Large-volume silicic eruptions are often evacuated from magma reservoirs which display gradients in composition, temperature, crystallinity, and volatile content. The 9.7 Ma Devine Canyon Tuff (DCT) of eastern Oregon represents such an eruption, with >300 km³ of compositionally zoned pyroclastic material deposited as a variably-welded ignimbrite. The ignimbrite displays homogenous bulk tuff major element compositions with a wide range of trace element compositions, allowing for the investigation of how these magmas were generated, stored, and modified in the magma reservoir by studying pumices which represent the primary magmas composing the DCT. Five pumices ranging from dacite to rhyolite bulk compositions were selected across the range of trace element compositions and were crushed and sieved to measure how crystallinity and mineral abundances change within each pumice at different particle size fractions. Single alkali feldspar and clinopyroxene crystals were analyzed using EMP and LA ICP-MS from each pumice. Physical results yielded a systematic decrease in crystallinity from 22% to 3% going from the dacite to the most evolved rhyolite composition, with the highest crystallinity occurring between <991-425 microns for all pumices analyzed. The dacite pumices displayed a glomerocrystic texture not observed in rhyolite pumices. Two populations of crystals were distinguished using single crystal chemical data, one belonging to the rhyolitic magmas and another belonging to the dacitic magma. Acquired mineral data have relevance for how strongly zoned with regard to trace elements the rhyolitic magmas of the DCT were, how these magmas were generated, and how they were stored within the magma reservoir. Applying melt extraction models to explain observed patterns in trace element compositions between pumices is problematic. In this model, the observed range of trace elements in rhyolite pumices would be attributed to two separate melt extraction events from an intermediate crystal mush where the first expulsion of melt from the mush produced the most evolved rhyolite composition and a second expulsion coupled with partial melting produced a second rhyolite with an indistinguishable major element composition but less evolved trace element composition. Mixing of these two rhyolite end members would then be needed for generating the range of intermediate rhyolite compositions. Magma mixing modeled using a mixing equation produced a poor fit for trace elements, suggesting the range of observed trace element compositions cannot be solely generated through the mixing of the extracted rhyolite melts but require processes that subsequently modify the mixed rhyolite compositions. The occurrence of crystal aggregates in the dacite may represent fragments of the crystal mush. However, the dacite was unlikely produced by partial remelting of the crystal mush, generating a less evolved, more intermediate bulk composition. In summary, mush extraction combined with partial melting of the crystal mush and mixing of compositional end members cannot fully explain the trace element patterns observed in the DCT pumices thus warranting further study.

The Dinner Creek Tuff erupted during a period of rhyolitic volcanism coeval to the flood volcanism associated with the Columbia River Basalt Group. The High Rock Caldera Complex, Lake Owyhee and McDermitt volcanic fields account for ~90% of the rhyolites erupted between 16.7-15.0 Ma. Situated at the northern end of the Lake Owyhee volcanic field, the Dinner Creek Tuff was originally mapped as a ~2,000 km2 single ignimbrite confined to the Malheur Gorge. Streck et al. (2015) correlated tuff outcrops previously mapped as generic Miocene welded tuff as well as local units such as the "Mascall" or "Pleasant Valley" tuff of eastern Oregon to individual cooling units that comprise the newly redefined Dinner Creek Tuff, enclosing an area of ~25,000 km2. Areal extents defined in this study show that all outcrops now determined to be Dinner Creek Tuff enclose an area of ~31,800 km2 not including any fallout deposits that likely extended beyond the defined area. Although Dinner Creek Tuff rhyolites have nearly identical compositions, different ages and subtle geochemical and mineralogical differences exist and were used to divide the Dinner Creek Tuff into four discrete cooling units. Except for unit 4, the units are lithologically very similar. Unit 1 is the Dinner Creek Tuff unit associated with the Malheur Gorge type section. The four cooling units have ages of 16.15-16 Ma (unit 1), 15.6-15.5 Ma (unit 2), 15.46 Ma (unit 3) and 15.0 Ma (unit 4). Areal extents were established for all four cooling units based on feldspar compositions along with lithological and bulk rock geochemical data. Minimal extents of individual units are as follows: ~22,590 km2 (unit 1), ~17,920 km2 (unit 2), ~14,170 km2 (unit 3) and ~8,370 km2 (unit 4). Using conservative thicknesses, determined erupted tuff volumes are ~170 km3 (unit 1), ~125 km3 (unit 2), ~99 km3 (unit 3) and ~46 km3 (unit 4), totaling ~440 km3 and dense rock equivalents are ~152 km3 (unit 1), ~96 km3 (unit 2), ~76 km3 (unit 3) and ~31 km3 (unit 4), totaling ~356 km3. These extents and volumes are the absolute minimum based solely on the locations of exposed tuff sections and the inclusion of the source. Centering eruptive units on source areas where they are known, expands the tuff extents into a more radial pattern as would be expected for low-aspect ratio, high energy ash-flow tuff eruptions. These probable extents increase the areal extents of the individual units to: ~36,900 km2 (unit 1), ~31,660 km2 (unit 2), ~17,290 km2 (unit 3) and ~10,150 km2 (unit 4) distributed over a ~43,490 km2 area. Likewise, erupted tuff volume and dense rock equivalents also increase: volume-- ~277 km3 (unit 1), ~222 km3 (unit 2), ~121 km3 (unit 3) and ~56 km3 (unit 4); DRE-- ~248 km3 (unit 1), ~170 km3 (unit 2), ~93 km3 (unit 3) and ~38 km3 (unit 4). New mapping confirms previous hypotheses that the Castle Rock caldera erupted unit 1 and identified the new Ironside Mountain caldera as the source for unit 2 while precise source areas for unit 3 and 4 are not yet known but are thought to lie within the Dinner Creek Eruptive Center. Minimal calculated caldera volumes for units 1 and 2 are ~98.5 km3 (unit 1) and ~31.1 km3 (unit 2). Adding the thick ponded intra caldera tuff volume to the determined and probable erupted tuff volumes determined in this study, increases the erupted volumes to ~268 km3 (determined) and ~375 km3 (probable) for unit 1 along with ~157 km3 (determined) and ~253 km3 (probable) for unit 2. DREs increase to ~251 km3 (determined) and ~347 km3 (probable) for unit 1 along with ~128 km3 (determined) and ~202 km3 (probable) for unit 2.

The Wildcat Creek Tuff is a thin (~3-12 m), rhyolite to andesitic ash-flow tuff with a minimal extent of 1500 km2 in Malheur county, eastern Oregon. The previously undated tuff yielded a single crystal, anorthoclase 40Ar/39Ar age of 15.49±0.02 Ma and thus is closely related to mafic and silicic volcanism of the Columbia River Province. The tuff texturally stands out by its high proportion of co-mingled mafic inclusions appearing as dark, scoriaceous, and phenocryst-poor fragments, and their proportion dictate bulk tuff compositions ranging from rhyolite (74% SiO2) to andesite (59% SiO2). Glass analyses confirm rhyolite end member at 74-75 wt.% SiO2 and two mafic members, one at 59-60 wt.% SiO2 and the other at 56-57 wt.% SiO2. Rare plagioclase and even rarer pyroxene phenocrysts with compositions clustering at An60-74 and An35-45, and Mg17-19 and Mg80-84, respectively, similarly suggest two andesitic magmas with the 60% member being the dominant mafic composition. It has distinctly lower TiO2 and CaO, slightly lower FeO, and comparable Al2O3, MgO, and alkalis. Eruption of crystal-poor dacitic to basaltic-andesitic cognate components is also observed in other Miocene ash-flow tuffs from eastern Oregon, like the Rattlesnake, Dinner Creek, and the Devine Canyon Tuffs, as well as other less voluminous tuffs. However, the high proportion of mafic components in the Wildcat Creek tuff seems currently unrivaled. The co-eruption of intermediate magmas with rhyolite implies that mafic magmas were tapped from a common reservoir, and these magmas increased in proportion during the course of the eruption(s). This continued up to the point where nearly all deposited tuff material consisted of andesite. This is consistent with progressively deeper magma withdrawal, in turn implying that mafic magmas resided below the rhyolites as a discrete magma batch. Dacitic components of voluminous rhyolitic tuffs have been recently interpreted as remelted samples of a crystal mush after crystal-poor rhyolites where extracted. Dacitic Wildcat Creek Tuff samples do not bear any evidence of this. To the contrary, small negative Eu anomalies, normal Ba and Sr concentrations, and nearly aphyric nature are consistent with a large portion of mixing between Wildcat Creek Tuff rhyolites and regional mid Miocene, Fe-rich, and crystal poor basaltic andesite magmas that occur ubiquitously as lava flows.

Two mid-Miocene (16.5-15 Ma) rhyolite volcanic centers in eastern Oregon, the Buchanan rhyolite complex and Dooley Mountain rhyolite complex, were investigated to characterize eruptive units through field and laboratory analysis. Results of petrographic and geochemical analysis add to field observations to differentiate and discriminate the eruptive units. Additionally, new geochemical data are used to correlate stratigraphically younger and older basalt and ash-flow tuff units with regional eruptive units to constrain the eruptive periods with modern Ar-Ar age dates. Previous work at the Buchanan rhyolite complex was limited to regional mapping (Piper et al., 1939; Greene et al., 1972) and brief mention of the possibility of multiple eruptive units (Walker, 1979). Observed stratigraphic relationships and geochemical analysis were used to identify eight distinct eruptive units and create a geologic map of their distribution. Slight differences in trace element enrichment are seen in mantle normalized values of Ba, Sr, P, Ti and Nd-Zr-Hf and are used to differentiate eruptive units. New geochemical analyses are used to correlate the overlying Buchanan ash-flow tuff (Brown and McLean, 1980) and two underlying mafic units to the Wildcat Creek ash-flow tuff (~15.9 Ma, Hooper et al., 2002) and flows of the Upper Steens Basalt (~16.57 Ma, Brueseke et al., 2007), respectively, bracketing the eruptive age of the Buchanan rhyolite complex to between ~16.5 and ~15.9 Ma (Brueseke et al., 2007; Hooper et al., 2002). The Dooley Mountain rhyolite complex was thoroughly mapped by the U.S. Geological Survey (Evans, 1992) and geochemically differentiated in a previous Portland State University M.S. thesis (Whitson, 1988); however, discrepancies between published interpretations and field observations necessitated modern geochemical data and revisions to geologic interpretations. Field and laboratory studies indicate that the Dooley Mountain rhyolite complex consists of multiple eruptive units that were effusive domes and flows with associated explosive eruptions subordinate in volume. At least four geochemically distinct eruptive units are described with variations in Ba, Sr, Zr and Nb. Picture Gorge Basalt flows and Dinner Creek Tuff units found within the study area both overlay and underlie the Dooley Mountain rhyolite complex. These stratigraphic relationships are consistent with the one existing Ar-Ar age date 15.59±0.04 Ma (Hess, 2014) for the Dooley rhyolite complex, bracketing the eruptive period between ~16.0 and ~15.2 Ma (Streck et al., 2015; Barry et al., 2013). The findings of this study indicate that the Buchanan rhyolite complex and the Dooley Mountain rhyolite complex are the westernmost and northernmost rhyolite complexes among the earliest (16-16.5 Ma) mid-Miocene rhyolites associated with initiation of Yellowstone hot spot related volcanism.