From volcanic hazard to valuable resource: the hidden potential of lava tubes

1 Introduction

Lava tubes are volcanic environments formed by the surface cooling of lava flows. As the uppermost layer solidifies, it develops an insulating crust that allows molten lava to continue flowing beneath it. When the molten interior eventually drains, an empty cavity remains. A lava tube is defined as a “roofed conduit of flowing lava—active, drained, or plugged” (Kauahikaua et al., 1998; Halliday, 2002), a term encompassing any lava pathway regardless of shape, size, or cavity preservation. When referring specifically to primary, non-erosive volcanic cavities, we use the term “lava tube caves”. This review excludes erosional volcanic cavities such as those produced by removal of the scoria layer of ‘a‘ā flows (common where lava meets the sea or rivers) or by the erosion and enlargement of ground fractures (Megerssa et al., 2019). Although lava tubes were once thought to form only within the smooth, low viscosity pāhoehoe lavas typical of Hawaii (Greeley, 1971a), they are now recognized as widespread features occurring in nearly all effusive volcanic environments, whether active or dormant (Figure 1).

Lava tube caves are found across many volcanic regions worldwide—including the western United States, Mexico, Easter Island, Iceland, the Canary and Azores islands, Cape Verde, the Galápagos, Japan, and Italy—testifying to their global distribution (Figure 1). They are common in pāhoehoe flows (e.g., Peterson and Swanson, 1974; Peterson et al., 1994; Hon et al., 1994; Kauahikaua et al., 1998) but also occur within the viscous and rough ‘a‘ā flows, such as those characteristics of Mt Etna in Italy (Calvari and Pinkerton, 1998; 1999; Guest et al., 2012). Owing to their distinctive morphologies and striking underground landscapes, many lava tube caves have become major geo-touristic attractions. Prominent examples include the Undara system (Australia), La Corona tube and Cueva del Viento (Canary Islands), Kazumura (Hawaii), Yongcheon Cave (Jeju Island, South Korea), the lava tubes of Mt. Etna, Vesuvius, and Stromboli (Italy), and Viðgelmir in Iceland. Newly identified tubes in Saudi Arabia (Harrat Kishb), Cape Verde, the Democratic Republic of Congo, Ethiopia, and Jordan further demonstrate that these features remain an active subject of exploration and documentation. Throughout history, lava tubes have served as dwellings, refuges, ceremonial sites and storage areas, benefitting from their natural insulation and stable internal microclimates (Calvari and Puglisi, 2007; Romio and Lobosco, 2025). Their stable conditions also favor the preservation of archaeological materials and faunal remains (Pint and Al-Shanti, 2003; Lan Cuong, 2022), making them important archives of cultural and paleo-biological information. Caves are important because they provide records of a planetary body’s geological, meteorological and environmental history, and on Mars, this may include evidence of past or even present microbial life (Titus et al., 2021). From a volcanological standpoint, lava tubes are critical because they enable lava to be transported efficiently over great distances, thereby increasing the reach and destructive potential of effusive eruptions (Atkinson et al., 1975; Greeley, 1987; Barberi et al., 1993; Calvari and Pinkerton, 1998; Kauahikaua et al., 1998). Their study has therefore become essential for understanding lava flow dynamics, emplacement mechanisms, and hazard assessment. In recent decades, research on lava tubes has expanded into fields such as planetary geology, astrobiology, engineering, and architecture, particularly due to their potential as natural shelters or bases for human activity on the Moon, Mars, and other rocky planetary bodies (Cushing et al., 2015; Blair et al., 2017; Sauro et al., 2020; Martin and Benaroya, 2023; Chanover et al., 2023; Carrer et al., 2024; Feng et al., 2024). The temporal evolution of scientific publications on lava tubes from 1940 to 2025 (Figure 2) reflects this growing interdisciplinary interest. Research output remained limited until the late 1970s, gradually increased through the 1990s, and surged after 2010 in parallel with advances in volcanology, geomorphology, cave science, planetary exploration, and subsurface analogue studies. Publication rates peak around 2021–2022 at roughly 100 papers per year, underscoring the increasing scientific relevance of lava tubes.

In this review, we synthesize current knowledge on terrestrial lava tubes, highlighting their formation mechanisms, structural and textural characteristics, and diagnostic relationships with lava flow surface morphologies such as skylights, breakouts, tumuli or shatter rings. Given their sizes (a few meters to tens of meters), these features might not be visible on the Moon’s surface, or their existence might be covered by the thick mantle of regolith that can reach thicknesses even of tens of kilometers (Hiesinger and Jaumann, 2014). We discuss how lava tube features record key dynamical processes—such as flow propagation, stagnation, inflation, and drainage–and how surface indicators can be used to infer the presence of underlying tube networks. We further examine volcanic hazards associated with lava tubes during active effusive eruptions, and their post-emplacement stability, accessibility, and potential role as geo-heritage sites. Because lava composition strongly influences tube formation (Wood, 1977), we also compile a dataset summarizing the compositions, ages, and dimensions of lava flows that host tubes and compare them to andesitic–rhyolitic flows that do not (Supplementary Table SM1). This provides insights into the compositional thresholds and environmental conditions favorable for tube development. The review focuses exclusively on primary volcanic cavities formed by magma transport at the surface, excluding features generated by erosional, gravitational, or tectonic processes, as these are not directly related to lava flow emplacement dynamics.

2 Dataset

To collect data for this review on lava tubes, we started with the online catalogue compiled by Paul Burger, updated as of 6 September 2025. This catalogue lists 837 cavities over 50 m (https://cave-exploring.com/index.php/long-and-deep-caves-of-the-world/world-lava-tubes/; accessed on 7 October 2025 at 11:48) located in 25 different countries, primarily the USA (Hawaii), Spain (Canary Islands), and Africa. From this list, we selected those for which major element composition data were available in the volcanology literature, and we added a few lava tubes not mentioned in the catalogue. Our final list examined in this review comprises 65 lava tube analyses from 25 countries, spanning in composition from carbonatitic through basaltic up to trachytic. When available, we added the age and size of the lava tubes to our dataset to investigate their longevity and gain insights into their persistence and stability.

3 Overview on effusive volcanism

Effusive eruptions primarily involve the emplacement of lava flows, either as single flows or as compound lava flow fields (Walker, 1971). Single flow units (Figure 3a) are commonly characterized by a rough ‘a‘ā surface, whereas compound lava flow fields (Figure 3b) generally include both ‘a‘ā and pāhoehoe but typically feature a more common smooth pāhoehoe surface texture (Byrnes and Crown, 2001).

A single lava flow unit is an individual flow with a distinct basal crust, an upper crust, and a molten inner core, together forming a separate cooling unit (Figure 3a). Such a unit can remain molten even after being buried by subsequent flows (Nichols, 1936; Walker, 1971) and may display either ‘a‘ā or pāhoehoe surface morphology, as well as transitional types such as spiny, slabby, toothpaste or sharkskin pāhoehoe (Self et al., 1998; Harris et al., 2017).

In contrast, a compound lava flow field consists of multiple lava flow units separated by time intervals ranging from minutes to years (Walker, 1971). These fields may include exclusively ‘a‘ā flows, exclusively pāhoehoe lobes, or a mixture of both. Typically, at Mt. Etna ‘a‘ā flows dominate the early phases of eruptions characterized by high effusion rates (Wadge, 1981) (Figure 3a), whereas pāhoehoe flows are more common during the final, low-effusion-rate stages (Figure 3b), whereas the opposite applies to Kilauea volcano (see Cashman et al., 2026 and references therein).

Whether an eruption produces single lava flows or compound flow fields depends primarily on the effusion rate (Walker, 1971): high effusion rates favor the formation of single flow units, whereas low effusion rates promote the development of compound flow fields. Additional controlling parameters include eruption duration and lava viscosity. Long-duration eruptions and low-viscosity lava tend to form compound lava fields, whereas short-lived eruptions and more viscous lava tend to produce single flow units (Walker, 1971).

As lava advances down the flanks of a volcano, it develops stationary lateral levees (Sparks et al., 1976; Lipman et al., 1987), channels, and other structural features whose morphology depends on magma physical properties (e.g., viscosity, density), topography (especially slope), effusion rate, and emplacement duration (Walker, 1971; 1973; Wadge, 1981; Lipman et al., 1987; Kilburn and Lopes, 1988; 1991; Rowland and Walker, 1990; Harris et al., 1998). Effusive eruptions can produce extensive flow fields that travel substantial distances from the vent. In Hawaii, lava flows have travelled 12–27 km from their source (Peterson and Swanson, 1974; Lipman et al., 1987; Mattox et al., 1993). On Mount Etna, the Alcantara flow reached 24 km (Branca et al., 2019), while the 1669 eruption responsible for significant destruction in Catania produced a lava flow that advanced 17 km before entering the sea (Branca et al., 2013). More recent Etna eruptions (1981, 1983, and 1991–93) generated flows extending 7–8 km (Guest et al., 1987; Calvari et al., 1994; Stevens et al., 1997).

In some cases, compound flow fields can reach even greater distances. Notable examples include the 35-km-long Laki flow field in Iceland (Thordarson and Self, 1993), the 52-km-long 1859 Mauna Loa flow in Hawaii (Kauahikaua et al., 2002), the 55- to 160-km-long Undara, Toomba, and Kinrara flows in Australia (Atkinson et al., 1975; Stephenson et al., 1998), the 75-km-long Carrizozo flow field in New Mexico (Keszthelyi and Pieri, 1993), and the exceptional 170-km-long Pampas Onduladas flow in Argentina (Pasquaré et al., 2008; Espanon et al., 2014). The most remarkable examples of long-distance emplacement are the continental flood basalts of the Columbia River, Deccan, and Etendeka–Paraná provinces, which inundated vast regions without constructing large volcanic edifices (Keszthelyi et al., 2006).

Across these settings, the main mechanism enabling lava to propagate over great distances—even in flood basalt provinces (Self et al., 1998; Keszthelyi and Self 1998)—is inflation, which leads to the development of lava tubes. Lava tubes act as thermal insulators, preserving the molten interior from rapid cooling (Malin, 1980; Hon et al., 1994). It is widely accepted that long lava flows are sustained by such tube-fed systems (Self et al., 1998; Keszthelyi and Self, 1998; Keszthelyi et al., 2006), which explains the considerable effort devoted to understanding the processes governing their formation and growth.

4 History of lava tube investigation

The descriptions of Anton Giulio de Amodeo, known as Filoteo, a speleologist from Castiglione in Sicily (Italy), are likely among the earliest written accounts on lava tubes. He analyzed several lava tubes from Etna in a book titled Topographia, published in 1591 (Brunelli and Scammacca, 1975). Other interesting accounts include Hamilton’s (1776) descriptions of the 1764 Vesuvius eruption, where he observed a lava channel crusting over, and of the lava tubes and cavities within eruptive fissures that he visited on Etna. Lyell (1847) also described Coan (1844) narrated the 1843 Mauna Loa eruption. Coan (1844) observed flowing lava through a skylight (an opening in the roof of a lava tube) and coined the term “pyroduct” (conduit of fire) for the conduit (Coan, 1844; Kauahikaua et al., 1998). Lucrezi (1977) asserted that the first observations on lava tube caves were made by Aristotle (384–322 BC), Seneca (4 BC), and Pliny the Elder (23–79 AC). However, it was the biologist Lazzaro Spallanzani (1792) who visited and conducted the first experiments on the lava tubes of Vesuvius and Etna, who was arguably the first scientist to investigate these phenomena. Spallanzani (1792) described a tube he noticed during his 1788 journey on Vesuvius and provided an estimation of lava flow’s velocity. In Sicily, the geologists Dolomieu and Sartorius von Waltershausen described and surveyed some of Etna’s lava tubes, offering some tentative explanations of their possible genesis, as did Georg Hartung in the Canary Islands in 1856, who described the Corona lava tube system (Romio and Lobosco, 2025).

Macdonald (1972) illustrated lava tubes formed by inflation of pāhoehoe lava flows and revealed their role in feeding the Columbia River flood basalt. However, their presence might not be evident because the flow’s interior, spreading on a gentle slope, did not drain. More recently, Peterson and Swanson (1974) and Peterson et al. (1994) explained how lava tubes form and distinguished three main ways of roofing over a lava channel, as detailed in the following paragraph. Many studies have focused on qualitatively describing the development of lava tubes in a lava flow field, but only a few accounts have quantitatively examined the physical process of lava tube formation (e.g., Valerio et al., 2008; Filippucci et al., 2011). There are even fewer papers that analyze the mechanisms responsible for the formation of lava tubes through petrological, geochemical and viscosity measurements (e.g., Corsaro et al., 2005; Cashman et al., 2006; Lanzafame and Ferlito, 2014). New interesting results were provided by Romio et al. (2026), which present a parametric 3D lava tube generator, based on a wide catalog of terrestrial lava tube cross-sections that enables the simulation of realistic cave environments for planetary research, while also supporting terrestrial applications, and offering a cost-effective method for reconstructing lava tubes’ 3D geometries for the evaluation of volcanic hazard and creation of virtual reality models. In the following, we describe the main lava tube features and the processes leading to lava tube formation and growth.

5 Lava tube caves and mechanisms of lava tube formation

Lava tube caves are fascinating volcanic structures, often representing beautiful, magical or even religious locations to explore when visiting volcanoes. Their study was recently reconsidered due to the idea that they might offer possible shelters for humans on other planets (Sauro et al., 2020; Martin and Benaroya, 2023), and this aspect has prompted further investigations (e.g., Carrer et al., 2024), although considerable technical and engineering challenges remain before lava tubes can realistically be used as human shelters (Titus et al., 2021). The initial reason that promoted scientists’ interest in studying lava tubes was their capacity to isolate active lavas from heat loss as they move on the ground surface. This isolation thus increases the expansion and threat posed by lava flows and lava flow fields (Peterson and Swanson, 1974). The only case where lava tubes have probably been reported to protect the land from further lava flow invasion is on Kilauea volcano (Hawaii Islands). Here, under stationary discharge rate conditions, lava tubes transport lava directly to the ocean, preserving the houses and infrastructure built near the active lava flow fields from further lava invasion (Koeppen et al., 2013).

Lava tubes are primary volcanic conduits of various sizes, extensions and shapes, which form due to the non-Newtonian rheology of the lava and of the yield stress of magma at sub-liquidus temperatures (Hulme, 1974; Pinkerton and Stevenson, 1992). The large difference in temperature between the lava (800 °C–1250 °C; Sigurdsson, 2015) and the environment (commonly between 0 °C and 30 °C) results in rapid cooling of the lava flow as soon as it exits a vent. This leads to the sudden formation of a solid crust, under which the lava can travel with virtually no additional cooling. In fact, lava flowing within tubes may lose just 0.5 °C/km (Keszthelyi, 1995), increasing its potential enormously for further lengthening and land destruction. It is crustal growth that promotes lava tube formation and accretion, and this process can occur through a variety of different mechanisms, as described in the following section.

The ability of a lava flow to solidify on a slope as soon as the supply from the feeding vent stops is clear evidence of the non-Newtonian behavior of lava, which can be assimilated to a Bingham fluid (Hulme, 1974). To create a Bingham fluid flow, its yield strength must be overcome. This can be caused by a continuous lava supply from behind or by a steep slope ahead of the flow front that promotes the drainage of the lava flow interior. As soon as a lava flow exits from a vent, its initial temperature decreases through irradiation (heat exchange with the surrounding atmosphere) and conduction (with the ground). This causes the formation of a solid crust at the base, margins, and top of the flow, which soon detaches from the fluid interior, forming a motionless envelope. This corresponds to the formation of a lava tube (Merle, 1998; 2000). The obvious consequence of the existence of a strain gradient is that any lava flow able to form lateral motionless levees should, in principle, be able to produce a lava tube, provided the supply rate continues for long enough to form a stable crust, and if drainage allows the formation of a cavity. However, the formation of a rooted upper crust stable enough to allow lava flow movement below requires: (i) time, (ii) gradual cooling, (iii) a stable supply and (iv) an appropriate viscosity. Thus, the main conditions for lava tube formation can be summarized as: (i) an eruption duration of at least 1 or 2 days, or sufficient to allow crust formation; (ii) a steady lava supply at the erupting vent that promotes gradual flow cooling and crust formation; (iii) an effusion of fluid lava that has not been greatly degassed, which allows for the separation of a solid crust from the hot fluid interior; and (iv) a lava viscosity gradient sufficient to cause the detachment of a solid crust from the flow interior (Greeley, 1987; Calvari and Pinkerton, 1998; Byrnes and Crown, 2001; Koeppen et al., 2013).

When an effusive eruption starts, the first stage of lava emplacement is normally that of sheet flows (Peterson and Swanson, 1974; Lipman et al., 1987; Calvari, 2019). These form either ‘a‘ā or pāhoehoe flows when lava is pouring out from an active vent at a high discharge rate (Wadge, 1981; Harris et al., 2005; 2007; 2011; Tonato et al., 2025). Sheet flows are characterized by wide, thin sheets of lava with a flat top and no apparent channels or other structures on the surface (Hon et al., 1994; Hoblitt et al., 2012). As lava flowage continues, regions of stationary liquid will develop along the sides of a flow, and lateral cooling promotes the formation of shear zones (Borgia et al., 1983; Merle, 1998). These result in the formation of lateral levees (Sparks et al., 1976), which delimit an inner channel (Hulme, 1974). The lava channel is characterized by a central portion, or plug, with equal velocity (Figure 4). The outer portion connecting the plug with the levees has a velocity gradient that increases outward until the velocity at the levees becomes zero (Rittmann, 1977; Borgia et al., 1983; Merle, 1998). This velocity distribution is the result of the Bingham rheology of the lava (Johnson, 1970; Kauahikaua et al., 1998), and of the shear stress increase from the middle of the channel to the boundary (Figure 4) and marks the plug region where no deformation occurs (Borgia et al., 1983). The accretion of lateral levees is the initial stage of lava tube buildup.

The formation of a stable crust requires the existence of a rheological boundary, which can be expressed as the ability to develop a shear strain between a hot plastic interior and an outer rigid shell or crust of a lava flow. If the lava is too viscous, such as rhyolitic (Gottsmann and Dingwell, 2001; Bullock et al., 2018) or andesitic flows (Borgia et al., 1983), then the lateral shear regions may not develop, or they may concentrate at the margins of the lava flow to define the channel zone (Borgia et al., 1983). However, an additional requirement necessary to form a lava tube cavity is the drainage of the flow interior. This can be promoted by the pressure of the feeding lava from behind or by the ground slope at the flow front. If this drainage happens after a stable outer carapace has formed, a cavity might then develop within the lava. Otherwise, the upper crust collapses onto the floor, and the collapse propagates to the flow front, closing the cavity and stopping the flow’s advance (Borgia et al., 1983).

A long pause during the eruption, with cutoff of supply, or a sustained decline in effusion rate allows the tube to cool off, contract and collapse, forming obstructions that make it difficult to reoccupy when the eruption resumes (Lipman et al., 1987; Mattox et al., 1993). Conversely, a significant increase in the effusion rate would cause large breakouts from the tube margins that could permanently divert the lava from it (Keszthelyi and Pieri, 1993; Koeppen et al., 2013). Large breakouts causing a sudden drainage may result in the collapse of the upper crust and in the obstruction of the newly formed cavity. Alternatively, the collapse can be partial and only reduce the tube section (Calvari and Pinkerton, 1998).

The drainage of a flow that forms a lava tube cave can be triggered by slight changes in the ground slope. It has been observed that the same lava issued from a tumulus (namely, an inflated lava surface forming a small hill, broken in some parts to feed secondary flows) to form several radial flows can produce either filled (non-drained) lava tubes, partially drained tubes, or even fully drained tubes (Figure 5), as seen during the 1983 lava flows on Etna. The example of Figure 5 shows that even a small difference in the topographic gradients of the ground surface caused the four flows and associated different lava tubes to develop from the same tumulus and during the same eruption, at just a few meters from each other. Thus, in addition to lava flow composition and rheology, topography is a paramount requirement for forming lava tube cavities.

Lava tubes develop within lava flows displaying both ‘a‘ā and pāhoehoe surface morphologies, but apparently not within blocky flows (Wood, 1977). The main requirement is not just a low discharge rate (Gregg, 2017), but rather a steady discharge rate for a sufficient duration to allow the formation of a stable, rigid, and cool outer crust (Greeley, 1987; Calvari and Pinkerton, 1998; 1999). However, since ‘a‘ā and pāhoehoe flows are emplaced under different discharge rate conditions, the size of the initial ‘a‘ā and pāhoehoe flows, which is a function of the initial discharge rate, affects the size and shape of the master lava tubes, or major tube distributaries or main lava pathways (Wadge, 1981). Lava tubes apparently form preferably in lava flows filling valleys and at the margins of lava flow fields. Following the path of rivers, they may bifurcate to form distinct networks (Greeley and Hyde, 1972).

Observations of active tubes in Hawaii led Greeley (1971a), 1987, Greeley and Hyde (1972), Peterson and Swanson (1974) and Peterson et al. (1994) to introduce one of the most accredited hypotheses, according to which tubes form by the roofing over of a lava channel (see Figure 6). Peterson and Swanson (1974) and Peterson et al. (1994) distinguished three main ways for lava tube formation (Figure 6): (1) the growth of a rooted crust (Figure 6a), (2) the aggregation of crustal rafts transported on the flow surface (Figure 6b), and (3) the growth of levees towards the inner portion of the lava channel by overflows to form an arch (Figure 6c).

Other processes independent of the existence of a lava channel are (4) flow budding (Peterson et al., 1994) and (5) the progressive extension of pāhoehoe lobes by inflation (Figure 6d; Hon et al., 1994). Both have also been observed in ‘a‘ā lava flows (Calvari and Pinkerton, 1998). When the discharge rate within a tube shows slight fluctuations, a short-lived increase may cause inflation of the roof, the formation of a tumulus (Figure 6e), and the opening of localized ephemeral vents. These act as pressure-release valves and give rise to small, short-lived secondary flows, mostly displaying pāhoehoe morphology. An additional, but less common, process of lava tube formation is through landslides (Figure 6f) that cover the active lava flow field with a breccia or volcanoclastic material, favoring upper crust formation. This happened at Stromboli during the 2002-03 eruption, when landslides from the walls of the Sciara del Fuoco scarp locally covered the proximal portion of flows that were spreading there and that were not yet crusted over (Calvari et al., 2005). The same process was also observed at the Tajogaite volcano (La Palma, Canary Islands) during the 2021 eruption, when the mobilization of part of the cinder cone improved lava tube formation in the proximal part of the lava flow field, close to the fissure vents (Calvari et al., 2026a).

5.1 Roofing over of a lava channel

The growth of rooted crust (Figure 6a) occurs when the crust attached to the lateral banks grows slowly across the stream and ultimately meets the crust growing from the opposite side, much like a closing zipper (Peterson and Swanson, 1974; Peterson et al., 1994). Such crust gradually extends down slope along the channel, sealing it from the vent to the flow front (Figure 6a). In other cases, the growing crust can be disrupted by the lava flow moving along the channel and transported along the flow, where it can jam at channel constrictions (Figure 6b). These broken crust pieces form aggregated crustal rafts, which are sealed by the fluid lava acting as a cement (Peterson and Swanson, 1974; Calvari and Pinkerton, 1998). Channelized flows characterized by a more constant surface level can generate a roof through the accumulation and welding of solidified lava shreds along the banks. Similarly, at Etna, in areas distal from the vent where the flow is marked by a more viscous and colder ‘a‘ā lava that flows on flat slopes, the channel is covered and insulated by the welding of lava clinkers on its free surface (Figure 6b; Calvari and Pinkerton, 1998). However, unlike the case of floating crustal rafts, which are typical of pāhoehoe lavas and develop from the center to the flow margins (Peterson et al., 1994), the presence of lava clinkers, which are distinctive of ‘a‘ā lavas, and their aggregation is more evenly distributed over the entire surface of the flow.

Another important process is the growth of levees by repeated overflows (Sparks et al., 1976), which increases their thickness and width towards the inner portion of the channel. This first forms an arch and then a stable crust that expands both upslope and downslope to give rise to a lava tube (Figure 6c). Many such roofs might be destroyed by surges in the lava current that break through the crust (Bailey et al., 2006), or by lava withdrawal when the crust is not yet fully solidified, which causes the roof to collapse. The last process for lava tube formation is lava flow inflation (Figure 6d; Peterson and Swanson, 1974; Hon et al., 1994). Initially described for pāhoehoe flows, the inflation process was also recognized within ‘a‘ā lava flows (Guest et al., 1984; Calvari and Pinkerton, 1998), along with the other three processes mentioned previously. A progression of the three processes–e.g. inward levees growth, crusting over of a wide channel, and flow front inflation (Figure 6)–from the source vent to the flow front was found to be characteristic of a decreasing ground slope and increasing channel width (Peterson and Swanson, 1974; Calvari and Pinkerton, 1998). After the pioneering work of Hon et al. (1994), the discovery of inflated features within thick lava sequences significantly changed the understanding of the emplacement processes responsible for flood basalts and caused a re-evaluation of their emplacement time (Keszthelyi and Self, 1998; Thordarson and Self, 1998).

Lava tubes usually develop on average slopes of less than a few tens of degrees, up to maximum values of almost 60°. Lava tubes on Mt. St. Helens (Washington, USA) are defined by a slope range of roughly 1.0°–4.5° (Greeley and Hyde, 1972). Values less than 1.0° characterize the Undara lava tubes in North Queensland, Australia (Atkinson et al., 1975) and of 0.4°–0.7° on average in Jeju Island, South Korea (Ahn and Hwang, 2008). Excavated debris levees have been observed to form during effusive eruptions at Stromboli volcano (Calvari et al., 2005), where recent lava flows spread along the Sciara del Fuoco’s steep slope of ∼30°. In this zone, the emplacement of debris flows during lava flow expansion also contributed to the formation of an upper solid crust and, subsequently, to lava tube formation (Figure 6f). In Hawaii, tubes even formed across the Hilina fault system, where the cliff’s slope was locally as much as 60° (Peterson and Swanson, 1974). Lava tubes that developed within pāhoehoe lava flow fields, such as those emplaced on Kilauea between 1986 and 1997, and within ‘a‘ā lava flow field on Etna in 1991–93, were typically formed by the roofing over of lava channels on steep slopes (4°–30°) and by inflated flow fronts on slopes of less than 1° (Calvari and Pinkerton, 1998; Kauahikaua et al., 1998).

5.2 Flow budding

Wentworth and Macdonald (1953) proposed that a pāhoehoe flow advances by budding lobes, with the size of the lobes apparently controlled by the stiffness of the bubble-laden lava. This implies that degassed lava forms centimeter-sized toes, while gas-rich lava forms larger lobes (Self et al., 1998). During growth of pāhoehoe flow lobes by budding, a small lobe develops a skin and is inflated by the lava pressure growing below the skin. The pressure builds up below the skin until it breaks in one or more places, and new lobes emerge from the breakouts (Grimes, 2002). Flow budding applies equally to inflated sheet lobes that can range from 1 to 100 m and from less than 100 m to more than 10 km in lateral dimensions (Self et al., 1998). The flow advances as a complex, anastomosing system of flows and tubes. Eventually, the pathways of molten lava concentrate in a few distinct tubes, while others stagnate (Kauahikaua et al., 1998). Pāhoehoe lobes can coalesce laterally (Figure 7) during inflation (Hon et al., 1994), with inflation gradually causing the digestion of the inter-lobe septa within the molten core (Kauahikaua et al., 1998; Self et al., 1998). Sometimes the inter-lobe septa might be digested only partially, leaving a cusp in the tube’s ceiling that reveals both the pāhoehoe nature of the cave and the lateral coalescence process. However, tube development within pāhoehoe lobes or toes does not necessarily require an anastomosing system, and pāhoehoe flows can spread as one continuous sheet extending almost entirely across the width of the flow (Kauahikaua et al., 1998).

5.3 Flow inflation

Studies of sheet flows from Kilauea volcano revealed their vertical growth without the need to develop lava channels. Sheet flow inflation was observed on sub-horizontal slopes of no more than 2°, and generally below 1°. These slopes inhibit channelization and promote the coalescence of liquid lava pathways, forming a distributed molten core and a flat, smooth upper surface (Hon et al., 1994). Flow margins are almost vertical scarps or steeply tilted slabs of crust, and sheet flow growth by inflation can lead to individual flow units several hundred meters wide and several kilometers long (Hon et al., 1994). Measurements of inflation rates show advancing pāhoehoe flows that rapidly become several times thicker than the original lobes. Electrical conductivity data confirms that the conductive lava within the advancing flow is a single continuous sheet, initially extending almost entirely across the width of the flow (Kauahikaua et al., 1998).

The process of flow inflation was originally observed within pāhoehoe flow units, forming features called pressure ridges or lava rises (Wentworth and Macdonald, 1953; Walker, 1991). Later, flow inflation was also recognized within ‘a‘ā flow fronts, causing a significant and often undetected thickness increase (Calvari and Pinkerton, 1998; Calvari et al., 2018). Inflation was identified as a ubiquitous primary process leading to long lava flows (Kauahikaua et al., 1998; Keszthelyi and Self, 1998; Stephenson et al., 1998) and extensive flood basalt on Earth, the Moon, Io, and Mars (Keszthelyi et al., 2006; Garry et al., 2012), even in submarine environments (Gregg and Chadwick, 1996).

5.4 Submarine lava tubes

Lava tubes have also been observed within pillow lavas, which are interconnected, elongated flows formed in the submarine environment. More specifically, pillow lavas can be considered analogous to sub-aerial tube-fed pāhoehoe, following the sheet flow phase and emplaced by more sustained, slower but steadier eruptive phases occurring after an internal plumbing system has been well established (Ballard et al., 1979). Submarine lava tubes are considered extremely important for distributing lavas on the ocean floor (Fornari, 1986). Cross-sections of tubes in pillow lavas resemble pillows with a convex upper and flat or V-shaped lower surface, modelled around the pre-existing topography (Staudigel and Koppers, 2015). Submarine inflated lava flows develop features like their sub-aerial counterparts, with inflation promoting the rise of the upper crust, tumuli formation, and causing the rigid surface crust to crack along the long axis of the lobe and rotate into an outward-dipping configuration (Fornari et al., 1978; Fornari, 1986; Appelgate and Embley, 1992). Submarine lava tubes often have broken and collapsed roofs, and the inner part contains accumulations of broken black glassy clasts derived from the cooling-contraction fragmentation of the pillows’ crust (Fornari et al., 1978). Submarine lava tubes, channels, and lava lakes seem to be found in greater abundance in Mid-Ocean Ridge environments where the spreading rate is medium to fast (Fornari, 1986).

5.5 Cavities within eruptive fissures

Although these volcanic cavities cannot be considered as lava “tubes” because they extend vertically rather than horizontally, they attract the interest of scientists because they are listed among possible candidates for human shelters on the Moon, Mars, and even Venus (Cushing et al., 2015). On Earth, they offer unique insights into the eruptive processes, revealing the movement of magma within eruptive fissures, its lateral drainage to form master tubes, and its final drain-back within the eruptive conduit. The features inside the cavities, which develop within eruptive fissures or feeder dikes, reveal that lava levels oscillated several times and moved in different directions (Grimes, 2002). The magma drain-back movement within an eruptive fissure is sometimes so rapid that it detaches the newly formed and still-plastic inner lining from the side walls (Figure 8). The detached lining can roll over to join in the middle of the cavity, causing its obstruction (Bonaccorso and Maugeri, 1999; Grimes, 2002; Hudak et al., 2022). Leotta and Liuzzo (1998) surveyed several cavities formed within eruptive fissures at Etna volcano and described these cavities as comprising several stacked hollows, separated at depth by “oven mouths”. This local name is used by speleologists for the cavities left by the drainage of ephemeral vents. These ephemeral vents opened at the lower end of the eruptive fissure at increasing depth as the level of magma within the fissure lowered. This movement is testified by both horizontal and vertical striations, as well as lateral benches, scratched and left by the magma on the sidewalls (Leotta and Liuzzo, 1998).

Extending to depths of hundreds of meters with vertical walls, caves within eruptive fissures are often difficult for non-expert speleologists to explore and are thus normally outside tourist paths. In a very few cases, they offer easy entrances and access, such as the Serracozzo cave on Etna (Figure 8b), which formed partially within the eruptive fissure and partially within the master tube that fed the final phase of the 1971 flank eruption (Bonaccorso and Maugeri, 1999). Another example is the Tre Livelli lava tube, which can be accessed from hornitos (small, steep conical hills formed along the eruptive fissure by mild explosive activity that accumulates lava spatter around a few vents) emplaced at Etna during the 1792-93 eruption (Calvari et al., 2024). These two eruptive fissure caves have in common their continuation with a master tube and are both characterized by an uppermost horizontal portion, some 20–30 m long, showing magma movement both upslope for a few meters and mainly downslope, where they join the master tube (Bonaccorso and Maugeri, 1999; Calvari et al., 2024). Eruptive fissures can be accessed through hornitos, skylights, or cinder and spatter cones formed above the fissure (Figure 8a), or when erosion has removed the outer portion of the cavity, allowing a view inside (Wadsworth et al., 2015; Hudak et al., 2022).

6 Lava tube expansion

Once formed, lava tube networks may remain active for months or even years if the eruption continues, supplying lava to the far distant end of the lava flow field (Peterson and Swanson, 1974; Calvari and Pinkerton, 1998). The level of lava within the tubes can fluctuate in response to changes in the flow rate (Lautze et al., 2004; Bailey et al., 2006). This can either cause blockages within the tube system from roof or wall collapse when the lava level suddenly drops, causing lateral instability, or lava spilling from skylights to feed secondary flows when there is a lava surge (Peterson and Swanson, 1974; Guest et al., 1980). Lava was reported to flow almost noiselessly within tubes in Hawaii (Peterson and Swanson, 1974). In contrast, while walking on the roof of master tubes at Etna during the initial phase of the 1991-93 eruption, it was possible (SC) to feel the vigorous flow of the lava, accompanied by a dull noise probably caused by lava balloons bouncing against the tube’s walls and shaking the ground above.

As soon as a lava channel becomes completely roofed, the flowing lava is hidden from view, although holes resulting from local collapses of the roof or incomplete sealing may form skylights (Rittmann, 1977). These are exceptional points from which to measure the lava flow speed and rate. However, while active, they must be approached with caution due to the high-temperature gases being released. The minimum time required to form a portion of a lava tube is just a few days, and the complete sealing of a master tube can occur in just 1 week (Peterson and Swanson, 1974; Calvari and Pinkerton, 1998; Calvari et al., 2026b). During long-lasting eruptions, the roofing over of lava channels may form complex tube distributaries, with lava channels branching and merging, giving rise to braided lava tubes (Peterson and Swanson, 1974), overlapped tube networks and tube capture (Wood and Mills, 1977), which can occur in both horizontal and/or vertical directions (Calvari and Pinkerton, 1998; 1999). When complex and overlapped tube networks form within a long-lasting lava flow field, the flow surface structures–such as channels, tumuli, or ephemeral vents–may not relate to the underground features of the lava tubes (Calvari et al., 2024). This is especially true in zones close to the eruptive fissure or effusive vents, where several stacked and piled-up lava flows may occur at different times and in various directions (Guest et al., 1980). In this context, the flow surface morphology will show the emplacement of the final phase of the lava flow field and of the shallower tubes emplaced there. This may also happen in extraterrestrial environment, and the Marius Hills region on the Moon, where mass deficit features are on the order of several kilometers in width, could represent a potential to host braided lava tube systems (Zhu et al., 2024).

Ponte (1953) observed the 1950-1951 eruption at Mt. Etna and described the roofing over of an ‘a‘ā lava channel, with a decreasing lava tube section with increasing distance from the main vent, as well as a gradual drop in lava effusion due to the obstruction of the lava channel. Even the large 1614-1624 Etna lava flow field, which mainly displays a pāhoehoe flow surface (Guest et al., 1984), comprises several large and stacked tubes emplaced during the earlier phases of the eruption. Indeed, the main outline of the lava flow field was determined by the emplacement of four arterial ‘a‘ā lava flows, along which the master lava tubes formed. These deep tubes were later reactivated by the process of vertical tube capture (Wood and Mills, 1977), when the upper-level tubes developed primarily within shallow pāhoehoe inflated flow lobes broke through the roof of the lower tubes, thus expanding at a deeper level within the lava flow field.

The largest and more complex lava tube systems developed on Etna within ‘a‘ā lava flows were probably those formed within the 1991–93 lava flow field. This long-lasting eruption was characterized by two phases of growth (Calvari et al., 1994; Calvari and Pinkerton, 1998): one before and one after an artificial diversion that was undertaken at the base of the proximal fissure system (Barberi et al., 1993). During its natural downslope expansion, the lava flow field was conditioned by an artificial earth barrier built at the exit of the valley for containment purposes (Barberi and Villari, 1994). This obstacle managed to halt the downslope expansion of the lava flow field for several months but increased the thickness of the lava behind the barrier, thus promoting the superposition of lava flows and the growth of a stacked tube system (Calvari and Pinkerton, 1998). The growth of the lava tube systems was eventually the trigger of the unexpected lava flows that overran the barrier and spread toward the nearby village of Zafferana Etnea, before being deactivated by the artificial diversion operated upslope along the fissure system (Barberi and Villari, 1994). Two stacked tube systems formed within the lava flow field during the first and second phases of the eruption. The first was also the most extensive because it formed within ‘a‘ā lava flows emplaced during the earlier phases of the eruption (Calvari and Pinkerton, 1998), when the effusion rate was greater (Wadge, 1981; Calvari et al., 1994) and master channels were forming within arterial ‘a‘ā lava flows (Kilburn and Guest, 1993).

Extensive lava tubes developed at Kilauea (Hawaii) during the 1990 eruption that emplaced the Kalapana flow field. During this eruption, it was observed that lava tubes established within primary arterial pāhoehoe inflated flows were longer-lasting than those formed within secondary breakouts fed by the primary inflated flows (Mattox et al., 1993). Tubes within primary flows were often reoccupied by lava after a brief eruptive pause, whereas tubes formed within secondary breakouts were never reoccupied once the supply was cut off (Mattox et al., 1993). Both primary flows and breakouts advanced erratically, with alternating periods of rapid advance of the flow front and periods when the flow front stagnated, and the flow inflated (Mattox et al., 1993).

It is important to consider that lava tube networks emplaced during long-lasting eruptions are the result of variable emplacement conditions that may change considerably from the start to the end of an eruption. These different conditions include the supply rate from the vent, temperature, viscosity, crystallinity, and gas content, but also topography. Thus, the 3D shape of a tube network can result from the overlap and possible merging of earlier, large, and straight lava tubes within ‘a‘ā lava flows emplaced at high flow rate, and late-stage, smaller, and more dendritic tube networks formed during the late stage of the eruption within pāhoehoe flows emplaced at low discharge rate. However, it is interesting to note that lava tubes developed either within ‘a‘ā or pāhoehoe flows normally display a different geometry in both plan view and longitudinal and transversal sections.

6.1 Plan view of lava tubes (‘A‘ā vs. pāhoehoe)

The planimetric shape of lava tubes mimics that of the host lava flow, with straight and simple shapes being typical of ‘a‘ā lava flows emplaced on steep (>4°) ground slope, and dendritic to complex shapes featuring pāhoehoe flows expanded on flat ground (<4°). The planimetric shape of lava flows and tubes is strongly influenced by topography, especially by the ground slope and by the existence of valleys or ravines that may confine their emplacement. An example of the diversity of tubes emplaced in similar topographic conditions and displaying a simple and branched planimetric extension as a function of their morphology, either ‘a‘ā or pāhoehoe, is provided by the Arcimis and Acqua Vitale lava tube caves emplaced during the 1792-93 eruption at Etna volcano described by Calvari et al. (2024) and reported in Figure 9. Here they are compared with a section of the pāhoehoe Kazumura lava tube (Hawaii) formed on a low gradient terrain (1.9°; Allred and Allred, 1997) and with the ‘a‘ā Corona lava tube in Lanzarote emplaced on a steep slope (28°–30°) close to the vent, but as low as 2° in some parts of the lava tube (Tomasi et al., 2022).

6.2 Longitudinal and transversal sections of lava tubes

The shape of longitudinal and transversal lava tube sections can tell us how the tube formed, under which topographic conditions, and the processes that occurred during its growth (Figure 10). The shape and size of lava tubes’ transversal sections follow the shape and size of their host lava flows, with the aspect ratio (h/w, where h is height and w is width; Walker, 1973) determined primarily by the lava composition, and then by ground slope (Kauahikaua et al., 1998). Keeping the lava composition constant, a high ground slope (>4°) would favor a tube section with an aspect ratio greater than 1 (h > w), resulting in tubes greater in height than width, whereas a low ground slope (<4°) would imply an aspect ratio lower than 1 (h < w) and tubes greater in width than height.

Longitudinal and transversal sections of lava tubes may reveal important processes that occurred during their life, such as, for example, lava flow front inflation and opening of ephemeral vents from the front of previous flows (Figure 10a), or vertical tube capture (Figure 10b). Vertical tube capture is revealed, for example, by the apophyses observed along the longitudinal section in Figure 10b (Calvari et al., 2024), and by key-hole transversal sections (Figures 10e,f 2, 3) (Calvari and Pinkerton, 1999). Gothic arch sections (Figure 10f 1) reveal that the tube originated as an open channel that roofed over (Guest et al., 1980). Chambers greater in width than height, displaying a transversal section like a horizontal figure-eight or an infinite symbol with a cusp in the middle, are caused by horizontal capture (Calvari and Pinkerton, 1999). This shape is typical of pāhoehoe flow emplacement, where several flanked lava lobes inflate to merge laterally by thermal erosion of the septum between them (Figure 10a).

Asymmetric lateral features, such as lateral benches and grooves (Figures 10b,c) inside lava tubes are common where the tube bends, causing erosion in the concave side and deposition on the opposite, a process very similar to that occurring along rivers (Wood, 1981), with the difference that here deposition is by solid lava. Wood (1981) proposed that, like a water river, a lava river may possess the capacity to ‘adjust’ its channel, through erosion and deposition, in a manner that minimizes thermal and mechanical energy losses. This process has been simulated in the laboratory by Cashman et al. (2006) who have examined the effects of channel irregularities (expansions, contractions, sinuosity, and bottom roughness) on crust formation and disruption. They found that such changes affect the local flow behavior and can thus change channel size. For example, gradual widening of a channel results in a decrease in flow velocity and may allow a down-flow transition from the mobile crust to the tube regime. In contrast, narrowing of the channel causes an increase in flow velocity, thus inhibiting tube formation (Cashman et al., 2006).

7 Lava flow field features revealing the pathway of lava tubes

Compound lava flow fields (Walker, 1971) present both positive constructional features and negative collapse forms. Positive features, such as tumuli, pressure ridges and inflation clefts, reveal endogenous growth and sometimes minor exogenous expansion when small lava flows emerge from cracks in tumuli and pressure ridges. Negative features, like skylights, rings and pit craters, reveal the drainage of voids and subsequent collapse along the lava tube pathway. Figure 11 displays some examples of these lava flow field features.

7.1 Skylights

Skylights are openings on the roof of lava tubes (Figure 11b), normally displaying overhanging rims and having a much smaller width than the tube below (Sauro et al., 2020). Many skylights form during an eruption, allowing gases from the fluid flow to escape. As such, they are often surrounded by precipitates of sulphur, chlorine, carbonates, etc. During an eruption, they allow one to view and measure the lava flowing inside the tube. However, great caution must be taken when approaching skylights because of the high temperature of the escaping gases and sudden changes in wind direction. The floor below skylights is normally flat and displays ‘a‘ā or pāhoehoe surfaces, formed by the last lava travelling within. It does not present accumulated debris since any debris would have been transported away by the flowing lava. Skylights often offer good opportunities for lava tube entry. They tend to form over lava falls or at changes in the tube’s direction, and not over straight sections of the tube (Kauahikaua et al., 1998). Sauro et al. (2020) introduces the difference between sinkholes and skylights, with sinkholes produced by collapse after the tube has drained and cooled, and thus the tube below might be obstructed by collapse breccia. By contrast, skylights form during active lava flow, and the breccia is either digested within the lava flow or carried away along the tube. Yang et al. (2026) have recently found a link between the shape of skylights and the concealed tube geometry, with a distinct “wide-upstream, narrow-downstream” diameter distribution coupled with an “upstream-ascending, downstream-descending” bottom elevation profile. Specifically, Yang et al. (2026) show that the tube segment immediately upstream of the skylight is characterized by an inflation of the host lava flow, whereas the downstream segment transitions into a tightening zone. This observation has important implications in the exploration of extraterrestrial tubes, because the tube upstream of skylights being wider is also more prone to be used as human habitation than the narrower downstream portion.

7.2 Tumuli

Tumuli are positive topographic features found on lava flows, especially in places of low ground slope (Figure 11a). They are formed by the tilting up of crustal slabs without any crustal shortening and have elliptical to circular shapes with axial cracks (Walker, 1991). They are distinguished from pressure ridges, which are deformed by laterally directed pressure (Walker, 1991). Tumuli generally form at breaks in the ground’s slope (Calvari and Pinkerton, 1998) due to the local hydrostatic pressure of the fluid lava beneath the chilled crust (Greeley and Hyde, 1972; Walker, 1991; Duncan et al., 2004). A few tumuli are hollow, forming dome-like cavities (Kempe, 2012), or they may be filled with viscous lava, sometimes forming squeeze-ups and toothpaste flows emerging from the cracks (Walker, 1991). They have been identified as indicators of lava tube location (e.g., Greeley and Hyde, 1972; Guest et al., 1984; Calvari and Pinkerton, 1998; Duraiswami et al., 2001; Duncan et al., 2004; Khalaf and Hammed, 2016) and are considered the normal evolution of inflated pāhoehoe sheet flows (Kauahikaua et al., 1998).

7.3 Shatter rings

The term “shatter rings” was first used by Kauahikaua et al. (2003) to indicate wide (20–50 m across), rounded depressions above a lava flow field. They are typical of both pāhoehoe (Kauahikaua et al., 1998; 2003; Orr, 2011) and ‘a‘ā (Calvari and Pinkerton, 1998; Calvari et al., 2024) lava flow fields. They form on top of lava tubes (Wood et al., 2002) in places where the tube covers a step or a sudden change in the ground slope. This may cause a local accumulation and overpressure of lava inside the tube, which eventually breaks the tube’s roof and raises it like a cork or a plug (Kauahikaua et al., 1998). Once the overpressure is released–for example, through a lava overflow from the breached roof or through ephemeral vents on the sides of the channel–the roof collapses within the tube, partially obstructing it. Shatter rings, sometimes also called raised-rim craters (Greeley and Hyde, 1972), often form from a sequence of fluctuating lava discharge within the feeder tube, with surges followed by declining phases that cause alternating phases of inflation and deflation of the tube roof (Orr, 2011). Shatter rings are normally surrounded by a rim of debris from the friction that causes breakage of the rock, and they often present debris at the bottom as well. The fractures inside the shatter rings can be intruded by late-stage squeeze-outs and toothpaste lobes and might display scratches and striations indicating the direction of displacement (Orr, 2011). Being essentially collapse features, they very rarely offer entrances to the lava tube system below. However, they are useful lava flow field features that reveal the presence and pathway of underlying lava tubes. In addition, as wide features, they are normally located above master tubes.

7.4 Ephemeral vents or breakouts

During the 1990 eruption at Kilauea (Hawaii) and the 2014-15 eruption of Fogo (Cape Verde), maps of the source points of breakouts (Figure 11a) provided an accurate trace of the lava pathway, thus revealing the position and extent of the active lava tubes (Mattox et al., 1993; Calvari et al., 2018). Calvari and Pinkerton (1998), (1999) distinguished three types of ephemeral vents or breakouts. First-order ephemeral vents are those that open at the front of cooling-limited arterial flows and contribute to the downslope lengthening of master lava tubes. Second-order ephemeral vents open above arterial flows and act as pressure-release valves during periods of tube blockage or temporary increases in effusion rate. The lava flows fed by these vents are normally short-lived. Sometimes they form tumuli or toothpaste lava that oozes from their fractures. Third-order ephemeral vents are located at the flow margins and last only for the time it takes to drain the flow interior when the flow is no longer fed, forming small pāhoehoe lobes and toes or toothpaste lava (Rowland and Walker, 1987).

7.5 Littoral cones

Littoral cones closely resemble the spatter and cinder cones surrounding the vents of basaltic volcanoes and form when ‘a‘ā flows and tubes enter the sea (Wentworth and Macdonald, 1953). Studies in Iceland (Morrissey and Thordarson, 1991; Thordarson and Self, 1993; Greeley and Fagents, 2001) revealed that complexes of rootless cones, which occur commonly within pāhoehoe lava, are associated with subsurface lava tube systems. They form when these tube systems encounter water ponds, triggering sustained explosive activity–a process very similar to how littoral cones form at Kilauea when lava tubes meet the ocean (Jurado-Chichay et al., 1996; Mattox and Mangan, 1997).

7.6 Hornitos

Hornitos are small mounds of welded spatter (lava clots, Figure 11c) built at rootless vents on the backs of lava flows and along eruptive fissures (Bullard, 1947; Wentworth and Macdonald, 1953). They commonly have a pipe at the center, although the pipe might be clogged or sealed by the last ejecta. They are built by the gradual accumulation of lava spatter around one or more vents. The spatter is thrown by weak explosive activity through a vent along an eruptive fissure or conduit, or through an opening in the roof of an underlying lava tube (Wentworth and Macdonald, 1953). Hornitos display a conical shape and are typically 1–10 m high and 1–10 m wide. When they are drained, hornitos offer access to the lava tube underneath.

8 Lava tube inner features

Lava tube inner features (Figure 12) can tell us of the processes occurring when the tube was still active and may allow inferring the duration that the tube was active, its rheology, and its role in feeding secondary flows (Calvari and Pinkerton, 1998; 1999). Observations of the Kazumura cave in Hawaii (Allred and Allred, 1997) showed the formation of flow levees and tubes in tubes by cooling of the surface of the lava flowing within a tube when passing below a skylight and crusting over very rapidly by heat loss (Peterson et al., 1994). Stacked tubes can also form by overlapped lava flows crusting over and merging vertically by tube capture (Calvari and Pinkerton, 1999). Vertical tube capture likely happens when the lower tube drains partially, leaving an empty space below the upper brittle crust and thus offering a lesser magmastatic support to the upper crust (Calvari et al., 2024).

8.1 Inner coatings

Ollier and Brown (1965) report tube inner linings or coatings (Figure 12a) of various thickness and sometimes displaying polygonal cracks affecting only the inner lining. Presence of a lining shows that the molten lava in contact with the tube wall was sticky enough for a layer to remain plastered onto the wall after withdrawal. When several linings are present, every accretionary layer represents a separate volcanic surge causing the tube to be full of lava (Wood, 1981), followed by a period of quiescence and drainage leading to the emptying or lowering of the lava level inside the tube. The thicknesses and number of lava coatings within a tube tell us much of its history, with several thin linings revealing a long activity of the tube by fluid lava feeding several lateral breakouts, and a few thick coatings suggesting a more viscous lava filling a few times the tube at the end of the eruption or at the margins of the lava flow field (Calvari and Pinkerton, 1999; Calvari et al., 2024). Lava tube linings range from less than 1 mm to more than 1 m thick and are generally somewhat more vesicular than the main body of the lava flow, with vesicles within the lining often increasing toward the tube, possibly representing outgassing into the tube (Greeley and Hyde, 1972). During or after drainage, parts of the lining may develop a glassy surface, possibly because of rapid cooling by air currents in the lava tube (Greeley and Hyde, 1972). If lava drainage occurs suddenly when the linings are still plastic and not well attached to the walls, they can partially or completely detach to form lava rolls on the floor (Calvari et al., 2024).

8.2 Striations

Flow marks, wrinkles and drip marks, testify to the original plastic state of the lava tube inner walls, where lava flowage left its trace especially through the scratches of the upper rigid crust on the plastic side walls (Figure 12b). Lava tube caves inner walls often show lateral horizontal striations, but sometimes these marks are very inclined following the tube slope, or even vertical, indicating the sudden collapse of the lava flow at a deeper level by vertical capture (Calvari and Pinkerton, 1999).

8.3 Lava stalactites and lava stalagmites

Lava stalactites (Figure 12c) can be found on the ceilings of lava tubes and sometimes on the sidewalls (Figure 12f). They are not usually distributed along the whole extension of a lava tube but are more common where the tube section is narrower and below lava shelves protruding from the wall (Gadanyi, 2008; Woo et al., 2019). In Iceland, longer stalactites are found in wider tube sections, and shorter stalactites in smaller tube sections (Gadanyi, 2008). When lava is dribbling along the tube wall, instead of forming a stalactite it forms lava runners (Figure 12f), which are similar to bulged human veins (Gadanyi, 2008). Jaggar (1931) described that some stalactites are products of splash phenomena and are lava drips, whereas others displaying a vermiform or rod-like stalactite and driblet spires formed by streams which crusted over and kept on flowing under a bridge of crust. Jaggar (1931) considers that gases escaping from the lava accumulate below the ceiling of the cavity, thus keeping a very high temperature (greater than 1200 °C) as within a furnace. Given that air is sucked in below as the hotter gases escape through cracks and windows of the roof, it allows melting of the tube roof to a glaze with quite a different crystallinity from the lava that forms very slowly. Partial melting up to 25% of the host rock has been observed within lava stalactites at Etna volcano (Corsaro et al., 2005). Stalactites do not trickle, but form by accretion (Jaggar, 1931). The stalagmites underneath are rare and must be created by dripping in some early and high-temperature stage of the stalactites formation above them (Jaggar, 1931; Wood, 1981; Gadanyi, 2008). However, most of the times lava stalagmites do not endure since carried out by fluid lava moving within the tube (Waters et al., 1990). Lava stalagmites are found only in very fluid lava, such as in Hawaii and Iceland (Jaggar, 1931; Wood, 1981; Gadanyi, 2008), and even at Tajogaite (La Palma, Canary Islands) but not on Etna, and most often at the sides of the tube where movement of the lava flowing into the tube was minimum or absent. Calvari and Pinkerton (1999) recognized at least four types of lava stalactites on Etna: (1) stalactites with very smooth surfaces and a conical shape, also called “shark’s teeth” (Allred, 1998), forming on ridges elongated along the flow direction, typically red in color, and believed to form by roof remelting due to gases accumulated below the roof (Jaggar, 1931; Kauahikaua et al., 1998). (2) Another type of stalactite is rough, grey in color and spiky, and is common at tube constrictions, formed when the lava was filling the tube and then drained, leaving this dribbling lava from the roof (Jaggar, 1931; Calvari and Pinkerton, 1999). The spiny nature of these stalactites is mainly due to plagioclase crystals and small amounts of interstitial glass. The third group (3) of stalactites form when the inner lining of lava tubes is partially peeling off, leaving rough “pull-apart” stalactites (Calvari and Pinkerton, 1999). The fourth (4) kind of stalactites is not very common, and has bulbous shapes, formed by several coatings of multiple layers over an initial stalactite, probably by the splashing lava within the tube (Allred, 1998; Calvari and Pinkerton, 1999). A special (5) kind of stalactites has been observed within the Kazumura lava tube of Hawaii, where their interior is commonly coated with a thin, smooth, vitreous surface usually associated with worm-like lava stalactites. These are called tubular lava stalactites or lava straws and can be straight, branching, eccentric, or even deflated (Allred and Allred, 1998). Lava straws are very fragile features extruded from the ceiling as gases escaped from the cooling lava. They usually have very thin walls and are extremely delicate (Woo et al., 2019; Brush, 2022). An unusual type of cave stalactite or lava straw has recently been found in a volcanic cave on Vesuvius, where on the lateral walls of a lava tube formed within a pāhoehoe flow, Lemaire et al. (2024) described centimeter-sized horizontal lava straws presumably formed by the detachment of the inner lining whose interior was still fluid.

8.4 Lava falls

Backward erosion of a lava flow happens when molten lava is flowing along a channel and expanding on a ground surface that has some irregularities. On Etna during the 1991-1993 eruption, it was directly observed that a lava flow confined within a channel with a gentle slope was increasingly becoming steeper. In 1 week, the flow was ∼1 m deeper below the lateral levees and the gentle slope became a vertical step with a turbulent backward movement of the lava flowing at its base within the channel. This flowage was exponentially increasing the background slope by mechanical and thermal erosion of the substratum. This process can evolve within lava tubes forming lava falls (Kempe, 2012). Swanson (1973) and Peterson and Swanson (1974) attribute the formation of lava falls to the dropping of the stream into an underlying, pre-existing lava tube, whereas Kauahikaua et al. (1998) found several lava falls along each tube on Kilauea and no evidence of vertical tube capture. Kauahikaua et al. (1998) also observed that lava falls form immediately below skylights in a time span of 2 months and remain in the same position for months without any evidence of backward erosion of the headwall (Kauahikaua et al., 2003).

8.5 Lateral benches

High-lava marks on the walls of a tube, like the high-water marks of a river in flood, record the position of lava at some former high stage in its flow (Figure 12d). If lava remains constant for a considerable time at a high level within a tube, the congealing of the lava surface inward from the walls may build a lava shelf or balcony; if ponding occurs lower on the walls, a lava bench may form (Waters et al., 1990). Sometimes lateral benches and balconies are present only on one side wall and not on the other. This happens when the tube is bending, thus lava flowage causes erosion on the concave side and deposition, with benches and balconies formation, on the convex side walls. If flowage is stationary for long enough, two lateral benches or balconies might eventually merge to form a stable roof and a tube-in-tube, in this case with the lower cavity being the more recent (Waters et al., 1990).

8.6 Gas blisters

The bubbles and blisters forming in molten lava (Figure 12e) are produced by the release of water and other gases from the molten rock. A gas blister is characterized by a small broken hole on the lava surface, which is a gas-escaped structure (Woo et al., 2019). When pressure is lowered by the rise of molten magma to the surface, or by the turbulence of tumbling over a cascade, the lava may froth just as the dissolved carbon dioxide in beer will froth and form bubbles on opening the can and tumble the beer into a glass (Waters et al., 1990). Gas blisters normally appear on the walls of tube sections (Figure 12e) close to the eruptive fissure or near the main vent (Calvari et al., 2024), where the magma erupted is still rich in gas. During its path on the surface, degassing releases most of the fluids contained within the magma, and gas blisters cannot form. Thus, the presence of gas blisters on the inner lining of a lava tube suggests its proximity with the source vent, as well as its formation during an early stage of lava emplacement (Cashman et al., 1994; Calvari et al., 2024; Lemaire et al., 2024).

9 Lava tube thermal and mechanical erosion

Early theoretical studies of thermal erosion by flowing lava showed that erosion was significant only in cases of turbulent flow (Hulme, 1973; Huppert et al., 1984; Jarvis, 1995). Carr (1974), using field observations by Swanson (1973) and Greeley and Hyde (1972), concluded that erosion is possible for laminar flows of Kilauea lavas, but is enhanced by turbulent flow. Pinkerton et al. (1990) observed that low temperature carbonatitic lavas from Oldoinyo Lengai (Tanzania) spreading in a laminar flow condition were eroding their substrate at a rate of 2 mm/min close to the main vent. Erosion was also observed where the flow surface velocity increased by a change in slope of the ground, and here the lava was also undercutting the channel walls (Pinkerton et al., 1990). Pinkerton et al. (1990) attribute this process to thermal erosion, as displayed by erosion rates systematically decreasing down slope with lava cooling. The persistence of lava tubes along the same path appears to be enhanced by the channel bottom erosion operated by flowing lava (Peterson and Swanson, 1974; Kauahikaua et al., 1998), but this process in other volcanoes appears much more limited in space, and possibly only concentrated at breaks in slope (Calvari and Pinkerton, 1998). When tubes show downcutting, they can result in asymmetric tube sections, and country rock digestion may result in inclusion of large blocks of country rock into the main flow (Greeley and Hyde, 1972). The thermal and mechanical erosion of the channel floor can be very fast - indeed, a rate of 10 cm/day was measured at one skylight for nearly 2 months on Mauna Loa (Kauahikaua et al., 1998). Erosion of the lava flow bottom is common and might lead to meanders with significant lateral migration for long-lasting flows (Swanson, 1973; Allred and Allred, 1997). Thermal erosion by a pāhoehoe flow showing downcutting results in a key-hole cross section of the lava tube (Kauahikaua et al., 2003; Figure 13), whereas at Etna this cross-section shape indicates vertical coalescence of staked ‘a‘ā lava tubes (Calvari and Pinkerton, 1999; Calvari et al., 2024).

Following bottom erosion from flowing lava within tubes, when tube capture occurs and the lava from the upper tube flows within the lower, the lower tube is also the last in being active (Allred and Allred, 1997; Calvari et al., 2024).

10 Lava tube composition

The main requirement for having a lava tube within a lava flow is the development of a shear surface separating the outer crust (rigid) from the molten core (plastic). It is conceivable that such a process occurs easily in basaltic (fluid) lavas where the development of an outer solid crust is favored by the great difference in temperature between the flow and the atmosphere and may be limited or even precluded in more viscous flows, such as the rhyolites. However, descriptions of the observed emplacement of a rhyolitic lava flow at Courdon Caulle volcano in Chile (Tuffen et al., 2013) report the development of a channel with marginal shear surfaces along which several ephemeral vents or breakouts opened, thereby draining the degassed flow interior. It is worth noting that the authors did not observe the presence of lava tubes within this lava flow, but it is remarkable that the flow continued to move down slope even when the supply at the main vent was cut off. Tuffen et al. (2013) described several breakouts opened at the lower margins of the lava flow, resulting in darker and smaller, degassed secondary flows very similar to those observed around master tube-fed ‘a‘ā lava flows of Etna (Calvari, 2019). These secondary flows must have been connected with and derived from the flow interior, thus evidencing a molten portion fed by thermally preferential pathways moving within a solid carapace that could represent an embryonic lava tube, perhaps not yet drained to form a cavity. Together with this example, Magnall et al. (2019), on describing ephemeral vents observed at the margins of a dacite lava flow on Santiaguito volcano in Guatemala, note that these ephemeral vents have only been identified originating from a collapsed portion of the lava flow margins. In addition, Harris et al. (2002) states that block flows have many thermal similarities with tube-confined flows, where the thick, cool crust insulates the flow, causing low heat losses and the capability of extending many kilometers at extremely low effusion rates and velocities before cooling brings about stagnation. Building on these observations, we could argue that lava tube caves in very viscous lava do not form because the upper crust is so thick and heavy that when the fluid interior is squeezed out, it collapses thereby obstructing the lava pathway. It is possible that the time needed to form an upper crust attached to the lateral flow margins is so long for viscous lavas, that eventually the whole flow will cool down impeding the formation of a cavity.

To test if a compositional threshold exists for the formation of lava tubes, we have selected from the available literature 65 compositions of lava tubes formed in lava flows from 26 countries, whose analyses of major elements and related references are listed in the Supplementary Material (Supplementary Table SM1). The SiO₂ vs. Alkali (Na₂O+ K₂O) graph is reported in the TAS (Total Alkali-Silica) diagram of Figure 14, and shows that SiO₂ mainly spans between 40 and 63.46 wt%, whereas alkali are comprised between about 1 and 12 wt%, with the highest in silica being the trachyte lava flow from Pantelleria (Italy; Civetta et al., 1984; White et al., 2009), and the lowest 0.22 wt% (against ∼38 wt% alkali) from the carbonatitic 1993 eruption of Oldoinyo Lengai in Africa (Simonetti et al., 1997; McFarlane et al., 2004; not reported in the graph, see Supplementary Table SM1).

Lava tubes have been found in extremely diverse compositions of lava flows, from the highly fluid carbonatitic lavas with the lowest silica content (0.22%; Oldoinyo Lengai, Tanzania; Simonetti et al., 1997; McFarlane et al., 2004) to the very viscous trachyte flows of Pantelleria (Sicily Channel, Italy; 62%–64% silica, Civetta et al., 1984). Tube caves could possibly form even within rhyolitic flows (such as Cordon Caulle in Chile; Tuffen et al., 2013) should the development of shear surfaces within the flow be the primary requirement for lava tubes formation. However, another requirement is a sufficient slope to allow the movement of the flow interior and a late-stage drainage that results in the formation of a cavity. Thus, it is plausible that lava tubes do not develop within rhyolitic flows (>70% silica) if not at an embryonic stage as that described by Tuffen et al. (2013). If this is correct, it seems that there is a rheology threshold for the development of shear zones within lava, possibly also related to other parameters, such as the gas content or the crystallinity of the lava. More investigation in this direction is necessary, especially using chemical analyses from directly collected samples of lava tubes analyzed in the same laboratory, considering crystal and bubble content, and crystal and bubble size distribution.

11 Hazard from lava tubes

Lava tubes have long been recognized as volcanic structures enabling a greater length of lava flows and a greater extension of lava flow fields both on sub-aerial and submarine environments (e.g., Fornari, 1986; Thordarson and Self, 1993; Calvari and Pinkerton, 1998; Stephenson et al., 1998; Kauahikaua et al., 2002; Keszthelyi et al., 2006). In order to monitor lava tube formation during a long-lasting effusive eruption, several techniques have been applied, such as daily aerial photographs to map the active lava flow field and its breakouts, geo-electrical techniques, telemetered webcams, time-lapse cameras and satellite thermal images to map the lava tubes, as well as field surveys using Unmanned Aerial Vehicles (UAVs) and structure-from-motion (SfM) techniques to obtain Digital Elevation Models (DEMs) in a short time (e.g., Kauahikaua et al., 1990; Mattox et al., 1993; James and Robson, 2012; Cappello et al., 2016; Calvari et al., 2018; Favalli et al., 2018; De Beni et al., 2021; Orr et al., 2022).

Lava tubes established within primary or arterial flows are longer-lasting than those formed within breakouts or secondary flows and are often reoccupied by lava after a brief hiatus (2–3 days) if further supply occurs (Mattox et al., 1993). On the contrary, lava tubes developed within secondary flows or breakouts are never reoccupied after a pause (Mattox et al., 1993). The cumulative area covered by breakouts from a single lava tube can exceed the area covered by the primary master flow, although at a much slower rate (Mattox et al., 1993).

Magma supply to master lava tubes can be significantly enhanced by the process of vertical capture, allowing surface flows to expand vertically feeding deep and older lava tubes formed earlier during the eruption, and as such being bigger and longer since they are formed during the initial high discharge rate of the eruption (e.g., Wadge, 1981; Harris et al., 2011). Vertical tube capture normally occurs along the same flow path when a more recent flow overlaps a previous one and keeping the same direction (Calvari and Pinkerton, 1998). However, sometimes divergent and overlapped lava tubes, in conditions of high ground slope, might superimpose each other in a way that a deeper one captures a shallower one or drives molten lava in opposite directions, as was observed for example, in a lava tube at El Hierro (Canary Islands; Montoriol-Pous and De Mier, 1977). Should this process occur during an eruption, molten lava could be led to a deeper tube and in a different direction when compared to the last active portion of the lava flow field, extending the hazard of lava flow invasion in areas considered inactive. Analyzing the longitudinal and transversal sections of several lava tubes surveyed within the 1792–93 lava flow field formed during a 1.5-year-long eruption of Mt. Etna (Italy), Calvari et al. (2024) observed that a very common process during this eruption was the vertical capture, with several small and shallow pāhoehoe flows emplaced during the final stages of the eruption being captured by the deepest and wider tubes formed during the initial, high-discharge rate stages of the eruption, and as such being longer and wider. This process may occur with no evident trace on the flow field surface, given that small pāhoehoe lobes apparently stop their advance on the ground, while they keep feeding deeper and longer arterial lava tubes, extending the supply of lava to the distal margins of the lava flow field. In addition, we should consider that current computer and numerical models simplify lava flow dynamics but cannot yet fully simulate lava flow behavior, such as tube and channel development and transitions from ʻaʻā to pāhoehoe modes and they should be used with caution (Kauahikaua and Tilling, 2014).

To assess the supply at the hidden lava tube pathways during an effusive eruption - and hence evaluate the potential hazard - it is important to obtain thermal maps of the whole flow field as frequently as possible. From these maps, it is possible to assess the distribution in space of secondary vents or breakouts, which surround and identify the lava tube pathways (Mattox et al., 1993; Calvari and Pinkerton, 1998; Calvari et al., 2018; 2026a; Orr et al., 2022). Lava tube pathways are also shown by the alignment of tumuli, skylights and shatter rings, among the most important morphological features to consider for revealing their extension. Once lava tube paths have been recognized, it is then possible to intervene with appropriate deviation plans, if feasible (Barberi et al., 1993).

12 Lava tube ages and stability

Lava tubes are fragile underground environments that allow exploring unique volcanic structures such as eruptive fissures, conduits and the inside of drained lava flows, with their lava stalactites and stalagmites, lateral benches, gas blisters, roll-over structures, grooves and so on, but also with their particular fauna. The contraction following lava flow cooling causes fractures along the tubes inner surface, starting as soon as the lava flow stops and continuing for decades (Stevens et al., 2001). When tubes are close to the ground surface, their fractures can be colonized by roots of bushes and trees whose growth may enlarge them thus increasing the instability of the tube’s roof.

Many lava tubes contain easily decipherable records of breakdown that occurred when molten lava was flowing through a tube (Waters et al., 1990). Careful examination of the congealed surface of the last flow of lava down a tube is likely to reveal both small and large blocks of rock that tumbled from the roof of the cave and were then rafted downstream on the molten flood until it, in turn, congealed into rock (Waters et al., 1990). If a collapse from the tube roof happens while the lava is flowing within, then the molten lava may penetrate between the fallen blocks and buoy them up sufficiently so that with the additional hydraulic energy of lava ponding behind the obstruction, the flow is able to entrain and bulldoze enough of the obstruction for the lava river to restore its former gradient (Waters et al., 1990). Much of this buoyed material is deposited downstream in alcoves, where the tube widens, or on the inside of curves, where the stream velocity slackens (Waters et al., 1990). Relief of the pressure in a backed-up lava tube can also come from collapse of the floor of the filled lava tube downward into an underlying lava tube, giving rise to a vertical tube capture (Calvari and Pinkerton, 1999).

Most lava tubes maintain remarkable stability irrespective of their age, such as for example, the Cueva de Halada which is part of the Payenia Volcanic Province in Argentina, which formed within an inflated pāhoehoe flow ∼0.2 Ma (Bernardi et al., 2019), the Jingfudong lava tube in the Tengchong region (China), formed from a roofed over ‘a‘ā lava channel and with an age of 0.16–0.23 Ma (Chen et al., 2016), or again the San Antonio Mountain Cave in North America, New Mexico, aged between 3.4 and 3.9 Ma (Rogers et al., 2000). The oldest described lava tubes are probably those found within the 65–66 Ma Deccan Flood Basalts in India (Duraiswami et al., 2004; Sen et al., 2026), the Paranà Basalts in Brazil (Waichel et al., 2013), which are early Cretaceous in age (Janasi et al., 2011), and the Roza Member in the Columbia River Basalt Group in U.S.A. (15 Ma; Thordarson and Self, 1998).

The persistence of lava tubes on Earth clearly indicates their remarkable stability. Some estimation of their strength can be furnished by the diversion experiments carried out on Etna to break the roof of the just formed lava tube during the 1991-93 effusive eruption (Barberi et al., 1993). The first attempt to use a 20-kg explosive charge failed to break the roof of a lava tube ∼8 km away from the eruptive fissure, and only when 8 charges of 20 kg each were used together, did they eventually manage to breach the tube’s roof (Barberi et al., 1993). No details of the thickness of the tube were available, but we know that the roof of the tube in the proximal region of the lava flow field, 750 m away from the eruptive fissure, was estimated at over 5 m, and impossible to destroy using explosives (Barberi et al., 1993). In this zone, the section of the tube was ∼40 m², and the tube was filled by 2/3 of lava running at a speed of ∼1 m s^-1 (Barberi et al., 1993). The stability of lava tubes also depends on lava flow vesicularity, and on the capacity of lava to compact after emplacement (Beccaro et al., 2023).

Given that many lava tubes are considered geosites and extensively visited by tourists, it may be worth to find appropriate systems to evaluate their structural stability. A recent study by Barile et al. (2026), in press successfully used acoustic emission testing for assessing the stability of a lava tube formed on Vesuvius during the 1858 eruption (Lemaire et al., 2024; 2026). Theinat et al. (2020) when dealing with the stability assessment of extraterrestrial lava tubes, used numerical modelling to show that the key parameters for the stability of the tubes are the roof thicknesses and the tensile strength of the lava. Chwała et al. (2024a) performed stability analyses on lunar lava tubes using finite element limit analysis which was found to be very effective for analyzing collapse geometries to constrain the relations between the collapse size extent and width of the lava tube. Chwała et al. (2024a) pointed out that thin roofs (<10 m) are very unlikely to be stable for the tubes of a few hundred meters in width under lunar conditions, whereas Chwała et al. (2024b) took into account the layering of tube’s roof as an evidence of tube instability that might involve the inner layers and be not visible on the surface. In addition, recent stability analyses conducted for real lava tubes such as Skull Cave and Valentine Cave (Chwała et al., 2025) indicate large stability margins (despite the assumed low rock mass strength) as well as significant variability in stability due to differences in cross-sectional geometry. Based on real data, Chwała et al. (2025) demonstrate that an irregular geometry can have a substantial influence on cave stability.

13 How to study active and inactive lava tubes

Although several long-lasting lava flow fields on Earth contain lava tube networks therein, to the best of our knowledge there are no systematic surveys carried out on a lava flow field devoted to reveal the presence and extent of every one of its underground cavities. Thus, as the only possibility to explore lava tubes and understand their processes of formation, we rely on natural collapses of the roof or anthropic erosion of the sides or top. The only few examples of field surveys on a lava flow field are those carried out by Torrese et al. (2021), (2022). For this purpose, these researchers successfully used the ambient seismic noise, cross-validated with electrical data and ground truth, resulting in qualitative indicators of the presence of cavities, as well as void-rich zones possibly due to partially filled tubes or cavities of small dimension. Positive results were also obtained using a GPR (ground-penetrating radar; Esmaeili et al., 2020) and a VLF (very low frequency) electromagnetic induction technique (Jackson et al., 1987; Hon et al., 1994), infrasonic tremor wavefield (Matoza et al., 2010), and more recently multispectral imaging and Interferometry of Synthetic Aperture Radar (InSAR) from satellite validated by stereo-photogrammetry (Hrysiewicz et al., 2025), and gravimetry (Montañez Muñoz et al., 2025). The detection of magnetic anomalies is another method for investigating and identifying lava tubes and their sizes (Budetta and Del Negro, 1995), and this approach could potentially be applied to the identification of extraterrestrial lava caves (Bell et al., 2022). Electrical Resistivity Tomography (ERT) surveys provide 3D resistivity models, and a cluster analysis of resistivity data identified anomalies associated with underground voids possibly representing inaccessible lava tubes (Grassi et al., 2026).

Most of the studies on lava tube formation were carried out by direct observations and field surveys (Mattox et al., 1993; Kauahikaua et al., 1996; 1998), or by using aerial photos and satellite images to infer their path and growth (Mattox et al., 1993; Calvari et al., 1994; 2005; 2018; 2026a). Magnetic or electromagnetic profile measurements were used to detect the size and section area of the molten lava within active lava tube (Kauahikaua et al., 1990; 1996; 1998; Mattox et al., 1993; Budetta and Del Negro, 1995) to monitor the feeding rate from the source region.

14 Geo-heritage and conservation of lava tubes

Studies on volcanic geo-heritage are relatively recent, with the first papers published in 2010 (Németh et al., 2017). Volcanic geo-heritage has become one of the main aspects of geo-heritage research globally through generating inventories, developing effective strategies for valorization, volcanic diversity estimates, and conservation strategies (Ruban, 2024). The topic has proved so relevant that the International Association of Volcanology and Chemistry of the Earth’s Interior (IAVCEI) has established the Commission on Volcano Geo-heritage and Protected Volcanic Landscapes dedicated to promoting volcanic geo-heritage research and management (https://www.iavceivolcano.org/commissions-networks/). It is well known that establishing geo-heritage sites enhances the interest of the research community in particular scientific problems (Ruban, 2024). Geo-heritage resources can be exploited for the purposes of science, education and tourism, and in the special case of lava tubes this might improve volcanic hazard awareness (Alessio and De Lucia, 2017). To do so, the first step is to produce an inventory, essential for identifying which items are relevant and need conservation. The inventory should be completed by a list of features, describing the unique or rare aspects of the site, its fragile environment, with suggestions on how to limit human disturbance and a reasonable exploitation associated with the sustainability of the touristic approach to the underground world. The main forms of the anthropogenic pressure on lava tubes include road cutting, maintaining spaces for massive outdoor recreation, as well as the impact on the natural beauty via installing trails, stairs, lights and other infrastructure and tourist facilities (Ruban et al., 2022). But every person entering a lava tube would affect the natural environment simply by breathing, thus changing the oxygen content of the cave, but also shedding skin or clothing fibers, walking on the floor and thus bringing microbes and organic matter on their footwear (Brush, 2022). Even if the impact of a single visit is negligible, the cumulative effects of many visitors over a period of time can be significant (Brush, 2022). Very delicate features such as lava straws or lava stalactites are also very fragile and once broken cannot regrow.

Lava tubes are mostly located within protected park zones or UNESCO geoparks, such as the Canary Island (Dóniz-Páez et al., 2020), Australia (Brush, 2022), Korea (Woo et al., 2019) and Sicily (Calvari et al., 2024). Among the volcanic caves of geological heritage value, the Yongcheon Cave in the Jeju Island (South Korea) is probably unique, as it offers along its more than 3 km length a rare combination of volcanic features and carbonate speleothems overlapped in time due to its changing environment during the Quaternary (Woo et al., 2019).

As they are often discovered by construction work for roads or infrastructures, many volcanic cavities are also located along public passages with no control, and tourist crowding may challenge the state of these geo-sites. In some cases, they are below buildings or expanding cities (Mexico City), and this may cause serious threats to the conservation of tube cave structures but also to the stability of expanding edifices above (Nieto-Torres et al., 2022).

In addition to natural causes, the tourist frequentation within lava tubes may alter the environment by changing their temperature and humidity, disturbing the populations of bats and insects. However, the possibility of organizing tours led by well-trained volcanology guides might raise volcanic hazard awareness in the local population and improve knowledge on volcanic activity, while also enhancing recognition of the invaluable geo-heritage associated to the territory (Alessio and De Lucia, 2017). Ultimately, lava tubes represent invaluable components of volcanic geo-heritage, offering insights into past eruptions, environmental changes, and subterranean ecosystems. Their conservation is therefore not only a matter of protecting geological features, but also of safeguarding irreplaceable natural archives that contribute to our understanding of volcanic systems.

15 Extra-terrestrial perspectives

The exploration of extraterrestrial planets is often coupled with the possibility of detecting and characterizing lava tubes, their size, extension and stability (e.g., Keil et al., 2026; Chwała et al., 2025). Several depressions on the surface of the Moon, Mars and Venus have been considered as possible lava tube candidates (Greeley, 1971b; 1973; 1977; Carrer et al., 2026; Recchuiti et al., 2026), and it is particularly intriguing to evaluate the depth of possible lava tubes as well as their size and stability from the surface features (Keil et al., 2026; Chwała et al., 2024a; 2024b; 2025). Caves are unique environments where both secondary mineral precipitation and microbial growth are enhanced by stable physico-chemical conditions (Léveillé and Datta, 2010). Lava tubes preserve stratigraphic, geochemical, and possibly biological records, making them valuable targets for planetary science, astrobiology, and future robotic or human missions (Léveillé and Datta, 2010; Losarcos and Dombard, 2025). However, to detect the presence of lava tubes in a planet it is necessary to use appropriate investigation techniques. On Venus for example, characterized by a thick CO₂ atmosphere, the use of satellite optical remote sensing instruments is precluded, whereas a radar system specifically developed for Venus allowed to detect the presence of several pit chains that might be related to a lava tube (Carrer et al., 2026).

Sauro et al. (2020) offer an extensive review of shapes, locations, and possible sites that should be explored in the future on Earth, Moon and Mars. In addition to lava tubes, pit craters and pit chains have also been considered for their potential as shelters for human exploration (Cushing et al., 2015). Caves on the Moon and Mars represent a new frontier in exploration, because the subsurface of another planet has never been explored (Sauro et al., 2023). A descent into an extraterrestrial cave system provides an exciting engineering challenge compared to surface exploration, including the development of mobility for extreme terrains, innovative communication and power architectures that reach the cave interior, and strategies to overcome darkness for perception and navigation (Sauro et al., 2023).

Lava tubes on the Moon and Mars are 1–3 orders of magnitude more voluminous than on Earth, with the same processes of inflation and crusting-over active on Mars, while deep inflation and thermal entrenchment were the predominant mechanisms of emplacement on the Moon (Sauro et al., 2020). However, to transform a lava tube into a human habitat, several architectural modifications would be required (Rose and Shaw, 2024). Three key stages to lava tube habitat construction have been identified, which will require unique design solutions: (1) stabilization of the surface surrounding the skylight rim, (2) access through the vertical shaft of the skylight, and (3) interior modifications (Rose and Shaw, 2024). In addition, due to the great distance from Earth, a colony on Mars, the Moon or Venus would benefit to be based to some extent, or entirely, on local resources, a paradigm that was introduced by NASA with the aim of increasing mission safety, decreasing costs and developing a more sustainable exploration (Romio, 2022).

Since lava tubes are difficult to explore remotely, extra-terrestrial features are often analyzed using Earth analogues, with the most common selected sites being Etna (Sicily, Italy), La Palma (Canary Islands, Spain) and Kilauea (Hawaii, USA) volcanoes (D'Incecco et al., 2025).

The extraterrestrial lava flow features are also studied because of the possibility of having injected into the atmosphere large water volumes during their eruption, and because these water volumes could be stored as ice close to the eruptive vent (Hamid et al., 2025). To be able to locate water on a planet like Mars would be extremely valuable for future human exploration, given that the amount of water dissolved into its atmosphere is about 10,000 smaller than on Earth (Hamid et al., 2025).

16 Discussion and conclusive remarks

Despite the term ‘tube’, implying a cylindric form, most of the explored lava tube cave systems worldwide are not in fact long cylinders, but rather have a ‘sausage-like’ shape, with alternations of wide chambers and narrow passages (e.g., Allred and Allred, 1997; Barnes et al., 2024; Calvari et al., 2024). This shape is due to the inflation process explained above, very common worldwide, and occurring in both ‘a‘ā and pāhoehoe lava flows. On flatter areas, the flow front enlarges and thickens by the increasing pressure of lava accumulating below the crust. If the inflation lasts long enough to allow the thickening and stabilization of the crust, a wide cavity will be left after the lava has withdrawn through the opening of a vent at the lower end of the chamber. Solidification of these secondary effusive breaks generates narrow corridors, which connect the wider lava tube cave chambers. Long-lasting effusive eruptions can generate complex, stacked tube systems that may extend along the whole length of the lava flow field, and this could be the real shape of the not yet explored lava tubes existing on other planets.

A key characteristic of long-lived lava flow fields is that they often contain stacked tube networks developed at different levels within the lava flow field and at various stages of the eruption (Thordarson and Self, 1993; Calvari et al., 1994; 2024; Calvari and Pinkerton, 1998; 1999). To date, no lava flow fields on Earth have been fully surveyed to map the entire extent of their internal tube networks. Past studies (Budetta and Del Negro, 1995; Kauahikaua et al., 1996; 1998) have focused primarily on specific sections where skylights or collapse entrances reveal an underlying system, or where a tube network was already known (Torrese et al., 2022). Furthermore, geophysical prospection has only rarely been deployed to discover new tube branches (e.g., Grassi et al., 2026).

A major gap in our current understanding of lava tubes is that most mapped volcanic cavities delineate only the path and dimensions of the subterranean environment, rarely correlating these structures with the surface topography. Consequently, the depth limits for lava tube persistence and structural stability remain poorly constrained, even on Earth. Such information is highly valuable for linking lava tube stability to its age and depth within the flow field, particularly because extraterrestrial counterparts appear to be located beneath thick layers of regolith and overlapping lava flows (Hiesinger and Jaumann, 2014; Sauro et al., 2020). To better understand lava tube stability and its relationship with depth, subterranean surveys must be integrated with the topographic profile and morphology of the ground surface. This integration is particularly critical for stacked lava tubes or complex, multi-level braided systems, where tube depth cannot be easily inferred from the surface. Having these data enables the precise calculation of the overburden thickness and weight, which are essential for analyzing how structural stability changes with depth. Ultimately, improving our understanding of terrestrial lava tube systems will better prepare us for the exploration of extraterrestrial volcanic caves.

When assessing the potential for lava tubes within a lava flow field, lava composition can help determine whether such structures are likely to exist. However, a compositional threshold for lava tube formation cannot be definitively ruled out based on the compiled data in Figure 14. This is because the reported data were not directly sampled by us, nor were they analyzed within a single laboratory using identical analytical procedures. Because these data originate from several distinct sources, their intrinsic and potentially unquantified variability must be considered. Furthermore, when the chemical composition is sourced from a publication different from the one documenting the presence of the lava tube, there is a risk that the reported chemical analyses may not correspond to the specific tube-bearing unit. More rigorous and comprehensive investigations are needed to clarify the role of lava composition, ideally incorporating data on crystal and vesicle content as well as their size distributions. The analysis presented here represents an initial step toward this objective. While we have shown that tube-bearing lavas display wide compositional variability, assessing tube-free flows requires accounting for emplacement duration and pre-existing topography, factors that are often poorly constrained. Indeed, short emplacement durations may prevent tube development even in compositionally favorable lavas. Conversely, if the underlying topography is too flat, active lava pathways may fail to drain, preventing cavity formation.

To definitively identify a compositional threshold, targeted field campaigns are required to sample lava directly from within tubes linked to historical eruptions with well-documented emplacement histories, durations, and pre-eruptive topographies. Other outstanding questions concern the potential for, and extent of, substrate erosion by lava flowing within tubes (Greeley et al., 1998). While well-documented examples exist in Hawaii (Keszthelyi and Self, 1998; Kauahikaua et al., 1998), as illustrated in Figure 13, the mechanics of thermal and mechanical erosion require further investigation in diverse geological settings. Such studies should ideally target areas unaffected by tectonic or volcanic deformations, such as fissure expansion and ground fracturing from dike injections (Calvari and Pinkerton, 1999).

Another open question pertains to whether distinct morphological differences exist between lava tubes developed within ‘a‘ā versus pāhoehoe flows. Answering this question requires expanding and refining global lava tube inventories and surface morphology mapping, a task that demands close collaboration between speleologists and volcanologists. Furthermore, morphological differences between terrestrial lava tubes and their planetary counterparts remain unknown—a question that can only be resolved once the direct exploration of extraterrestrial volcanic caves begins.

Research on lava tubes has accelerated significantly, driven by their potential to serve as subterranean shelters during the exploration of rocky planets. Using terrestrial lava tubes as accessible analogs for extra-terrestrial environments has catalyzed local research, yielding key insights into their emplacement mechanisms, subterranean dimensions, hazard potential, and conservation requirements.

To date, no comprehensive studies have integrated multi-method geophysical surveys to locate and quantify the volume, geometry, and extent of all internal lava tubes across an entire flow field. Instead, existing research has primarily focused on reconstructing the local formation and growth of lava tubes during active effusive eruptions (e.g., Mattox et al., 1993; Calvari and Pinkerton, 1998; Kauahikaua et al., 1998; Pedersen et al., 2017; Calvari et al., 2018; 2026b). For instance, during the 1990 Kīlauea (Hawaii), 2014–15 Fogo (Cape Verde), and 2021 La Palma (Canary Islands) eruptions, mapping the source points of breakouts (Figure 11a) provided an accurate trace of active lava pathways, revealing the positions and extents of developing tube networks (Mattox et al., 1993; Calvari et al., 2018; 2026a). However, these studies lack a direct comparison with subterranean environments and the internal architecture of the lava pathways within the flow field. This comparison is critical because active lava tube pathways may fail to form permanent caves if the conduits do not drain, or if the overburden pressure of the overlying lava triggers the collapse of existing cavities. Although speleologists worldwide have made exceptional contributions by surveying the shallow portions of lava flow fields as soon as thermal conditions permit exploration, long-duration eruptions spanning years or decades—such as the Dammusi eruption on the northern flank of Mount Etna (1614–24; Guest et al., 2012) or the Puu O’o eruption at Kilauea (1983–2011; Orr et al., 2015)—can produce complex, stacked tube networks. Consequently, the deepest components of these systems often remain undiscovered beneath the flow field. A preliminary effort to quantify the volume ratio between a lava tube cave and its host flow was conducted by Calvari et al. (2026b). Extending such analyses to an entire lava flow field is essential, particularly for long-lived systems. Furthermore, monitoring tube evolution at regular intervals within flow fields that were actively observed during syn-eruptive emplacement is of paramount importance. This approach would elucidate the temporal evolution of lava flow field compaction (e.g., Murray and Guest, 1982; Richter et al., 2016) and its long-term impact on tube stability.

In this review, we have summarized the primary outcomes of this research, highlighting remaining gaps in knowledge and identifying critical avenues for future investigation. While lava flow emplacement has been widely modeled and simulated to assist in hazard assessment during ongoing effusive crises (e.g., Harris and Rowland, 2001; Del Negro et al., 2008; Cappello et al., 2016; Solana et al., 2017; Chevrel et al., 2018; Ganci et al., 2021), robust numerical models capable of inferring lava tube formation and development remain unavailable—excepting early, simplified models designed to simulate fundamental formation processes and thermal distributions (Dragoni et al., 1995; Sakimoto and Zuber, 1998; Valerio et al., 2008). Developing reliable models for lava tube expansion requires a significantly deeper understanding of primary genetic processes and the total extent of tube networks within a given flow field. Future research should prioritize comprehensive surveys dedicated to identifying and mapping the full dimensions of all lava tubes within historical flow fields. These efforts must detail both the shallow, surface-accessible systems and the deep-seated networks buried beneath tens of meters of lava. Comparing three-dimensional (3D) reconstructions of these tube networks with the 3D geometry of their host flows and syn-eruptive dynamics will vastly enhance our understanding of these unique volcanic features, greatly advance hazard assessment capabilities, and better inform strategies for their conservation, management, and public appreciation.

In memoriam

We began discussing this paper in collaboration with Jim (James) Kauahikaua, a good friend and a researcher expert on lava flows and lava tubes from Kilauea volcano. Jim was a former Director of the USGS Hawaiian Volcano Observatory and unexpectedly passed away on 8 October 2023. This paper is dedicated to his memory. It is also dedicated to Riccardo Pozzobon, a researcher at the University of Padova, with whom we shared the surveys of some lava tubes on Etna and Lanzarote. Riccardo disappeared suddenly while working on a glacier in Alaska on 2 September 2025.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/feart.2026.1914385/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Nick Varley, University of Colima, Mexico

Reviewed by: Gabriele Lanzafame, University of Catania, Italy

Marcin Chwała, Wrocław University of Science and Technology, Poland

Rahul Sirvi, Government Motilal Vigyan Mahavidyalaya Bhopal, India