Introduction

In lava flows and welded tuffs with well-developed columnar joints, two types of tiers are commonly observed, a colonnade and an entablature, with different column characteristics. The colonnade is composed of regular straight columns, whereas the entablature is composed of curved smaller columns. In lava flows and welded tuffs that are composed of both types of tiers, the entablature may lie on the colonnade, or the entablature may occur between lower and upper colonnades. Entablatures typically have thinner columns and a higher amount of dendritic crystals and glassy mesostasis than in the colonnade. Based on these facts, the entablature has been thought to have solidified under a greater cooling rate than the colonnade (Iddings 1909; Saemundsson 1970; Long and Wood 1986). One of the possible mechanisms proposed for the greater cooling rate is the convective cooling caused by the downward migration of external water through cracks (Saemundsson 1970; Long and Wood 1986; Forbes et al. 2014). However, some previous studies suggest that cooling from all directions in the center of a lava flow causes a complex stress distribution and the formation of irregular and curved columns (Spry 1962), as well as an increase in cooling rate compared to the margin of the flow (Grossenbacher and McDuffie 1995). In this case, the involvement of external water is not required for entablature formation and increased cooling rates in the central part of the lava flow. Hamada and Toramaru (2020) argued that when a crack is formed perpendicular to the upper surface of a lava flow, isotherms develop parallel to the crack, resulting in the formation of many radial cracks near the crack tip, as can be observed in some entablatures. Their model also does not require external water infiltration.

Thus, while previous studies have reached a consensus that the entablature forms due to a greater cooling rate, it is still debated whether water or vapor infiltration is necessary to cause the increased cooling rate. This uncertainty is at least partly because the chemical signature of water involved in cooling has rarely been found in lava flows with an entablature. In addition, as Forbes et al. (2014) pointed out, there are fewer data to help estimate the cooling rate in an entablature compared with a colonnade (e.g., column widths, chisel mark spacing, etc.). Therefore, we investigated the internal structures and microtextures of the Genbudo basaltic-andesitic lava flow (Iwate volcano, northeastern Japan), which has a well-developed entablature and where external water appears to have played a significant role in its formation.

Geological setting

The Amihari Group in the western part of Iwate Volcano, is a group of andesitic to basaltic volcanic deposits distributed over an area of about 12 km from east to west that formed between about 0.7–0.1 Ma (Itoh et al. 2017). The volcanic group can be divided into two stages: an older and younger (Nakagawa 1987). The Younger Amihari Group (ca. 0.2–0.1 Ma) consists of relatively thick lava flows, and their original topography is relatively well preserved.

The Genbudo lava flow, the target of this study, is one of the lava flows of the Younger Amihari Group, as described in Nakagawa (1987), or by Doi (2000) as within the Iwate Volcano “2nd Group”. The lava flow extends about 3 km south from the source area of the Amihari Group, and the flow front of the lava flow is exposed along the Kakkonda River (Fig. 1a). At the "Kakkonda no Ooiwaya", which is designated as a Japanese natural monument, a complete vertical cross-section of the lava flow is exposed, allowing macroscopic observation of the internal structures including the colonnade and entablature.

Fig. 1
figure 1

A Geologic map of the distal end of the Genbudo lava flow, showing outcrop locations. b Cross-sections of the Genbudo lava flow at the two outcrop locations. At Outcrop #1, the Genbudo lava flow is fully exposed from the top to bottom. At Outcrop #2, only part of the entablature is exposed, several meters above the boundary between the colonnade and the entablature

Full size image

Methodology

In this study, we mainly observed the Genbudo basalt lava flow at two outcrops. In Outcrop #1 (Kakkonda-no-Ooiwaya), where the vertical cross section of the lava flow was completely exposed, we mainly observed the macroscopic structure of the flow (Fig. 1b). However, it was difficult to access the upper part of the lava flow there due to its height of 70 m above the ground surface. Therefore, in order to observe the structure of the upper part of the lava flow in more detail, we took pictures of the outcrop with a drone (DJI SPARK) equipped with a digital single-lens reflex camera with a telephoto lens.

In Outcrop #2 (400 m northwest of Outcrop #1; Fig. 1), we investigated the three-dimensional distribution of the fractures where we could closely observe the entablature from two almost orthogonal directions. Rock samples were collected from the entablature and colonnade while paying attention to the characteristics and distribution of the observed fractures (Table 1). Whole-rock major and minor element compositions in 4 samples from the entablature and 2 samples from the colonnade collected at the Outcrop #2 were determined by X-ray fluorescence (XRF) analysis using a Rigaku ZSX Primus II (Rh X-ray tube, 50 kV, 60 mA) at the Faculty of International Resource Sciences, Akita University. Thin sections of 9 samples from the entablature and 7 samples from the colonnade were also prepared for microscopic observation using SEM–EDS.

Table 1 Petrographical imformation of Genbudo lava samples
Full size table

Results

The structure of the Genbudo lava flow

The thickness of the Genbudo lava in Outcrop #1 is about 70 m (Fig. 1b). The flow is composed of three layers from the bottom to the top: a colonnade, an entablature, and the uppermost part (Fig. 2a). The boundary between the colonnade and the entablature is sharp, but the boundary between the entablature and the uppermost part is not as distinct. The thicknesses of the colonnade and entablature are 14–17 m and 32–38 m, respectively. The uppermost part is 8–14 m thick and consists of finger-like structures extending from the main lava body with monomict breccias filling the spaces between the fingers.

Fig. 2
figure 2

Photographs of Outcrop #1 in the Genbudo lava flow. a Overview of Outcrop. b Close-up of the bottom of the lava flow. c A column of the colonnade showing rough and smooth bands with plumose structures (dashed lines) within striae on the columnar joint surface. Note increase in striae thickness from bottom to top in the photo

Full size image

The shape of the colonnade columns in horizontal section is commonly hexagonal. The diameters of colonnade columns range from about 65 cm to 90 cm. The columns are nearly straight, but slightly bent near the entablature (Fig. 2a). Striae Ryan and Sammis (1978) are observed on the sides of the colonnade columns, indicating intermittent crack extension (e.g., Goehring and Morris 2008) (Fig. 2b). A single layer of stria is typically composed of a relatively smooth lower surface and a rougher upper surface (Fig. 2c). The thickness of individual stria layers increases upward. At a height of 1.5 m from the base of the lava flow, the thickness of a single stria layer is about 20–30 cm. On the smooth surface of striae, plumose structures can often be observed that indicate the direction of crack propagation. In Outcrop #2 (total height about 5 m), only the uppermost 50 cm of the colonnade is visible above the ground surface (Fig. 1b).

The entablature columns have more irregular and complex shapes than the colonnade columns, due to the development of different types of fractures. Two types of primary fractures can be recognized in the entablature by observing the Outcrop #2 from multiple directions. The first type is relatively continuous sheet-like fractures (Fig. 3a) (Lescinsky and Fink 2000). As slightly curved sheet-like fractures developed nearly parallel to each other, the columns between the sheet-like fractures formed slightly bent plates about 15–40 cm thick (Fig. 2). The second type is pseudopillow fractures (Watanabe and Katsui 1976; Lescinsky and Fink 2000; Forbes et al. 2012; Forbes et al. 2014), which develop in several networks throughout the entablature (Fig. 3b). Pseudopillow fractures are arcuate fractures that extend several meters and occur roughly parallel to each other, intersecting each other in places (Fig. 3b, c). Short fractures extend almost perpendicular to the pseudopillow fractures at intervals of a few centimeters, which is the feature that gives them their name (Fig. 3d). The short fractures are filled by pale yellowish material. Where this occurs, lava within 20 cm from the pseudopillow fractures is darker and more glossy than in other parts of the flow.

Fig. 3
figure 3

A Photograph of the boundary between the entablature and the colonnade at Outcrop #2. The sheet-like fractures (s.f) parallel to each other can be observed in the entablature. b Photograph of the boundary between the entablature and the colonnade at Outcrop #1. The surfaces of some sheet-like fractures are visible in the photograph. Arcuate pseudopillow fractures (p.p.f; yellow dashed lines) extend like tree branches throughout the entire entablature. c Pseudopillow fractures (highlighted in yellow) developed almost perpendicular to the sheet-like fractures. An arrow in the lower left indicates the view direction in Fig. 3a. d Photograph of a sheet-like fracture (d in Fig. 3c) viewed from directly above. Short fractures (yellow arrows) are observed almost perpendicular to the pseudopillow fracture. The lava within 20 cm of the pseudopillow fracture is glassy. Pale yellowish materials fill the short fractures. e Photograph of the uppermost part of the Genbudo lava flow. f Close-up of the upper right of Fig. 3e. Pseudopillow fractures are shown by yellow dashed lines. A block indicated by the red arrow has a curviplanar surface

Full size image

The uppermost part of the lava flow consists of finger-like projections of lava (4–10 m in length, 1–3 m in width) that extend upward from the entablature with cogenetic monomict breccias between the fingers (Fig. 3e). The direction of finger elongation is sub-parallel to each other and nearly perpendicular to the top surface of the lava flow (Figs. 2a, 3e). Short fractures extend almost perpendicular to the outline of the fingers. The marginal part of each finger is darker in color than the interior. Just below where the fingers project outward, pseudopillow fractures extend downward in the "interdigital space" between the fingers and short fractures develop perpendicular to the pseudopillow fractures (Fig. 3f). The monomict breccias (lapilli and blocks) between the fingers are angular and blocky in shape and have planar to curviplanar surfaces (e.g. Scutter 1999; Otterloo et al. 2015). They do not have sedimentary structures like bedding. They occasionally exhibit a macroscopically visible jigsaw-fit of adjacent blocks. Clast-supported breccia is found next to the fingers while matrix-supported breccia is observed away from them. The characteristics of the autoclastic rocks between the lava fingers are very similar to those of the type-B hyaloclastite, including polyhedral fragments without glassy rims, described by Yamagishi (1987) and those of the hyaloclastites summarized by Otterloo et al. (2015).

Microscopic characteristics of the colonnade and entablature

Both the colonnade and the entablature of the Genbudo lava flow contain plagioclase, olivine, augite, orthopyroxene, quartz, and opaque minerals (titanomagnetite and ilmenite) as phenocrysts. Cr-spinel microcrystals (Cr# =  ~ 0.47) are included in olivine phenocrysts. Augite, orthopyroxene, plagioclase, and opaque minerals form glomerophyric textures. The whole-rock chemical compositions of the samples collected from the colonnade and entablature are all classified as basaltic andesite (Fig. 4). There is no systematic difference in whole-rock composition between the colonnade and the entablature.

Fig. 4
figure 4

The total alkali-silica diagram (after Le Maitre 2002) for the Genbudo lava flow

Full size image

The groundmass textures vary at the boundary between the colonnade and the entablature (Table 1). The groundmass of the colonnade near the boundary between the two is almost holocrystalline and is composed of plagioclase (average size ~ 50 µm), clinopyroxene (augite and pigeonite; average size ~ 30 µm), orthopyroxene, olivine, magnetite, and quartz crystals (Fig. 5a). Small amounts of silicic glasses (SiO2 =  ~ 80 wt%) are rarely distributed among the groundmass crystals (Fig. 5b).

Fig. 5
figure 5

Backscatter electron images of samples from the Genbudo lava flow. a Colonnade (sample GL5: 3 m below the colonnade-entablature boundary). The groundmass is almost holocrystalline. Quartz and magnetite are crystallized. b Colonnade (sample GL6: 0.5 m below the colonnade-entablature boundary). Quartz and magnetite occur and the interstitial glasses are highly silicic (SiO2 =  ~ 80 wt%). Fe-rich blebs occur in the glasses. c Entablature (sample GL12: 20 cm from a pseudopillow fracture). Interstitial glasses occur in the samples near the pseudopillow fractures. Quartz and magnetite are not crystallized. The glasses (SiO2 = 67–69 wt%) are poorer in SiO2 than those in the colonnade. d Entablature (sample GL10: 30 cm from a pseudopillow fracture). Glass does not occur in the samples away from the pseudopillow fractures. e Entablature (sample GL13: 10 cm from a pseudopillow fracture). Networks of micro-fractures partly filled with palagonite are distributed in the groundmass. f Entablature (sample GL13: 10 cm from a pseudopillow fracture). Spherical bubbles occur in plane symmetry with the micro-fracture as the symmetry plane. Note that there are two forms of groundmass clinopyroxene: larger prismatic crystals and smaller dendritic crystals. g Entablature (sample GL13: 10 cm from a pseudopillow fracture). Palagonite also occurs on the surface of micro-fractures within the plagioclase phenocryst. h, i Entablature (sample GL14: less than 10 cm from a pseudopillow fracture). Fe-rich blebs do not tend to occur in the Si-rich glasses in the vicinity of dendritic clinopyroxenes. Pl, plagioclase. Cpx, clinopyroxene. Qtz, quartz. Mgt, magnetite. Gl, groundmass glass. Fe-rich dp, Fe-rich droplet. Si-rich gl, Si-rich glass around dendritic clinopyroxene. P, palagonite

Full size image

As contrasted with the colonnade, the groundmass of the entablature near the boundary between the two is commonly hypocrystalline. The average grain sizes of plagioclase and clinopyroxene are about 50 µm and 30 µm, respectively. No quartz or magnetite are crystallized in the groundmass; instead, substantial amounts of volcanic glass (SiO2 = 67–69 wt%) is found (Fig. 5c). However, the samples taken more than 30 cm from the pseudopillow fractures occasionally exhibit holocrystalline textures like those of the colonnades (GL10; Fig. 5d).

The dark glassy parts very close to the pseudopillow fractures (0–20 cm from the fractures) have networks of micro-fractures ranging from a few to 10 µm in width (Fig. 5e; Table 1). The micro-fractures also penetrate plagioclase, clinopyroxene, and glass in the groundmass. The groundmass on either side of the micro-fracture exhibits jig-saw fit textures. The micro-fractures are partly filled with palagonite. Bubbles are distributed in the groundmass glass along the micro-fractures (Fig. 5f). Bubbles do not occur in the groundmass away from the fractures (Fig. 5e). Irregularly shaped voids occur within phenocrysts penetrated by the micro-fractures (Fig. 5g). The palagonite-filled micro-fractures are not distributed more than 30 cm away from the pseudopillow fractures (Fig. 5d). There are two types of groundmass augite crystals very close to the pseudopillow fractures: relatively larger prismatic microlites (average size ~ 30 µm) and smaller dendrites with larger aspect ratios (Fig. 5f, h). Some prismatic augites have dendritic overgrowth rims. Fe-rich blebs with a diameter of less than 1 µm are distributed in the interstitial glass in which dendritic augite has precipitated (Fig. 5i). The compositions of the glasses adjacent to the Fe-rich blebs are richer in SiO2 and poorer in CaO, MgO, FeO, and TiO2 than those of the interstitial glasses without Fe-rich blebs (Fig. 6). The Fe-rich blebs tend to be less abundant in the Si-rich glass adjacent to the dendritic augites. In contrast, there is neither a network of micro-fractures nor dendritic augite in the colonnade lava samples (Fig. 5a, b) or in the entablature more than 30 cm away from the pseudopillow fractures (Fig. 5d).

Fig. 6
figure 6

Major element compositions of groundmass glasses and Si-rich glasses around Fe-rich droplets in the entablature near a pseudopillow fracture (sample GL14) plotted on the pseudo-ternary diagram showing the two immiscibility fields in the system leucite-fayalite-SiO2 (Roedder 1951; Roedder 1978)

Full size image

Discussion

Evidence of increased cooling rate by water ingress

Occurrences of dendritic crystals and glassy groundmass only in the entablature like in the Genbudo lava have already been reported from other columnar jointed lava flows and are considered as evidence of rapid cooling (e.g. Columbia River basalt: Long and Wood 1986; DeGraff et al. 1989; Lyle 2000). These studies interpreted that the rapid cooling of the entablature occurred by the interaction between the hot lava and external water.

In addition, the network of pseudopillow fractures developed in the entablature of the Genbudo lava are similar to pseudopillow fractures that have been reported from other columnar jointed lava flows and have been interpreted as a good evidence of coolant (liquid water or steam) infiltration into lavas (Lescinsky and Fink 2000; Mee et al. 2006; Lodge and Lescinsky 2009; Forbes et al. 2014). Forbes et al. (2014) considered that the master fractures in the pseudopillow fracture systems are the main conduits bringing steam and/or water into the entablature, based on the presence and abundance of dendritic oxides and glassy mesostasis. Below, we discuss the cooling process of the entablature in the Genbudo lava on the basis of the observed petrographic textures surrounding the pseudopillow fractures.

The presence of palagonite-filled micro-fractures in the dark glassy parts of the entablature indicates that it was subjected to external water or steam infiltration through the micro-fractures. When water or steam comes into contact with hot lava, the expansion by a phase transition from water to steam or the thermal expansion of steam can occur as the temperature increases. The rounded shape of the bubbles along these micro-fractures indicates that the water or steam infiltration occurred while the interstitial melt was still hot enough to deform by the volume increase of water or steam.

Faure et al. (2007) observed a change in the morphology of crystallized olivine from polyhedral to dendritic with increasing cooling rate during a crystallization experiment of olivine in a basaltic melt. Similarly, the two forms of polyhedral and dendritic pyroxene found near the pseudopillow fractures of the entablature within the Genbudo flow (Fig. 5h, i) indicate a rapid increase in cooling rate during the crystallization of the lava groundmass. The amount of pyroxene dendrite increases with proximity to pseudopillow fractures (Fig. 5c, f). In addition, the palagonite-filled micro-fractures are distributed only in narrow zones within 20 cm of the pseudopillow fractures. In the narrow zones, the short fractures filled by the pale yellowish material (palagonite?) also extend perpendicularly from the pseudopillow fractures inward (Fig. 3d). These observations suggest that water entering the palagonite-filled micro-fractures was introduced through the pseudopillow fractures and the infiltration of water caused an increase in the cooling rate of the crystallizing lava. The pseudopillow fractures are distributed in a network throughout the entablature, suggesting that water circulated throughout the entablature horizon.

Formation of hyaloclastite in the uppermost part of the flow

The characteristics of the autoclastic breccias between the finger-like projections of lava in the uppermost part of the Genbudo flow suggest that they are hyaloclastites which are the products of quench fragmentation (Rittmann 1960; Pichler 1965; Kokelaar 1986; Yamagishi 1987; Cas and Wright 1988; Cas 1992). The jig-saw fit textures in some of the breccias indicate that these are in situ hyaloclastites (Otterloo et al. 2015). Quench fragmentation occurs when magma is super-cooled to glass by contact with water, contracts and is fragmented in situ by the propagation of a network of cooling fractures. In the “interdigital space” of the uppermost part of the flow, the pseudopillow fractures extend downward toward the interior of the entablature, suggesting that the water circulating within the entablature came from the uppermost part of the flow. This interpretation is consistent with the occurrence of hyaloclastite in the space between the fingers within the uppermost part of the flow. Similar occurrences of hyaloclastite around finger-like coherent lava covered by hackly joints have also been reported in the Santa Gertrudis lava, Chile, formed by the interaction of magma and snow or ice (Mee et al. 2006).

At the top of the Genbudo lava flow, the alternating distribution of finger-like projections of coherent lava and hyaloclastite indicates that the generation of fragmented lava was localized. The elongation direction of the finger-like lava projections is almost perpendicular to the top surface of the lava flow (Figs. 2a and 3e), suggesting that lava fragmentation was caused by water infiltration along columnar joints that developed almost perpendicular to the top surface of the lava flow (Fig. 7). The water may have been brought by the temporary damming of a river in the paleo valley into which the lava flowed (e.g. Long and Wood 1986). In other words, lava fragmentation did not equally continue on the entire upper surface of the lava in contact with water, but it occurred locally within the columnar joints along which water penetrated. The initial fracture spacing on the original lava surface before the hyaloclastite formation may have been comparable to the joint spacing in the lower colonnade, or perhaps smaller due to the larger temperature gradient between water and the original lava surface. However, as water percolates into fractures, superheating of the percolating water is likely to reduce the temperature gradient within the lava and lead to more widely spaced fractures (Otterloo et al. 2015). This effect may have resulted in the spacing of the finger-like projections of lava being much wider than the spacing of the colonnade columnar joints. Bahr et al. (1986) experimentally confirmed that curviplanar cracks are formed by propagation at low ΔT (< 217 K). The occurrence of blocks with curviplanar surfaces between the finger-like projections of lava also supports this interpretation.

Fig. 7
figure 7

Conceptual model of columnar joint formation in the Genbudo lava flow. a Cooling occurs rapidly above and slowly from below the lava flow and columnar joints begin to form. River water, held back by the lava flow, covers the surface of the lava flow during its emplacement. b Within the joints formed on the upper surface of the lava, fracturing and quench fragmentation occur due to rapid cooling of lava by cold water to form hyaloclastite. c Pseudopillow fractures develop downward from the hyaloclastite formed in the “interdigital spaces”, and water further penetrates into the lava through pseudopillow fractures. Later, sheet-like fractures also occur perpendicular to the pseudopillow fractures

Full size image

Cooling rate of the entablature

The groundmass glasses without Fe-rich blebs near the pseudopillow fractures have compositions within the field of silicate liquid immiscibility (Fig. 6), indicating that the interstitial melts were differentiated to chemical compositions that could cause unmixing. The compositions of the glasses adjacent to the Fe-rich blebs suggest that they may be Si-rich glasses produced by unmixing. Honour et al. (2019) observed immiscible Fe-rich droplets in experiments using natural ferrobasalt as a starting material, and found that the onset of unmixing was localized due to compositional inhomogeneity in the interstitial melt caused by a compositional boundary layer formed during crystallization. They observed that Fe-rich droplets were distributed adjacent to plagioclase, but that Fe-rich droplet-free zones were formed around pyroxene and magnetite. The Fe-rich blebs observed near the pseudopillow fractures of the Genbudo lava flow also tend not to be distributed in the Si-rich glass around the dendritic pyroxenes. These observations suggest that the Fe-rich blebs we observed near the pseudopillow fractures were caused by silicate liquid immiscibility during cooling.

Honour et al. (2019) showed that the relationship between the average of the five largest diameter Fe-rich droplets in basaltic dykes and the cooling rate calculated by a one-dimensional conductive cooling model can be approximated by the following power-law regression curve:

$${\text{y}}=84.681{{\text{x}}}^{-2.461}$$

where x is the average of the five largest diameter (µm) Fe-rich droplets in each sample and y is cooling rate (°C/h). Using the regression curve, we estimated the cooling rate of colonnade and entablature in the Genbudo lava flow. As the average diameter of the five largest Fe-rich blebs in the colonnade near the colonnade-entablature boundary is 1.2 µm (Table 1), the cooling rate in the colonnade is estimated to be about 49 °C/h (Fig. 8). On the other hand, the sizes of the Fe-rich blebs near pseudopillow fractures of the entablature are so small (0.4 µm) that they deviate from the range of this regression curve, but extrapolation of the regression curve gives a cooling rate of 642 °C/h (Fig. 8). We note that the Genbudo lava is a little richer in SiO2 than the rocks studied by Honour et al. (2019). Nevertheless, there is no doubt from these results that there was a large contrast in cooling rates at the colonnade-entablature boundary within the Genbudo flow. This large contrast in cooling rates indicates that the entablature and the colonnade had very different cooling regimes. The colonnade-entablature boundary must represent a horizon where a solidification front due to cooling that progressed upward from the ground by slow thermal conduction interfaced with a solidification front that progressed downward by rapid convective cooling due to water ingress.

Fig. 8
figure 8

Estimation of cooling rates for the colonnade and the entablature using Fe-rich droplet diameters produced by liquid immiscibility, based on laboratory results of Honour et al. (2019)

Full size image

Conclusions

  1. (a)

    The uppermost part of the Genbudo lava flow consists of finger-like projections of coherent lava and the cogenetic monomict breccias in between the projections. The breccias are angular and blocky in shape and have planar to curviplanar surfaces. They occasionally exhibit jig-saw fit textures. These characteristics show that they are hyaloclastites formed by in situ brecciation of the lava. The hyaloclastites were produced by cooling contraction caused by water infiltrating along cooling joints that developed vertically on the upper surface of the flow.

  2. (b)

    Based on a previously determined relationship between the immiscible Fe-rich bleb diameter and cooling rate, the cooling rate near the pseudopillow fractures in the entablature of the Genbudo lava flow is estimated to have been 642 °C/h, an order of magnitude larger than that of the colonnade (49 °C/h).

  3. (c)

    Palagonite-filled micro-fractures near pseudopillow fractures in the entablature of the Genbudo flow are direct evidence of external water infiltration during rapid cooling. The presence of palagonite-filled micro-fractures and associated bubbles are thus important criteria for determining whether external water was involved in entablature formation at other columnar jointed lava flows.