Hawaiian lava flows in the third dimension: Identification and interpretation of pahoehoe and ′a′a distribution in the KP-1 and SOH-4 cores



[1] Hawaiian lava flows are classified as pahoehoe or 'a'a by their surface morphology. As surface morphology reflects flow emplacement conditions, the surface distribution of morphologic flow types has been used to study the evolution and eruptive history of basaltic volcanoes. We extend this analysis to the third dimension by determining the distribution of flow types in two deep drill cores, the Scientific Observation Hole-4 (SOH-4) core, drilled near Kilauea's East Rift Zone (ERZ), and the pilot hole (Kahi Puka-1 (KP-1)) for the Hawaiian Scientific Drilling Project (HSDP), drilled through distal flows from Mauna Loa and Mauna Kea. Flows are classified using both internal structures and groundmass textures, with the latter useful when identification based on mesoscopic flow features (e.g., surface morphology and vesicle content and distribution) is ambiguous. We then examine the temporal distribution of pahoehoe and ′a′a flows in proximal (SOH-4) and distal (KP-1) settings. Sequence analysis shows that the two flow types are not randomly distributed in either core but instead are strongly clustered. The proximal SOH-4 core is dominated by thin pahoehoe flows (∼60% by volume), consistent with the common occurrence of surface-fed pahoehoe flows in near-vent settings. The distal KP-1 core has a high proportion of ′a′a (∼58% by volume), although the proportion of pahoehoe and ′a′a varies dramatically throughout the Mauna Kea sequence. Thick inflated pahoehoe flows dominate when the drill site was near sea level, consistent with the numerous inflated pahoehoe fields on the current coastal plains of Kilauea and Mauna Loa. ′A′a flows are abundant when the site was far above sea level. As slope increases from the coastal plains to Mauna Kea's flank, this correlation may reflect the combined effect of long transport distances and increased slopes on flow emplacement. These results demonstrate that flow type and thickness variations in cores provide valuable information about both vent location and local site environment. Observed variations in flow type within the KP-1 core raise interesting questions about feedback between volcano evolution and flow morphology and suggest that flow type is an important variable in models of volcano growth and related models for lava flow hazard assessment.

1. Introduction

[2] The surface morphologies of basaltic lava flows record a dynamical balance between deformation (through advection) and cooling of the flow surface during emplacement [Fink and Griffiths, 1990, 1992; Griffiths, 2000]. For this reason, classification of terrestrial flows as pahoehoe or ′a′a has genetic, as well as descriptive, implications, and the surface distribution of flow types has been used to study the evolution [e.g., Holcomb, 1987; Lockwood and Lipman, 1987; Rowland, 1996] and future hazards [Kauahikaua et al., 1995; Trusdell, 1995] of individual volcanoes. Stratigraphic studies of morphologic flow types in drill cores add the third dimension to the physical history of Hawaiian volcanoes, providing a time sequence of the flow history at a single location [e.g., Lipman and Moore, 1996; Stolper et al., 1996]. However, unambiguous identification of flow types in drill cores can be difficult, particularly as flow tops (flow surfaces) are often difficult to recover. Additionally, inferences about the physical conditions of emplacement (e.g., vent proximity, topography, and eruption rate) are often poorly constrained, particularly where core locations cannot be linked directly to subaerial exposures (e.g., in large igneous provinces such as the North Atlantic Volcanic Province) [Duncan et al., 1996; Eldholm et al., 1987; Larsen et al., 1994].

[3] Here we examine the distribution of flow types in two drill cores from Hawaiì (Figure 1) to evaluate the physical attributes of basaltic lava flows with respect to their environment of emplacement. The Scientific Observation Hole-4 (SOH-4) was drilled at 366 m elevation on Kilauea's East Rift Zone (ERZ) as part of a project to evaluate geothermal energy potential [Novak and Evans, 1991; Olson and Deymonaz, 1992]. The upper 1000 m of core contains subaerial lavas and shallow intrusive rocks from Kilauea Volcano. The Hawaiìan Scientific Drilling Project (HSDP) Kahi Puka-1 (KP-1) hole was drilled near Hilo Bay as a pilot hole to assess the viability of deep drilling for studying the evolution of Hawaiian volcanoes [Stolper et al., 1996]. It penetrates distal subaerial lavas from Mauna Loa and much of the distal subaerial Mauna Kea lava sequence. We compare flow type and thickness distributions in the two drill cores with the surface distribution of lava flow types on Kilauea [Holcomb, 1987] and Mauna Loa [Lockwood and Lipman, 1987] to examine the stratigraphic record at sites proximal and distal to eruptive rift zones. We then link the physical interpretation of flow distributions at these two sites to general studies of the evolution of Hawaiian volcanoes [e.g., DePaolo and Stolper, 1996]. Our detailed analysis of flow type distribution in drill cores, in locations where both proximity to source vents and general emplacement styles are known provides both a methodology and a baseline for description and interpretation of cores from more poorly constrained locations.

Figure 1.

Location map showing the island of Hawaii and its five constituent volcanoes. Drill holes SOH-4 and KP-1 are marked with stars. Rift zones (RZ) of Kilauea and Mauna Loa volcanoes are labeled, summit calderas shown in black.

2. Classification of Active Basalt Flows

[4] Hawaiian lava flows are classified by their surface morphologies as pahoehoe or ′a′a [e.g., Alexander, 1859; Dutton, 1884; Macdonald, 1953]. Surface morphology reflects conditions of flow advance and emplacement, and varies with eruptive style and distance of transport [e.g., Cashman et al., 1999; Holcomb, 1987; Hon et al., 1994; Kilburn, 1981, 1990, 1993; Macdonald, 1953; Peterson and Tilling, 1980; Rowland and Walker, 1990; Swanson, 1973]. The surface distribution of flow types in Hawaiì is determined by slope [e.g., Peterson and Tilling, 1980], eruption duration [e.g., Holcomb, 1987; Lockwood and Lipman, 1987], eruption rate [Rowland and Walker, 1990], and vent proximity and location [Holcomb, 1987; Swanson, 1973].

[5] Although a complete gradation occurs between morphologic flow types, end-member pahoehoe and ′a′a flow morphologies are distinctly different, as are conditions of their emplacement. Pahoehoe flows have smooth flow surfaces that form glassy ropes, cords, and festoons [Wentworth and MacDonald, 1953]. Pahoehoe flow fields in Hawaiì develop when eruptions are sustained over weeks to years, and lava supply is steady [Holcomb, 1987; Kauahikaua et al., 1998]. Under these conditions, robust lava tubes permit nearly isothermal transport of lava from rift zone or summit vents to the coast [e.g., Peterson et al., 1994; Cashman et al., 1994; Helz et al., 1995; Clague et al., 1999]. On low slopes, radial spreading of advancing pahoehoe sheets slows flow advance, and flows often inflate to several meters, although breakouts and resulting surface flows are also common [Mattox et al., 1993; Hon et al., 1994]. Pahoehoe flow fields may contain subsidiary ′a′a when surface-fed flows travel sufficiently far from the vent, particularly as breakouts from lava tubes on moderate slopes [e.g., Peterson and Tilling, 1980].

[6] In contrast, ′a′a flow surfaces are rubbly and composed of jagged crustal fragments. Hawaiian ′a′a flow fields form when emplacement is rapid (days to weeks); average effusion rates under these conditions are often >10 m3/s [Rowland and Walker, 1990]. Flows are commonly fed from fissure vents and transported through open channels [e.g., Lipman and Banks, 1987; Wolfe et al., 1988]. Flows may widen and thicken toward flow fronts as a consequence of both rheological changes that occur during early stages of flow [e.g., Cashman et al., 1999; Kilburn, 1993], and decreasing slopes [e.g., Walker, 1967; Kilburn, 1996]. An important characteristic of Hawaiian ′a′a flows is that surface morphology changes along the flow length, resulting in mixtures and intergradations of flow types. These flows often initiate as pahoehoe fed from fissure vents. Flows focus into open channels and transform to ′a′a during transport; as a result, overflows of pahoehoe from ′a′a channels are common in medial regions [Lipman and Banks, 1987; Wolfe et al., 1988]. Less common are pahoehoe breakouts from ′a′a flow fronts [Calvari and Pinkerton, 1998; Jurado-Chichay and Rowland, 1995].

[7] Pahoehoe and ′a′a flows have different internal structures [e.g., Macdonald, 1953; Wentworth and Macdonald, 1953] and groundmass textures [Emerson, 1926; Kouchi et al., 1986; Sato, 1995] as a consequence of differences in flow emplacement conditions. Pahoehoe flows are typically vesicular, although the vesicles vary in size and concentration. Thin pahoehoe flows commonly have spherical vesicles throughout [Cashman et al., 1994; Mangan et al., 1993; Walker, 1989], while thick (inflated) pahoehoe flows have well developed upper and lower vesicular crusts that form during emplacement by an inward-propagating thermal boundary [e.g., Hon et al., 1994]. The thickness of the upper vesicular crust can thus be used to infer emplacement duration [Self et al., 1996; Cashman and Kauahikaua, 1997]. The dense interior represents lava that solidified after flow ceased, and preserves features related to postemplacement processes [Goff, 1996; Self et al., 1998]. Thermally efficient lava transport results in limited syntransport crystallization, thus samples collected hot from active pahoehoe flows, or from the glassy (quenched) surfaces of solidified flows, typically have low crystal abundances (Figure 2).

Figure 2.

Plagioclase number densities (Na in number/mm2) as a function of plagioclase volume fraction crystallized (ϕp) measured in active flows from Mauna Loa and Kilauea (total crystallinity ϕ is approximately 2ϕp in all samples). Filled triangles are transitional and ′a′a samples from channelized flows. Open triangles are proximal samples from channelized flows and have pahoehoe surface morphologies. Red squares are samples from tube-fed pahoehoe flows. Open squares show the sequence of crystallization in Makaopuhi lava lake. Data are given by Cashman and Marsh [1988], Cashman et al. [1999], Folley [1999], Montierth [1999], and Polacci et al. [1999].

[8] ′a′a flows are generally less vesicular than pahoehoe flows, although thin flows may be moderately vesicular in near-vent regions. Vesicles within the interiors of thick flows are commonly flattened [Macdonald, 1953], while vesicles in small pasty breakouts from ′a′a channels may be deformed and oriented in concentric rings parallel to the flow margins [Canon-Tapia et al., 1996; Polacci and Papale, 1997]. Thicker flows often incorporate fragments of recycled ′a′a crust near upper surfaces [Crisp and Baloga, 1994], but are typically dense throughout much of their thickness. Samples collected along active ′a′a channels show a rapid increase in both plagioclase number density and crystallinity with increasing transport distance (Figure 2). The observed crystallinity increase indicates rapid cooling and crystallization during early stages of ′a′a flow advance [Cashman et al., 1999], probably enhanced by stirring during flow [Emerson, 1926; Kouchi et al., 1986; Sato, 1995].

[9] The surface distribution of flow types can be related to vent proximity, slope, and eruption duration, as documented for Kilauea Volcano by Holcomb [1987] (Figure 3). 81% of Kilauea's surface is covered by pahoehoe, emplaced as either tube-fed flow fields produced during long-lived eruptions from Kilauea's summit [e.g., Clague et al., 1999], or surface-fed flows along Kilauea's rift zones. ‘A‘a is abundant on the SE flanks of the ERZ and Southwest Rift Zones (SWRZ), where slopes are steep and surface-fed pahoehoe transforms to ‘a‘a. These steep slopes are covered by pahoehoe only where robust lava tubes develop during sustained eruptions, such as those of Mauna Ulu 1969–1974 and Pu‘u ‘O‘o (1983 to present). In contrast, ‘a‘a and pahoehoe are equally abundant on larger Mauna Loa volcano. ‘A‘a predominates on steep slopes of the rift zones (particularly to the SW), while pahoehoe covers proximal rift zone environments and accumulates in flatter saddle regions and coastal plains [Lockwood and Lipman, 1987; Rowland and Garbeil, 2000]. These patterns of flow type distribution suggest that vent proximal cores (such as SOH-4) should be dominated by thin, surface-fed pahoehoe flows, while the distal KP-1 core might be expected to show more variation in both flow type and thickness, reflecting the distance of transport, topography traversed, and variations in eruption duration.

Figure 3.

Surface distribution of lava flow types on Kilauea Volcano. Note correspondence of surface-fed pahoehoe flows with the ERZ and SWRZ (on Figure 1) and abundance of ′a′a on the steep southeastern flanks of the rift zones (modified from the study of Holcomb [1987]).

3. Methods

3.1. Core Description

[10] The SOH-4 drill core is stored at the University of Hawaii's Research Facility in Snug Harbor, Honolulu. We examined the upper 925 m of the core (boxes 1–299) with the aid of a logbook that provided preliminary flow type and thickness [Trusdell et al., 1992]. This half of the 1985 m core contains most of the subaerial flows. No samples were collected from this core.

[11] We examined the KP-1 core, stored at Caltech, from boxes 1 to 181 (0–495 m depth) and from boxes 334 to 354 (995 m to the bottom of the core at ∼1055 m). The core sampled flows from both Mauna Loa (upper 280 m) and Mauna Kea. A HSDP logbook provided preliminary flow unit designations, thickness, contact locations, internal flow unit boundaries, and general flow descriptions. Core morphologies between 495 and 995 m depth were classified using photographs of the core (on CD; on loan from E. Stolper, Caltech). Specific cores were later reviewed at Caltech to clarify ambiguities based on flow designations from photographs alone. Thirty samples were collected from 10 flow units selected to cover the observed range of flow morphologies (Table 1).

Table 1. List of Sample Thin Sections Examined From the KP-1 Corea
Run NumberDepth (m)UnitThickness (m)VolcanoFlow TypeLocation in FlowGroundmassComments
  • a

    The run numbers, depths, unit numbers, and volcano (ML = Mauna Loa, MK = Mauna Kea) are from the HSDP logbook. Flow types are described as pahoehoe (P), ′a′a (A), transitional (T), massive pahoehoe (PM), and transitional pahoehoe (P/T). Groundmass textures are qualitatively described as coarse grained (C), medium grained (M), fine grained (F), and very fine grained (VF). Plagioclase number densities were measured in the starred samples.

R67-3.192.99135.67MLPTop thirdCDense zone
R67-3.393.06135.67MLPTop thirdMSegregation vein
R125-0.8197.452910.18MLATop quarterVF*Large olivine laths
R126-8.35202.82910.18MLABottom thirdVF*Dense zone
R127-5.1204.862910.18MLANear baseVF*Quenched base?
R129-9.0208.5325.91MLTMiddleVF*Rubbly top, moderately vesicular
R151-8.2265.544318.26MLPTop quarterC*Contains thin segregation bands
R153-3.7269.54318.26MLPTop halfFMostly segregation material
R154-4.25271.84318.26MLPBottom halfC* 
R155-0.6273.594318.26MLPBottom halfC* 
R156-3.6277.554318.26MLPBottom quarterVFGroundmass altered
R159-7.3287.82474.53MKPMTopFFinely vesicular UVZ
R159-8.5288.19474.53MKPMBottomFFinely vesicular LVZ
R177-5.3332.485810.12MKATop halfVF* 
R178-4.5335.285810.12MKABottom quarterVF*Plagioclase-rich pull-aparts
R233-2.8484.88962.71MKTTop thirdVF*Altered
R233-4.35485.33962.71MKTTop thirdVF*Altered
R233-9.7486.98962.71MKTBaseVF*Finely vesicular LVZ
R399-1.1868.11762.96MKPTopFFinely vesicular UVZ
R399-2.1868.411762.96MKPTop quarterF* 
R399-3.7868.891762.96MKPTop halfM*Contains segregation
R399-5.8869.531762.96MKPBottom quarterC*Dense zone
R463-3.51042.162246.16MKP/TMiddleM*FV top and base
R467-1.41054.122274.75MKPBottom quarterFContains F segregation band

[12] We identified flows by unit boundaries (glassy or rubbly flow tops, weathered zones), although we subdivided some thick units identified in the HSDP logbook to better define the character of individual flows comprising larger compound flow units. We recorded the internal structure and texture of each unit, including internal flow boundaries, vesicle zones, vesicularity, vesicle size and shape, color, and unusual features (such as segregation material). Flow surface morphologies, groundmass textures, and vesicle characteristics provide the primary basis for flow classification. Other useful features include the abundance, form, and spatial distribution of late-stage segregations.

3.2. Classification of Flow Types

[13] Flows are interpreted to be pahoehoe when they preserve glassy crusts and ropy flow tops, and have abundant spherical vesicles. Preserved pahoehoe surfaces are most common in packages of thin flows emplaced during a single eruption (compound flow sequences) (Figure 4a) [e.g., Walker, 1972]. Inflated flows (Figure 4b) are identified by their thickness (commonly >2–3 m), characteristic vesicle distributions [Cashman and Kauahikaua, 1997; Self et al., 1998], and vesicular evolved melt segregations [Goff, 1996; Philpotts et al., 1996; Self et al., 1996]. Thicker flows have crystalline interiors of moderate grain size, in striking contrast to their glassy margins.

Figure 4.

Core photographs. Core diameter is 6.2 cm, arrows drawn on core denote the up direction. (a) Thin pahoehoe flow showing the flow unit boundaries (FUB), finely vesicular zones (FV), and medium-sized vesicles in the flow interior (MV). (b) Inflated pahoehoe flow with finely vesicular upper and lower zones (UVZ and LVZ, respectively) and a dense flow interior. (c) ′a′a flow with a brecciated flow top, dense interior (not all is shown), zone of irregular vesicles, and finely vesicular flow base. (d) Transitional flow with a flow top breccia, finely and moderately vesicular zones (FV and MV), and brecciated flow base.

Figure 4.


[14] ′a′a flows are easily identified by rubbly, brecciated surfaces (Figure 4c), by oxidation and weathering of the upper and lower surfaces, and by entrainment of surface breccia fragments near the flow top [e.g., Crisp and Baloga, 1994]. Other flow features characteristic of ′a′a include vesicle-poor interiors (particularly in thick flows), highly deformed and irregular vesicles, and finely crystalline textures throughout. Thin horizontal segregation bands are common, and may denote planes of late-stage shearing-induced dilation [Smith, 1997, 1998, 2000].

[15] Flows that could not be simply categorized as either pahoehoe or ′a′a are labeled transitional (Figure 4d). Most often, transitional flows have rubbly, brecciated flow tops and bases, like ′a′a, but contain abundant spherical vesicles, like pahoehoe. Unlike pahoehoe, however, vesicles in transitional flows vary greatly in size and number density and are rarely segregated into discrete upper and lower vesicular zones. All gradations exist between transitional and ′a′a, and transitional and pahoehoe, flow types.

3.3. Textural Analysis

[16] We examined 28 samples from the KP-1 core in thin section (Table 1) and selected a subset of 18 thin sections (from seven flows encompassing all three morphologic types) for textural analysis. Backscattered electron (BSE) images, collected with a JEOL 6300 Scanning Electron Microscope (SEM) at 10 keV acceleration voltage, 5 nA beam current, and a working distance of 15 mm, allowed identification of the primary groundmass phases (plagioclase and pyroxene) (Figure 5). Flow textures were quantified through measurement of plagioclase abundance (areal number density and proportion), as plagioclase is easily imaged and represents ∼50% of the groundmass in all flows studied. We optimized image magnification to include 100–250 plagioclase crystals in each image (1000–3000 crystals imaged per thin section). SEM images were enhanced using Adobe Photoshop and analyzed using NIH Image 1.61. Individual crystals were distinguished by grain boundaries and orientation differences. Measured crystal number density (per area, Na) and area fraction (ϕ) define the average crystal size (d) as

display math

[Cashman et al., 1999; Hammer et al., 1999], which can then be used to estimate volumetric number densities (Nv) as

display math

[Underwood, 1970].

Figure 5.

BSE image of a pahoehoe flow interior. Field of view 1.5 cm (horizontal). Px is pyroxene and Pl is plagioclase.

4. Results

4.1. Distribution of Flow Types in SOH-4 and KP-1 Cores

[17] The upper 1000 m of the proximal SOH-4 core contain abundant thin pahoehoe flows and dikes. Of the 370 flow units examined, 67% are pahoehoe (213 single and 35 compound flows), and they represent 56% of the total flow volume (Figure 6a). In contrast, only 25% of the flows are ′a′a, and 8% are transitional. All flows are thin, and flow thickness distributions are approximately log normal for each flow type (Figure 7). The flow thickness distributions show that pahoehoe flows are somewhat thinner than ′a′a flows (the means of the lognormal distributions are 1.1 and 1.86 m, respectively), although the thickness distributions show considerable overlap. Transitional flows are of intermediate thickness (mean 1.26 m).

Figure 6.

Distribution of flow types and thickness as a function of depth in (a) SOH-4 and (b) KP-1 cores. Green squares are pahoehoe, yellow triangles are ′a′a, and red circles are transitional flows. Note log scale on the horizontal axis.

Figure 7.

Thickness frequency distributions of different flow types shown on a log scale. In each histogram, open bars are flows from the SOH-4 core and solid bars are from KP-1.

[18] Flow type distributions in the distal KP-1 core are both similar to, and different from, those in the SOH-4 core. Of 284 flows that we identified, 61% are pahoehoe, 28% are ′a′a, and 11% are transitional (Figure 6b). Thus, the proportion of different flow types is similar in proximal and distal cores. However, pahoehoe flows are substantially thinner than ′a′a flows in the KP-1 core, and represent only 40% of the total volume. Additionally, while the mean pahoehoe flow thickness is similar in both cores, the KP-1 core contains several thick pahoehoe flows, particularly in the upper (Mauna Loa) sequence (note the tail on the frequency distribution in Figure 7). ′A′a flows are thicker than their SOH-4 counterparts, with 20 flows >8 m thick; the mean of the lognormal distribution 4.68 m. Transitional flow units, which are dispersed throughout the core, have mean thickness of 3.31 m, intermediate between pahoehoe and ′a′a.

[19] To examine the temporal distribution of flow types, we plot the flow data using both the frequency (by number) (Figures 8a and 8b) and volumetric proportion (Figures 9a and 9b) of each flow type in 50 m intervals. This thickness interval is arbitrary, and was chosen to provide sufficiently detailed flow sequence descriptions while still revealing patterns of flow variations with depth. Pahoehoe and ′a′a are fairly evenly distributed throughout the SOH-4 core (Figures 8a and 9a), although pahoehoe is more abundant in the middle of the depth range (∼350–550 m). In contrast, flows in the KP-1 core appear clustered, with discrete core intervals dominated by a single flow type (Figures 8b and 9b). Thick pahoehoe flows comprise 72% of the Mauna Loa flow volume, while the underlying Mauna Kea sequence contains only 30% pahoehoe, primarily as inflated flows at depths >900 m. ′A′a dominates the middle section of the KP-1 core (i.e., much of the Mauna Kea sequence).

Figure 8.

Distribution of flow types (by number) with depth in (a) SOH-4 and (b) KP-1 cores. Color coding of flow types is the same as in Figures 5 and 6. The Mauna Loa (ML)–Mauna Kea (MK) boundary in the KP-1 core is marked in purple.

Figure 9.

Distribution of flow types (by volume) with depth in the (a) SOH-4 and (b) KP-1 cores. Colors are the same as in Figure 8.

[20] In summary, a qualitative examination of flow type distribution in the two cores indicates differences that can be related to the emplacement environment. The rift zone setting of the SOH-4 core preserves numerous thin pahoehoe flows and less numerous thin ′a′a flows, flow facies resulting primarily from brief eruptions of Kilauea's ERZ [e.g., Holcomb, 1987]. In contrast, thick (>2–3 m) pahoehoe flows in the upper sequence of the KP-1 core have internal structures characteristic of the tube-fed inflated sheet flows that reflect accumulation on a coastal plain, consistent with the current location of the KP-1 site [Lipman and Moore, 1996]. The prevalence of thick ′a′a flows throughout much of the Mauna Kea sequence is anticipated given the inferred distance of the site from rift zone vents [e.g., Wentworth and Macdonald, 1953; Lockwood and Lipman, 1987].

4.2. Textural Measurements

[21] Preliminary classification of flows in the upper 925 m of the SOH-4 core and the 1055 m KP-1 core allowed unambiguous identification of end-member pahoehoe and ′a′a flow types, but required classification of some flows as transitional. This ambiguity is not surprising given the occurrence of transitional flow types in active flow fields [e.g., Macdonald, 1953; Rowland and Walker, 1987], and in cores from large igneous provinces [e.g., Coffin et al., 2000; Duncan et al., 1996; Eldholm et al., 1987; Larsen et al., 1994]. In Hawaiì, transitional flows are associated with the emplacement of both ′a′a and pahoehoe flow fields, and should ideally be assigned to one of these eruption styles. Textural analysis of active flows suggests that groundmass textures may provide a means of classifying transitional flow types (e.g., Figure 2).

[22] Inspection of thin section images shows that pahoehoe flow margins are sparsely crystalline, with plagioclase and pyroxene crystals in a glassy or dendritic matrix (Figure 10a). Flow interiors are holocrystalline and of moderate grain size (Figures 10d and 10g). In contrast, ′a′a flows have extremely fine-grained margins (Figure 10b), and show only a small inward increase in grain size, regardless of flow thickness (Figures 10e and 10h). Most transitional flow samples that we examined are texturally similar to ′a′a, with fine-grained margins and interiors, even though the spherical vesicles in some samples are pahoehoe-like (Figures 10c, 10f, and 10i). The only exceptions are flows that we classified as pahoehoe/transitional in our examination of the core (flows originally classified as “massive” by the HSDP team) (e.g., unit 63) which had pahoehoe-like coarse-grained groundmass textures.

Figure 10.

Textures of different flow types. Images (a)–(f) are photomicrographs at the same scale (width 4.5 mm) and images (g)–(i) are BSE images (width 1.2 mm): (a) pahoehoe flow top of unit 47, (b) ′a′a base of unit 29, (c) transitional top of unit 96, (d) pahoehoe interior of unit 43, (e) ′a′a interior of unit 58, (f) transitional interior of unit 96, (g) pahoehoe interior of unit 43, (h) ′a′a interior of unit 58, and (i) transitional interior of unit 96.

[23] Textures are quantified by measuring the aerial number densities (Na) of plagioclase crystals in the groundmass (Figure 11). As groundmass plagioclase represents approximately 50% of the total groundmass population in all samples, measured crystal number densities (Na) are a direct measure of variations in average crystal size (cf. (1)) and volumetric number density (Nd) (2). Pahoehoe flow margin textures are difficult to quantify because of their extremely fine-grained, often dendritic matrix, and thus are not included in Figure 11. Pahoehoe flow interiors have relatively low crystal number densities (Na = 100 mm−2; Nv = 1.4 × 103 mm−3), which imply an average crystal size d = 70 μm. A ‘a flows have margins with plagioclase Na ∼ 2000–3000 mm−2 (d = 13–16 μm; Nv = (1.2–2.3) × 105 mm−3). Àà flow interiors have plagioclase number densities only slightly lower than those of flow margins, with Na ∼ 1400 mm−2 (d = 19 μm; Nv = 0.75 × 105 mm−3). The transitional flow samples that we analyzed closely resemble àà in texture, with uniform groundmass textures throughout flow thicknesses, and Na = 1000–3000 mm−2 (d = 13–22 μm; Nv = (0.4–2.2) × 105 mm−3).

Figure 11.

Average plagioclase number densities (mm−2) as a function of normalized depth in each flow (1 is the flow top, 0 is the flow base). Symbols as in Figure 6; samples starred in Table 1. Standard deviations are small relative to symbol size.

5. Discussion

[24] The distinctions between pahoehoe and ′a′a flow types described above allow identification of the two primary flow types in Hawaiian drill cores. This classification, in turn, permits analysis of pahoehoe and ′a′a distributions throughout the proximal SOH-4 and distal KP-1 cores, allowing flow emplacement style to be added as a parameter in models of volcano growth [e.g., Lipman and Moore, 1996; Stolper et al., 1996]. Flow type distribution and thickness are analyzed in the context of known proximity to rift zone vents to provide a basis for the interpretation of flow emplacement conditions in drill cores where vent location and preeruption regional gradient are unknown. Finally, observed variations in flow type with changes in the proximity of the KP-1 site to sea level [Sharp et al., 1996] are examined in the context of observed changes in slope as a function of elevation on modern Hawaiian volcanoes.

5.1. Interpretation of Flow Textures

[25] The observed order-of-magnitude difference in groundmass plagioclase size (a factor of ∼100 difference in volumetric number density) between pahoehoe and ′a′a core samples (Figure 11) is consistent with textural differences described in samples collected from active flows (e.g., Figure 2), and can be explained by differences in cooling and crystallization histories of the two flow types. Inward coarsening of crystal textures in pahoehoe flows results when flows are emplaced hot (with a low crystal content), and subsequently cool conductively, as is common in dikes [Cashman, 1993; Gray, 1978; Ikeda, 1977; Winkler, 1949] and lava lakes [Cashman and Marsh, 1988; Kirkpatrick, 1977; Moore and Evans, 1967]. In contrast, early radiative cooling and stirring of lava in ′a′a channels creates numerous crystal nuclei throughout the flow [Cashman et al., 1999; Emerson, 1926; Kouchi et al., 1986; Sato, 1995], and flows achieve a high crystallinity during emplacement. High initial crystal number densities restrict the final grain size of the solidified flow interiors, thus limiting the effects of postemplacement conductive cooling.

[26] In our experience, flows with transitional or indeterminate morphologies have interior groundmass textures that can be unambiguously classified as pahoehoe or ′a′a, thus proving useful for identifying flows of uncertain origin [e.g., Buecker et al., 1999; Lesher et al., 1999]. Most of the transitional samples that we examined appeared ′a′a-like, with either uniformly fine-grained interiors or bimodal groundmass textures containing both microphenocrysts and microlites. The latter textures may reflect syneruptive (degassing-driven) crystallization, such as that observed during the 1984 eruption of Mauna Loa [Crisp et al., 1994; Lipman and Banks, 1987]. The common textural features of rubbly or slab pahoehoe surfaces and near-spherical vesicles, combined with fine-grained groundmass textures, make it likely that many of the transitional flows are overflows or breakouts from ′a′a channels [e.g., Jurado-Chichay and Rowland, 1995; Cashman et al., 1999]. Based on these characteristics, we assume that most of the flows identified as transitional are from ′a′a flow fields, rather than a distinct lava type [e.g., Keszthelyi and Thordarson, 2000]. However, transitional flows may also form in pahoehoe flow fields as breakouts from temporary ponds or from beneath stagnant tumuli [Peterson and Tilling, 1980]. For this reason, in the analysis below we assume that transitional flows are ′a′a unless they are single flows within a sequence of pahoehoe, in which case we assume them to be pahoehoe.

5.2. Flow Sequences

[27] To evaluate the apparent clustering of flow types, we must first test the hypothesis that the observed sequence of flow types (pahoehoe or ′a′a, with transitional flows assigned as above) is not random. To do this, we use a runs test where we determine the number of runs, or uninterrupted sequences of the same type, at each site [Davis, 1973; Moore, 2000]. From this measurement, we calculate the probability that a given sequence of runs was created by a random process. The expected mean number of runs (μ) in a randomly generated sequence of m items of state 1 and n items of state 2 is

display math

The expected variance of the mean is

display math

and permits creation of a Z test for randomness

display math

where x is the observed number of runs. Using a 5% level of significance, the sequence is random if the observed number of runs falls within the 95% confidence interval of the expected, that is, if −1.96 ≤ Z ≤ + 1.96.

[28] Both holes fail the test for randomness, with Z values of −9.4 for SOH-4 and −10.2 for KP-1 (Table 2). Strongly negative values of Z indicate that there are many fewer runs than expected for a random sequence, and thus that the flows are strongly clustered by type. To insure that our results are not dependent on the correct identification of transitional flows, we did a run test ignoring those flows. This results in somewhat lower Z values of −6.5 and −7.4, respectively, but still indicates strong clustering of flow types.

[29] Flow sequencing can also be measured by examining the length of individual runs. The longest run in SOH-4 is of pahoehoe flows, with 56 flows (or up to 66 if all intervening transitional flows are pasty pahoehoe). These flows span 60 m between 403 and 463 m depth (or 105 m from 403 to 507 m depth). The longest run of ′a′a in this core is 11 flows over 47 m (875–922 m). In contrast, the longest run in the KP-1 core is ′a′a, with 28 flows over 115 m (453–568 m). These flows lie within the main tholeiitic shield-building phase of Mauna Kea, where 86% of the core volume is ′a′a. The longest run of pahoehoe involves 25 flows that span 82 m (210–292 m) across the boundary between Mauna Loa and Mauna Kea. These flows accumulated on a broad coastal plain, and are predominantly inflated pahoehoe flows and accompanying pahoehoe breakouts [Lipman and Moore, 1996]. That they span the Mauna Loa–Mauna Kea boundary suggests that at this location, flow type was influenced by the local environment in addition to magma composition, eruption conditions, and vent proximity. The longest run of transitional flows in either hole is only 4 (<8 m, in KP-1), and lies within the 28 flow run of ′a′a. The short run length of transitional flows lends credence to our designation of transitional flows as associated with either ′a′a or pahoehoe emplacement styles.

5.3. Comparison of Proximal (SOH-4) and Distal (KP-1) Cores

[30] The distribution of flow types in drill cores from large igneous provinces has been used primarily to infer the proximity of a site to feeder vents [e.g., Duncan et al., 1996; Eldholm et al., 1987; Larsen et al., 1994]. In these studies, pahoehoe is considered a proximal facies, and ′a′a representative of distal sites. On Kilauea and Mauna Loa, this distinction works well for the rift zones, but is complicated by extensive tube-fed pahoehoe flows and coastal plains built from inflated pahoehoe [Holcomb, 1987; Lipman and Moore, 1996; Lockwood and Lipman, 1987]. Additionally, very long lava flows are commonly inflated pahoehoe [e.g., Cashman et al., 1998; Coffin et al., 2000; Self et al., 1996, 1998; Stephenson et al., 1998]. Here we examine flow type distributions in the SOH-4 and KP-1 cores, where vent proximity and age distributions are known, to evaluate the uniqueness of flow sequences from proximal and distal sites, and to assess possible controls on flow type distributions.

[31] The SOH-4 and KP-1 drill cores contain similar proportions of pahoehoe (69% versus 63% by number, when transitional flows are reclassified using the criteria above). When classified by volume, however, the flow distributions are quite different, with 60% of the SOH-4 core pahoehoe, compared to 42% of the KP-1 core. The longest run in the SOH-4 core is also pahoehoe. Both the prevalence of pahoehoe flows, and the limited thickness of both flow types in the SOH-4 core, are consistent with expectations for rift zone facies, and suggest that these attributes may be diagnostic of near-vent locations. In contrast, the distal KP-1 core has a high proportion of ′a′a. ′A′a flows are thicker on average than in the proximal SOH-4 core (Figure 7b), and form the longest run (115 m). These attributes suggest that numerous thick ′a′a flows may typify distal locations, with inflated pahoehoe flows recording deposition on coastal plains [Hon et al., 1994; Lipman and Moore, 1996], or other regions of low slope such as saddles between volcanic edifices [Rowland and Garbeil, 2000].

[32] Additional information comes from the distribution of flow types in another rift zone core, SOH-1 [Quane et al., 2000]. The SOH-1 core is located 8 km down rift from SOH-4 at an elevation of ∼200 m, but has a dramatically different stratigraphy. The subaerial flow sequence is 740 m thick, and comprises 215 flow units, of which 62% are ′a′a (76% by volume) [Quane et al., 2000]. Average flow thickness can be estimated using reported number-based and volume-based flow type distributions, yielding ∼4 m for ′a′a and ∼2 m for pahoehoe. Flow sequencing can be analyzed using Figure 2 in the study of Quane et al. [2000]; a runs test shows that the sequence of pahoehoe and ′a′a flows in the SOH-1 core also fails the test for randomness, with an estimated Z of −8.0.

[33] There are two interesting aspects of flow type distributions in the SOH-1 core. First, the dominance of ′a′a in the core is surprising given both the prevalence of pahoehoe 8 km up-rift at SOH-4, and the predominance of pahoehoe on Kilauea's surface [Holcomb, 1987]. One possible explanation is that eruptive vents have been less common in the past on the lower portions of the ERZ (at SOH-1) than at SOH-4. If true, SOH-1 may provide an example of a medial rather than proximal site, consistent with its abundant ′a′a and intermediate flow thicknesses. Second, there is one section of the SOH-1 core (180–380 m) that is dominantly pahoehoe. This pahoehoe sequence may indicate nearby vents during this time period. While it is difficult to match this sequence directly to the flow sequence in SOH-4 because of problems in dating [Guillou et al., 1997; Quane et al., 2000], it is tempting to correlate this interval with a similar abundance of pahoehoe observed between 350 and 550 m in the SOH-4 core.

[34] In summary, flow distributions in the SOH-4 and KP-1 cores generally fit expectations for proximal and distal locations, with the proximal core having more abundant pahoehoe and generally thinner flows than the distal core. Flow types and thicknesses are easily mapped in cores, and may provide important temporal information about processes such as the subaerial seafloor spreading inferred for the North Atlantic Volcanic Province [e.g., Duncan et al., 1996; Larsen et al., 1994]. However, the SOH-1 core illustrates the complexity of relating flow type to local environment, and suggests that environmental interpretations should be based on more than one core locality.

5.4. Controls on Flow Type Distributions in the KP-1 Core

[35] Important controls on the surface morphology of basaltic lava flows are vent proximity, effusion rate, eruption duration, and slope [Holcomb, 1987; Hon et al., 1994; Lipman and Moore, 1996; Lockwood and Lipman, 1987; Peterson and Tilling, 1980; Rowland, 1996; Rowland and Walker, 1990; Rowland and Garbeil, 2000; Wadge, 1978; Walker, 1967]. Of these parameters, only vent proximity is known for the core samples. However, known sea level curves and detailed dating of the KP-1 hole can be used to determine the elevation of the KP-1 site through time [Sharp et al., 1996]. Elevation, in turn, can be used to approximate slope, when compared with data for modern Hawaiian volcanoes. We suggest that variations in elevation (slope) can be used to explain the two most striking features of flow type distributions in the KP-1 core: (1) the clustering of thick (inflated) pahoehoe flows in the Mauna Loa and basal Mauna Kea sequences and (2) the abundance of ′a′a between depths of 350 and 850 m.

[36] To test the possibility that flow types may reflect changes in KP-1 site elevation, we group flows in the KP-1 core into three depth intervals based on the proximity of the site to sea level (Figure 12). Lava accumulated near sea level at drill core depths of 0–280 and 950–1055 m. Flows accumulated in these depth intervals are dominantly pahoehoe (>75% by volume). Additionally, these depth intervals are the only locations in the KP-1 core with thick inflated pahoehoe sheet flows. Lipman and Moore [1996] interpreted the shallow (Mauna Loa) pahoehoe flow sequence to be a thick lava delta that reflected a long-term (∼100 ka) balance between lava accumulation rate, volcano subsidence, and sea level rise. We suggest that early subaerial eruptives of Mauna Kea volcano accumulated in a similar coastal setting. The presence of thick inflated flows provides independent evidence of this site's coastal location at that time (∼400 ka), supporting DePaolo and Stolper's [1996] prediction that the pilot hole nearly reached submarine Mauna Kea (the submarine transition was reached at 1079 m in the main drill hole) [DePaolo et al., 2001].

Figure 12.

(a) Ages of selected samples plotted as a function of depth in the KP-1 core (red dots). The wavy blue line shows local sea level at the drill site, assuming a subsidence rate of 2.6 m/kyr and allowing for eustatic sea level changes inferred from the marine isotope record (modified from the study of Sharp et al. [1996]). (b) Percentage of pahoehoe (green) and ′a′a (yellow) flow types, by volume, over the time increments shown in (a). The inset shows a depth range of 350–850 m and represents the time that the drill site showed its greatest departure from sea level.

[37] Between depths of ∼280 and 950 m, lava flowed into the KP-1 site from Mauna Kea volcano, and rates of lava accumulation outpaced those of sea level rise [Sharp et al., 1996]. The KP-1 site was far from sea level over the depth interval of 355–850 m (380–150 ka), and reached a maximum elevation of 400 m at approximately 320 ka (at a core depth of 416 m). This depth (time) interval has an anomalously high proportion of ′a′a (86%) (Figure 12), and contains the longest continuous run of ′a′a in either core (from 453 to 568 m). As the Hilo site was not only well above sea level but also far from Mauna Kea's summit, emplacement at the drill core site probably required both long transport distances and flow over slopes of 5°–10° [Mark and Moore, 1987; Rowland and Garbeil, 2000]. In summary, moderate slopes and the KP-1 site's distance from eruptive vents produced a preponderance of ′a′a flows on Mauna Kea's SE flank. The subsequent change to pahoehoe in the uppermost part of the Mauna Kea section probably reflects the combined effects of waning rates of lava production from Mauna Kea and changes in depositional environment from resulting sea level rise [Lipman and Moore, 1996].

5.5. The Growth of Hawaiian Volcanoes

[38] When inferences from flow distributions are combined with accumulation rate and sea level curves, the physical development of the Hilo drill core site may be described (Figure 13). Mauna Kea first appeared on the ocean floor at approximately 750 ka and reached sea level before 400 ka [Frey et al., 1990; Moore and Clague, 1992]. Lavas at the base of the KP-1 drill core (380–400 ka) reflect their nearshore (coastal plain) environment by the prevalence of inflated pahoehoe (Figure 13a). An increase in lava accumulation rate led to substantial subaerial edifice construction by 320 ka, and formation of the steeper slopes characteristic of moderate elevations on Mauna Kea [Mark and Moore, 1987; Rowland and Garbeil, 2000]. These steeper slopes accumulated primarily ′a′a flows (Figure 13b). Lava accumulation rates at the Hilo site then slowed as a consequence of the combined effects of young Mauna Loa blocking flows from Mauna Kea's ERZ [Moore and Clague, 1992] and plate movement over the Hawaiian plume [DePaolo and Stolper, 1996]. The reduction in lava accumulation rate allowed reestablishment of balance between subsidence, sea level rise, and lava accumulation, and an increase in the proportion of pahoehoe. By ∼100 ka, inflated pahoehoe flows from Mauna Loa dominated the Hilo site (Figure 13c).

Figure 13.

Cartoon illustrating development of the observed flow type distributions in the Mauna Kea sequence of the KP-1 core. (a) Mauna Kea emerges above sea level and flows at the drill site location accumulate as inflated pahoehoe on a coastal plain. (b) As a consequence of high lava accumulation rates, the drill site emerges above sea level to reach a maximum elevation of ∼400 m. Here, flows erupted from Mauna Kea's rift zone had to traverse moderate slopes. The combination of transport distance and topography probably led to the thick accumulations of ′a′a. (c) The magma supply rate to the drill core site slows as Mauna Kea passes over the plume and emerging Mauna Loa blocks flow from Mauna Kea. Reestablishment of a coastal plain at the site permits accumulation of inflated pahoehoe flows, first from Mauna Kea and later from Mauna Loa.

[39] The drill core data provide vertical (temporal) information on patterns of flow morphology coverage that can be used in conjunction with surface mapping to understand the evolution of Hawaiian volcanoes. Kilauea has low slopes (<4°, except in the region of the Hilina Pali) [Rowland and Garbeil, 2000] and is surfaced largely by pahoehoe [Holcomb, 1987]. Mauna Loa has steeper slopes (>5° and up to 10° at higher elevations) [Rowland and Garbeil, 2000], and proportionately more ′a′a [Lockwood and Lipman, 1987]. Comparison of these two volcanoes, when combined with data from the KP-1 core, suggests that as Hawaiian volcanoes grow, slopes increase, particularly along rift zone vents [e.g., Moore and Mark, 1992]. This slope increase leads to a progression from largely pahoehoe to largely ′a′a flow surface morphologies. Thus, the high abundance of pahoehoe surfacing of Kilauea relative to Mauna Loa may reflect the lower average slopes, and younger age, of Kilauea. A similar progression from early pahoehoe to late ′a′a has been recognized on Fernandina Volcano in the Galapagos [Rowland, 1996], although Rowland interprets this pattern to reflect an increase in eruption rate through time. That eruption rate was not the only factor in 'a'a formation on Mauna Kea is suggested by the great range in average accumulation rates (shown by the changing slope of the accumulation rate curve in Figure 12) over the depth interval in which 'a'a is the dominant flow type. Interactions between slope-forming processes and the effect of those slopes on flow morphology may thus be important in the integrated development history of basaltic volcanoes.

[40] Tracking flow emplacement style through the growth histories of Hawaiian volcanoes may also aid assessment of future volcanic hazards for the island of Hawaiì [e.g., Wright et al., 1992]. Current volcanic hazard assessment relies on historical data and digital maps of flow coverage to determine both the probability that a single lava flow will enter a particular region, or the probability that a single point will be covered by lava flows in a given period of time [Kauahikaua et al., 1995]. However, as 'a'a and pahoehoe flows have different styles of flow advance and different patterns of local surface coverage, hazard assessments for specific locations may be improved by consideration of the type of flow expected.

6. Conclusions

[41] Interpreting conditions of lava flow emplacement over space and time is crucial for assessing lava flow hazards, for understanding the evolution of basaltic shield volcanoes, and for deciphering the history of large igneous provinces. Our results provide some methodologies for analyzing flow sequences in cores, and for interpreting those sequences in the context of historic Hawaiian eruptions.

[42] End-member pahoehoe and 'a'a have not only distinctly different internal structures [e.g., Canon-Tapia et al., 1996; Cashman and Kauahikaua, 1997; Polacci and Papale, 1997; Self et al., 1998; Wilmoth and Walker, 1993] but also distinctly different groundmass textures that can be explained by their different transport and cooling histories. Thus, analysis of groundmass textures may allow identification of flow morphology even when surface features are not preserved, or when flow characteristics are ambiguous.

[43] Our data indicate that flow types are clustered in both core sequences, and that both flow type and thickness distributions vary with vent proximity. The proximal SOH-4 core is dominated by long sequences of thin pahoehoe flows, punctuated by limited thin 'a'a flows. Both thin flows and abundant pahoehoe are anticipated for the rift zone setting of SOH-4 [e.g., Holcomb, 1987]. In contrast, thick 'a'a flows are volumetrically dominant in the KP-1 core and form much of the Mauna Kea sequence, with thick sequences of pahoehoe found only in the uppermost (Mauna Loa) and lowermost (early Mauna Kea) core sections. Generally thicker flow units and an increase in 'a'a abundance are consistent with the distal setting of this site. While limited, these data suggest that the combined use of flow type and thickness characterization may help to establish vent proximity in cores where the vent location is not known.

[44] Observed flow type sequences provide information on flow emplacement conditions. Lipman and Moore [1996] ascribe the abundance of inflated pahoehoe in the younger Mauna Loa sequence to a prolonged balance between the relative rates of sea level rise, island subsidence, and lava accumulation in the Hilo area. We suggest that the same is true for the lower part of the Mauna Kea sequence. In contrast, 'a'a flows predominate during time periods when high rates of lava accumulation relative to sea level rise raised the site well above sea level. Observed changes in dominant flow type probably reflect the combined effects of long transport distances and increased slopes. A dependence of surface morphology on topography has long been recognized [e.g., Macdonald, 1953], but has not yet been incorporated into general models of volcano evolution. Our work suggests that slope increases during edifice construction should lead to an increasing abundance of 'a'a through time, and may partially explain spatial and temporal patterns of flow distribution on basaltic volcanoes in Hawaiì [e.g., Holcomb, 1987] and the Galapagos [e.g., Rowland, 1996]. Analysis of flow sequences in drill cores thus constrains aspects of the local environment (both vent proximity and slope) that may aid both in interpreting the evolution of ancient basaltic flow fields and assessing the future volcanic hazards of active volcanoes.


[45] We gratefully acknowledge the help of Ed Stolper and Mike Baker with our examination of the KP-1 core at Caltech and of Jim Kauahikaua for field constraints on flow interpretations. Reviews by D. Pyle and L. Keszthelyi greatly improved the manuscript. This work was supported by NSF EAR9508144 and by USSSP funding.