Revealing emplacement dynamics of a simple flood basalt eruption unit using systematic compositional heterogeneities
Published:February 07, 2019
- PDF LinkChapter PDF
C. Vye-Brown*, T.L. Barry†, S. Self§, 2019. "Revealing emplacement dynamics of a simple flood basalt eruption unit using systematic compositional heterogeneities", Field Volcanology: A Tribute to the Distinguished Career of Don Swanson, Michael P. Poland, Michael O. Garcia, Victor E. Camp, Anita Grunder
Download citation file:
Intralava geochemical variations resulting from subtle changes in magma composition are used here to provide insights into the spatial-temporal development of large basalt lava flow fields. Recognition that flood basalt lavas are emplaced by inflation processes, akin to modern pāhoehoe lava, provides a spatial and temporal frame-work, both vertically at single locations and laterally between locations, in which to examine lava flow emplacement and lava flow field development. Assuming the lava inflation model, we combined detailed field mapping with analysis of compositional profiles across a single flow field to determine the internal spatio-temporal development of the Palouse Falls flow field, a lava produced by an individual Columbia River flood basalt eruption.
Geochemical analyses of samples from constituent lobes of the Palouse Falls lava field demonstrate that systematic compositional whole-rock variations can be traced throughout the flow field from the area of the vent to the distal limits. Chemical heterogeneity within individual lava lobes (and outcrops) shows an increase from lava crusts to cores, e.g., MgO = 3.24–4.23 wt%, Fe2O3 = 14.71–16.05 wt%, Cr = 29–52 ppm, and TiO2 = 2.83–3.14 wt%. This is accompanied by a decrease in incompatible elements, e.g., Y = 46.1–43.4 ppm, Zr = 207–172 ppm, and V = 397–367 ppm. Systematic compositional variations from the source to distal areas are observed through constituent lobes of the Palouse Falls flow field. However, compositional heterogeneity in any one lobe appears less variable in the middle of the flow field as compared to more proximal and distal margins. Excursions from the general progressive trend from vent to distal limits are also observed and may reflect lateral spread of the flow field during emplacement, resulting in the juxtaposition of lobes of different composition.
Transport of magma through connected sheet lobe cores, acting as internal flow pathways to reach the flow front, is interpreted as the method of lava transport. Additionally, this can explain the general paucity of lava tubes within flood basalt provinces. In general, flow field development by a network of lava lobes may account for the occurrence of compositionally similar glasses noted at the proximal and distal ends of some flood basalt lavas.
Continental flood basalts record some of the largest volcanic eruptions on Earth and require the accumulation of large volumes of eruptible magma. Typically, these volcanic provinces are constructed of extensive (~104–105 km2) pāhoehoe lava flow fields (Self et al., 1997; Bondre et al., 2004; Bryan et al., 2010). The mechanisms of lava flow field development have important implications for understanding the timing and durations of volcanism (Self et al., 1997, 1998), as well as any resultant climatic or environmental impacts from these voluminous basaltic provinces (e.g., Thordarson and Self, 1996; Chenet et al., 2008; Self et al., 2015). However, absolute time scales and periodicity of individual eruptions are presently beyond the precision of dating methods (Barry et al., 2010). Broad, province-scale characterization of flood basalt flow fields requires detailed correlations of the physical features and parameters of lava, e.g., lobes, lava transport pathways across individual eruptive units, as well as diagnostic geochemical signatures of multiple lavas through the province succession, to identify individual eruptive units and constrain flow emplacement processes (e.g., Vye-Brown et al., 2013a).
Voluminous lava flow fields in flood basalt provinces show many physical characteristics similar to smaller, historical lava flows and pāhoehoe (Hon et al., 1994; Self et al., 1997). Structural and morphological evidence for an inflation mechanism of emplacement includes compound lavas with thick crusts, massive cores, and internal vesicular layering (Hon et al., 1994; Self et al., 1998; Thordarson and Self, 1998; Bondre et al., 2004). Such evidence results from the endogenous growth of each lava lobe and would have enabled the propagation of lava through insulated pathways to new lobes at an advancing flow front. By this mechanism, lavas inflate and therefore thicken as cooling accompanies progressive emplacement of magma between brittle upper and lower crusts (Hon et al., 1994; Self et al., 1996, 1998). Later-emplaced lava within a lobe forms the massive, central zone, or core, which commonly cools to display columnar jointing. Several additional lines of evidence supporting emplacement of the majority of lobes in a flood basalt flow field by inflation include: anisotropy of magnetic susceptibility (AMS; Cañon-Tapia and Coe, 2002); quantitative fluid dynamic and thermal constraints (Keszthelyi and Self, 1998); and within-lava geochemical variations (e.g., Philpotts, 1998; Maclennan et al., 2003; Reidel, 1998, 2005; Passmore et al., 2012; Vye-Brown et al., 2013b). These features have been applied to quantitatively assess the emplacement style of some extensive flood basalt and historical lava flows to reveal that variations in lava character can be related to the spatiotemporal development of a lava flow. In particular, compositional variation in magma output during an eruption should be systematically recorded vertically and laterally within a flow field due to the nature of inflation as an emplacement mechanism. Here, we explore the potential insights gained from examining compositional variations within the products of a single, large-volume basalt eruption.
An inherent problem in associating observed geochemical variations to an eruption sequence in flood basalt lavas has been the lack of a single, suitably well-defined eruption unit or flow field. Only a few studies have attempted this (e.g., Martin, 1989, 1991; Vye, 2009). The difficulty lies in unambiguously identifying the constituent lobes of a single flow field within an apparently monotonous succession of similar-looking basalt lavas. To resolve this, a well-established stratigraphy, and a program of detailed mapping and logging are needed. Having physically identified a traceable single flow field over a wide area, principles of the inflation model can be applied to investigate its chronological development (Vye-Brown et al., 2013a). The relative temporal relationship within individual lobes can be extrapolated from single lobes to an entire flow field. For example, in general, more distal lobes are likely to be emplaced later than proximal lobes, and the cores of lobes may be synchronously linked by a molten core. However, this does not take into account potential complexities caused by anastomosing lava flows, which may result in more distal lobes forming prior to proximal ones (Vye-Brown et al., 2013a). The model thus relates chronologic development within a lobe to the emplacement of adjacent lobes. Therefore, physical characterization of a flow field offers the opportunity to further investigate intralava geochemical variation and relate any geochemical signature to emplacement of the flow field as a tool for investigations elsewhere. Here, we compared the relatively simple, well-constrained physical emplacement sequence of the Palouse Falls flow field of the Wanapum Basalt in the Columbia River Basalt Group with compositional heterogeneity recorded in its constituent lobes.
Palouse Falls Flow Field
The Basalt of Palouse Falls (hereafter Palouse Falls) could be termed a simple flow field, typically consisting of just one lobe at each observed location, consistent with previous use of the term “simple” lava (Walker, 1971). The Palouse Falls is the oldest lava field of the Frenchman Springs Member (Beeson et al., 1985). The stratigraphic position of the Palouse Falls makes it relatively easy to identify in the field; in many locations, it marks the first lava of the Wanapum Basalt following a significant hiatus commonly marked by a widespread saprolite horizon (temporally equivalent to the Vantage interbeds). In some proximal locations, the Palouse Falls overlies petrographically distinct Eckler Mountain lavas (Tolan et al., 1989). The Palouse Falls is one of the smallest-volume flow fields of the Columbia River Basalt Group, at only 233 km3 (Martin et al., 2013), and it is calculated to have been emplaced over a minimum of 19 yr (Vye-Brown et al., 2013a). It is typically sparsely phyric (<10%), with small, tabular, and equant plagioclase phenocrysts up to 5 mm long. The flow field was traced along near-continuous exposure over ~30 km along the banks of the Snake River from the presumed vent area near Palouse Falls (approximate position based on the lack of more eastern-lying exposures; S.P. Reidel, 2009, personal commun.) to Lower Monumental Dam (46°39.828ʹN, 118°13.377ʹW). Individual lobes vary in size from 58 m thick in the proximal area to just 2 m thick at the southern margin of the flow field (Fig. 1). There, the flow field consists of two overlying lobes and is only 4 m thick in total. Lobes studied in borehole cores from the distal reaches of the flow field, in the Pasco Basin (~70 to ~85 km from the presumed vent area), are up to 50 m thick. Significant thinning of the flow field occurs with increasing distance from vent, apart from within the Pasco Basin fill. There, ponding of lava likely occurred in the topographic depression of the Pasco Basin. Upper crusts of the sheet lobes typically exhibit abundant vesicular horizons and multiple megavesicle horizons (dome-shaped voids with flat floors and arched to dome-shaped roofs, with dimensions ranging from several to tens of centimeters, floored by moderately vesicular to nonvesicular glassy segregated material; Thordarson and Self, 1998). Lobe cores are typically separated from the upper crust by a zone of horizontal jointing. Cores are well jointed, with either a radial or hackly jointed section in the center. Chatter marks occur on some of the thick columns (>80 cm diameter). Individual sheet lobes within the Palouse Falls flow field range from <1 to ~4 km long and on average cover an area of ~4–5 km2. The total areal extent of the Palouse Falls is 10,495 km2 (Martin et al., 2013), implying that there may be at least 2000–2600 constituent lobes, as this flow field is largely a single layer. In this study, we characterized, in detail, seven of the constituent lobes to indicate a minimum range of physical and compositional variations between lobes.
Six vertical sections representing seven different sheet lobes were logged and sampled, of which five sections occur along a near-continuous outcrop, from a source-proximal position to the distal reaches of the Palouse Falls flow field (Fig. 1). One additional lobe was logged from borehole core records from the Pasco Basin, but it was not sampled (gray log in Fig. 1). Samples were taken at intervals within each vertical section, representing a single lobe. The intervals typically ranged up to 10 m according to the physical features that characterized each locality (Fig. 2). High-intensity sampling was conducted within one lobe to ensure no bias was introduced by the sampling interval (PF_4, Fig. 1). The physical lava features (Fig. 2) are summarized on compositional plots so that compositional variations can be considered relative to structures resulting from the inflation process.
The average sample size collected was 500 g, and altered rock was avoided. Samples were split to remove weathered or joint faces and powdered in an agate mill. Sample sizes of at least 300 g were used to produce a homogeneous powder representative of the whole-rock sample. The quenched, glass-rich Palouse Fall samples are finely crystalline in nature, which further minimized sample heterogeneities, ensuring analytical reproducibility.
Major-and trace-element analyses are presented for 42 samples from the seven outcrop sections (Fig. 1; see also supplementary data1). Major-and trace-element analyses were conducted on whole-rock powders, using the X-ray fluorescence (XRF) technique, at the Open University as described by Potts et al. (1984). Both major-and trace-element analyses were carried out using an ARL 8420+ dual goniometer wavelength-dispersive XRF spectrometer. The heterogeneity index (HI) was applied to quantify the significance of compositional heterogeneity within each sample suite (Table 1; Rhodes, 1983; Rubin et al., 2001). Precision and accuracy were calculated by repeat runs of U.S. Geological Survey (USGS) standards BHVO-1 and WS-E, as well as study samples DCy5 and FSc7 (Table 2). Furthermore, a suite of samples was analyzed at the Washington State University laboratories to enable cross-laboratory comparisons between samples of Columbia River Basalt Group results from the Open University and those reported in the literature (Table 3). Correlation coefficients for this comparison were a good fit at R2 = 0.9991 for all the major and trace elements used in this study.
Chemostratigraphic subdivisions of the Columbia River Basalt Group at formation and group levels are well established (Hooper, 2000). Plots of TiO2 versus P2O5 emphasize the principal differences between the Columbia River Basalt Group formations, such as the “Ti gap” separating the Wanapum Basalt from older formations (Fig. 3; Siems et al., 1974). While further plots such as SiO2/K2O, Zr/Sr, SiO2/P2O5, and Cr/TiO2 have been used to subdivide some formations to the level of individual flow fields, it is acknowledged that eruption units of the Wanapum Basalt are difficult to subdivide on this basis (Hooper, 2000; Martin et al., 2013). Compositional variability greater than analytical error may be problematic for stratigraphic correlations across the Columbia River Basalt Group based on chemistry alone. The results of this study reveal that the composition of the Palouse Falls lava lies at the lower end of the compositional range for the Wanapum Basalt for TiO2 and P2O5, with values approaching the “Ti gap” separating the Grande Ronde and Wanapum Basalts (Fig. 3).
Palouse Falls Intralobe Variation
Intralobe compositional variations within a single flow field can be assessed by analyzing suites of samples taken at varying heights within lobes of that field. Alongside detailed logging, any compositional variation can be compared to the physical structure of the lobe. Intralobe geochemical variation can occur vertically (e.g., Fig. 4), but it also can occur laterally across the constituent lobes of the flow field. Tie lines between sample points have been used to more-readily illustrate the vertical intralobe variation and to identify core and crustal zone divisions between logged sections. However, these plots are not a reflection of the absolute compositional variability, and the plots may have a more steplike appearance at zone boundaries if the sampling frequency was greater. Segregation features within the Palouse Falls were not sampled as part of this study. This was done to avoid known localized perturbations to bulk composition (Goff, 1996; Hartley and Thordarson, 2009).
Vertical Geochemical Variations
Generally, within individual lobes, there are distinct variations in some oxides and compatible elements. Some of the concentrations increase from lava crusts toward the cores, e.g., MgO = 3.24–4.23 wt%, Fe2O3 = 14.71–16.05 wt%, Cr = 29–52 ppm, and TiO2 = 2.83–3.14 wt%. Where this occurs, there is similarly a decrease in incompatible elements, e.g., Y = 46.1–43.4 ppm, Zr = 207–172 ppm, and V = 397–367 ppm. Variation in some elements (e.g., Cu, Ni, Th, Pb, and Ga) within individual lobes is rarely reconcilable outside analytical error.
Within these data, it is assumed that the most vent-proximal section is within a sheet lobe exposed in the lower Winn Lake Canyon (PF_1, Fig. 1). Relative to other sampled lobes, the Winn Lake Canyon section displays a highly variable composition (Fig. 4). Sequential decreases in some incompatible elements and oxides occur from the lobe crust toward the core (e.g., TiO2, Al2O3, and V). This is accompanied by increases in some compatible elements (e.g., Fe2O3, Sr, Cr, and MnO). It is apparent that the upper crust to core boundary, and the lower contact of a finely jointed area of the lobe core, correlate with changes in the compositional profile. Overall, there appears to be a shift from slightly more-evolved upper and lower crusts to a less-evolved lobe core.
The second section away from the assumed vent area is a sheet lobe exposed along the Snake River (PF_2, Fig. 1). This lobe reveals fewer oscillations in the compositional profile than PF_1 (Fig. 5). This may be the result of larger sampling intervals; only one sample was acquired from the lobe core in this section, because the core is massive, lacking the finely jointed zone so clearly displayed in PF_1. However, despite the low sampling frequency, there is a similar apparent progressive decrease and increase from crust to core in the incompatible and compatible elements, respectively, as in section PF_1. In some elements, the section reveals limited variation outside of 98% certainty for analytical error (Fig. DR-A [see footnote 1]). PF_2 is the first occurrence in the Palouse Falls flow field where many of the samples are within analytical error of each other. This is reflected in the calculated heterogeneity index (HI; Table 1).
The next sampled lobe, 22 km away from the assumed vent area, is PF_3 (Figs. 1 and 2), and it shows a transition in composition between the lobe core and a finely vertically jointed zone. Inflections in chemical composition into this zone are similar to the middle of the lobe core in PF_2, although no finely vertically jointed horizon was noted in PF_2 (Fig. 5). Compared to the finely jointed zone, samples to either side differ from it, although they are compositionally similar to one another for many elements. The signature of the upper crust suggests a progressive increase in compatible elements (e.g., V, Sr, MgO, Na2O) toward the lobe core, with either gradational or stepped decreases in incompatible elements (Fig. DR-B [see footnote 1]).
The next section is 30 km away from the assumed vent area, near Lower Monumental Dam (PF_4, Fig. 1). Here, samples were collected at ~2 m intervals to reduce the possibility of any bias imposed on the compositional profiles by sample preference (Fig. DR-C [see footnote 1]). The resulting compositional profiles give the most detailed characterization of intralobe variation in this study. Clear positive and negative infl ections in incompatible and compatible elements, respectively (with few exceptions), are present at the crust to lobe core boundary and at the finely jointed zone noted in the middle of the lobe core. The lower-most contact of the lobe was not exposed, although the upper part of the lower crust was accessible. Omission of the lowermost sample(s) is reflected in the restricted compositional range of the lower crust by comparison with other lobes in the Palouse Falls flow field (e.g., PF_2, Fig. 5).
The furthest outcrop exposure from the assumed vent area is the Ginkgo dike area (PF_5, Fig. 1), ~40–50 km from source. The chemical profiles of this lobe (Fig. DR-D [see footnote 1]) are similar to those from Snake River profile, PF_2 (Fig. 5). While care was taken not to overinterpret profiles with fewer sampling points, variations with height within this lobe bear similarities to PF_2, such as systematic enrichment/depletion of elements and oxides (e.g., SiO2, MgO, CaO, Zr) rather than the stepped compositional variation seen in the intervening lobes.
The final section sampled in the distal reaches of the Palouse Falls flow field was borehole DC8 from the Pasco Basin (PF_7, Fig. 1). A distinct contrast emerged when this logged section was compared with the exposed sections along the Snake River (PF_1–5). Other than the principal core to crust divisions, few of the physical characteristics identified in outcrop were observable within the borehole lobe. However, the lobe sampled within the borehole was confirmed to belong to the Palouse Falls flow field, because it overlies the Vantage interbeds, does not display any of the petrographically distinct features of the Eckler Mountain Basalt, and is overlain by the plagioclase-phyric Ginkgo flow field. Progressive increases in the abundance of oxides and some trace elements (e.g., MgO, CaO, Sr, and Cr) occur with height from the basal core to the upper crust in conjunction with progressive decreases in the abundance of Zr, P2O5, Y, and V (Fig. DR-E [see footnote 1]).
Interlobe Compositional Variations in the Palouse Falls Flow Field
If we compare the analytical results from lobes of the entire flow field, using each sampled lobe as a single locality, the range of values in proximal lobes is greater than in distal lobes, e.g., MgO ranges up to 0.75 wt%, with reduced variation in the middle of the field (Fig. 5). Compositional heterogeneity between adjacent lobes and across the flow field reveals progressive variations with distance from the vent. Average values for samples from: (1) within lava cores, (2) within lava crusts, and (3) relative to the total range of sample values for the whole flow field demonstrate similar patterns in the mean compositional ranges. There are decreases in compatible elements (e.g., Cr = 52–32 ppm) and indices of fractionation (Mg number = 35.8–31.8), as well as increases in incompatible elements (e.g., TiO2 = 2.87–3.12 wt%) with increasing distance from the assumed vent area (Fig. 6).
Lateral Geochemical Variations
In addition to vertical compositional variation, there is also the potential for lateral local compositional variations within a lobe, as shown from a sampling suite from the Sand Hollow flow field, outcropping on the State Highway 26 at 46°46.821′N, 118°05.603′W (Fig. 7). Three horizons were sampled on either side of the core to upper crust transition zone, with three samples taken at 1 m intervals laterally at each horizon. The results show variation outside of analytical error for some elements. This suggests that significant compositional variation exists within the same laterally continuous horizon, and variation may also exist within individual samples. However, in this case, the mean average analysis at each horizon was generally distinct from the sample ranges vertically.
Vertical Small-Scale Geochemical Variations
Small-scale sampling suites, consisting of closely spaced samples, offer an opportunity to investigate the degree and source of variation. Further, it is important to assess the scale at which heterogeneity correlates with emplacement features. Compositional profiles from high-resolution sampling at 20 cm intervals in vesicle-rich to vesicle-poor bands within the complex upper crust of a lobe from the Grande Ronde Basalt at Lyons Ferry Marina (46°35.174′N, 118°13.345′W) show similar characteristics to whole intralobe profiles (cf. Figs. 4 and 8). Degassed vesicular bands have slightly less-evolved compositions than the overlying nonvesicular bands within the upper crust, e.g., MgO = 4.95 wt% in vesicular band 1 as compared to 4.60 wt% in nonvesicular band 1. This suggests that the process(es) responsible for the heterogeneity occurs at various scales. Similar small-scale sample investigations at 10 cm intervals from the lower crust into the core of a lobe from the Sand Hollow flow field (within the Wanapum Basalt, at 46°40.032′N, 118°13.380′W) revealed a range of compositions as extensive as those observed in the entire lobe (Fig. 9). The degree of heterogeneity in this small zone is outside analytical error for most elements, but the high-resolution sampling does not provide a greater insight to the cause of such variable compositions.
Origin of Compositional Variability
On all scales, the major-and trace-element compositional variations in the Palouse Falls flow field lobes appear to be coupled to intralava volcanological features. Such heterogeneity in lavas in other provinces has been attributed to either random variation (Lindstrom and Haskin, 1981), or a variety of causative processes, including crystal accumulation (Philpotts et al., 1999; Philpotts and Philpotts, 2005; Passmore et al., 2012), surface mixing and thermal erosion during emplacement of the lava (Reidel and Fecht, 1987; Reidel, 2005; Hooper et al., 2007), and weathering (Wimpenny et al., 2007). There is no evidence for crystal accumulation within the Palouse Falls flow field due to the uniformly fine-grained, quench-cooled texture of the basalt. Lobe growth by either surface mixing or thermal erosion of disparate flows is also precluded because features found on lobe margins and surfaces, such as glassy selvages and pāhoehoe ropes, support emplacement of the lava as a single cooling unit. The effects of weathering may result in compositional profiles in a vertical section through a basalt lava, where weathering intensity decreases with depth from the surface and/or is amplified in highly vesicular zones. Such variability would be preferentially seen within mobile elements, e.g., Ca, Mg, Na, Ba, and lacking in relatively immobile elements, e.g., Ti, Nb, Zr (e.g., Nesbitt and Wilson, 1992), but no such patterns are seen within the Columbia River Basalt Group results here. We propose that the semisystematic variation in the lobe profiles is a result of magmatic heterogeneity (Rubin et al., 2001) rather than any postemission processes.
Mass balance calculations were run using PETROLOG software (Danyushevsky, 2001) to assess the extent to which the compositions observed can be explained by magmatic processes. The calculations assumed a volatile content of ~0.3 wt% H2O, which is in agreement with values for Columbia River Basalt magmas (Thordarson and Self, 1996; Hartley and Thordarson, 2009) and olivine-plagioclase-clinopyroxene phases for the fractionation assemblage. The first-run calculations used the least-evolved composition with the highest MgO% for the Palouse Falls flow field. With such parameters, the most-evolved sample compositions can be attained within 5% fractional crystallization. A second run used an average composition for the Imnaha Basalts, which are suggested to represent the most plume-like composition of the Columbia River Basalt Group (e.g., Hooper, 2000). Application of the model, using parameters identical to those used in the first run, revealed that between 34% and 46% fractional crystallization of an Imnaha magma produces the Palouse Falls compositions. With either of these run outcomes, there is scatter in the composition of the samples from the flow field results relative to the model results, suggesting that processes other than simple fractional crystallization were involved in generating the compositional profiles. However, identification of responsible process(es) is difficult to resolve within the compositional range. Results from osmium isotopes within another eruptive unit in the Columbia River Basalt Group, the Sand Hollow flow field, support variable degrees of crustal contamination of magma erupted to form a single flood basalt flow (Vye-Brown et al., 2013b).
Intralobe compositional variations appear to be coupled to physical, emplacement-related features, both within individual lobes and with distance from source across the flow field. Thus, intralobe variations record pre-emission magmatic compositional differences. Consideration of the features inherent to inflated sheet lobes such as vesicle-rich horizons and massive cores suggests that systematic variations from more-evolved lobe crusts to less-evolved lobe cores, as well as more-evolved compositions at the distal edges of the flow, reflect the presence of discretely different, compositionally distinct magmas that were available at the same time. The currently available data do not allow us to distinguish between the existence of more than one distinct magma type or a compositional gradational spectrum within a single magmatic body. However, we can identify that the eruption of the Palouse Falls magma must have occurred from either a stratified, periodically replenished, magma chamber or a network of separate chambers deepening toward more primitive compositions with time, sequentially tapped during the ongoing eruption. We now consider how this magmatic compositional variability can be used as a tool to map out the emplacement of large flow fields.
Implications for Emplacement Models
The physical similarity between voluminous flood basalts and modern active and historical basaltic lavas results from the emplacement style (Self et al., 1996; Ho and Cashman, 1997; Keszthelyi and Self, 1998; Thordarson and Self, 1998; Waichel et al., 2006; Passey and Bell, 2007). The variation in volume by several orders of magnitude between lava flows such as those on Hawai‘i or Iceland and continental flood basalts has given rise to speculation about whether emplacement models for small-volume lavas (<1 km3) can be applied as analogues for large-volume lavas (up to 6000 km3; Reidel and Fecht, 1987). However, the similarities in morphologies, surface features, and internal zonation of pāhoehoe sheet lobes and inflated pāhoehoe sheet lobes in Hawai‘i (Hon et al., 1994; Self et al., 1996) have led to an increasing recognition of pāhoehoe inflation as an important process in emplacement of flood basalts. While we may consider, on this basis, that the vertical growth of a lobe in any one location is understood, the degree of connectivity between lava lobes remains poorly constrained, along with other questions about how the intralobe lava flow pathways facilitate the formation and growth of new lobes, and what connective pathways may look like over time.
Compositional Evidence for Flow Field Development
The results presented here record systematic compositional variation both vertically and horizontally within a large-volume flood basalt pāhoehoe flow field. Decreases in compatible element abundance with enrichment in incompatible elements from the crusts to the cores of individual lobes appear to be fairly consistent throughout the flow field. In some localities, there is limited compositional variation within a lobe (e.g., PF_2). Here, the composition of the lava that was initially emplaced varies little (within analytical error) from subsequent magma that was injected into, and inflated, the lobe. In addition to the vertical changes in chemistry, there is also a lateral correlation between lobes, with slightly less-evolved crust compositions found at greater distances from vent. This supports the field evidence for sequential emplacement of lobes from the vent to the distal reaches of the flow field, accompanied by a shift in magma composition. The evidence corroborates the intralobe temporal relationship of less-evolved melts emplaced later in the eruption, represented in the lobe cores.
The distribution of compositional variations within the Palouse Falls flow field supports laminar lava flow emplacement through a connected network of lobe cores. Such a pattern corroborates lobe cores forming from the last emplaced lava within each profile. Furthermore, it is likely that the most distal lobe core was emplaced before all the more proximal lobes had become stagnant (with an infill of molten magma) and crystallized. This is significant, because the observed near-constant lobe core thicknesses are a result of the mechanism of emplacement (Vye-Brown et al., 2013a), and the cores of proximal lobes are interpreted to act as feeders for more distal lobes.
Lava Transport Methods
The transport of magma to the propagating front of the flow field through lobe cores could be sheet-like, through lava tubes, or through a preferential series of pathways within connected lobe cores. There is a notable absence of lava tubes within the Palouse Falls flow field and within flood basalt provinces generally (e.g., Kauahikaua et al., 1998; Bondre et al., 2004). Radiating joints within the elliptical masses of lava (e.g., Rosalia flow field in The Dalles area of the Columbia River province, 45°41.911′N, 121°23.693′W) have been proposed to be similar to lava tubes observed in flow fields on Hawai‘i (Waters, 1960; Greeley, 1971; Greeley et al., 1998; Halliday, 2002). Laboratory studies and field data indicate that lava tube development results from an increase in frictional resistance to lava flow as the distance between cooled crusts is reduced due to advance of the solidification fronts. According to the documented evolution of tubes on Kīlauea Volcano, Hawai‘i (Zablocki, 1978; Hon et al., 1994), the resulting decrease in cross-sectional area increases the lava flow velocity, and lava crust growth is retarded due to the influx of hot lava. This produces well-developed tubes on time scales suggested to be within 2–4 wk of sheet lobe formation (Hon et al., 1994). The rapid transport of lava through tubes, or sheet lobe cores acting as internal flow pathways, may account for the occurrence of compositionally similar glasses at the proximal and distal ends of some Columbia River Basalt Group flow units (Swanson et al., 1975; Mangan et al., 1986). However, tubes may be misidentified elongated or channel-confined lobes, because such features are present but scarce in flood basalt provinces. Propagation of lava through connected sheet lobes remains a preferred method for flow field development (e.g., Self et al., 1997, 1998; Thordarson and Self, 1998; Keszthelyi et al., 2006).
In detail, flow field formation must be highly complex. However, during the emplacement of the last distal lobe, there must be at least one active proximal near-vent lobe and a series of linked lobes all the way through the flow field in order to transfer the same-composition magma through the lobe cores and into the most distal sheet lobe core. Thus, based on our inflation model, preferential pathways of lava flow through lobe cores may be expected to have similar chemistry. Progressive variations in composition from the source to distal areas would be observed through the constituent lobes of the flow field over time. Excursions from this general trend may reflect lateral spread of the flow field in addition to longitudinal development. This would result in adjacent lobes within a flow field exhibiting differing compositional variations, which is observed in the Palouse Falls flow field.
Developments of the Inflation Model
Our results show that there are significant variations in the character of the compositional profile of any single vertical section through a lobe. Such variability may be further affected by: (1) the very outermost few centimeters of the upper and lower crusts often not being sufficiently well preserved to provide samples for analysis; and (2) the possibility that the upper crust may not be the compositional mirror of the lower crust due to inhibited development of the lower crust during lobe emplacement and thickening (see Keszthelyi and Self, 1998). The latter feature may also be affected by rafting of cooled sections of the upper crust on magma flowing within the core. Furthermore, variation in the lower crust may be affected by variations in the flow dynamics at the base of the lava core, including either thermal or mechanical erosion; turbulence caused by topographic variations in the substrate; and shear deformation of crystal lattice caused by flow of overlying lava. While thermal erosion may be theoretically possible, it appears that such a physical process is both unsupported by field observation and unlikely within this province (Greeley et al., 1998; Kerr, 2001). Evidence for variations in the stress regime within inflated lobes is provided by AMS studies, which have identified variable shear rates and isolated late-stage shearing at the lava core-crust boundary (Cañon-Tapia and Coe, 2002). Shear deformation within the lower-crustal zone would complicate the simplistic model of laminar flow associated with the pāhoehoe inflation model, but it may enable some propagation of magma toward an advancing flow front once a crystal lattice has formed.
Implications for Magmatism
Investigation of the origin of compositional variations in flood basalts and other magmatic and volcanic bodies may provide new insight to the assembly and extraction of large bodies of eruptible magma. Heterogeneities in ignimbrites are common, including the Fish Canyon Tuff, Colorado (Bachmann and Bergantz, 2008), Bishop Tuff, California (Hildreth and Wilson, 2007), Valley of Ten Thousand Smokes, Alaska (Fierstein and Wilson, 2005), and Zaragoza, Mexico (Carrasco-Núñez and Branney, 2005). In all of these examples, the heterogeneity has been interpreted as spatial and temporal variability preserved in the deposits through the mechanism of emplacement, resulting from complex mixing and withdrawal from a density-stratified magma chamber. However, long-term heterogeneities may be induced by: mixing, assimilation, internal phase changes, and decompression (Bachmann and Bergantz, 2008). While basalt magmatic systems may not preserve heterogeneity to the same extent as more-evolved magmas, the same processes generating compositional heterogeneity are otherwise applicable. The active processes and the time scales over which such processes occur (i.e., the length of repose periods between eruptions) may influence the degree of compositional heterogeneity within a single flow field. Long eruption durations (calculated to be a minimum of 19.3 yr for the Palouse Falls flow field; Vye-Brown et al., 2013a) may also increase the possibility of extracting compositionally different magmas during the lifetime of an eruption.
In comparison to this study on lava, compositional variability in sills (Latypov, 2003) has previously been interpreted as resulting from: style and duration of emplacement (Gibb and Henderson, 1996; Gibb and Henderson, 2006); assimilation and contamination of melt with wall rock (DePaolo, 1981); and magmatic heterogeneities preserved as a function of emplacement (Richardson, 1979). However, subhorizontally emplaced sills have many similarities to thick lava lobes or sheet lobes; the outer portions of each body are emplaced first, with relatively younger magma emplaced into the center, creating an age profile younging from the outer margins to the middle. The differences between flood basalt lavas and sills come from the different cooling rates and style. While flood basalt sequences have an asymmetric cooling profile displayed in a thick upper crust and thin lower crust, sills are more likely to have a symmetrical cooling profile, as both the upper and lower contacts have a similar thermal regime with the surrounding country rocks. However, detailed studies on the timing and style of emplacement of flood basalt lavas may offer useful insights into the mechanisms of sill emplacement.
Systematic geochemical compositional variations within the Palouse Falls flow field provide supporting evidence for the emplacement of individual lobes by pāhoehoe-type inflation. The composition of the Palouse Falls lava is variable both with height within individual lobes and between lobes, typically having more-evolved lobe crusts enriched in incompatible elements and less-evolved mafic cores. The range of composition across the flow field questions the validity of the chemostratigraphic methods used to identify flow fields within flood basalt provinces, unless there is a large sample suite. Meter-scale sampling and compositional analysis within single flow fields offer a significant advance in the understanding of emplacement mechanisms that are not revealed by lower-resolution data sets. Intralobe compositional variations appear to be coupled to physical, emplacement-related features, both within individual lobes and between different lobes, with distance from source. Such compositional variation is a result of magmatic heterogeneity from the point of emission, and these data provide a record of sequential development of the magmatic system over time.
Compositional variations from the source to distal areas were observed through the constituent lobes of the Palouse Falls flow field. Excursions from the general progressive trend may reflect lateral spread of the flow field during emplacement, in addition to longitudinal development. This would result in lobes of differing composition being juxtaposed against each other within a flow field, which is what we observed in the case of the Palouse Falls flow field. Transport of magma through sheet lobe cores, which act as internal flow pathways, may account for the occurrence of compositionally similar glasses noted at the proximal and distal ends of some Columbia River Basalt Group flow units. Propagation of magma through a series of linked sheet lobes remains the preferred interpretation of the method of transport of lava during flow field development. Lateral variability through the Palouse Falls flow field reflects the subsurface presence of discretely different magmas that were available at the same time. Compositional variability could be the result of an eruption that sequentially taps either a stratified periodically replenished magma chamber or a network of separate chambers deepening toward more primitive compositions with time. If sufficient compositional variability exists, then compositional mapping may enable identification of flow pathways and temporal development within a lava flow field.
We thank Steve Reidel for discussions and assistance in the field, and Steve Blake and the late Peter Hooper for useful discussions and their comments on an early version of the manuscript. We are grateful to two anonymous reviewers for their thorough reviews, which improved the manuscript, and to Vic Camp and Mike Poland for considerate editorial handling of the paper. S. Self and T. Barry were supported by Natural Environment Research Council grant NER/A/S/2003/00444. Studentship funding for C. Vye-Brown was provided by The Open University Ph.D. program and fieldwork funding from: the Daniel Pidgeon Award (Geological Society of London), Mineralogical Society, and the Peter Francis Bursary Fund. C. Vye-Brown publishes with permission of the executive director of the British Geological Survey (Natural Environment Research Council).