Compound and simple lava flow terminology in continental flood basalt provinces

Morphological types of continental flood basalts (CFBs) have been described with various terminology that hints at emplacement mechanisms and eruption dynamics. Walker (1971) introduced the notion of “compound” and “simple” lava flows, using descriptions of lava flows of the Deccan Traps CFB province in west-central India (Fig. 1). Nichols (1936), Walker (1971), Self et al. (1997), and Harris and Rowland (2009) define a flow-unit, or lobe, as the smallest component of a lava body, i.e., as having “a top which cooled significantly and solidified before another flow-unit was superimposed on it” so that “each flow-unit is a separate cooling unit”. These lobes are not completely cooled and so can fuse together, but they have cooled enough to retain structural boundaries between lobes. Walker’s (1971) simple flows consist of flow-units that are meters to tens of meters thick and kilometers in lateral extent, and thus, several orders of magnitude larger than the typical outcrop; for such units, we adopt the term “sheet lobe” after Thordarson and Self (1998). By contrast, Walker’s (1971) compound lava flows are composed of many flow-units, typically represented by stacks of pāhoehoe toes and lobes ~ 0.5 to 2 m in size. As per Self et al. (2022), we here prefer the term “small-scale pāhoehoe” (shortened to small-scale phh) to describe such morphology.

Fig 1
figure 1

a, b Cross-sectional and c plan views of the two end-member morphologies in continental flood basalts and their internal features. a A sheet lobe with well-formed colonnade and entablature tiers in the Gangakhed-Ambajogai section, southeastern Deccan Traps. Danielle Moyer provides a scale. b A compound flow field made up of numerous flow-units (lobes and toes) exposed in the Ellora Caves, central Deccan Traps. Dani Moyer for scale. c Squeeze-ups exposed on the eroded surface of a compound flow field in the western Deccan Traps. Hammer (33-cm long) and car provide a scale. The squeeze-ups represent small rootless conduits that lead to intraflow resurfacing in such flow fields

In CFBs, sheet lobes are common and may have formed during phases of high effusion rate (Walker 1971; Self et al. 1997). Sheet lobes also appear to be common on Mars and other terrestrial bodies in the solar system (e.g., Greeley and Bunch 1976; Theilig and Greeley 1986; Keszthelyi et al. 2006). Many CFB provinces show an interesting variability in their proportions of sheet lobes and small-scale phh flows across the stratigraphy. In the North Atlantic Igneous Province exposed in the Faroe Islands, and in the Ethiopian Traps, lava morphology shows a stratigraphic dependence (Kieffer et al. 2004; Passey and Bell 2007). Similarly, the lower part of the Etendeka (NW Namibia) CFB stratigraphy appears dominated by small-scale phh flows, whereas sheet lobes occur mostly in the upper part (Jerram et al. 1999a, b). On the other hand, small-scale phh-dominated morphology has only rarely been documented in the Columbia River Basalt province (Hooper 1997; Thordarson and Self 1998). In the Deccan stratigraphic sequence in the Western Ghats escarpment, the lower and middle formations contain both small-scale phh flows and sheet lobes, with some formations consisting almost entirely of one of the two types; by contrast, upper formations are largely composed of sheet lobes (e.g., Vanderkluysen et al. 2014, 2015; Self et al. 2021, 2022).

Morphological transitions in lava flow fields have long been attributed to changes in eruption dynamics (e.g., Hon et al. 1994; Anderson et al. 1999; Bailey et al. 2006), but the specifics of the effects of eruption dynamics on the morphology and growth of lava flow fields warrant further study. Wax simulations of flow fields can isolate the effects of various eruption parameters on morphological development. Here, we discuss the constraints on eruption dynamics that contribute to the emplacement of small-scale phh or sheet lobes.

Wax analog experiments

Polyethelyne glycol wax (PEG) has been used to simulate lava flow dynamics and morphologies which are specifically affected by the solidification process (e.g., Hallsworth et al. 1987; Fink and Griffiths 1990, 1992; Gregg and Fink 2000; Rader et al. 2017; Lev et al. 2019; Peters et al. 2022). PEG wax is a useful analog for lava due to its readiness to form a thin viscoelastic crust that becomes brittle as it cools and has a temperature-dependent viscosity (Fink and Griffiths 1990; Soule and Cashman 2004). The wax solidifies at ~ 17 °C and thus is typically extruded into a cold fluid to speed up the process of crust formation. The dimensionless parameter Ψ, derived originally by Fink and Griffiths (1990), relates crust formation to viscous flow and allows the discrimination of morphologies produced under different flow regimes (Table 1). Examinations of PEG wax crust strength that establish the analogous formation of a thin viscoelastic layer below the brittle crust have noted that PEG crust is stronger than lava crust and thus there are typically fewer fractures formed in PEG flows (Soule & Cashman 2004).

Table 1. Experimental scaling factors. Strong crusts support lobate inflation. Weak crusts result in spread out flow features

Previous experiments investigating lava flow morphology using PEG wax have demonstrated that the crust morphology reflects the cooling regime and emplacement style (e.g., Fink and Griffiths 1990, 1992; Gregg and Fink 2000; Rader et al. 2017; Peters et al. 2022). These studies focused on linking emplacement style to effusion rate with morphologies such as levees, folds, rifts, and eventually pillows resulting from progressively decreasing effusion rate (Fink and Griffiths 1990). Note that in the lava analog literature, the term “pillow” has no connotation of subaqueous emplacement and refers only to lobate morphological shapes made by the crust. Additionally, we use ‘extrusion rate’ when referring to experiments as opposed to ‘effusion rate’ for eruption measurements as discussed in Harris et al. (2007).

Additional analog studies have found that crust formation and growth influence the development of small-scale phh lava flows, unconfined lava channels, and lava tubes emplaced on a slope (e.g., Hallsworth et al. 1987; Blake and Bruno 2000; Cashman et al. 2006; Kerr et al. 2006). With the exception of Kerr et al. (2006), Rader et al. (2017), and Peters et al. (2022), the previous set of experiments did not investigate the effect of long travel distances on surface morphology, nor did they account for changes in flow morphology and emplacement style due to changes in effusion rate with time. In the Deccan Traps, lava flows appear to have flowed for tens to hundreds of kilometers, much further than flows typically observed in places like Hawai‘i or Etna (Walker 1973; Jay and Widdowson 2008; Vanderkluysen et al. 2011; Fendley et al. 2020). Variable effusion rates are typical for effusive eruptions and affect lava flow emplacement via inflation, breakouts, and surges (Wadge 1981; Baloga and Pieri 1986; Anderson et al. 1999; Bailey et al. 2006; Rader et al. 2017; Peters et al. 2022). In this study, we use PEG 600 wax to explore if aspects of lava morphology of the Deccan and other CFBs may have been controlled by effusion rate variations, travel distance from the source, or a combination of factors. We aim to constrain the possibilities which may have led to the abundance or lack of small-scale phh flow features in flood basalt fields.

Lava flow morphology preserves evidence of flow propagation

Slow-moving flow fronts of basaltic lava typically propagate as lobes or toes which, over time, coalesce and inflate creating a liquid core, with lines of vesicles that run parallel to the top and bottom surfaces of the flow (Self et al. 1998). The vesicles become trapped against the lobe’s chilled margin, thus emphasizing cooling boundaries between lobes. Persistent effusion rates favor the development and maintenance of sub-crustal flow networks (such as lava tubes), allowing lava to be transported to the edge of a flow field instead of overtopping and resurfacing previously emplaced portions of the active flow field (Rader et al. 2017; Peters et al. 2022). Thus, flow fields with sub-crustal networks often solidify with drained, empty lava tubes or with massive flow interiors, depending on slope (Hon et al. 1994; Self et al. 1998). Occasionally, lines of vesicles that parallel a lobe’s chilled rind can be preserved in situations where the lobe develops a rigid crust that is not disrupted by additional lava injections (Self et al. 1998; Jay et al. 2018; Nikkola et al. 2019). These conditions are likely to be met if lava supply rates are low, however, quantifying how low is an ongoing area of research (Lev et al. 2019). All previous wax studies examined effusion conditions that preserve surface morphology, but this study extends that work to include a cross-sectional evaluation of lobe preservation to simulate the patterns seen in CFB outcrops. By extending our analysis to cross-sections, we aim to broaden the applicability of wax analog experiments to older flows, which may have eroded or buried surface morphology.



Experiments were conducted in a circular, 130-cm-diameter inflatable pool filled with 12-cm-deep ice-water mixture buffered at 0 °C (Fig. 2). The base of the tank was flat and had a metal fixture from which PEG was extruded, anchored to a 0.3-cm thick sheet of plexiglass (18 × 27 cm) that created a slight rectangular rise around the wax extrusion point (dubbed the ‘source’ for experimental discussions). Tubing 1.5 cm in diameter connected the metal fixture to the programmable peristaltic Syringe-pump 9000. The base of the pool was coated in medium-grained sand secured with a spray adhesive in order to add roughness along the bottom of the pool. Three video cameras were placed around the tank to capture a side, top, and oblique view of each experiment (Supplementary material Video 1). PEG 600 wax mixed with food coloring was maintained at a constant temperature (see Table 1 for specific conditions of each experiment) in a 500–1000 cm3 beaker suspended in a water bath until use. For group 1 (non-pulsed experiments), the pump was programmed to run at a constant rate (300 cm3/min) for 2–3 min, pause for 40 or 120 sec, then continue running at the same rate. For group 2 (pulsed experiments), the pump was programmed to run at 300 cm3/min for 10 sec, and a subsequent lower rate (50 cm3/min) for 10–50 sec, providing a pulsation interval and magnitude for each experiment. The color of the wax was changed after the pause (group 1) or after 80–120 sec of pulsatory flow (group 2).

Fig. 2
figure 2

Experimental set up for larger-scale wax flows

Upon completion of each experiment, the surface was photographed using a Nikon P900 digital camera and cross sections were cut through areas of interest (Fig. 3). We used the cross sections to evaluate the thickness of crust and the internal structure of the experiment before the wax melted or dissolved.

Fig. 3
figure 3

Internally preserved structures of two experiments. a Experiment 8 had a Ψ of 1.1 and a P value of 0.26 and shows features like those of the small-scale phh Deccan lava flows. b Experiment 33 had a Ψ of 20.3 and P value of 0.58 and shows internal structures more common in Deccan sheet lobes


The two major scaling factors in this paper are described in this section but can also be quickly referred to in Table 1. The non-dimensional value Ψ of each experiment was calculated using the equation Ψ= Πτss) where Π is the modified Peclet number and τss) is a non-dimensional solidification timescale as derived by Fink and Griffiths (1990), and subsequently adjusted by Gregg and Fink (1996). This scaling factor compares the rate of viscous flow to the rate of crust formation such that Ψ is high when crust formation is much slower than spreading due to gravity (further discussion of the calculation of Ψ can be found in Fink and Griffiths 1990; 1992; Griffiths and Fink 1992; Gregg and Fink 1996; Rader et al. 2017). Pulsation intervals (P) targets the relative amount of time close to steady-state extrusion and is expressed using a scaling factor, P = Qavg/Qmax where Qavg is the average extrusion rate during the entire experiment and Qmax is the highest extrusion rate during the experiment. High values of P correspond to experiments in which the low extrusion-rate period was short (e.g., 10 s, fast pulsation), whereas low values of P correspond to experiments in which the low extrusion-rate period was long (e.g., 50 s, slow pulsation). These data are presented in Table 2 for all experiments. Original top-down photos of the experiments can be found in Supplementary Materials 2.

Table 2 Long-extrusion experiments

Image processing

The boundaries between different flow emplacement styles were marked on a final overview image of each experiment after correlating surface features with flow type from the video. Based on the original nomenclature from Fink and Griffiths (1990), levees were identified where flow was so fast (high Ψ) that crust could only form along channel margins. Folds are designated when a very thin crust formed, quickly forming little folds perpendicular to flow. Rifts are defined as regions where new crust formed as older crust broke and was transported away, resulting in a concentric zone of crust that spread radially away from the fracture. Pillows are defined as a randomly lumpy surface texture. We used video footage to make these distinctions because as flows progressed, some aspects of surface morphology changed, as discussed in the results. We measured the distances in the direction of flow between the morphological zones (levees, folds, rifts, and pillows) from these images.

Inclusion or exclusion from the dataset

Experiments presented in this manuscript were conducted during two experimental sessions utilizing different camera arrangements. This fact, plus the occasional equipment failure, resulted in some experiments not having the appropriate angles or high-enough resolution to make resurfacing calculations. However, Tables 2 and 3 present the other aspects of data that were unaffected by equipment issues. Some aspects of experiments #7–23 were previously published in Rader et al. (2017); however, we made new measurements and calculations with those data and those results are included in Table 3, as well as the figures. Additional experimental notes can be found in Rader et al. (2017).

Table 3 New measurements on smaller-extrusion experiments from Rader et al. 2017


Flow parameters and flow footprint

A total of fourteen new experiments were run with Ψ values from 1 to 138 and pulsation intervals of 0.31 to 1 (Table 2). Experiments with P values > 0.8 were considered to be ‘non-pulsating.’ The starting extrusion rate for all experiments was 300 cm3/min. Seven experiments had complete pauses (0 cm3/min) in extrusion rate while the other seven had reduced extrusion rate (50 cm3/min) for between 10 and 50 sec.

We compare the new experiments with eleven experiments published in Rader et al. (2017; Table 3). The older data set was re-analyzed with regard to intraflow resurfacing and proportion of marginal breakouts (terms used in Rader et al. 2017, Peters et al. 2022). The Ψ values from previous experiments in Rader et al. (2017) ranged from 1 to 82 with pulsation intervals of 0.26 to 1.

The footprints of the experiments are often elongated in one or two directions due to the rectangular platform from where wax emitted (the source). This shape resulted because a preferred pathway would randomly be established along wherever the flow first reached the edge of the small platform, which would reduce the flow rate in other directions. All experiments except #34 stopped flowing before impacting the side of the pool.

Morphomaps and emplacement style

Outlining the regions of the Fink and Griffith (1990) emplacement style on each flow illustrates how those emplacement styles change with distance from the source in larger flow fields (Fig. 4). Emplacement morphology tends to follow a concentric pattern that radiated out from the source with faster-flowing morphologies preserved at the center and the slowest flow morphology (pillows) preserved at the margins. However, disruptions in flow rate initially reset the progression from fast to slow morphologies but were followed by faster morphologies propagating to greater distances from the source, causing islands or repetitive patterns of differing morphologies. For all flows, as surface area grew, pulses in wax supply became distributed over an enlarging active footprint. Thus flow rate changes in those pulses had a more muted effect on the morphology at greater distances. The result of this muting effect was a more consistent propagation of pillows at the margins even with longer pauses between pulses, e.g., lower P (Fig. 5).

Fig. 4
figure 4

Outlines of emplacement style regimes after Fink and Griffiths (1991) of each large-scale experiment. Experiments are overlaying the region where they plot for Ψ and P. Experiments with boxes around their numbers are categorized as ‘not-pulsed.’ Base images of experiments can be found in Supplementary Materials 2

Fig. 5
figure 5

a Graphs showing the area created by Fink and Griffiths (1991) emplacement types and distance when emplacement type changes between pulsed vs. not pulsed experiments. b Pulsed flows were more likely to produce lower-energy emplacement styles as well as have those transitions occur closer to the vent

Emplacement style transitions

All experiments initially produced the morphological type associated with the maximum Ψ value as defined by Fink and Griffiths (1990) and refined by Gregg and Fink (1996) (pillows < 0.65; 0.65–2.8 rifts; 2.8–6.4 folds; > 6.4 smooth with eventual levees). However, experiments with low P often ended up quickly reverting to an emplacement style consistent with a lower Ψ value. Low P was also associated with a more disjointed resurfacing style of emplacement with islands of morphologies surrounded by lower Ψ value emplacement styles (e.g.., experiments 34, 36, and 37; Fig. 4) whereas faster pulsation (higher P) led to higher proportions of faster-emplacement morphologies at similar Ψ (e.g., experiments 27, 29, and 30 in Figs. 4 and 5). Slow pulsation (low P) also reduced the distance that lower-Ψ emplacement morphologies would be found from a source.

Resurfacing and marginal breakouts

As the number of marginal breakout events increased, and the number of resurfacing events decreased (Fig. 6), more material was emanating at the margins of the flow. Resurfacing is a process necessary to preserve internal lobate boundaries whereas marginal breakouts tend to cause lateral flow growth, which can allow for inflation and the disruption of internal lobate boundaries (Peters et al. 2022). Higher values of Ψ and P both tended to decrease the amount of resurfacing (Fig. 6). Resurfacing only became the dominant process at Ψ < 5 and P < 0.3 (Fig. 7).

Fig. 6
figure 6

a All graphs show that flows which propagate laterally via marginal breakout events have fewer resurfacing events. The diameter of symbols has been changed in the bottom two graphs to be proportional to b P or c Ψ. Higher P and Ψ tend to decrease the area that is resurfaced

Fig. 7
figure 7

Preservation of internal chilled boundaries preservation is more likely in a resurfacing-dominated regime. Our experiments illustrate that low Ψ and P that favor the style of eruption that likely produced the small-scale phh Deccan flow units (left of the green line), whereas higher P and Ψ values to the right of the line strongly favor marginal breakouts


When compared with lava, PEG wax has a proportionally stronger crust relative to its density that tends to fracture less frequently and in fewer locations (Soule & Cashman 2004). This suggests that our findings that compare the frequency of rupturing (breakouts or resurfacing) might provide a generous estimate of conditions that can preserve lobate structures.

Conditions that favor resurfacing

Once a flow surface is established, new crust can be created in four ways: 1) emanating from the toe of a flow that already has crust formed (new lobe creation), 2) filling in a crack in existing crust that rifts the solidified margins further downslope, 3) expanding the active front before enough crust has formed for a marginal breakout to occur (unconstrained flow), and 4) resurfacing. Resurfacing is different from the other crust-forming mechanisms because it increases the thickness of a flow field while having the potential to preserve internal structures with distinct lobes. Therefore, we examined the conditions that make resurfacing common in our experiments. The experiments with the lowest Ψ and the lowest P (greatest pauses between pulses) produced the most resurfacing, which allowed the preservation of internal structures with multiple distinct lobes.

Extended extrusion disrupts morphological transitions

Previous PEG wax experiments illustrated how effusion rate is linked to surface morphology in a progression from high to low effusion rate of levees → folds → rifts → pillows (Fink and Griffiths 1990, 1992, 1998). However, as flow fields grow, local extrusion rates diverge from the source extrusion rate as material is distributed across a broader area with increasingly more complex networks of preferred pathways and topographic constraints. Broadly speaking, the local effusion rate will decrease as material covers a larger area, leading to very voluminous eruptions producing lava morphologies that are typically associated with low effusion rates. Observations of flow rates in distal portions of flow fields indicate decreases in velocity and therefore Ψ value (Lipman and Banks 1987; Hon et al. 1994; Soule et al. 2004; Dietterich and Cashman 2014). Thus, the surface morphology will also reflect that decrease in kinetic energy with increasing distance from the source. In this study, we simulate larger flow fields than any previous published study with PEG wax and, in doing so, we confirm that flow surface morphology progresses to lower energy states with distance from the source and that these transitions occur at greater source distances for experiments with higher effusion rates. The relationship is most stark for the lowest energy transition of rifts to pillows. A fivefold increase in Ψ was found to increase the distance in the rift-pillow transition by 10%. A similar change extends the fold-to-rift transition by only 5%.

Pulsation impacts surface morphology patterns

Pulses in source effusion rate can also lead to complex patterns in surface morphology due to shifts in emplacement style. Pulsating effusion rate can create repetitive patterns of crustal textures such as concentric bands of inflated lobate surfaces. All but two experiments with pulsation values of ≤ 0.6 produced repetitive flow morphology whereas only one experiment with a P > 0.6 did. The effect was more evident for lower Ψ experiments as well. Thus, the surface patterns of lava flows may be useful indicators of eruption dynamics such as high effusion rates maintained for a short time vs. low effusion rates for a long time. Similarly, the degree of steadiness of the effusion rate may also be preserved in the patterns on the surface of flow fields.

Scaling and effusion rate correlation in lava flows

Our experiments aimed to evaluate what conditions are required to preserve lobate morphology on the surface and in cross-section in large flow fields. Large flow fields such as those found in CFBs have been hypothesized to grow laterally as small-scale phh flows (pillow surface morphology) that can grow vertically through inflation and become thick lava flow-units (sheet lobes) whose interiors contain features such as columnar jointing (Self et al. 1996, 1997, 1998). To preserve an internal structure of stacked small-scale phh lobes, the dominant mechanism of vertical growth of the lava pile must occur via resurfacing as new flow-units or lobes overtop those emplaced minutes to hours ago, rather than through endogenous inflation of individual lobes. The Deccan, Etendeka, North Atlantic, or Ethiopian CFBs contain flow fields with both types of internal morphologies without significant geochemical differences that may dominate morphological transitions (Kale et al. 2020); we therefore conclude that the eruption conditions may have been the main control on the production of small-scale phh lobes as opposed to thick single-unit flow interiors.

Our experiments indicate that Ψ values had to be < 1 and increase to > 5 to elicit the shift from resurfacing to marginal breakouts. This corresponds to an increase in effusion rate by a factor of 125 for linear sources and 625 for a point source as determined from solving the Ψ equation (Eq. 9a and b) from Griffiths and Fink (1992). For the purposes of illustrating the calculation procedure below, we assume the point source value (which is the equivalent of the linear source value with a fissure length of 5 m) since the lengths of flood basalt eruptive fissures active at any single instant in time are unknown.

Thus far, neither measured effusion rates at an erupting vent, nor local flow rates associated with pāhoehoe lobe formation for flow fields dominated by resurfacing instead of inflation, have been published. Similarly, we could find no measured effusion rates linked to internal lobe preservation in any witnessed eruption. However, an effusion rate of 0.2 m3/s at the vent resulted in individual pāhoehoe lobes with minimal flow inflation, documented in 1987–1990 Kīlauea eruptions (Hon et al. 1994). Similarly, effusion rates of 0.1–0.5 m3/s resulted in single lobes with little inflation at Kīlauea and Etna (Blake and Bruno 2000). Assuming that the lobes produced in these low-effusion-rate eruptions would have retained their glassy rinds and lobate vesicle structures once covered with new lava, we can use 0.2–0.5 m3/s as the effusion rate for the small-scale phh units of the Deccan, which have internal lobe margins preserved. An increase by a factor of 625 brings the effusion rate up to 125–312 m3/s, which would give an emplacement time of ~ 320 years for a flood basalt flow the size of the Roza flow (~ 1300 km3). That time scale is similar to the calculated eruption time interval for the Deccan Traps based on the amount of Hg preserved in sediment from that time period (Fendley et al. 2019).

Recent basaltic lava eruptions have effusion rate estimates of 100 m3/s on average and peaking at > 500 m3/s (Holuhraun) and even upwards of 1000s m3/s at Kīlauea (Coppola et al. 2017; Bonny et al. 2018; Patrick et al. 2019; Plank et al. 2021; Dietterich et al. 2021). Thus far, no interior morphologies with small-scale lobes have been reported (perhaps not yet exposed) in the Holuhraun or 2018 Kīlauea eruptions supporting the concept that the high effusion rates measured are consistent with massive lava flow interiors. Therefore, it is reasonable to assume that the two effusion rates needed to produce the two end-member interior morphologies in the Deccan Traps were not outside the ranges of historically observed eruptions. This calculation suggests that the duration of eruption was likely the factor that made the individual Deccan flood basalt eruptions surpass the volumes of the Kīlauea and Holuhraun eruptions.


In investigating the relationship between effusion rate, surface morphology, and interior structure of large CFB lava flows, using PEG wax experiments, we came to conclusions that are grouped into two categories; those which pertain to fundamental mechanisms of lava flow emplacement (1–4) and those conclusions that apply our findings to eruption dynamics of the Deccan Traps (5, 6).

  1. 1)

    Wax experiments support the idea that ‘simple’ and ‘compound’ pāhoehoe lava flows are emplaced from batches of similar material, and thus, their morphological difference is primarily a function of the local cooling rate which dictates crust formation. In our experiments, cooling rate is strongly tied to extrusion rate; however, there could be additional environmental conditions that contribute to widely differing local cooling rates.

  2. 2)

    Resurfacing is negatively correlated with breakouts at the margin of a flow in a topographically unconstrained setting; however, the correlation is not perfect as lava can also create new crust in rifts, folds, and channels. Resurfacing is favored over these other mechanisms only when pulsation rate is low and Ψ is low. Increasing either of those parameters results in far less resurfacing, a greater number of marginal breakouts, and larger areal extent. This suggests that expansive eruptions likely reached their great lateral extent due to higher and/or more steady effusion rates when compared to smaller lava flow fields.

  3. 3)

    Preservation of cooling boundaries between lobes requires very low P (greatest pauses between pulses) and Ψ- Constraining the exact values required for lobe boundary preservation requires additional data from active lava flows, but here we estimate that effusion rates of less than 0.5 m3/s may lead to these features.

  4. 4)

    Surface morphology associated with low Ψ value (pillows) is still produced in experiments with high source Ψ values but is formed at progressively greater distances from the source as Ψ value increases. Our experiments show that pillow-like surface morphology will begin to appear 10% farther away from the source given a fivefold increase in average effusion rate. This has implications for interpreting eruption unsteadiness as well as magma output rate from the surface morphology of lava flow fields.

  5. 5)

    To ensure that no small-scale internal lobe structures would be preserved in the Deccan sheet lobes, the average Ψ value of the eruption would have had to increase from < 1 to > 5. This is the equivalent of a 625-fold increase in effusion rate. If we use comparable effusion rates from flow fields that did not produce highly inflated flows, with 0.2–0.5 m3/s as a starting point, then we arrive at 125 m3/s. The latter value is a reasonable effusion rate which has been measured at two large basaltic eruptions at Holuhraun (100–500 m3/s) and the 2018 Kilauea eruption at Leilani Estates (1000s m3/s), in recent history.

  6. 6)

    The changes in effusion rates for the large Deccan lava flow fields estimated by this study are realistic and possible to achieve without special mechanisms such as bolide impacts, as changes of this magnitude have been recorded at other basaltic lava flow fields formed during historic times. Additionally, we conclude that a long duration of active eruption, not an unusually high effusion rate, is the major factor for creating flood basalts.