Many volcanoes are sites of interaction between the hottest (magma) and coldest (ice) natural materials on Earth’s surface. Magmatic and volcanic activity in the presence of ice or snow gives rise to eruptive activity defined under the term of glaciovolcanism (Kelman et al., 2002; Smellie, 2006). In addition to the obvious candidates of Antarctica and Iceland, glaciovolcanism has also occurred extensively at volcanoes that line the edges of continents around the Pacific Ocean (e.g., Mee et al., 2009; Cole et al., 2018; Wilson et al., 2019; Kataoka et al., 2021; Loewen et al., 2021; Figure 1). Subduction zone volcanism throughout Japan, Kamchatka, north-western United States, Central and South America, New Zealand, Papua New Guinea, Indonesia, and the Philippines has typically produced tall composite cones that form as a result of frequent eruptions over long time periods (Davidson and de Silva, 2000). The lofty heights of these volcanic features mean that many of them experience seasonal snowfall under current climate conditions, and ones that are located in the high-latitudes host permanent ice caps and glaciers (e.g., Mount Veniaminof, Alaska). In this “Pacific ring of fire and ice”, even volcanoes in tropical regions such as Papua New Guinea (e.g., Mount Giluwe; Barrows et al., 2011) and Mexico (e.g., Nevado de Toluca; Capra et al., 2013) have experienced glaciation and therefore may have been affected by glaciovolcanism.

The abundance and extent of glaciers on volcanoes have fluctuated in response to 100,000 years-long global climate cycles over the last ∼700,000 years (Goñi et al., 2019; Coombs and Jicha, 2020). During the Last Glacial Maximum (LGM), around 25,000–20,000 years ago, when the average global temperature was 6°C colder than today (Tierney et al., 2020), large glaciers were present on many volcanoes, including ones that are currently free of permanent ice (Eaves et al., 2016). Signs of past glaciation exist as u-shaped valleys, moraines, and striated lavas (Figure 1B). Within the traditional view of the evolution of stratovolcanoes affected by glaciers (e.g., Hobden et al., 1996), these features record periods of erosion (during relatively cold climate stages) that were distinct in timing from periods of volcanic growth (during relatively warm climate stages). Under this paradigm, ridges of lava on the sides of valleys are inferred to have been carved out by glaciers. This has certainly occurred at numerous volcanoes (e.g., Singer et al., 1997); however, ⁴⁰Ar/³⁹Ar and K-Ar dating of lavas has also shown that eruptions have occurred during periods when glaciers were extensive in many cases (Fiertsein et al., 2011; Conway et al., 2016; Calvert et al., 2018; Pure et al., 2020). These examples challenge the dichotomy between volcanic growth and glacial erosion and present a key question: what happens when lava is erupted at a stratovolcano that is covered in ice?

Building on pioneering work by Mathews (1952) in British Columbia, a landmark publication by Lescinsky and Sisson (1998) showed that thick lavas located on the tops and sides of ridges at Mount Rainier in Washington, United States (Figure 1C), formed when they flowed down the flanks of the volcano and were confined along the margins of valley-filling glaciers during past cold periods of Earth’s history. The ice deflected the lavas from flowing into valleys, forcing them to pond into thick units along high-elevation parts of the volcano. The glaciers retreated after thousands of years, and left the lavas “perched” on the tops and sides of ridges. Similar landforms have been described at Mount Ruapehu in New Zealand (Figures 1, 2; Spörli and Rowland, 2006; Conway et al., 2015). The steep sides of such outcrops display glassy textures and characteristic fractures that indicate they cooled quickly against water and steam produced by melting of the ice (Figure 2C). Rapid cooling against vertical walls of ice has produced horizontal or curved “columns” on many ice-bounded lavas (Figure 2D), which are formed by the propagation of fractures away from (often complex) cooling surfaces (Lodge and Lescinsky, 2009). The intersection of the fractures at angles of ∼60° forms the characteristic cross-sectional polygons with diameters of ∼10–20 cm (Figure 2C).

As forecast by Lescinsky and Sisson (1998), ice-bounded lavas have now been identified at several stratovolcanoes around the Pacific (e.g., Mee et al., 2009; Conway et al., 2015; Lachowycz et al., 2015; Coombs and Jicha, 2020; Mixon et al., 2021). Although the mechanism of their formation is well-defined in theory, it remains difficult to grasp the dynamics of lava-ice interaction without direct observation of such an eruption. Emplacement of lava flows onto snow (Edwards et al., 2015) and ice caps (Loewen et al., 2021) has been documented, and large-scale experiments using basaltic melt have been undertaken in the last decade (Edwards et al., 2013). Even with these reference points, identification of ice-bounded lava flows during field surveys requires a certain level of intuitive assessment via real-time reconstruction of past environments. There are many circum-Pacific stratovolcanoes with overlapping eruption and glacial histories where lava-ice interaction has not been documented. The experiment described in this contribution is designed to convey the stratigraphic and morphological characteristics of ice-bounded lavas, which will assist field geologists with identifying these features in such settings.

Active learning methods have been shown to be a successful teaching tool for secondary and tertiary school science subjects (Hake, 1998). Food-based experiments have been particularly popular within this approach, likely because they use inexpensive items and kitchens can be portrayed as accurate representations of scientific laboratories, where hypotheses are tested by carrying out reproducible methods. In Japan, the recipes of “Volcano Kitchen” have been a popular source for outreach activities that utilize common ingredients to teach lessons in volcanology (Hayashi, 2006). A recently published children’s book based on the August 2021 eruption of Fukutoku-Okanoba volcano has also incorporated cooking to introduce primary school students to the petrology of volcanic rocks (Maruya, 2022). Baked goods and candies that look like rocks and minerals are commonly found at geopark visitor centres throughout Japan, and have been successful for introducing domestic and foreign tourists to the textures of rocks in the geologically diverse country. Other examples from around the world include the M&M^® magma chamber for simulating fractional crystallization (Wirth, 2003) and the use of fudge and syrup for explaining lava flow rheology (Rust et al., 2008). Whereas precise scaling of natural phenomena is required for analogue modelling (Galland et al., 2006), analogy experiments (after Ichihara, 2023), including the one presented in this contribution, can still convey valuable information with approximate scaling of physical parameters.

Providing a pathway for students to see geoscience as an option for further study and work is indeed a worthwhile activity. Building sustainable and resilient societies will rely on future generations tackling the major issues of climate change mitigation, renewable energy generation, mineral extraction, and natural hazard assessment. Introducing geoscience research to primary and secondary school students (ages 5–18 years) may encourage them to pursue further study of science, technology, engineering, and mathematics (STEM) subjects at university and a career aimed at tackling these challenges. The aim of this contribution is to showcase an analogy experiment of a common glaciovolcanic process (i.e., lava-ice interaction), that can be utilized to introduce the topics of climate change and volcanic hazards to students and non-specialist audiences. The common ingredients and basic kitchen tools make this exercise widely applicable, and the hands-on format is designed to encourage laboratory research via active participation in school classrooms and outreach demonstrations for all ages.

A basic soda bread is used as the stratovolcano analogy (hereafter, Volcano) in the experiment. Sift 3 cups (360 g) of wheat flour and 3 teaspoons (12 g) of baking powder into a large bowl and mix. Pour 100 mL of soda water (also known as sparkling water) into the bowl at first, and stir into the dry ingredients. Add more soda water by increments of 10 mL until the mixture becomes flaky. Push the flaky parts together with your hands to form a dough, and knead by pressing and folding it over repeatedly. If the dough is not sticking together when you fold it, add some more soda water until it becomes a cohesive ball. If the dough is sticking to your hands, add some more flour until you can handle it easily. Spread the dough out into a broad cone, by pushing down the edges. Place the dough on a lightly greased oven-proof plate. Score the top of the dough by making four ∼1 cm-deep incisions from near the edges to within 2 cm of the centre, at 90° to each other. Melt the butter and rub it over the top of the dough. The oven or microwave (using the oven setting) should be pre-heated to 200°C, prior to putting in the plate with the dough to bake for 30 min at 200°C. Take out the bread and leave to cool for approximately 10 min. Cut around the score marks to form “valleys” that are 2–3 cm-wide and ∼5 cm-long.

Two gelatinous chocolate sauces are used as lava flow analogies in the experiment. To make the first lava flow analogy (hereafter, Lava 1), place 5 g of thickener (powdered collagen gelatine) in a cup and add ∼10 mL of hot water, plus 5 g of dark chocolate and 15 g of white chocolate. Heat in a microwave for 40 s (∼700 W power), then stir thoroughly to form a thick sauce with even consistency. Repeat the steps using 20 mg of dark chocolate, 5 g of thickener and 10–15 mL of hot water to make Lava 2 in a separate cup. Lava 2 should be slightly less viscous than Lava 1 for the purposes of the experiment.

Buy a small container of light-colored plain milk-based ice cream (e.g., vanilla) to use as the glacier analogies (hereafter, Glaciers) in the experiment.

Any carbonated drink can be used as an alternative to soda water. The authors have successfully made the bread volcano using ginger ale, cola, and beer (including non-alcoholic beer). To modify the taste of the bread, add 1–2 teaspoons of salt, and/or 1–2 tablespoons of sugar to the dry ingredients. Milk can be used as an alternative to water when making the sauces.

The ingredients contain gluten, dairy, and sugar. Please check the dietary restrictions of participants if you choose to share the bread, ice cream and chocolate as snacks to eat after the experiment. We do not recommend eating the sauce or ice cream used during the experiment. Instead, consume the remaining chocolate and ice cream ingredients.

The experiment can be undertaken after the analogy materials have been constructed (Volcano, Lava 1 and Lava 2, and Glaciers). There are 3 main steps within the experiment, each representing a key stage in the evolution of a glaciated stratovolcano: 1) Syn-glacial; 2) deglacial; and 3) post-glacial.

The aim of syn-glacial stage of the experiment is to reproduce lava flow eruptions that occurred during glacial stages when ice extent was advanced on stratovolcanoes (e.g., during the LGM). Add Glaciers to the valleys on the Volcano. The Glaciers should be thickest in the middle of the valleys, and sit just below the height of the flanks at the heads and sides of the valleys. Leave an area with no Glaciers on the top of the Volcano to represent the summit vent area. Pour Lava 1 onto this summit area at an approximate rate of 2–4 mL s⁻¹ _, from a height of 2 cm. Vary the starting point of these lava flows on the summit so that Lava 1 moves down onto different sides of the Volcano. Lava 1 should flow down along the ice-free parts of the Volcano, and into the gap between the sides of the valleys and their Glaciers (Figures 3A, B). Leave the Lava 1 flows to cool down and harden.

Whilst making Lava 2, leave the Glaciers to melt and then scoop away the remainder carefully from the valleys. This represents the deglacial stage of the experient. Try not to disturb the Lava 1 flows that have set along the tops and sides of the valleys.

The post-glacial stage of the experiment is designed to replicate the effusion of lava flows on an ice-free stratovolcano following deglaciation. Pour Lava 2 at the same rate onto the top of the volcano in increments and from different starting points again. This time, Lava 2 flows should travel over the syn-glacial stage Lava 1 flows and move into the bottom of the valleys (Figures 3C, D).

There are 138 active arc stratovolcanoes with glaciers on Earth (Edwards et al., 2020). Many more stratovolcanoes hosted glaciers during past cold periods such as the LGM, but are currently free of permanent ice (e.g., Eaves et al., 2016). Glaciers have therefore been a key part of the life cycles of many volcanoes, and will continue to impact eruptions in the future. As fluctuations in the volumes of glaciers on human timescales become apparent (Vargo et al., 2020), it is pertinent to be aware of how glacier extents have changed over longer timescales in the past. Examining the links between volcanism and the cryosphere that arose from these past changes is important for forecasting future eruptive activity.

Stratovolcanoes grow into large constructional structures over time periods of several 10² to 10⁵ years due to the repetitive emplacement of lava and tephra onto their flanks (Davidson and de Silva, 2000; Yamamoto et al., 2018; Christiansen et al., 2020). Whereas symmetrical cones such as Mount Fuji are often presented as icons of arc stratovolcanoes, many are in fact asymmetrical and complex structures (Figures 1B, C). The latter types exist because volcanoes also undergo destruction throughout their lifetimes, via sector collapse and erosion. Ice is a key agent for both collapse and erosion at volcanoes that have been affected by glaciers. Retreating glaciers promote edifice collapse by destabilising steep and weak units of volcanic rock (Capra, 2006), whereas advancing glaciers scour out valleys and push eroded material onto the ring plains that surround volcanoes (Eaves et al., 2016). Ice also acts to modify the preservation and distribution of volcanic rocks on stratovolcanoes during eruptions, as shown by the experiment (Figure 3). Understanding how glaciers have affected the preservation of volcanic deposits is vital, because these deposits provide a necessary record of past volcanic activity and, therefore, a framework for the possible timing and types of future eruptions at active volcanoes.

The most important message conveyed by the experiment is that many of the lavas observed on the sides of valleys exist because glaciers deflected them onto those locations during eruptions, rather than having exposed them via erosion after they were emplaced onto ice-free flanks. Syn-glacial lavas exhibit an increase in thickness towards the valley and characteristic cooling fractures on their margins that indicate they were impounded and chilled by valley-filling ice (Figure 2; Figure 5A). Post-glacial lavas generally exhibit more uniform thicknesses, with no evidence for impedance of their flow towards lower elevations by a morphological barrier, and will lack evidence for rapid cooling (Figure 5A). A major implication of identifying ice-bounded lavas is that stratovolcanoes can be continuously active during their lifetimes. This differs from the traditional model of stratovolcano evolution that states that eruptions and edifice growth occur during interglacial stages and relative inactivity and erosion occur during glacial stages.

Age data for ice-bounded lava flows record more than just the timing of an eruption; they provide in situ constraints of the past presence and extent of ice in alpine settings (Lescinsky and Fink, 2000). Conversely, ages for post-glacial lavas that flowed into valleys provide constraints for the timing of glacier retreat (Figure 5B; Conway et al., 2015). These constraints are valuable for reconstructing the past climate and environments of Earth. Despite these useful products, the influence of ice on the distribution of lava flows during eruptions also produces complex stratigraphic relationships that can be difficult to unravel. Post-glacial lava flows may appear to be stratigraphically lower than syn-glacial ice-bounded lavas in the field, although they are younger in age (Figure 5C). This apparent inversion of stratigraphy is important to recognize when documenting the eruptive history of a stratovolcano, and using geochronology to measure absolute ages for volcanic units is useful for testing such stratigraphic relationships in the field.

Studying the differences in volumes and compositions of syn-versus post-glacial lavas and tephras (explosive eruption deposits) also offers the opportunity to investigate whether the addition/removal of ice masses on volcanoes affect their underlying magmatic systems. Geological and geochemical evidence from Iceland indicates that deglaciation after the LGM resulted in an increase in rates of volcanism because the removal of ice on the crust resulted in increased production and eruption of new magma (Maclennan et al., 2002). It remains unclear whether deglaciation leads to increased eruption rates at arc stratovolcanoes (Watt et al., 2013; Conway et al., 2016; Pure et al., 2020), although a few studies have interpreted that eruption dynamics at some volcanoes were influenced by the retreat of ice following the LGM in Chile (Rawson et al., 2016). With glacier volumes in many locations currently decreasing in response to anthropogenic climate change (Vargo et al., 2020), it is critical to understand the links between deglaciation and volcanism in more detail so that we can forecast how the melting of glaciers and ice sheets will affect eruptive activity in the future (Tuffen, 2010).

Additional hazards related to volcano-ice interaction include sector collapses and lahars. A large landslide in 2010 at Mount Meager in southwest Canada was prompted by glacial retreat and melting of snow (Roberti et al., 2018). Meltwater lahars are an extremely hazardous phenomenon produced at snow- and ice-clad stratovolcanoes (Pierson et al., 1990; Uesawa, 2014; Smellie, 2022). Snowy volcanoes are popular tourist locations for winter activities, so constraining rates of meltwater production (Cole et al., 2021) and modelling the travel pathways for lahars (Kataoka et al., 2021) represent vital research objectives and hazard mitigation tools in volcanology.

The coexistence of magma and ice at glaciated stratovolcanoes make them hotspots to learn about cool geological ideas. Lava-ice interaction has produced stunning outcrops of volcanic rock (Figure 2), which record valuable information about the evolution of volcanoes and Earth’s climate. These visually striking features can serve as icons for geoscience education, to increase future generations’ interest in mitigating climate change and natural hazards. Background information and the key lessons about volcanoes and glaciers that the experiment highlights are summarized below.

1 Global climate variability has produced repetitive 100,000 years-long cycles of glacial advance and retreat over the last ∼700,000 years. Ice sheets and glaciers grew during the last glacial period, and had well-advanced extents during the LGM around 20 ka, when the average global temperature was 6°C lower than today.