaltic andesite judging from the compositions of opx cores that partly equilibrated with the host dacite prior to each of the Plinian eruptions. Distinct groups of opx compositions together with amphibole barometry further reveal that the plumbing system of Nevado de Toluca consists of a complex system of separated magma storage possibly located at different crustal levels. Plinian eruptions are preceded by homogenization processes. Diffusion modelling of Mg in plagioclase indicates that differentiation cycles between input of mafic magma at shallow depth occur over millennia, while Fe-Mg interdiffusion in orthopyroxene constraints the timing of pre-eruptive magma injection in the range of years to less than two centuries. Based on our petrological results and modeling, we conclude that future voluminous explosive eruptions of Nevado de Toluca could potentially be foreseen by geophysical monitoring.

5 T ^HERMO - ^CHEMICAL

PERSPECTIVES ON MAGMA DIVERSITY

Abstract

The chemistry of magmas erupted by volcanoes is a message from deep within the Earth’s crust, which if decrypted, can provide essential information on magmatic processes occurring at inaccessible depths. While some volcanoes are prone to erupt magmas of a wide compositional variety, others sample rather monotonous chemistries through time. Whether such differences are a consequence of physical filtering or reflect intrinsic properties of different magmatic systems remains unclear. Here we show, using thermal and petrological modelling, that magma flux and the thermal structure of the crust modulate diversity and temporal evolution of magma chemistry in mid to deep crustal reservoirs. Our analysis shows that constant rates of magma input leads to extractable magma compositions that tend to evolve from felsic to more mafic in time. Low magma injection rates into hot or deep crust produces less chemical variability of extractable magma compared to the injection of large batches in colder or shallower crust. Our calculations predict a correlation between magma fluxes and compositional diversity that resembles trends observed in volcanic deposits. Our approach allows retrieval of quantitative information about magma input and the thermal architecture of magmatic systems from the chemical diversity and temporal evolution of volcanic products.

Gregor Weber, Guy Simpson, Luca Caricchi

Department of Earth Sciences, University of Geneva, Geneva, Switzerland.

Submitted to Scientific Reports on the 17 February 2020.

5.1 Introduction

Understanding the variability and temporal evolution of erupted magma chemistry is critical to quantify magmatic and ore forming processes (Hedenquist and Lowenstern, 1994; Huber et al., 2009; Caricchi and Blundy, 2015), and to anticipate the potential future activity of volcanoes (Hildreth, 2007). Detailed age-resolved geochemical records show that some volcanic centres erupt a wide variety of magma compositions, while others produce restricted chemical diversity throughout their lifetime (Frey et al., 2004; Bacon and Lanphere, 2006; Hora et al., 2007; Singer et al., 2008; Gertisser and Keller, 2003; Kent et al., 2010; Muir et al., 2015)(Fig. 5.1).

Additionally, for systems that sample a large variability of magma types, the composition and variety of erupted magmas change over time (Fig. 5.1e, f; Fig. S5.3). Differences in

compositional diversity and temporal trends between individual volcanoes have been attributed to different mechanisms. The evolving rheological properties of the crustal rocks hosting magma reservoirs can modify the capacity of magmas to rise to the surface and erupt or accumulate at depth (Jellinek and DePaolo, 2003; Gregg et al., 2013; Degruyter and Huber, 2014; Karlstrom et al., 2017), which may impact on compositional diversity. The physico-chemical properties of the magma itself are a first order control in this respect. Density or viscosity barriers may prevent magmas of specific chemistry from erupting, leading to erupted magmas with a rather

monotonous composition (Stolper and Walker, 1980; Marsh, 1981). Such barriers might be overcome by mixing of mafic and silicic compositions (Kent et al., 2010; Kent, 2013) or develop in parallel with the construction and destruction of large volcanic edifices (Pinel and Jaupart, 2000). In these models, volcanoes are depicted as physical property filters that sample only a part of the compositional spectrum present in their plumbing system. It is not clear, however, why such filters would be effective in volcanoes with restricted chemical variability and not be as effective in systems where erupted magmas exhibit a variety of compositions. Alternatively, differences in the range of erupted magma compositions could reflect contrasting recharge regimes and thermal states of crustal magma reservoirs feeding volcanoes(Wörner et al., 2018).

In this study we further test this hypothesis, exploring the thermochemical evolution of mid to deep crustal magmatic systems subjected to injection of hydrous basaltic magma.

Figure 5.1. Differences in the geochemical variability of Arc volcanoes. Total alkali (K2O+Na2O wt.%) versus silica (SiO2 wt.%) plots with indicated compositional fields of whole rock data representing the long-term eruptive histories of well-studied volcanic systems. a) Nevado de Toluca (Trans Mexican Volcanic Belt). b) Merapi (Sunda Arc, Indonesia). c) Mazama-Crater Lake (Cascades, Oregon). d) Puyehue-Cordón Caulle (Southern Volcanic Zone, Chile). e) SiO2

(wt.%) versus age for Mazama modified from Bacon and Lanphere (2006). f) SiO2 (wt.%) of dated eruptive products for Puyehue-Cordón Caulle modified from Singer et al., (2008). Orange lines in e) and f) are moving averages, grey shading indicates range of compositions. Data was taken from the GEOROC database (http://georoc.mpch-mainz.gwdg.de/georoc/) and new whole rock analysis for Nevado de Toluca (Methods, Electronic table E5.2). The data exemplify that volcanic systems show large differences in the variability of erupted compositions.