Report

VOLCANIC ERUPTION MECHANISM MODELING WORKSHOP

November 14-16, 2002
Durham, NH

The Volcanic eruption mechanism modeling workshop was convened to bring together the growing community of volcanic eruption modellers along with experimentalists and observationalists to accomplish two specific objectives:

1. Ensure that modellers are provided the most up-to-date values of magmatic parameters on the basis of recent experimental results and observations, and

2. Establish a standardized set of model "experiments" with identical magma and conduit characteristics and with a common protocol so that model results can be subsequently meaningfully compared and evaluated.

Workshop participants explored the sensitivity of modeled volcanic conduit systems to magma parametric values, overall system geometry, and most critically, fundamental physical formulation and underlying assumptions. In parallel with the efforts of the modeling community, a growing number of experimental results are emerging that bear directly on the parameterization of volcanic eruption models. It is thus possible to abandon many of the "if-then" scenarios traditionally resorted to by modellers, and use real-world values for magma parameters in an attempt to more realistically simulate natural magmatic conditions and thus better understand the processes involved in actual eruptions. Recent advances in observational techniques and analysis of natural volcanic products has also added to the arsenal of information that can be used to better calibrate, constrain and/or evaluate the performance of numerical models.

It is imperative that all models use the same values for fundamental parameters so that their individual performance can be assessed. By running models in this way, the differences in results can accurately reflect contrasting formulations, and thus provide insights into the processes being modeled. A specific set of "experiments" was devised at the workshop based on a standard set of parameters to be used by all models. With this result, modeling participants have returned to their labs, and in the coming months, will perform the full suite of model runs, providing the specific set of output required by the standard protocol. The results will be subsequently compiled and evaluated to determine the sources of differences in model output. The results of the model runs will be brought together at a special session in April, 2003 at the joint AGU-EGS-EUG meeting. This will be the venue for displaying the results and highlighting the critical differences in model formulation that bear on our understanding of magmatic processes prior to and during volcanic eruption. Modellers will then be able to revisit their models to explore the sensitivity of their formulations to variations in functionalization so that the key differences can be pinpointed. It is expected that once this is done, it will point to specific magma characteristics that are insufficiently understood. This, in turn, will spur experimentalists to return to their labs to focus attention on the most critical parameters identified by the model intercomparison results.

At the workshop, a number of critical conceptual, experimental, and observational gaps were identified. It is hoped that the research community can now move forward to address these gaps within the experimental and observational, as well as modeling sectors.

Volcanic eruptions occur by processes that are not fully understood, and are thus as unpredictable as they are devastating. In an attempt to better understand eruption mechanisms and thus begin to accumulate the body of knowledge necessary before eruption prediction can be considered, numerical models are being developed by a growing number of research groups globally. Numerical models are particularly useful because magmatic processes cannot be observed directly, and the complexities of volcanic systems cannot be solved analytically. In addition, numerical methods provide effective visualization tools that help explore the multidimensional parameterizations involved in analyzing volcanic systems. Although the various existing models are based on very different mathematical formulations, they are designed to accomplish the same goal- that of realistically accounting for the processes that drive volcanic eruptions. However, because the models use incompatible sets of magmatic parameters and simulated conduit environments, it has been impossible to compare and evaluate model results and identify fundamental conceptual weakness that warrant further exploration by the research community as a whole. The volcanic eruption mechanism modeling workshop was convened because state of the art has finally progressed to the point where i is now possible to explore the sensitivity of modeled volcanic systems to magma parametric values, overall system geometry, and most critically, fundamental physical formulation and underlying assumptions.

In parallel with the efforts of the modeling community, a growing number of experimental results have been emerging that bear directly on the parameterization of volcanic eruption models. It is thus possible to now abandon many of the "if-then" scenarios traditionally resorted to by modellers, and use real-world values for magma parameters in an attempt to more realistically simulate natural magmatic conditions and thus better understand the processes involved in actual eruptions. Recent advances in observational techniques and analysis of natural volcanic products has also added to the arsenal of information that can be used to better calibrate, constrain and/or evaluate the performance of numerical models. Although the modeling community is not yet prepared to simulate natural eruptions, the laboratory and field constraints on various critical model input parameters have now reached a point that it was possible at the workshop to develop a common set of input values so that model results could be meaningfully compared, and the modeled processes can be diagnosed in terms of functionality and sensitivity to variation of input parameters. The latter is particularly important because it can identify those parameters that need further clarification or experimental/observational constraints.

Although a growing number of volcanic eruption models is being developed using a variety of conceptual approaches, it is imperative that all models use the same values for these fundamental parameters so that their individual performance can be assessed. By running models in this way, the differences in results will reflect contrasting formulations, and thus provide insights into the processes being modeled. A specific set of experiments was devised by the modellers at the workshop using a common protocol (but with individual approaches and formulations). The objectives of the workshop were thus to produce a standard set of parameters and a set of model "experiments" to be performed by the all models. The workshop brought together the growing community of volcanic eruption modellers along with experimentalists and observationalists to accomplish two specific objectives:

1. Ensure that modellers are provided the most up-to-date values of magmatic parameters on the basis of recent experimental results and observations, and

2. Establish a standardized set of model "experiments" with identical magma and conduit characteristics and with a common protocol so that model results can be subsequently meaningfully compared and evaluated.

Participants at the workshop set up a suite of "experiments" to assess model performance using the growing body of experimental data and observations as a common set of input parameters. With this result, modeling participants have returned to their labs, and in the coming months, will perform the full suite of model runs, providing the specific set of output required by the standard protocol. The results will be subsequently compiled and evaluated to determine the sources of differences in model output. Modellers will then be able to revisit their models to explore the sensitivity of their formulations to variations in functionalization so that the key differences can be pinpointed. It is expected that once this is done, it will point to specific magma characteristics that are insufficiently understood. This, in turn, will spur experimentalists to return to their labs to focus attention on the most critical parameters identified by the model intercomparison results.

The three days of the workshop were arranged so that on the first day, the experimentalists described the state of the art in laboratory investigations of a suite of potential model input parameters. The second day was devoted to descriptions of each participating model, including purpose and intent, capabilities, input requirements, and run output. On the third day, the workshop culminated with the construction of a common set of input parameters and their values or formulations, and a set of output required of each model for the purpose of subsequent intercomparison.

The workshop and the model runs resulting from it focused on a specific interval in volcanic eruptions, starting with host magma with nucleated bubbles, and ending at the vent or at the point that the ascent rate reaches the speed of sound in magma (whichever comes first). Atmospheric eruption clouds depend largely on pre-eruptive magma dynamics, but are a separate and complex issue, so were not be addressed that the workshop. Also, magma generation and lithospheric migration was not considered, but the existence of volatile-saturated magma was the standard starting point for the workshop.

Table 1 tabulates the various input parameters defined as standards to be used by all model participating in the intercomparison.

The models are being run to produce specific output that can be meaningfully compared. All models will use the stand set of parameter values above for this comparison. A standard set of output is required of ALL models. A second set is to be produced only by a subset of models that include formulations that deal with certain parameters, and a third by another subset as indicated below. Each of these sets of model output will be compared and assessed and that the sources of any differences can be identified within the models. Significant differences are expected. By pinpointing the aspects of the model formulation that lead to the differences resulting from the intercomparison, the modellers can re-examine their formulations to explore various approaches to address the physical processes being addressed. In this way, it is expected that the modellers will be in a better position to develop more realistic formulations of actual volcanic processes.

The models will each be run in a series of "experiments" and produce the output above for comparison. In addition, certain parameters will be varied by each model to test the sensitivity of the model to such parameters.

The sensitivity experiments will include variations in the following:

This sensitivity study will entail a large number of model runs, an each model will have different sensitivity to each of the various parameters. This will be a second intercomparison, and will complement the results from the standard input set. The specifics of parameter variations values are left to the individual modellers, as appropriate for each model.

The standard input parameters were chosen a the workshop on the basis of the best presently available information. However, it was acknowledged that in many critical areas, better constraints on these model input parameters will be necessary before models can be more realistically applied to natural eruptions. To provide these constraints will require experimental efforts in the laboratory, and observational programs in the field. In addition, further theoretical development would be helpful in certain key areas. The various constraints that could be provided by various parts of the volcanological research community are listed below.

The parameter values and model protocol developed at the workshop will be used by modellers to run the specific set of "experiments" as indicated above. The results will be compared in April, 2003 at the joint AGU-EGS-EUG meeting in Nice, France (Abstract deadline Jan. 15, 2003). An open scientific session will provide a venue for all models to be presented along with their results for the standard runs as well as for the sensitivity "experiments." This session will be open to all abstract contributors regardless of attendance at the initial workshop in Durham, NH in November, 2002. After the regular scientific session, there will be a focused workshop in a "splinter session" during which modellers will have the opportunity to discuss differences within the standard and sensitivity "experiments," and will devise the next step in model development. In addition, experimentalists and observationalists will discuss strategies for obtaining and providing the necessary input and output parameter constraints as identified in the Durham workshop.

Send mail to alex.proussevitch@unh.edu with questions or comments about this web site.
Last modified: Tuesday May 08, 2007