Columbia River flood basalt flow emplacement rates—Fast, slow, or variable?
Published:February 07, 2019
- PDF LinkChapter PDF
Stephen Reidel, Terry Tolan, Victor Camp, 2019. "Columbia River flood basalt flow emplacement rates—Fast, slow, or variable?", 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:
Emplacement models for voluminous sheet flows of the Columbia River flood basalts vary significantly in style and duration, with the latter ranging from as little as one week to decades and even centuries. Testing the efficacy of such models requires detailed field studies and close examination of each stratigraphic unit. The Steens Basalt, the oldest formation of the Columbia River flood basalts, differs from the later formations in that it is composed of stacked successions of thin, commonly inflated flow lobes combined into thicker compound flows, or flow fields. These flow lobes are of limited geographic extent, with relatively high emplacement rates, but they are otherwise similar to modern examples. Evidence for flow inflation in the much larger sheet flows of the Grande Ronde Basalt, Wanapum Basalt, and Saddle Mountains Basalt is also apparent, but with more variable rates of emplacement. For example, the Asotin and Umatilla Members (Saddle Mountains Basalt) and Sentinel Bluffs Member flows (Grande Ronde Basalt) erupted distinct compositions along their linear vent systems, but over 200 km west of their vents, these flows are no longer distinct. Instead, they exist as compositional zones of a single, moderately mixed lava flow. Such flows must have been emplaced rapidly, in perhaps weeks to months, while others have been shown to erupt over much longer time periods. We conclude that emplacement rates may be quite variable throughout the Columbia River flood basalt province, with thin flow units of Steens Basalt erupting continuously and rapidly, and larger inflated sheet flows erupting over variable time spans, some from a few weeks to months, and others over a duration of years.
There have been no historic eruptions of basalt flows similar in size to those of the Columbia River Basalt Group (Fig. 1). The largest historic basaltic fissure eruption was the 1783–1784 Lakagigar (Laki) eruption in Iceland, which produced ~15 km3 of basalt (Thordarson and Self, 1998); however, the Lakagigar eruption was several orders of magnitude smaller than the massive sheet flows of the Columbia River Basalt Group. Without modern analogs for these voluminous flood basalt flows, we are restricted to studying the well-preserved Columbia River Basalt Group feeder dike systems and flows for evidence of their overall durations, and their mode(s) and rate(s) of emplacement.
In this paper, we first review the factors that control flow emplacement and then discuss field and analytical evidence bearing on the mode and rate of emplacement. Field evidence on the nature, duration, and rate of the eruptive activity comes from the physical dimensions of the linear feeder-dike systems, the volume and distribution of preserved vent deposits, and intraflow structures found within individual sheet flows. Such observations provide important constraints on the eruption style and emplacement history of large-volume Columbia River Basalt Group flows (e.g., Shaw and Swanson, 1970; Swanson et al., 1975, 1979; Mangan et al., 1986; Swanson and Wright, 1981; Wright et al., 1989; Reidel and Fecht, 1987; Beeson et al., 1989; Reidel et al., 1989a, 1994; Reidel and Tolan, 1992; Self et al., 1993, 1997; Thordarson and Self, 1998; Vye-Brown and Self, 2013; Brown et al., 2014).
FACTORS CONTROLLING THE EMPLACEMENT OF COLUMBIA RIVER BASALT GROUP FLOWS
The conditions that make it favorable for the production and emplacement of large-volume Columbia River Basalt Group flows result from the interplay of several factors: (1) the regional tectonic setting of the feeder-dike systems; (2) the paleotopography; (3) paleoenvironmental setting encountered by advancing flows; and (4) the rate and duration of lava discharge from the linear fissure system. In the following sections, we discuss the importance of these factors as they relate to the eruption and emplacement of Columbia River Basalt Group flows.
Columbia River Basalt Group and Regional Tectonic Setting
The Columbia River Basalt Group is the youngest and best-preserved large igneous province on Earth (Fig. 2). The flood basalts comprising the province were erupted between ca. 16.7 and 5.5 Ma, with over 210,000 km3 of lavas emplaced in the Pacific Northwest of the United States (Tolan et al., 1989; Camp and Ross, 2004; Reidel et al., 2013a). This 11 m.y. interval is misleading, however, as the far greater volume of the Columbia River Basalt Group erupted over a much shorter time span. Hooper and Hawksworth (1993) described the “main phase” of the Columbia River Basalt Group eruptions as the eruptive activity that encompassed Imnaha through Wanapum Basalts. Camp and Ross (2004) modified this definition to include only those formations that represent a near-continuous outpouring basalt without any evidence of significant or regional time gaps. Under this definition, the main phase encompasses the initial Steens Basalt through the end of Grande Ronde Basalt eruptions, an interval that generated 93% of the Columbia River Basalt Group volume in ~1 m.y. The peak of the main phase occurred during the Grande Ronde Basalt eruptions, when 71% of the flood basalt volume erupted in ~400,000 yr, from ca. 16 to 15.6 Ma (Barry et al., 2013).
This voluminous period of eruption occurred in a back-arc setting between the Cascade volcanic arc and the Rocky Mountains along the western edge of the North American craton. The flood basalt lavas cover basement rocks that record a long and complex geologic history, beginning in the Proterozoic with the breakup of the supercontinent Rodina, followed by the suturing of Mesozoic accreted terranes, and continuing with deposition and deformation of Paleogene and Neogene sediments and volcanic rocks. Basement structures associated with these events became the template for reactivation and the development of several younger geologic structures now superimposed on the flood basalt province (Reidel et al., 2013b).
The Columbia River flood basalt province consists of two diverse tectonic settings: the southern area, sometimes referred to as the Oregon Plateau (Carlson and Hart, 1987; Brueseke et al., 2007; Camp and Ross, 2004), and the Columbia Basin to the north (Fig. 2). The Oregon Plateau can be subdivided into four structural-tectonic regions: (1) the northern Basin and Range, (2) the Oregon High Lava Plains, (3) the Owyhee Plateau, and (4) the Oregon-Idaho graben, with each having distinct structural styles and topographic expressions (Reidel et al., 2013b). The Columbia Basin covers a broader region, dominated by a subsiding basin superimposed on a westward-dipping paleoslope that continued to develop during the flood basalt volcanism (Reidel et al., 2013b).
Volcanism initially began in the Oregon Plateau and quickly spread north to the Columbia Basin (Camp, 1995; Camp and Ross, 2004) through a broad, linear fissure system. In the Oregon Plateau, flood basalt eruptions at ca. 16.7 Ma were nearly contemporaneous with early caldera formation and the eruption of rhyolitic ignimbrites that began ca. 16.5 Ma at the western end of the Snake River Plain hotspot track (Coble and Mahood, 2016; Benson et al., 2017; Henry et al., 2017). In the Columbia Basin, the structural regime was distinctly different, with volcanism accompanied by rapid subsidence and contemporaneous folding and faulting during and after the basalt eruptions (Reidel et al., 1989b, 2013b).
Topography plays an important role in controlling the extent of historic basalt flows, and it also had a significant influence on the distribution of flood basalt flows in the Columbia Basin, allowing them to travel more than several hundred kilometers from their source vents (Shaw and Swanson, 1970; Swanson et al., 1979, 1980; Swanson and Wright, 1981; Fecht et al., 1987; Beeson et al., 1979, 1989; Reidel and Tolan, 2013a, 2013b). Topographic features that influenced the extent of great flows can be classified as: (1) tectonic, (2) erosional, or (3) constructional.
Topography in the Columbia Basin played a more important role in flow distribution than it did in the Oregon Plateau. In southeastern Oregon, there is little evidence for the development of significant deformational structures prior to the eruption of the Steens Basalt. Instead, the topography appears rather subdued, with no significant topographic feature or regional paleoslope, but instead is broken in part by localized elevated volcanic edifices associated with Paleogene volcanism in the Oregon Plateau (Meigs et al., 2009; Scarberry et al., 2010). This allowed the Steens Mountain volcano, a shield-type volcanic center for the earliest flows, to erupt numerous flow lobes that spread away from the volcanic edifice without significant topographic interference (Camp et al., 2013). This lack of topographic control lies in stark contrast to the Columbia Basin, where the flood basalt eruptions spread westward from their linear vent systems in the east into a subsiding basin where high-volume sheet flows accumulated to a thickness of ~5 km (Reidel et al., 1989b). It was not until the Steens Basalt eruptions ceased when Miocene deformation of the Oregon Plateau began in the form of northerly trending normal faults associated with Basin and Range extension and numerous small northwest-trending faults associated with the Oregon High Lava Plains.
The fundamental topographic features of the Columbia Basin are tectonic, with the most significant being the broad Columbia Basin itself lying between the Rocky Mountains to the east, the Cascade Range to the west, Blue Mountains to the south, and the Okanogan Highlands to the north. Within the Columbia Basin, topographic features include the Palouse Slope, the Yakima fold belt, with intervening synclines and basins (e.g., Pasco Basin), and the Columbia trans-arc lowland (Fig. 2). These features combined to produce a regional, downgradient pathway from the post–Steens Basalt vent areas in northeastern Oregon and southeastern Washington, through the Columbia trans-arc lowland in the Miocene Cascade Range, to the Pacific Ocean throughout Columbia River Basalt Group time (Beeson et al., 1985, 1989; Beeson and Tolan, 1990; Reidel et al., 1989b, 2013b; Reidel and Tolan, 1992; Reidel et al., 2013b). Subsidence of the Columbia Basin and Columbia trans-arc lowland kept pace with the emplacement of the Columbia River Basalt Group flows (Reidel, 1984; Reidel et al., 1989b; Beeson et al., 1985, 1989; Beeson and Tolan, 1990; Reidel et al., 2013a, 2013b) and assured the viability of these regional pathways. Without these regional tectonic features, Columbia River Basalt Group flows would have ponded against the eastern slope of the Miocene Cascade Range and been confined to the Columbia Basin, perhaps forming megashield volcanoes like that exposed at Steens Mountain (Fig. 2).
The Yakima Fold Belt (Fig. 2) is a regional structural feature that consists of a series of northeast-to northwest-trending, narrow, faulted anticlinal ridges separated by broad synclines and basins that are superimposed on the western part of the Columbia Basin and Columbia trans-arc lowland (Reidel, 1984; Beeson et al., 1985, 1989; Reidel et al., 1989b, 2013a; Reidel and Tolan, 2013a). The influence of these folds in controlling the distribution of Columbia River Basalt Group flows becomes more pronounced as the interval length between successive Columbia River Basalt Group flows increases after Grande Ronde time (Beeson et al., 1989; Reidel et al., 1989b, 1994, 2013a).
Erosional features, such as river canyons, had only minor influence on controlling the extent of flows during the main phase of Columbia River Basalt Group volcanism (16.7–15.5 Ma) due to the fact that large-volume flows were erupted so frequently that rivers did not have sufficient time to incise significant canyons between eruptions (Fecht et al., 1987; Beeson et al., 1985, 1989; Reidel et al., 1989a, 1994; Reidel and Tolan, 2013b). As the frequency of Columbia River Basalt Group eruptions decreased in Wanapum and Saddle Mountains Basalt time, longer periods of time between eruptions (thousands to millions of years) allowed rivers to establish and erode major canyons, which provided pathways for advancing Columbia River Basalt Group flows and probably allowed them to travel greater distances than they might have traveled as simple sheet flows.
Emplacement of large-volume Columbia River Basalt Group flows also created constructional topographic features along their flow fronts and margins that influenced, in part, the distribution of subsequent lavas. The emplacement of the first flow typically creates a convex upward fill along the existing pathway. Subsequent flows preferentially advanced along the low area defined by the margin of the previous flow and the ground surface. This emplacement scenario is clearly demonstrated by the Basalt of Rosalia and Basalt of Lolo of the Priest Rapids Member, the Asotin and Wilbur Creek Members, and in the basalt of Martindale of the Ice Harbor Member (Fig. 1; Reidel and Tolan, 2013b).
Environmental conditions refer to the nature of the paleoground surface (e.g., sediment vs. lava flow top, wet ground vs. dry, etc.) that the flow traversed. The composition of the ground surface, and whether it is wet or dry, appears to be the primary environmental factor that influenced flow character (Swanson and Wright, 1981; Beeson et al., 1979, 1989). A dry basalt flow top had the least impact, but wet sediment had the greatest impact and could rapidly reduce the temperature of the lava and increase its viscosity. Beeson et al. (1989), for example, found field evidence suggesting that such adverse paleoenvironmental conditions influenced the distribution and morphology of the distal portions of some Columbia River Basalt Group flows in western Oregon and Washington.
The outward morphology of flows that advanced across wet sediments commonly resembles that of a compound flow, but the individual lobes are of much greater size (10 to >30 m thick); these larger lobes (>15 m thick) often display complex internal jointing that is suggestive of lava that was injected into the lobes (inflated) even after the lavas came to rest (M.H. Beeson and T.L. Tolan, 1980, personal observation [unpublished field data]). In this case, it appears that environmental factors resulted in the formation of flow lobes rather than being produced simply as the result of low lava discharge.
Wet, unconsolidated sediment can also create an environment where the lava can invade and burrow into the sediment, as first recognized by Byerly and Swanson (1978). This has been shown on both small and large scales. In the eastern Columbia Basin, several Saddle Mountain Basalt flows, such as the Asotin Member, have locally burrowed into Ellensburg Formation sediments, which were deposited by the ancestral Salmon and Clearwater Rivers during the large time gaps between eruptions (Camp, 1976). In the central Columbia Basin, where thick accumulations of sediment were deposited, nearly every Saddle Mountains Basalt flow is found to have invaded unconsolidated sediment (Reidel and Tolan, 2013b). West of the Cascade Mountains, where thick accumulations of Eocene to Miocene sediments were deposited along the ancestral Pacific Ocean shoreline, Columbia River basalt flows invaded and traveled for miles through these sediments (Beeson et al., 1979; Pfaff and Beeson, 1989; Wells et al., 2009).
Rate and Duration of Eruptive Activity
In addition to the importance of the regional paleoslope recognized by Swanson and Wright (1976), Shaw and Swanson (1970) argued that important criteria for explaining the great extent of the Columbia River Basalt Group flows are (1) existence of an adequate volume of magma, (2) sufficient magmatic head pressure to maintain adequate magma supply to the surface, (3) length and width of the actively erupting fissure system, and (4) the duration of the eruption.
Adequate Volume of Magma
The spatial extent of Columbia River Basalt Group flows clearly demonstrates that there was an adequate volume of magma available to produce voluminous sheet flows such as the Grande Ronde Basalt flows, which consist of thousands of cubic kilometers of basalt. Such high-volume basalt flows are consistent with other evidence that support a mantle plume source for the Columbia River Basalt Group (e.g., Hooper and Hawkeworth, 1993; Wolff and Ramos, 2013; Camp, 2013), which would provide adequate volumes necessary to transport magma through the lithospheric mantle and crust.
Magma Supply Rate
The greatest production of magma for the Columbia River Basalt Group occurred during Grande Ronde eruption time (Tolan et al., 1989; Reidel et al., 2013a). Prior to and after the eruption of the Grande Ronde Basalt, magma volumes were relatively low (Fig. 1). The volumes of individual eruptions, however, were still quite large compared to the eruptive volume of modern flows. Swanson and colleagues estimated the overall range in magma supply rate to be comparable to Hawai‘i (0.075 km3/yr—Swanson et al., 1975; 0.1 km3/yr—Swanson et al., 1979). The principal problem with estimating magma supply rates is that they are just that: estimates; however, there can be little doubt that magma supply rates for the Columbia River Basalt Group were sufficient to produce huge lavas. In the next section, we discuss emplacement rates, which can provide constraints on magma supply.
Length and Width of Fissure System
Swanson et al. (1975) indicated that the dike system for the Columbia River Basalt Group was over 200 km wide and 450 km long. Studies of the length and width of individual dikes within the fissure systems suggest that most fissures are from 60 km to nearly 200 km long. Swanson et al. (1975) showed that the Roza fissure system is at least 120 km long and 15 km wide, with individual dikes several meters in width. The length has since been extended to nearly 200 km based on the work of Martin (1989), Brown et al., (2014), and this study. We have found that the widths of dikes range from a few meters to between 10 m and ~20 m, and the lengths can be as much as 200 km (Reidel and Tolan, 2013, personal observation). Thus, the Columbia River Basalt Group dike system was more than adequate to supply basalt for the flows.
Duration of Eruption
The duration of Columbia River Basalt Group eruptions encompasses both the mode and rate of emplacement and has proven to be a controversial topic. The first study was by Shaw and Swanson (1970), followed by Swanson et al. (1975), who suggested the Roza Member was emplaced in a short period of time. They suggested that the entire Roza Member could have been emplaced in a few hundred years, and individual flows could have been emplaced in a matter of days. Since those studies, estimates have varied. In the following sections, we will address this issue.
ESTIMATES OF FLOW EMPLACEMENT RATES
The mode and rate of emplacement for large-volume flood basalts can be subdivided into two end-members types: fast emplacement and slow emplacement, with a complete spectrum of possible rates between.
The first attempt to quantify the flow emplacement rates was by Shaw and Swanson (1970), who concluded that most flows resulted from rapid, turbulent emplacement. Their estimate was based on Swanson’s extensive field experience with the Columbia River Basalt Group and modern Hawaiian flows, and Shaw’s modeling skills. They proposed that a typical Columbia River Basalt Group flow of 100 km3, erupting from a 5-m-wide dike and 3-km-long fissure would be erupted in about 1 wk. This would suggest that a magma supply rate of 5200 km3/yr would be possible.
Two decades later, Martin (1989) argued that his six Roza Member units, consisting of 1300 km3 of the Wanapum Basalt, were emplaced at a slower rate (130 km3/yr to ~40 km3/yr), suggesting decades based on cooling models (Martin, 1989, p. 102). Furthermore, he suggested that the time between his Roza flow units could have been between 10 and 30 yr.
In an analogy to inflated pahoehoe flows in Hawai‘i, Hon et al. (1994) and Self et al. (1996) proposed that Columbia River Basalt Group flows were emplaced as large, inflated flow fields over many decades to centuries. They suggested an alternative view of much slower laminar flow emplacement based on the idea of thermal insulation associated with flow inflation. Flow inflation occurs as centimeter-scale lobes of lava develop a chilled, viscoelastic skin, and then expand with continued injection of fluid lava (e.g., four Roza Member flow units in a minimum of 14 yr, with individual lobes in months to years; Thordarson and Self, 1998). Furthermore, Thordarson and Self (1998), Self et al. (1993), Finneamore et al. (1993), and Vye-Brown and Self (2013) suggested that Columbia River Basalt Group flows were erupted over very long periods, ranging from tens to hundreds of years, with low lava discharge. Keszthelyi et al. (2006) estimated eruption rates of ~166 km3/yr for the Roza Member lavas, which was supported by thermal models (Keszthelyi and Self, 1998).
These models of slow laminar flow lie in contrast to the work of Ho and Cashman (1997), who used glass compositions in the Ginkgo flow (Frenchman Springs Member, Wanapum Basalt, Columbia River Basalt Group) to estimate minor heat loss (20 °C) over a flow distance 370 km. They concluded that rapid emplacement would not require turbulent flow and could have been emplaced instead by fast-laminar flow. However, they did not provide an estimate of the time required for emplacement of the Ginkgo flow, so either slow or rapid emplacement could fit their data.
Petrographic examination of quenched Columbia River Basalt Group lava (e.g., rinds from pillow lavas) from the medial and distal parts of Columbia River Basalt Group flows has shown that the crystallinity is no greater than that of the glassy selvage zone of the feeder dike. This indicates that little or no crystal nucleation and growth occurred from the time the lava was erupted to when it reached its most distal point (Shaw and Swanson, 1970; Swanson et al., 1975; Mangan et al., 1986; Wright et al., 1989; Ho and Cashman, 1997). This lack of significant crystal growth has been interpreted to have resulted from very rapid eruption and emplacement of the lava, with flow-emplacement times on the order of a few days to several weeks to several months (Swanson et al., 1975; Mangan et al., 1986; Wright et al., 1989; Reidel and Tolan, 2013a; Reidel, 2015). Reidel and Fecht (1987) estimated ~1.08 × 109 m3/yr or greater for recharge of the magma chamber supplying the Asotin and Wilbur Creek Members, while Reidel (1998, 2015) estimated as much as 720 km3 in 10 d for the Umatilla Member.
FIELD AND PETROCHEMICAL EVIDENCE FOR MORE RAPID EMPLACEMENT
In this section, we will provide additional evidence for a more rapid emplacement of specific Columbia River Basalt Group flows. These data come from field observations and examination of the compositional variations observed in flood basalt flows; together, these data provide an excellent measure of how individual parts of the flows were emplaced. The compositions we use came from pristine, unweathered or unaltered samples based on thin sections and thus are the highest quality of sample available.
Umatilla Member (Saddle Mountains Basalt)
The Umatilla Member (Fig. 1) consists of two flows, the Umatilla flow followed by the Sillusi flow, with both having distinct compositions (Fig. 3; Reidel, 1998). The two Umatilla Member flows were named by Laval (1956) for the exposures at McNary Dam on the Columbia River near the Oregon-Washington border (Fig. 3). Both flows originated over 200 km to the east in the Lewiston Basin (Reidel, 1998). The Sillusi flow erupted from the Puffer Butte Volcano, which was fed by a dike that can be traced from the base of the Grande Ronde River canyon to the vent (Reidel et al., 1992). The Sillusi dike is composed completely of the Sillusi composition. The vent and dike for the Umatilla flow are not known, but the presence of the Umatilla flow in the Lewiston Basin, underlying the Sillusi flow, demonstrates that the Umatilla flow had to have originated in the Lewiston Basin or nearby.
As the Umatilla flow erupted, it filled the Lewiston Basin and followed the ancestral Salmon-Clearwater River channel across the present Blue Mountains and into the Umatilla Basin, where it spread north and west into the Pasco Basin (Fig. 3). Soon after, it was followed by eruption of the Sillusi flow, which also filled the Lewiston Basin. Like the Umatilla flow, the Sillusi flow also advanced down the ancestral Salmon-Clearwater channel to the Pasco Basin over 200 km west. Both flows are present at McNary Dam, on the border of the Umatilla and Pasco Basin (Fig. 3). However, in the Pasco Basin, there is only one flow present. At Finley Quarry (Fig. 3), the composition of the flow is complex, with the younger Sillusi flow present at the bottom of the quarry, but with a mixture of Umatilla and Sillusi compositions above. Erosion has removed the uppermost part of the flow there.
Farther north, on the U.S. Department of Energy’s Hanford Site in the center of the Pasco Basin, it is clear that the younger Sillusi flow invaded and inflated the older Umatilla flow, with variable mixing from top to bottom. At the time the Sillusi flow invaded the Umatilla flow, solidification of the top of the flow had progressed to a thickness of less than 0.5 m (Reidel, 1998). Figure 3 shows that Umatilla-Sillusi compositional mixing is wide-spread throughout the Pasco Basin. In these profiles (Fig. 3), individual compositions were obtained from nine continuous cores, with mixing relationships determined from numerical modeling of major-and trace-element analyses. The mixing occurs within 0.5 m of the top and extends from the base to the top surface, indicating little time had passed before the Sillusi flow mixed with the Umatilla flow. Mixing between the two produced a composite, single flow. The importance of this relationship to flow emplacement is that the Umatilla Member at this locality is a single flow derived from two distinct eruptions, each with its own distinct composition. These separate flows flowed, one after the other, 200 km westward to the Pasco Basin, where the Sillusi invaded, inflated, and mixed with the older Umatilla lava that had ponded in the Pasco Basin without any significant solidification. Such mixing can only occur if the eruptions are rapidly emplaced. Reidel (2015) estimated that if the flows had the same velocity as a channelized Hawaiian flow, they could have been emplaced in as little as a 30–60 d.
Asotin–Wilbur Creek Members (Saddle Mountains Basalt)
Swanson et al. (1979, 1980) recognized three petrogenetically and stratigraphically related Saddle Mountains Basalt flows, from oldest to youngest, the Wilbur Creek, Lapwai, and Asotin Members (Fig. 1). Hooper (1985) conclusively showed that the Lapwai flow is a mixture of the Wilbur Creek and Asotin Members. He showed that after the Wilbur Creek flow erupted, its magma chamber was recharged with Asotin-composition magma, which then mixed and erupted as the Lapwai flow. The magma chamber continued to recharge, and finally the Asotin flow erupted. The Lapwai flow is largely confined to the Lewiston Basin, but the Wilbur Creek and Asotin Members followed different channels for over 200 km from the Lewiston Basin to the Pasco Basin (Fig. 4), thus showing that the flows were separate lavas upon entering the Pasco Basin. The mapped Asotin flow in the Pasco Basin and areas farther to the west is actually a composite of the Wilbur Creek and Asotin Members (Reidel and Fecht, 1987), as demonstrated in the mixing profiles of continuous cores from the U.S. Department of Energy’s Hanford Site (Fig. 4).
The Wilbur Creek flow was the first of the two canyon-filling flows to be emplaced as a ponded flow in the Pasco Basin (Reidel and Fecht, 1987). The Asotin flow then followed a separate channel between the Lewiston Basin and Pasco Basin and entered the Pasco Basin at nearly the same location as the Wilbur Creek flow. The Asotin flow then inflated the Wilbur Creek flow and mixing ensued. Like the Umatilla Member flows described previously, mixing occurred a short distance from the top and bottom of the flow. At the northeast corner of Figure 4, the Asotin flow clearly invaded and inflated the Wilbur Creek flow, but as the flow progressed westward, there was more mixing between the two flow compositions. As the composite flow moved westward out of the Pasco Basin toward the Oregon-Washington coast (Beeson and Tolan 2002), the composition takes on the character of a well-mixed flow composition (~50% Asotin, ~50% Wilbur Creek). This, like the Umatilla Member flows, can only occur if the flows were emplaced relatively fast. The mixing between the two flows is a regional phenomenon and not a local occurrence, suggesting little time had passed for the Wilbur Creek Member flow to solidify.
Pomona Member (Saddle Mountains Basalt, Columbia River Basalt Group)
The 12 Ma Pomona Member (Fig. 1) intracanyon lava flow provides additional evidence for the relatively rapid emplacement of Saddle Mountains Basalt flows. The Pomona flow is 760 km3, covers 20,550 km2, and extends ~550 km to the Pacific Ocean at Grays Harbor, Washington (Fig. 5). The Pomona intracanyon lava flow lacks evidence that it encountered or interacted with the ancestral river that carved its canyon. This is in contrast with lavas that did encounter an ancestral river and that show extensive development of pillow lava, hyaloclastic debris, and bedded hyaloclastite (Tolan and Beeson, 1984; Reidel and Tolan, 2013b). It is clear that the Pomona lava had sufficient volume to dam the ancestral Salmon-Clearwater River where it entered the canyon in the Lewiston Basin; the lava dam allowed the lava to advance westward down a “dry” canyon. The Pomona flow was also able to dam the ancestral Columbia River in the northern Pasco Basin area, which allowed it to advance down a dry Columbia River canyon (Tolan and Beeson, 1984; Anderson and Vogt, 1987; Reidel et al., 1994; Reidel and Tolan, 2013b). Using the height of the Pomona dam in the Lewiston Basin and the present flow rate of the combined Salmon and Clearwater Rivers, the dam could not have completely impounded the rivers for more than several months before the water would have overtopped the dam and flowed unimpeded down the Salmon-Clearwater river channel. This is a reasonable assumption based on the constraints of paleobotanical studies for regional climate during the Miocene. Smiley (1989, p. 43) noted that the sedimentology, taphonomy, and assemblage analyses at the Miocene Clarkia Fossil beds in northern Idaho suggest “a climate of humid-warm temperate conditions similar to those of the present southeastern United States and eastern Asia (short mild winters and long wet summers). Such a climatic regime is indicated also by the Clarkia fish, insects, mollusks, the rich fungal component of the forest vegetation, and the aquatic microfossils.” Although the Clarkia Fossil beds were deposited in an impoundment behind the older Priest Rapids Member, the climate was probably not significantly different during eruption of the Pomona lava. The lava dam of the Pomona lava would have been overtopped within a few months, consistent with the higher stream flows inferred from the humid-warm temperate paleoclimate noted by Smiley (1989).
Intracanyon exposures of the Esquatzel, Pomona, and Elephant Mountain Members (Saddle Mountains Basalt) at Devils Canyon (Fig. 6; ~110 km to the west of their source area) demonstrate that the river was repeatedly dammed but reestablished its channel in the same river canyon. If the Pomona lava was emplaced over a long period (years or decades), there should be evidence that the ancestral Salmon-Clearwater and Columbia Rivers reestablished their presence within the canyons before the Pomona flow was completely emplaced. Lakes behind the lava dams created by the Pomona flow (on the order of 10–30 m high) would have probably been filled within a period of a few months, and the dams would have been overtopped long before the Pomona reached its most distal point (Miocene Washington coast), and certainly long before it was inflated to its final thickness. Because these rivers reoccupied their canyons, we would expect to find the point where the river waters overtook the slowly advancing flow front. The consequences of the river encountering an active Pomona flow front should include the creation of large quantities of hyaloclastic debris and the presence of intraflow structures indicative of lava and water interaction. Hyaloclastic debris should have been continuously created as the Pomona flow slowly advanced; the hyaloclastic debris created by this process also would be transported and deposited downstream in advance of the lava flow in a process documented by Tolan and Beeson (1984) for the Priest Rapids Member (Fig. 1).
Studies of the Pomona intracanyon flow (Snavely et al., 1973; Anderson, 1980; Swanson and Wright, 1981; Tolan and Beeson, 1984; Anderson and Vogt, 1987) do not find the extensive hyaloclastite deposits and intraflow structures that would be a consequence of a very long emplacement history. Because of an absence of these features, we conclude that the Pomona flow was rapidly emplaced in months and not decades.
Cohasset Flow (Grande Ronde Basalt)
The next flow we consider is the Cohassett flow of the 10,000 km3 Sentinel Bluffs Member (Figs. 1, 7, and 8), which was emplaced near the end of the Grande Ronde Basalt eruptions. The Sentinel Bluffs Member consists of six flows (Reidel, 2005). The first, the McCoy Canyon, and the last, the Museum, are compositionally well constrained. In contrast, the other four (the California Creek, Airway Heights, Stember Creek, and Spokane Falls flows) have been traced from their dikes and vents to the Pasco Basin, where they form a complex mixture that has been named the Cohassett flow by the U.S. Department of Energy (U.S. DOE, 1988); the Cohassett flow was once considered as a potential host repository for high-level nuclear waste because of its great thickness (>100 m).
Figure 8 covers over 250 km from near Spokane, Washington, southwest to the U.S. Department of Energy’s Hanford Site. The diagram shows compositional profiles from measured sections in continuous core. The exception is the Spokane County core near Cheney, Washington, where three distinct flows are present, from oldest to youngest: the California Creek, Airway Heights, and a hybrid of Spokane Falls and Stember Creek. There, the earliest flow is the Stember Creek flow, which was then inflated by invasion of the Spokane Falls flow composition. In the Pasco Basin, the California Creek flow was the first flow to be emplaced, but this was quickly followed by the invasion of a flow with Airway Heights composition, and then one of Stember Creek composition, and finally a flow with Spokane Falls composition. These multiple invasive relationships are shown best in cores DC-14 and DC-6, near where these flows entered the Pasco Basin. The pattern is present in other cores, but farther west in the basin, the pattern is the more complex due to greater transport distance in the inflation-mixing zone.
Figure 8 shows that as each flow was erupted, it flowed into the Pasco Basin, where it ponded and then was inflated by the next flow to enter the basin, the pattern being: California Creek → Airway Heights → Stember Creek and finally Spokane Falls. The “IVZ” designation in Figure 8 is a very prominent internal vesicular zone at the top of the Spokane Falls composition in the center of the Cohassett flow that can be traced throughout the Pasco Basin.
Figure 9 provides a more regional perspective of the Cohassett flow in three cross sections based on core and measured sections. Section A–Aʹ shows the northeast to southwest compositional zonation of the Cohassett flow. The Spokane County core lies to the northeast of the section, and the central portion lies across the Pasco Basin. Note on the far west, the Spokane Falls flow composition “breaks out” and covers the margin of the flow. In section B–Bʹ, the far-east portion is dominated by the Stember Creek flow; in the Pasco Basin portion, the Spokane Falls flow lies in the core of the flow, but, like section A–Aʹ, the Spokane Falls flow breaks out and actually covers the southern portion of the Cohassett flow. Section C−Cʹ lies across the southernmost portion of the Cohassett flow. As in section B–Bʹ, the Stember Creek dominates the eastern part, but only the Spokane Falls is present at the southern margin of the Cohassett flow. These invasive, commingling, breakout, and mixing relationships are similar to those of the Umatilla and Asotin–Wilbur Creek Member flows described earlier herein and require rapid emplacement of each composite type found in the Cohassett flow.
ESTIMATING COLUMBIA RIVER BASALT GROUP EMPLACEMENT RATES FROM DIKE–WALL-ROCK RELATIONSHIPS
Petcovic and Grunder (2003) analyzed wall-rock melting reactions in tonalite adjacent to the Maxwell Lake dike, a likely feeder to the Wapshilla Ridge Member (Fig. 1), the most voluminous of all Grande Ronde Basalt (Columbia River Basalt Group) eruptions, consisting of more than 41,000 km3 (Reidel et al., 1989a; Reidel and Tolan, 2013b). Petcovic and Dufek (2005) used two types of models constrained by the field example of the Maxwell Lake dike in order to assess the development of wall-rock melting due to basalt intrusion. In a series of preliminary static conduction simulations, the timing between dike injection(s) and duration of basalt flow in the dike was varied. The static conduction simulations were developed to explore a range of conditions that could have given rise to the observed wall-rock melt zones. In order to refine initial calculations and determine magma flow rates in the dike, advective transport simulations were developed. In all simulations, a 10-m-wide dike at 1145 °C (liquidus temperature) was instantaneously injected into 50 °C tonalite wall rock in a vertical orientation. This approach used a finite-volume numerical method, with a predictor-corrector algorithm to compute the partitioning of enthalpy into sensible and latent heat during phase change processes for nonlinear melt fraction to temperature relationships. One principal difference between this approach and previous calculations (Bruce and Huppert, 1990; Fialko and Rubin, 1999) is the incorporation of nonlinear melt fraction to temperature relationships specific to the magma/solid composition.
Results of static conduction modeling constrained by the Maxwell Lake dike field example suggest that a single injection followed by stagnation and cooling would have been incapable of producing the observed wall-rock melt. Repeated injection of basalt (such as inferred from sheeted dike complexes) would have generated wall-rock melt zones comparable to those observed in the field, but the regular textural progression across the dike and its wall rock is inconsistent with repeated brittle injection. Instead, static conduction results suggest that sustained basalt flow for 3–4 yr caused the development of the melt zones observed in the dike. Advective transport simulations suggest that the initial basalt velocity in the dike center was ~10 m/s, but that basalt at the dike margin solidified, causing constriction and slowing of flow. After ~60 d, magma in the dike would have reached a sustained velocity of ~2 m/s for the duration of flow. Wall-rock melting was initiated after ~1 yr of flow, and the wall rock had dropped below its solidus temperature within ~2 yr after flow ceased. The thickness, distribution, and fractions of wall-rock melt zones produced by advective transport modeling closely approximate field observations of the Maxwell Lake dike (Petcovic and Dufek, 2005). Model results suggest that the dike was active for 3–4 yr and likely represented a long-lived point source in flood basalt eruptions. Furthermore, this suggests that over 1000 km3 of lava were erupted per month, which is consistent with estimates of eruptions rates made by Reidel et al. (1994), Reidel and Tolan (1992), and Reidel (1998, 2005).
Results of the simulations allow limits to be estimated for eruption rates of the Wapshilla Ridge flow. Assuming that the Maxwell Lake dike fed the entire Wapshilla Ridge Member of 41,000 km3 over a period of 3–4 yr yields an average eruption rate ~25–40 km3/d. The eruption rate would have been lower if flow in the dike were intermittent rather than continuous. Intermittent flow has been documented in historical basalt eruptions and during eruption of the Teepee Butte Member, Grande Ronde Basalt (Reidel and Tolan, 1992). Some pauses during eruption cannot be ruled out, yet the lack of internal contacts and the regular textural progression across the dike and wall-rock melt zones are more consistent with nearly continuous flow and a single cooling history.
Volumetric eruption rates calculated on the basis of the thermal models for a typical Wapshilla Ridge Member flow are within the range reported for other Columbia River Basalt Group eruptions. Minimum eruption rates are comparable to rates estimated from slow emplacement models for the Roza Member flow (Self et al., 1997; Thordarson and Self, 1998). Maximum eruption rates are an order of magnitude lower than rates calculated using rapid emplacement models (Swanson et al., 1975; Reidel and Tolan, 1992), yet they are an order of magnitude higher than slow emplacement estimates. Wapshilla Ridge eruption rates are similar to the maximum eruption rate of 0.2 km3/d reported for the 1783 Laki (Skaftár Fires) eruption, the largest historical fissure eruption (Thordarson and Self, 1998). Although our maximum calculated volumetric eruption rate is consistent with models of rapid flow emplacement, the calculated minimum eruption rates and longevity of the Maxwell Lake dike (3–4 yr) support a somewhat slower emplacement model.
Model results suggest that the Maxwell Lake dike sustained high magma flux for at least several years. The transition from fissure eruption to localized vents during basaltic volcanism is often explained as a function of cooling in narrow portions of dikes coupled with enhanced flow in thicker portions, resulting in isolated, long-lived vents (e.g., Delaney and Pollard, 1982; Bruce and Huppert, 1990). This process may also explain the presence of wall-rock melt zones only along two portions of the Maxwell Lake dike, which experienced higher mass and heat flux as surrounding portions of the dike solidified. Wall-rock melt zones adjacent to the Maxwell Lake dike provide evidence for the existence of long-lived point sources playing an important role in flood basalt eruptions.
MODEL FOR THE EMPLACEMENT OF LARGE-VOLUME COLUMBIA RIVER BASALT GROUP FLOWS
The style of flow emplacement for the initial Columbia River Basalt Group eruptions of the Steens Basalt differs significantly from the emplacement of the high-volume pahoehoe sheet flows that typify the remainder of the province. The Steens Basalt eruptions generated much thinner flow lobes, ~1–5 m thick, with significantly smaller volumes (Bondre and Hart, 2008; Camp et al., 2013). Many of these flow lobes occur as stacked sequences that erupted in rapid succession to form petrochemically coherent compound flows, or flow fields, which vary in thickness from 10 to 50 m (Camp et al., 2013). Based on an estimated eruption duration of 47,500 yr for the Steens Mountain shield volcano, Camp et al. (2013) suggested that the bulk of the Steens Basalt succession may have erupted at an effusion rate of 0.67 km3/yr, which is approximately six times higher than the effusion rate of the current Pu‘u ‘Ō‘ō eruption of Kīlauea Volcano, Hawai‘i (Pietruszka and Garcia, 1999). The near-continuous outpouring of the Steens eruptions lies in stark contrast to the intermittent eruption of high-volume pahoehoe sheet flows that typifies the rest of the province. Even during the peak eruptions of Grande Ronde Basalt, the outpouring of these high-volume sheet flows appears to have been separated in time by 102 to 104 yr of volcanic repose (Barry et al., 2013).
The Steens eruptions are similar in style to the eruption of thin flow lobes typical of modern shield volcanoes, but perhaps at greater effusion rates. In contrast, the eruption of voluminous Columbia River Basalt Group sheet flows is distinct in style, volume, and emplacement rate when compared to modern examples. Our field studies suggest that most Columbia River Basalt Group sheet flows were derived from linear fissure systems that were segmented, with each segment having periods of intense activity and periods of relative quiescence. Segmentation could allow the flows to be emplaced in stages. During periods of high lava discharge, pulses would rapidly advance away from the fissure as fast-laminar flow. As discharge declined, velocity would decrease, accompanied by a transition from fast laminar to a slower laminar flow, allowing a crust to form over the flow. The waning stage of eruption is often marked by the development of shelly pahoehoe and spatter-fed flows, as exemplified in the end-stage eruption of the Teepee Butte Member of the Grande Ronde Basalt (Reidel and Tolan, 1992) and the Roza Member (Brown et al., 2014). As activity at one segment of the fissure declined, another segment could become active, or the entire fissure might become dormant.
Field mapping of post–Steens Basalt flows demonstrates a consistent westward flow direction for most Columbia River Basalt Group sheet flows, where a lava pulse advanced away from its active dike segment, moving relatively unhindered down the Palouse Slope to the Pasco Basin. Here, lava temporarily ponded or stagnated in this structurally low area and thickened or inflated as more lava entered the basin. This scenario is described in several examples discussed herein. Such temporary lava “lakes” may well have been fed by one or more active segments of the flow’s fissure system. Field data are consistent with flow-front stagnation and deepening of these lava “lakes,” filling the Pasco Basin until lava overflowed the basin rim. Because of the irregular basin topography created by the shape and size of the bounding anticlinal structures, continued westward advance of these flows rarely occurred as a single sheet, but rather as multiple flows, with each exiting through one or more low spots in the basin. These flow units then continued as separate flows or merged back together as mixed or commingled flows.
Columbia River Basalt Group flows that approached the Cascade Range were funneled into the Columbia trans-arc lowland (Fig. 1). Directing one or more flows into this relatively narrow lowland could also produce multiple flow units like that observed at Dog Mountain in the Columbia Gorge (Fig. 1). The advancing flow might repeat this process at other basins. This is exemplified in flows that encountered the Portland Basin before advancing toward the Pacific Ocean (Beeson et al., 1979; Beeson and Tolan, 1990).
Intermittent discharge along the fissure system could also produce smaller localized flow lobes along the length of the larger flows. The increased hydraulic pressure from intermittent discharge could breach stagnated flow margins anywhere along the length of the lava flow and could not, in general, be traced back to the vent.
Data from studies of Columbia River Basalt Group flows and feeder dike systems indicate a wide range of emplacement rates. The estimates range from rapid rates for the small-volume pahoehoe flow lobes of the Steens Basalt to more variable rates for voluminous pahoehoe sheet flows. Data presented here and elsewhere demonstrate that the emplacement rates for sheet flows could be as rapid as 1 to 2 mo, or as slow as years to decades. Examples of flows emplaced over longer periods of time include flows like the Roza Member (Martin, 1989; Self et al., 1993, 1996) and the Palouse Falls flow (Vye-Brown and Self, 2013). Those with rapid emplacement rates include the Sentinel Bluffs Member (Grande Ronde Basalt), and the various Saddle Mountains flows discussed herein. Huge flows like the Wapshilla Ridge Member (~41,000 km3) that erupted in 3–4 yr fall between the two extremes. It seems probable the Columbia River Basalt Group flows fall along a continuous spectrum of emplacement times rather than just the two extremes. Still uncertain are the mechanical and thermodynamic details of the magmatic system that controlled such variable rates of emplacement. It is our contention, however, that variable emplacement rates cannot be readily attributed to a single mechanism, but instead each basalt flow must be evaluated independently based on its own conditions of emplacement.
We wish to thank Martin Ross, Dennis Geist, and Mike Poland for thoughtful and valuable reviews. We would particularly like to thank Don Swanson and Tom Wright for their early work on the petrochemical stratigraphy and emplacement characteristics of the Columbia River Basalt flows, and for the encouragement, mentoring, and inspiration that they provided to us as budding students and young workers.