Explosive volcanic eruptions often produce pyroclastic density currents, ground-hugging mixtures of particles and hot gas that are denser than the ambient atmosphere. Pyroclastic density currents rapidly propagate in the lowest levels of the atmosphere where they can directly impact human population centers and structures, and are some of the most hazardous volcanic phenomena. Pyroclastic density currents erupted at Cotopaxi in 1877 and at Tungurahua in 2006 and 2008 were produced by a little understood process described as ?boiling-over?. During a boiling-over eruption, a dense froth of pyroclasts and gas pours over the crater rim or through a notch in the crater, creating a pyroclastic density current. The mechanism for the formation of boiling-over pyroclastic density currents differs in important ways from those that result from the collapse of lava-domes or high convective columns, and the deposits are distinctive. Boiling-over eruptions are likely common but under-recognized in the geologic record, as most are associated with relatively small volume eruptions of volatile-rich mafic to intermediate magmas. This project, which is supported by the Petrology&Geochemistry program and the Americas Program (OISE) is a multidisciplinary, international collaborative effort in which numerical modeling and field studies will be integrated to characterize boiling-over deposits, determine why this style of eruption occurs, and understand the transport of the resulting density currents.

In particular, this proposal focuses on two fundamental questions designed to advance our understanding of pyroclastic density currents: 1. What conduit conditions result in boiling-over dynamics? 2. What are the relative contributions of eroded substrate and entrained air in boiling-over eruptions, and how does this influence pyroclastic density current dynamics? It is proposed to use 2D multiphase numerical models to test whether the presence of an enclosed summit crater enhances the decompression rate, allowing for a rapid sequence of exsolution and microlite crystallization over a limited spatial domain. The results of this study will have general applicability to volcanoes worldwide. The research team will also use petrographic and volatile content observations from the field studies at Cotopaxi and Tungurahua as a crucial test of the fidelity of the modeling program. They will test the hypothesis that after exiting the summit crater, a particle-dense, bed load region in the pyroclastic flows develops rapidly. This bed load region, which is constituted of particles that make multiple and enduring contacts with the substrate, may enhance erosion and gas pore pressure, and suppress ambient air-entrainment resulting in longer runout distances of these flows. Detailed analysis of deposit architecture, paleomagnetic measurements, and granulometry along with a 3D granular multiphase computational approach will be used to examine the role of entrainment in the propagation of these flows. One of the many novel aspects of the proposed investigation is the use of thermal proxies, including thermoremanent magnetization, in conjunction with the numerical models to constrain the thermal consequences of entrainment.

Project Report

The study of pyroclastic density currents (PDCs) and their deposits has been conducted by geologists for generations, because of their importance in the geologic record and the extreme hazard that they present. For most historical eruptions, two dominant mechanisms for the initiation of PDCs have been described: column collapse and dome collapse. In the first, a dense convective particle-laden mixture collapses under the force of gravity, after it has been thrust up into the atmosphere by explosive fragmentation within the conduit. Column collapse can occur both during sustained (Plinian) or discrete (Vulcanian) explosions. In a second type of PDC producing mechanism, the solidified crust of a lava flow or dome fails, suddenly releasing pressure on the hot, volatile-saturated interior of the flow. This material then quickly becomes a mobile, two-phase mixture of hot pyroclasts and gas. Intermediate, between these examples are ‘boiling-over’ eruptions that produce a distinct source mechanism for generating PDCs, but the process is only cursorily described. During a boiling-over eruption, a dense froth of pyroclasts and gas pours over the crater rim or through a notch in the crater, creating a PDC . The aim of this project was to describe the source mechanisms for boiling-over PDCs, examine the physical processes that account for transport in PDCs (and that influence their runout), and in particular to understand the transformation from dense currents to dilute currents. This project incorporated geochemical, remnant magnetization, granulometery and numerical modeling to approach these problems comprehensively. For our field examples we selected the well-preserved deposits from Cotopaxi, Ecuador, and the recent (2006) eruptive products from Tungurahua, Ecuador. Over two field seasons we documented the morphology and extent of these flows and collected samples. The boiling over flows are characterized by large, bread-crust pyroclasts on the slopes of the volcano, often forming distinct levees. At channel margins and at topographic breaks in slope or obstacles, these flows often form dune-fields and are much more mobile, sometime propagating several hundred meters up a slope. To understand the dynamics of these flows and to better constrain entrainment processes in pyroclastic flows, we developed an Eulerian-Eulerian-Lagrangian (EEL) modeling approach that particularly focused on the cooling of individual clasts. While in the field we documented that the breadcrust clasts typically have vesicular interiors because the magmas are saturated in volatiles at the time of the eruption and bubbles grow as the volatiles diffuse through the melt and the clast decompresses. However, near the margins of the clast, glassy rinds form where the clast has cooled quickly. The bubble size distribution and rind thickness provide useful constraints on the cooling history of individual clasts. In the EEL modeling approach we kept track of the location and cooling history of individual clasts and also developed and implemented a coupled bubble growth model so that the simulations could give a detailed accounting of the path of individual clasts, their cooling histories, and their predicted bubble size distributions. Glassy rinds are predicted in the model, and have much smaller bubble sizes than the interior of the clasts due to passing through the glass transition temperature during transport. Rind thicknesses calculated for ballistic and pyroclastic flow clasts are significantly different for most geologically relevant conditions, with ballistic clasts having larger rind thicknesses than flows transported in pyroclastic flows. The one exception being clasts with very high water content and low crystallinity that could give potentially ambiguous rind thicknesses. However in most cases the regime diagram we have developed should be useful in interpreting deposits that formed without witnesses and may lead to greater recognition of boiling-over events in the geologic past. Documenting the overall prevalence of these features is particularly necessary to better understanding the risk they present at andesitic volcanoes. Data from these simulations could also be readily compared to a remnant magnetization study that we conducted on these flows that showed that the cooling history of confined versus channelized flows were significantly different and that entrainment in unconfined flows is much greater. We used a similar approach to predict the flow transformation from dense to dilute flows, and importantly predict the amount of mass and extent of each type of flow. Information on runout provided new ways of interpreting deposits, and has given us a better understanding of entrainment and flow propagation. This information is important in making better hazard assessments near active volcanoes.

National Science Foundation (NSF)
Division of Earth Sciences (EAR)
Application #
Program Officer
Sonia Esperanca
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Georgia Tech Research Corporation
United States
Zip Code