Search USGS⌕ Search

USGS · 70177891

Eruptive history of Mount Katmai, Alaska

Abstract

Mount Katmai has long been recognized for its caldera collapse during the great pyroclastic eruption of 1912 (which vented 10 km away at Novarupta in the Valley of Ten Thousand Smokes), but little has previously been reported about the geology of the remote ice-clad stratovolcano itself. Over several seasons, we reconnoitered all parts of the edifice and sampled most of the lava flows exposed on its flanks and caldera rim. The precipitous inner walls of the 1912 caldera remain too unstable for systematic sampling; so we provide instead a photographic and interpretive record of the wall sequences exposed. In contrast to the several andesite-dacite stratovolcanoes nearby, products of Mount Katmai range from basalt to rhyolite. Before collapse in 1912, there were two overlapping cones with separate vent complexes and craters; their products are here divided into eight sequences of lava flows, agglutinates, and phreatomagmatic ejecta. Latest Pleistocene and Holocene eruptive units include rhyodacite and rhyolite lava flows along the south rim; a major 22.8-ka rhyolitic plinian fall and ignimbrite deposit; a dacite-andesite zoned scoria fall; a thick sheet of dacite agglutinate that filled a paleocrater and draped the west side of the edifice; unglaciated leveed dacite lava flows on the southeast slope; and the Horseshoe Island dacite dome that extruded on the caldera floor after collapse. Pre-collapse volume of the glaciated Katmai edifice was ∼30 km 3 , and eruptive volume is estimated to have been 57±13 km 3 . The latter figure includes ∼40±6 km 3 for the edifice, 5±2 km 3 for off-edifice dacite pyroclastic deposits, and 12±5 km 3 for the 22.8-ka rhyolitic pyroclastic deposits. To these can be added 13.5 km 3 of magma that erupted at Novarupta in 1912, all or much of which is inferred to have been withdrawn from beneath Mount Katmai. The oldest part of the edifice exposed is a basaltic cone, which gave a 40 Ar/ 39 Ar plateau age of 89 ± 25 ka. The seismic record of caldera collapse includes 14 earthquakes of magnitude 6.0–7.0. By combining the times of earthquakes, the hours of downwind plinian-fall episodes from Novarupta, and the stratigraphic record of hydrothermal explosion breccia and phreatic mud layers ejected around the caldera rim and intercalated within the Novarupta pumice-fall sequence, it can be inferred that collapse began in the 11th hour of the 60-h-long eruption and continued fitfully for 3.5 days. Several big landslides and pumiceous debris flows shaken loose by the collapse-related seismicity are bracketed in time by their levels of intercalation within the Novarupta pumice-fall sequence. An intracaldera lake was ∼10 m deep by 1916, drained away in 1923, and has since deepened progressively to ∼250 m today. Compositionally, products of Mount Katmai represent an ordinary medium-K arc array, both tholeiitic and calcalkaline, that extends from 51.6% to 72.3% SiO 2 . Values of 87 Sr/ 86 Sr range from 0.70335 to 0.70372, correlating loosely with fractionation indices. The 5–6 km 3 of continuously zoned andesite-dacite magma (58%–68% SiO 2 ) that erupted at Novarupta in 1912 was withdrawn from beneath Mount Katmai and bears close compositional affinity with products of that edifice, not with pre-1912 products of the adjacent Trident cluster. Evidence is presented that the 7–8 km 3 of high-silica rhyolite (77% SiO 2 ) released in 1912 is unlikely to have been stored under Novarupta or Trident. Pre-eruptive contiguity with the andesite-dacite reservoir is suggested by (1) eruption of ∼3 km 3 of rhyolite magma first, followed by mutual mingling in fluctuating proportions; (2) thermal and redox continuity of the whole zoned sequence despite the wide compositional gap; (3) Nd, Sr, O isotopic, and rare earth element (REE) affinities of the whole array; (4) compositional continuity of the nearly aphyric rhyolite with the glass (melt) phase of the phenocryst-rich dacite; and (5) phase-equilibrium experiments that indicate similar shallow pre-eruptive storage depths (3–6 km) for rhyolite, dacite, and andesite.

Explore related subjects

90° N90° S · 180° W ← longitude → 180° E
Source-reported bounding extent: 57.70414723434193° to 58.87058467868075° latitude; -156.20361328125° to -153.6767578125° longitude. This indicates report coverage, not an exact sampling location. View area on OpenStreetMap.

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Edward Hildreth, Judith Fierstein. 2012. Eruptive history of Mount Katmai, Alaska. https://doi.org/10.1130/ges00817.1

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

Ignimbrites of the Sour Creek dome, Yellowstone volcano, USA: Implications for the Lava Creek Tuff eruption and Yellowstone Caldera

Field observations plus sanidine compositional and 40 Ar/ 39 Ar age data collected on rocks from the Sour Creek dome (SCD) area of Yellowstone volcano (United States) serve to identify five new ignimbrite packages with ages (average 631.5 ± 0.9 ka) that are analytically indistinguishable from those of the Lava Creek Tuff (LCT), challenging the current framework of two LCT eruptive members. Newly recognized package 1 occurs as LCT-aged recycled clasts within a proximal lag breccia, reflecting a major time break in the LCT eruption sequence. Conformably above, the scoria-bearing ignimbrite of package 2 is extensively found in the dome, having been previously mis-mapped as Huck-leberry Ridge Tuff and LCT member A. To the south, package 3 is a crystal-rich, densely welded ignimbrite (inferred to be younger than package 2) that is conformably overlain by package 4, a fine-grained ignimbrite that contains abundant clasts of recycled tuff. Lastly, the topographically highest parts of the dome are made up of package 5, a multicomponent ignimbrite containing scoria, recycled clasts, and pumices, potentially correlative with package 2 and/or package 4. Cathodoluminescence imaging and zoning assessment was integrated with sanidine geochemistry from all five packages, including their recycled clasts. Four of the five packages were sourced from separate magmatic bodies within a larger plumbing system, each defined by their sanidine Ba distributions and componentry. Although the SCD was originally mapped as a resurgent dome and is actively deforming today, faults cutting the region show comparable offsets for NW-SE “resurgent faulting” as for orthogonal NE-SW faulting, the latter of which also cuts the 162 ka Elephant Back lava flow. We infer that faulting in the SCD well postdates eruption and that the SCD instead represents a domical accumulation of ignimbrite, not a resurgent structure. The abundance of recycled materials (and absence of Absaroka country rocks) in the SCD ignimbrites implies that vent locations were well inboard of the mapped caldera boundary, with some potentially along the line of the Yellowstone River, requiring a revision of the currently accepted LCT caldera outline. The LCT eruptive events were thus far more complex than currently recognized, with multiple deposits fed from multiple magma bodies over a time scale that cannot be resolved by radiometric dating.

Wyoming↗

Syn-magmatic subsidence during the early stages of continental rifting in the Mesoproterozoic—A reanalysis of legacy data for the Midcontinent Rift, western Lake Superior

The Midcontinent Rift system (ca. 1.1 Ga) is a 2000-km-long series of elongated volcanic and sedimentary troughs and associated intrusive centers exposed chiefly in the Lake Superior region of North America. The rift system represents a long history of intense magmatism and subsequent sedimentation that was arrested by far-field tectonic events before sea-floor spreading was established. The premature cessation preserved a record of processes related to the beginning of continental rifting. The rift system under Lake Superior has been long studied using seismic-reflection data collected as part of the Great Lakes International Multidisciplinary Program on Crustal Evolution (GLIMPCE). We reexamine GLIMPCE Line C by developing a detailed velocity model for time to depth conversion constrained by other legacy data. We corroborate the model and develop a geologic interpretation using gravity and magnetic modeling and ties to geology mapped onshore. We recognize superposed subsiding sedimentary and volcanic basins for the southern half of the Line C depth section. This interpretation differs from previous paradigms that show major crustal faults that bound half-grabens or full grabens. We conclude that high-velocity (6.9 km/s) intrusive zones rather than major crustal faults border the sides of the basins. We speculate that the volcanic basin represents the initiation of seaward dipping reflectors. The syn-magmatic subsidence can be explained by dike injection and volcanic loading. Discrete lava basins throughout the region likely subsided at different times in a disorganized manner along the rift trend, raising questions about the long-term role of lithospheric thinning and melt generation.

Michigan, Minnesota, Ontario, Wisconsin↗

Characterizing changes in postfire debris-flow hazard as burned areas recover

Emergency assessments of postfire debris-flow hazards that are performed by the U.S. Geological Survey (USGS) provide estimates of debris-flow likelihood and rainfall triggering conditions that are used for evaluating and managing runoff-generated debris-flow hazards in recently burned areas throughout the western United States. Although the immediate postfire period, within roughly one year after fire, is typically the most susceptible to runoff-generated debris flows, the hazard evolves in time and space as the burned area recovers. The recovery trajectory a given burned area will take depends on local climate and weather and can be difficult to predict. Some burned areas recover quickly, whereas others experience debris flows for multiple years after fire. As a result, extending our ability to update debris-flow likelihood estimates and rainfall thresholds based on observed recovery of the burned area would be beneficial. We present a method for multi-year runoff-generated debris-flow hazard assessment that leverages the USGS “M1” debris-flow likelihood model and integrates updated, satellite-derived, normalized burn ratio data to estimate vegetation recovery. We predict recovery-aware rainfall thresholds and validate them against a multi-year debris-flow hazard prediction and could be adapted for use with other debris-flow models that incorporate burn severity data.

Arizona, California, Colorado, New Mexico, Washing↗