VHub Project Description
Collaborative Research: CDI-Type II Proposal. VHub – Cyberinfrastructure for Volcano Eruption and Hazards Modeling and Simulation
1. Background
Tens of millions of people around the world are at risk from volcanic eruptions. The risk is increasing rapidly as populations grow in, and as air traffic increases above, active volcanic regions, and as national economies become increasingly intertwined. In addition to their significance to risk, volcanic eruption processes form a class of multiphase fluid dynamics with rich physics on many scales. Risk significance, physics complexity, and the coupling of models to complex, dynamic spatial datasets, demand the development of advanced computational techniques and interdisciplinary approaches to understand and forecast eruption dynamics. Innovative cyberinfrastructure is needed to enable global collaboration and creative new science, while simultaneously enabling computational thinking in real-world risk mitigation decisions – an environment where quality control, documentation, and traceability are key. This is a problem that is global in scale and cuts across resource levels – including developing nations – and can, realistically, only be solved with development of a robust cyberinfrastructure.
The role of computation in modeling volcanic activity has advanced in two main stages. First, the late 1970s and early 1980s saw the development of predominantly one-dimensional, steady state, single fluid models for eruption plumes and pyroclastic flows (e.g., Wilson, 1976; Sparks et al., 1978; Suzuki, 1983; Woods, 1988). Second, in the late 1980s through mid 1990s, time dependent, multiphase, two-dimensional computational fluid dynamics (CFD) approaches, using high-performance computing platforms, were applied to volcanic flows (Wohletz et al., 1984; Valentine and Wohletz, 1989; Dobran, 1992; Neri and Dobran, 1994; Oberhuber et al, 1998). This CFD work was largely an extension of techniques developed in the fields of mechanical and chemical engineering, demonstrating the importance of interdisciplinary exchange in volcanology. Researchers continue to build upon these developments with ongoing refinements in the basic approaches (e.g., Ishimine, 2006; Bonadonna et al., 1998), and by continued stepwise addition of complexity (e.g., Dartevelle, 2004; Suzuki et al., 2005; Dufek and Bergantz, 2007, 2008; Espositi-Ongaro et al., 2008; Suzuki and Koyaguchi, 2009). These continued refinements are very important, but integration of modeling and simulation techniques and know-how into the volcanology community has been gradual, reflecting the limitations of the fairly small teams involved in modeling complex volcano eruption dynamics (note that volcanology is a relatively young, qualitative, and dispersed discipline, with maximum local work group sizes of ~3-5 researchers in the area of modeling and simulation).
As our modeling capabilities advance gradually, huge populations around the world are vulnerable to the effects of volcanic eruptions (Valentine, 2003; Huppert and Sparks, 2006); the human and economic cascade effects of eruptions mean that their consequences can extend to the global scale. Most countries have established agencies and observatories that are charged with volcano monitoring, forecasting, and risk mitigation. (Note that we will use the term “observatories” to generically refer to organizations that are charged with volcanic risk mitigation by their governments. In some cases these might be universities, for example.) However, the economic limitations and cultural barriers of individual countries and their observatories often limit the access that workers have to the best modeling tools. In fact, a serious chasm exists between the most advanced and readily applicable modeling tools used within the academic sector and the real-time practical approaches used in volcano observatories where real hazards are addressed, especially in developing countries. This means that the observatories’ ability to plan and reduce risk lags behind currently available expertise, leaving lives and property at unnecessary risk. Failure to identify and adequately disseminate applicable tools to communities who could benefit from them most is a major and glaring shortfall within volcanology. A root cause of this is the multi-disciplinary nature of the volcanological community. The fundamental physical models arise from collaborations between applied mathematicians, computational modelers and physical volcanologists, while the gathering of detailed field data and eruption constraints is commonly the domain of either field geologists and/or observatory staff who have expertise in geophysical monitoring techniques. A mechanism, long overdue, is needed to promote integration of these approaches.
Advances in the observation of volcanic phenomena and hazards, for example in satellite remote sensing of volcanic eruption clouds, have occurred in parallel with modeling developments. Volcanic clouds are the products of explosive eruptions and drift in the stratosphere for 10-40 days each year. They are a major threat to aviation safety and a window into the science of volcano-atmosphere interaction. In July-August 2008 two Alaskan volcanoes (Okmok and Kasatochi) spread eruption clouds over the continental USA for several weeks, threatening air traffic. As this proposal is being written, Mt. Redoubt volcano, also in Alaska, is sending ash plumes to 10-15 km altitude, disrupting air traffic between North America and Asia and putting at risk a major petroleum storage facility near the volcano. These incidents demonstrate an urgent need for real-time integration of volcanic plume modeling, trajectory forecasts, and direct observations for model validation and refinement, in order to more effectively assess aircraft hazards. This is an example of the need for a medium, such as proposed here, in which to integrate models and datasets from widely scattered sources.
On top of all these considerations is the challenge of assessing the consequences of eruptions, especially the need for scaling up our experience with low level activity to possible – or even probable – much larger events. We need a global perspective that incorporates the entire academic community and government agencies, all the sciences and social sciences to grasp the certainties and identify the uncertainties before we are confronted with a major crisis. Real time assimilation, via a globally accessible cyberinfrastructure, of information and forecasts of rapidly changing environments is a way to engage the broadest group of investigators.
2. Vision and Desired Outcomes
The time is now ripe for a major transformation in volcano modeling to greatly accelerate the integration of modeling and simulation into both volcanological research and applied hazards mitigation, to expand beyond the limitations of local work groups in advancing modeling capabilities, and to make modeling tools and related data sets readily accessible to volcano observatories and other user groups around the world. We propose to achieve this through the establishment of a virtual organization, which we refer to as VHub.
VHub will incubate, accelerate, and disseminate a myriad of solutions and advances in volcanology as it enables the integration of multidisciplinary computational thinking into research and applications. The VHub cyberinfrastructure will provide a mechanism for globally collaborative research and development of computational models of volcanic processes and their integration with complex geospatial, observational, and experimental data. VHub will provide seamless accessibility of appropriate models and data to organizations around the world charged with assessing and reducing risk, reaching across resource levels and cultural boundaries. The cyberinfrastructure challenges in this effort arise from a combination of solving difficult research problems involving multi-scale physics and integration of complex spatial-temporal dynamic data sets, and the structures needed to ensure accessibility of tools to civil protection workers for real-world decision making in a manner that ensures full traceability and transparency of results. The unification of these needs will drive advances in the design and implementation of virtual organizations.
This bold, high-risk agenda addresses needs identified by the community and will be supported by six fundamental desired outcomes:
- Dissemination. Make advanced modeling and simulation capabilities and key data sets readily available to researchers, students, and practitioners around the world.
- Collaboration. Provide a mechanism for participants not only to be users but also co-developers of modeling capabilities, and contributors of experimental and observational data sets for use in modeling and simulation, in a collaborative environment that reaches far beyond local work groups.
- Comparison. Facilitate comparison between different models in order to provide the practitioners with guidance for choosing the "right" model, depending upon the intended use, and provide a platform for multi-model analysis of specific problems and incorporation into probabilistic assessments.
- Application. Greatly accelerate access and application of a wide range of modeling tools and related data sets to agencies around the world that are charged with hazard planning, mitigation, and response.
- Education. Provide resources that will promote the training of the next generation of volcanologists and hazards specialists such that modeling and simulation form part of a tripartite foundation of approaches, alongside observational data and experimentation.
- Adaptation. Conduct ongoing, rigorous self-assessment to study the impact of the virtual organization and promote continual adaptation to optimize its impact, as well as to support the broad goals of the CDI program to understand emergent collective learning and collaborative patterns.
Why will VHub’s intended outcomes transform volcanology research and hazard mitigation and achieve the vision described above?
(1) While computing has had a positive impact in volcanology, it remains the domain of only a subset of researchers. With a few exceptions, workers focused on field data collection/ interpretation or on laboratory experiments are not practicing modelers, and vice versa. We envision VHub as enabling workers who focus on data collection and laboratory experimentation to more readily include modeling and simulation in their own work, while also strengthening the integration of observational and experimental data into model development. Of course, codes that are highly complex and/or are under development will remain, principally, in the domain of workers whose main focus is modeling. Nevertheless, well-tested codes will be made readily accessible with online execution such that, for example, a field researcher can use models and simulations to test interpretations of her/his observational data (hypothesis testing), a capability that is potentially revolutionary in volcanology. Additionally, the ability to tap VHub for simulations for teaching use will ensure that future volcanologists view modeling as an integral part of how research is conducted (see Section 6).
(2) The potential for greatly expanded multidisciplinary collaboration in model development and testing is very high with VHub. Model development is often done in comparative isolation, primarily by individuals or by small working groups. Naturally this limits the range of creative solutions that can be brought to bear when new problems arise in model development. Often, creative solutions can come from workers in disciplines outside volcanology, such as mechanical engineering, meteorology, and applied mathematics. Similarly, because of the wide range of physical regimes that are modeled in volcanology (such as flows that are multiphase, turbulent, cover a wide range of Mach numbers, and involve many different length and time scales), there is much that volcanology can offer to these other disciplines. For example volcanic plumes represent a class of strongly driven buoyant plumes in the atmosphere (e.g., Chakraborty et al. 2009) with mesocyclone dynamics, that transition from dominating the flow field in the proximal region around an eruption, to being passive “tracers” as they are transported far downwind. Similarly, explosive eruptions involve highly complex multiphase, compressible, turbulent flow that is analogous to many processes of interest to engineers, including man-made explosions and chemical engineering processes (e.g., Zhang and VanderHeyden, 2001; Benyahia et al., 2005). VHub will provide mechanisms to facilitate such interdisciplinary exchange (see Section 4).
(3) VHub will provide a platform to increase the rigor and consistency of model validation and benchmarking for specific eruption processes, and an organizational center for the community to develop criteria and test cases. Validation and benchmarking are conducted only on a limited basis and with heterogeneous criteria at this time in volcanology, and are often poorly documented. This currently limits the scientific impact of modeling research due to the poor knowledge of model limitations and uncertainties, which is very detrimental both to research and to applications that form the basis for risk mitigation decisions.
(4) VHub will be a platform for development of management approaches to standardize the documentation associated with input data, source code, and output data in collaborative work (see Section 4). This will ensure that model calculations are reproducible and that consistent information is available for analysis of results and extension of previous runs with new parameter values. This is also critically important if modeling and simulation gain regular use in risk mitigation.
(5) VHub will address the major need, identified in the preceding section, to make modeling tools available to organizations such as volcano observatories. For example, a poorly equipped observatory in a developing nation will be able to use VHub-based tools online with a basic personal computer and an internet connection. Furthermore, VHub will enable interaction between observatory personnel and modeling specialists around the world to help address time-critical problems as well as disaster planning.
(6) VHub will promote the development of formal methods for inter-model comparisons and benchmarking, and multi-model forecasts using approaches similar to those being developed in the field of hydrology and petroleum reservoir management (Poeter and Hill, 2007; Diks and Vrugt, in review), and will provide mechanisms to help users select appropriate modeling tools for specific intended uses.
(7) VHub will lead to new understanding and innovation in cyberinfrastructure. VHub will build on existing “hub” technologies but will extend them by developing approaches that deal with web-based visualization, mining, and assimilation of large, complex, time-varying geospatial datasets and linking them to modeling and simulation tools. VHub will need to be developed in a manner that is robust for users with a range of expertise from field-based to computation-based, and across a large number of cultures and resource levels, and that dually supports the sometimes-conflicting needs of basic research and real-world application (see Sections 3, 5).
3. Modeling and Simulation Hierarchy and Advances in Virtual Organizations
Understanding and forecasting volcanic processes and their consequences requires a hierarchy of approaches (Fig. 1; Valentine and Perry, 2009) that includes: (1) fundamental research via the collection and interpretation of experimental data and data on volcanoes (collected onsite or remotely), and complex CFD, chemical, and solid mechanics models that attempt to capture as much of the underlying physics and chemistry of a process as possible; (2) simplified, or abstracted, models that capture the essence of a complex process but in a way that can be run quickly and in stochastic mode (e.g., Monte Carlo) in order to test a range of uncertain parameters and hypotheses, and that can be coupled with geospatial data that are key to understanding eruption consequences (e.g., topography, distribution of people and infrastructure, evacuation routes); and (3) simple risk models, the output of which can be used by (often non-technical) decision makers; for example, the probability that an area will experience a defined level of damage during an eruption, and the uncertainty in that estimate. Note that level 2 and 3 models, often in the form of the simplest empirical relationships and without a strong underpinning from the first level of the hierarchy, are used at the large majority of volcano observatories due to economic, time, and personnel limitations. Ideally, the most accurate forecasts and risk assessments would involve iteration between all levels of the hierarchy; for example, a risk model might have a high sensitivity to a certain parameter, the uncertainty of which can be reduced by further data collection or CFD modeling. This in turn often results in new fundamental knowledge of volcanic processes. VHub will make this iteration possible not just by increasing access that observatories and agencies have to modeling tools, but also access by the broad community of contributors to models and data development. Framing VHub in the context of the modeling hierarchy therefore promotes core goals: (1) substantially improving risk mitigation, and (2) rapidly advancing basic volcanological research.
The different, yet integrated, approaches that are needed for the attainment of these two goals will in turn support the third core goal of advancing cyberinfrastructure/virtual organization science itself. A virtual organization (VO) that supports basic research needs to promote creativity and collaboration and to be dynamic in the sense that it is difficult to predict how research might drive the evolution of the VO. On the other hand, the tools and data that are accessible through the VO for risk mitigation and education must be strictly controlled and tested, such that there is full transparency and traceability, and that uncertainties are well documented. On the research side, evolution of the VO is strongly bottom-up, driven by the model and data developers. On the application side, the VO needs to be managed in a more top-down manner in order to ensure that end users properly implement data and modeling tools that have been made available to them. We argue that, properly implemented, the research and application sides of VHub can not only co-exist within the same framework, but that they will benefit each other. Application tools will continually expand, and their uncertainties progressively reduce, as newly developed modeling approaches mature on the research side. The research side will benefit from work aimed to make application tools transparent, traceable, and rigorously tested. Central to both research and application is the integration of modeling and simulation with complex, geospatial data sets, many of which will be temporally dynamic (for example, the linking of plume forecasting tools with real-time meteorological and remote sensing data).
4. VHub Cyberinfrastructure Capabilities and Development
VHub will be structured in a manner to support all levels of the hierarchy described above, including the integration between large, complex data sets and models (Table 1 and following text). The VHub research team and collaborators will initially populate (and oversee the maintenance of) the model and data warehouses, focusing on modeling tools and data sets related to volcanic mass flows (pyroclastic flows, mudflows, avalanches), surges and blasts, eruption plumes and fallout, and probabilistic assessment. The initial development of VHub will provide a platform for later inclusion of a full spectrum of volcanic process models including lava flows, sector collapse, built infrastructure response, evacuation models, and various other impact models (such as economic impact), as VHub continues and expands beyond the lifetime of this initial developmental project. VHub visualization (for both simulation results and datasets) will be heavily map-based; a substantial effort in VHub development will be exploration of geospatial tools and implementation in the hub environment.
Table 1. Capabilities of VHub
VHub capability |
Desired outcomes addressed |
Capability description |
Model warehouse |
Dissemination |
Library of simulation codes, organized (in part) according to where they fit in model hierarchy, user information, users’ discussion forum, and verification & validation test cases |
Online simulation and visualization |
Dissemination |
Subset of simulation codes that can be executed online |
Interactive Simulation Tools & Tool Development |
Collaboration |
Home for development of new codes or enhancements to codes already in the model warehouse, including configuration management. |
Data warehouse |
Collaboration |
Data sets on eruption phenomena and deposits, links to existing and developing databases, experimental data sets, and tools for visualization of large data sets. |
Online resources |
Application |
Online workshops, training on modeling tools including bounds of applicability for individual tools, webinars, virtual discussions and blogs, news & events. |
Teaching materials |
Education |
Materials for both online and classroom (high school, undergraduate, and graduate level) use emphasizing the use of modeling in understanding volcanic phenomena and predicting risk. |
Feedback and suggestions forum, usage tracking |
Adaptation |
Including modules for collection on VHub usage and impacts, supporting analysis of virtual organization outcomes and continual adaptation and optimization of the system. |
The virtual organization-building software HUBzero will serve as the starting point for the implementation of VHub. HUBzero has served as the foundation not only of the highly successful nanohub project but also for a collection of additional virtual communities (see HUBzero.org for an up-to-date list of associated hubs built on the HUBzero platform) built around common interests and goals. Hubs are an emerging technological infrastructure built on open-source software (under the general label of “web 2.0”) that include web serving and scripting, databases, and simulation tools, as well as collaborative user forums, training, and access to distributed computing resources. Particularly advantageous is that the hub serves as a collecting area where many different authors can contribute, edit, and rank both informational resources (e.g. hazard models, geographic databases) and simulation tools. Thus the hub serves as a comprehensive “collaboratory” for researchers, educators, and students. Here we briefly list the major efforts in the cyberinfrastructure development of VHub with respect to the primary hub feature set.
Interactive Simulation Tools & Tool Development. We will use VHub to display algorithms already in use and to encourage community-based prototyping of new algorithms for modeling volcanic processes. HUBzero software provides a virtual network console mechanism for deploying existing GUI-enabled simulation tools within a web browser framework, reducing the need to develop and deploy specialized web-enabled application frameworks. For non-GUI applications an interface must be created (HUBzero has such an affiliated project called Rappture). The simulation package for geophysical mass flows, TITAN2D, and TEPHRA2, a model for volcanic plume dispersion and fallout, will serve as the initial “seeds” in our simulation tool repository. These two models sit at the intermediate tier of the modeling hierarchy (Fig. 1). TITAN2D has an existing Python-based GUI as well as an experimental grid-enabled portal that will serve to readily provide the TITAN2D package to a wide audience (the grid-enabled version in particular already incorporates grid authentication based on DOE certificates and a proxy service). TEPHRA2, written in C and MPI, contains numerous modules that describe details of tephra fallout. Even for the code authors, it is currently laborious to develop alternative modules based on different physical assumptions, embed these alternative modules in the code, re-make the code, and test the effects of alternative assumptions. VHub tools will be used to disassemble this code, so that researchers and advanced students can explore aspects of tephra dispersion in isolation, identifying the effects of various assumptions in alternative models, and then quickly link this isolated bit of code to the overall model to test the ultimate effect of alternative models on particle dispersion. VHub will provide the flexibility for models to be tested using different languages. Entirely new tools will utilize the integrated tool-building framework available within HUBzero that will facilitate the uploading, compiling, testing, and deployment of new simulation tools. Included in this framework for each tool is its own project space, source code management, ticketing system for support and bug tracking, and a wiki site for documentation. We will use these tools to bring the multiphase hydrodynamics code GMFIX (Dartevelle, 2004; Dartevelle and Valentine, 2007), which fits in the bottom tier of the modeling hierarchy, into VHub. Our international collaborators will also contribute content to VHub ranging from experimental data (partly for use in validation of simulators) to high-level risk models, as will a wide range of workers who have already expressed commitment to use or contribute to VHub (supplementary materials). These features will focus code development in volcanology, encourage model verification and perhaps model validation, and act as a catalyst for models of volcanic processes to evolve at an accelerated rate.
International collaborators and others from the broad community will contribute datasets and modeling tools as part of this initial “seeding” process for VHub, and these will range across the Figure 1 hierarchy. Examples include Bayesian Event Tree tools, atmospheric transport codes, laboratory analog experimental data, datasets on key eruption parameters used in plume modeling, mass flow codes, and multiphase fluid dynamics codes.
It is vital that VHub engage a broader, multidisciplinary community to develop and improve models. For example we anticipate that 3-D atmospheric chemistry models such as GEOS-Chem (http://www.as.harvard.edu/ctm/geos/) that provide near real-time chemical forecasts and engage key atmospheric scientists in volcanic plume modeling.
Data warehouse. High quality satellite observations of volcanic phenomena are buried in extensive online archives at agencies such as NASA and NOAA, but there is currently no single archive dedicated to collating volcanic datasets from multiple satellite sensors and from various trajectories, for either atmospheric chemistry or fallout models. VHub will collate available datasets for multiple eruption case studies, along with visualization tools, in a single archive for direct and rapid application to model validation and refinement. One of the core partners (Michigan Technological University; MTU) will gather a primary archive of eruption data, with a mirror at the central VHub server. Links to relevant existing online databases will also be included, such as the World Volcano Observatory database (wovodat.org) and atmospheric SO2 data (satepsanone.nesdis.noaa.gov/pub/OMI/OMISO2/index.html). Data will be subsetted (to limit extraneous data) and tagged (e.g., by location, eruption type, magma composition, plume altitude) to facilitate data mining. See discussion below for a more detailed discussion of how large datasets will be handled.
Online Presentations. Online presentations based on PowerPoint are hosted using Macromedia Breeze, providing both video and audio of the presentation to any Flash-based viewing client (much more friendly for slow speed Internet connections). Capabilities will support live presentations, online archives, and podcasts. VHub will initially contain presentations covering training and application of TITAN2D, TEPHRA2, and GMFIX and their supporting simulation tools, as well as software tools and datasets that will be contributed by the collaborators and by members of the broad community. This supports the Online Resources and Teaching Material components (Table 1)
Uploading New Resources: A standard feature of the HUBzero package is a “wizard” for uploading new informational resources (tools, data, presentations, etc.) that guides the user through the procedure for each content area. Hub administrators approve all submissions, which upon approval are automatically added to the news and RSS feed. VHub personnel will serve as primary administrators for this service, including representation from the international collaborators. Of particular interest to VHub is hosting and maintaining a collaborative set of geographic databases.
Ratings/Citations: Registered users have the ability to rate, comment upon, and provide citations for the content of each resource, which together with the access statistics are used to form an overall rating for each resource. VHub will serve as a focus for information on particular hazard locations, providing a key coordinating resource for simulations, citations, and field studies.
Content Tagging: User and administrator defined key strings that are catalogued for definition with links to content areas. VHub will utilize this tagging to provide an enhanced repository of past simulations, tagged by location and initial flow information. This repository will be very useful as a starting point for new users as well as experienced users of simulation tools when first approaching a new geographic location.
Group Management for Private Collaboration: Any registered user to the hub can create (and manage) a private group for special collaborative purposes (e.g., for early development work on a new tool or new database). VHub is expecting to make use of such private groups for developing experimental versions of models such as TITAN2D as well as developing new geographic databases.
User Support/Feedback: HUBzero has an included help ticketing system whereby users can use a simple web form to submit help requests or problem reports to the hub administration team. Also included is a general support forum for requests that go beyond the expertise of the administrators. Of particular use to VHub is the knowledge base that is created through the use of this forum, as the accumulated expertise of the broader community is brought to bear on hazards applications.
News/Events: Registered users can post events to a community-driven hub calendar, and administrators can push items of general interest to the VHub community to the news area.
Usage Statistics: Comprehensive usage statistics are reported for each content area and simulation tool. VHub will utilize these statistics not only for feedback but also to focus development and enhancement, particularly with respect to the educational outreach goals (see Section 5).
We have budgeted for hosting VHub directly at HUBzero. Using HUBzero’s hosting service is cost effective in several ways, first in that we could not independently set up and maintain the HUBzero software at such a price point (less than the cost of 1 FTE), and second that we can leverage the experience of HUBzero staff to vastly decrease the time to achieve an initial functioning hub. We anticipate approximately 0.25 FTE for daily maintenance of the VHub site, in addition to the regular usage (along with regular PI/co-I administrative access and use). We will dedicate more time to developing tools within the hub framework for managing and visualizing geospatial data.
Our initial focus on visualization tools will be a hierarchical approach using Google Earth for dynamic spatial viewing (including features and highlights using overlays using the Keyhole Markup Language, or KML), the Generic Mapping Tools (GMT; gmt.soest.hawaii.edu), and more interactive tools based on GMT like iGMT and SEATREE for customized static maps, and ParaView for detailed visualization of complex two and three dimensional data sets (mass flows, vector fields and profiles). An especially promising tool for the integration of volcanological data with models and simulation is based on the Revolution Enterprise tool, as developed already for standalone dynamic digital maps (DDMs) by Condit (1999, 2005). DDMs currently use georeferenced maps and data in a web-enabled environment to facilitate and present map-based research. There is great potential to expand this already successful approach in the context of VHub by externally linking DDMs to simulation codes, allowing for example, the rapid comparison of model results and mapped data. Individual researchers within the membership of the VHub collaboration are already using all of these tools, hence the decision to initially utilize them for standards of VHub visualization. All of these packages are open source, and any improvements made through VHub will be provided back to the community. A more complex issue lies with remote visualization through VHub, potentially of significant importance when viewing large data sets over low bandwidth networks. Addressing this issue will be an ongoing effort, but we note that ParaView, the most graphically intensive visualization application in the initial VHub portfolio, has significant client-server capabilities in which only geometric data need to be sent to the viewing clients, thereby relying on the clients to perform the most intensive rendering tasks.
Data assimilation and propagation issues and challenges. VHub can house approximately 0.7TB of data associated with the primary hub at HUBzero.org, which includes presentations and training materials, software, and key shared data resources. VHub as a collaborative VO, however, will need an extensible method for adding and participating in new data repositories, especially those being built by other laboratories and research teams. It is not anticipated that all groups will want to have their data mirrored either by the main VHub site, or the University at Buffalo (UB) data repository, due to considerations of privacy and control as well as size, as synchronizing multi-TB data sets over available (non-TeraGrid) networks would be prohibitively slow. We propose to adopt data “bricks” that support data sharing through established (grid-based) middleware, the leading contender for which is iRODS (Integrated Rule-Oriented Data System) (www.irods.org). iRODS is gaining traction as a follow up to the Storage Resource Broker (SRB, http://www.sdsc.edu/srb/index.php/Main_Page), and has been developed by a subset of the developers of SRB. iRODS (and SRB) serves as middleware, providing distributed data services in grid-based environments. Such data grids are designed to provide virtualized access to disperse data collections, particularly in collaborative environments, and provide both directory and file specific access controls. Freeing users from the semantics of the underlying hardware, iRODS provides numerous client interfaces to facilitate access via methods familiar to users including webDAV, Windows explorer and UNIX-like command-line tools. iRODS and also allows individual data elements to be augmented/tagged with metadata that can later be used to locate desired elements. A rules-based system allows iRODS to provide not only data sharing, but also policy enforcement, validation, and workflow management (hence data security, privacy and integrity are still maintained). Metadata associated with the iRODS discovery process will also greatly assist in the automated tagging/tracking of data elements within VHub (such that all data will have an associated “pedigree”).
iRODS has been successfully deployed by a variety of funded projects both in the US and internationally, including both the UK e-Science data grid and the NSF Teragrid. Leveraging the established flexibility of iRODS will greatly facilitate VHub’s integration and collaboration with other grid/e-science initiatives worldwide (e.g., Rajasekar et al., 2007; see also agility.jpl.nasa.gov/documents/irods-testing/irods-summary).
The initial brick will be deployed at UB (approximately 12TB in size) and integrated with VHub. Any group wishing to set up a similar repository with access to and from VHub while allowing them full control of their own research data can then use this initial brick as a model. Alternatively, data hosted at UB can be expanded to accommodate VHub contributed data.
Metadata tracking and management. Models and simulations of different volcanic process such as lava and pyroclastic flows generate datasets that are the result of running a code using specific parameters for material properties, composition, initial and boundary conditions, as well as specific constitutive models. These resulting datasets do not always describe all the parameters used for their generation. For this reason, the VHub team will work to design and implement a database using a common metadata schema in order to describe and track the modeling components, model sources and input and output datasets, building upon work such as that described by Dunlap et al. (2008). Results of models produced in different centers also can be stored in this common structure, which will allow making comparisons between them for obtaining a feedback, which can help to improve the further source code developments. We will also work with the volcanology community to develop different “grades” or levels of metadata certification for observational and/or experimental datasets. For example a dataset that is used via VHub for a benchmarking exercise for codes that model the flow and deposition from pyroclastic density currents would need to have a high “grade” in terms of the detail of metadata, such as measurement locations and uncertainties in the data. Similarly high-grade data would be most desirable for use in risk mitigation, when possible. Lower grade datasets can still be useful but the grade will ensure that a user understands the uncertainties and limits to reproducibility for the data.
GeoProMT implementation. Use of VHub-based modeling and data tools for risk mitigation efforts at observatories should, ideally, be conducted in a systematic manner that supports traceability, transparency, reproducibility and sharing of information; ideally, basic research efforts should have the same goals. Optimum decision support requires careful and coordinated consideration of how the natural hazard processes, the gathered observations, the modeling algorithms and related uncertainties are represented in data and simulation models used. GeoProMT - the Geospatial Project Management Tool - is an internet-based interface (developed by Renschler in collaboration with the UB Educational Technology Center) that allows collaborative researchers, observatory personnel, and students to systematically share and investigate representations of earth systems' properties and processes at various scales. Scaling is here referred to as the transformation of information from one spatial/temporal scale to another (e.g. an interpolation, aggregation, disaggregation). Usually data transformation in the digital domain occurs in the following sequence: (1) Process Scale, (2) Measurement Scale, (3) Database Scale, (4) Modeling Scale, (5) Prediction Scale, (6) Assessment Scale, and again (1) Process/Validation Scale (Renschler, 2003, 2005). In the GeoProMT framework, each one of these steps is documented and validated before moving to the next step, insuring traceability, the capture of uncertainties, and reproducibility. The information shared is not only limited to geospatial data (process models, GIS, DDMs, or remote sensing), since the GeoProMT framework can also be used to share transformation of information of other documents such as literature (English and other languages) and images. The team will implement GeoProMT within VHub, including training on its use, and will evolve GeoProMT as users learn lessons.
5. VHub as a Case Study in Virtual Organizations
Part of the project will be a “study of VHub” and this will occur on two levels. One level will focus on learning from and optimizing the VO for the research community and for education. We will use data from the usage statistics, user support & feedback, and ratings/citations functions (Section 4) to address questions such as (but not limited to): (1) Who is using VHub and where are they from? (2) How did they find out about VHub? (3) How are they using VHub, what components do they visit? (4) How frequently do individuals use VHub and how long is each session? (5) How does VHub usage vary with time and with events such as when a particular volcano is erupting? We will also conduct surveys aimed at helping the team understand the thought processes and reasoning of those using VHub. This work will build upon research on virtual learning communities (e.g., Nachmias et al., 2000; Johnson, 2001). During the four-year project we will analyze this information at regular intervals (e.g., quarterly) and use the results to continually improve the VO. Additionally, the results will allow us to explore fundamental issues and models of cyberinfrastructure and virtual organizations that we will publish in peer reviewed literature. One issue of interest mentioned in Section 2 is the interplay between the bottom-up research side and the top-down application side.
As described in Sections 2 and 3, we also intend for VHub to support risk mitigation at volcano observatories around the world, representing a wide range of cultures, political and social situations, and resource levels. We envision VHub as a vehicle for educating populations and decision makers about potential scenarios, as it enables visualization of large volcano datasets and complex simulations and forecasts, as well as providing a platform for decision-analysis tools such as Bayesian event trees, expert elicitation, and high level risk models (Figure 1). These aspects of application and education reach beyond those of other science and engineering VOs that we are aware of. Therefore, a second component of the “study of VHub” will be an ongoing, formative assessment of the impact of the VO on the way observatories conduct the business of risk mitigation and communication. A small sample, probably between two to four, of observatories will be invited to participate in this assessment, which will involve annual visits by team members to the observatories. Team members will observe the work of the observatory and conduct face-to-face interviews to learn how a diverse set of users are incorporating VHub into their work. For example, which components and features are most useful, which need some modification to improve their impact, and which need to be redesigned or replaced in order to be effective. We will use data from these site visits to test concepts such as the technology acceptance model (TAM; Davis, 1989), and the task-technology fit model (TTF; Goodhue and Thompson, 1995; Zigurs and Buckland, 1998). The TAM suggests that there are two specific variables that are fundamental determinants of user acceptance of technology: perceived ease of use and perceived usefulness. The TTF model suggests that information systems can have a positive impact on user utilization and performance when there is a correlation between a task and the functionality of the technology used by the system. We also have set aside a small portion of the budget to support exchanges with other VO/cyberinfrastructure projects in order to exchange lessons learned and build upon others’ experiences.
6. Education and training
VHub education and training will take place on two levels. The first level focuses on educating workers (researchers, observatory personnel, graduate student conducting research) on the use and capabilities of VHub itself and on specific modeling and data tools. Three approaches will be used for this: (1) Online seminars and tutorials (real-time, archived online, and podcasts; see Section 4). (2) Small workshops or seminars led by VHub participants in conjunction with major conferences (e.g., American Geophysical Union, International Volcanological Congresses) and in locations that are easily accessible by a number of observatories. (3) “Go-to” workshops where VHub participants travel to a site, for example a volcano observatory or to a location that is easily accessible by multiple observatories, to introduce the workers to VHub and available tools. Some of these workshops will be done in conjunction with formative assessment visits to a subset of observatories (Section 5).
The second level is the use of VHub modeling and simulation tools and data sets to teach about volcanic processes in university and high school settings. A portfolio of simulations and datasets will be developed, along with background materials, for use in introductory undergraduate and high school courses (e.g., Introductory Geology, or an entry-level Natural Disasters course) and in special workshops. For example, Pitman organizes an annual workshop in computational science for Buffalo area high school students (ccr.buffalo.edu/display/WEB/Summer+Workshop). Over the years, topics have changed – computational chemistry, visualization, bioinformatics and computational biology. New modules can be developed and used as part of a geosciences workshop series; as the workshop students extend the original modules, these can be integrated into the program. All of the code and presentation materials developed will be made available to the VHub community. For more advanced coursework (upper level undergraduate or graduate) we will develop exercises that involve students directly using modeling tools to test hypotheses and sensitivity to initial and boundary conditions. Some of these exercises will draw on the successful experience of developing problem-solving skills as part of a numeracy focus in geoscience courses (e.g., the "spreadsheets across the curriculum" model of Vacher , 2001) and map-based problem-solving (e.g., the DDM approach of Boundy and Condit, 2004). For graduate students learning to develop models and/or work with complex data sets, VHub will provide an excellent online laboratory where they can work in a global collaborative environment to develop their skills and bring new modeling capabilities to the community. Specific lessons will be provided to NSF’s Digital Library for Earth System Education (www.dlese.org). VHub-developed or –based educational materials will be presented at conferences such as the National Science Teachers of America, American Geophysical Union, and Geological Society of America annual conferences.
References cited
Benyahia, S., Syamlal, M., O’Brien, T.J. (2005) Evaluation of boundary conditions used to model dilute turbulent gas/solid flows in a pipe. Powder Technology 156, 68-77.
Bluth, G.J.S., Carn, S.A. (2008) Exceptional sulphur degassing from Nyamuragira volcano, 1979-2005. International Journal of Remote Sensing 29, 6667-6685.
Bonadonna, C., Ernst, G.G.J., Sparks, R.S.J. (1998) Thickness variations and volume estimates of tephra fall deposits: the importance of particle Reynolds number. Journal of Volcanology and Geothermal Research 81, 173-187.
Bonadonna, C., Connor, C.B., Houghton, B.F., Connor, L., Byrne, M., Laing, A., Hincks, T.K. (2005) Probabilistic modeling of tephra-fall dispersal: hazard assessment of a multiphase rhyolitic eruption at Tarawera, New Zealand. Journal of Geophysical Research 110, B0320310.1029/2003JB002896.
Boundy, T.M and Condit, C.D. (2004) Bringing the field into the classroom by using dynamic digital maps to engage undergraduate students in petrology research. Journal of Geoscience Education 52, 313-319.
Bursik, M., Rogova, G. (2006) Use of neural networks and decision fusion for lithostratigraphic correlation with sparse data, Mono-Inyo Craters, California. Computers and Geosciences 32, 1564-1572, doi:10.1016/j.cageo.2006.02.012.
Bursik, M., Kobs, S., Burns, A., Braitseva, O., Bazanova, L., Melekestsev, I., Kurbatov, A., D. Pieri, D. (2009) Volcanic plumes and the wind: jetstream interaction examples and implications for air traffic. Journal of Volcanology and Geothermal Research, doi:10.1016/j.jvolgeores.2009.01.021.
Byrne, M.A., Laing, A.J., Connor, C.B. (2007) Predicting tephra dispersion axes with a mesoscale meteorological model and a particle fall model: Application to Cerro Negro, Nicaragua. Journal of Applied Meteorology 46, 121-135.
Carn, S.A., Bluth, G.J.S. (2003) Prodigious sulfur dioxide emissions from Nyamuragira volcano, D.R. Congo. Geophysical Research Letters 30, doi:10.1029/2003GL018465.
Carn, S.A., Bluth, G.J.S., Sawyer, G.M., Oppenheimer, C., Gross, E.L., GVO staff (2006) Continental volcanic degassing at Nyamuragira and Nyiragongo (DR Congo). Presented at the IAVCEI Conference on Continental Volcanism, Guangzhou, China, May 14-18, 2006.
Carn, S.A., Kervyn, F., Mitangala, P., Sawyer, G.M., van Overbeke, A.-C. (2007) The health impacts of persistent degassing at Nyiragongo volcano (DR Congo). Presented at Cities on Volcanoes 5, Shimabara, Japan, November 19-23.
Chakraborty, P., Gioia, G., Kieffer, S.W. (2009) Volcanic mesocyclones. Nature 458, 497-500, doi:10.1038/nature07866
Condit, C.D. (1999) Components of dynamic digital maps. Computers & Geosciences 25, 511-522.
Condit, C.D. (2005) Dynamic Digital Maps: A Means to Distribute Maps and Associated Media via Web and CD. In D.R. Soller (ed.) Digital Mapping Techniques '05 -- Workshop Proceedings. U.S. Geological Survey Open-file Report 2005-1428, 16 pp.
Connor, L.J., Connor, C.B. (2006) Inversion is the solution to dispersion: Modeling tephra fallout from the 1992 eruption of Cerro Negro volcano, Nicaragua. In: (H.M. Mader, S. Coles and C.B. Connor, eds.) Statistics in Volcanology, IAVCEI Series Volume 1, Geological Society of London.
Connor, C.B., Sparks, R.S.J., Mason, R.M., Bonadonna, C., Young, S.R. (2003) Exploring links between physical and probabilistic models of volcanic eruptions: The Soufriere Hills Volcano, Montserrat. Geophysical Research Letters 30, doi:10.1029/2003GL017384.
Dartevelle, S. (2004) Numerical modeling of geophysical granular flows: 1. A comprehensive approach to granular rheologies and geophysical multiphase flows. Geochemistry Geophysics Geosystems 5, Q08003, doi: 10.1029/2003GC000636.
Dartevelle, S., Valentine, G.A. (2007) Transient multiphase processes during the explosive eruption of basalt through a geothermal borehole (Námafjall, Iceland, 1977) and implications for natural volcanic flows. Earth and Planetary Science Letters 262, 363-384.
Davis, F. D. (1989) Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13, 319 - 340.
Diez, M. (2006) Solution and parametric study of a combined conduit and eruption column model. In: (H.M. Mader, S. Coles and C.B. Connor, eds.) Statistics in Volcanology, IAVCEI Series Volume 1, Geological Society of London.
Diez, M., La Femina, P.C., Connor, C.B., Strauch, W., Tenorio, V. (2005) Evidence of static stress changes triggering the 1999 eruption of Cerro Negro, Nicaragua, and regional aftershock sequences. Geophysical Research Letters 32, L04309 10.1029/2004GL021788.
Diks C.G.H., Vrugt, J.A. (in review) Comparison of model averaging methods for forecasting hydrologic systems. Stochastic Environmental Research and Risk Assessment.
Dobran, F. (1992) Nonequilibrium flow in volcanic conduits and applications to the eruptions of Mt. St. Helens on May 18, 1980, and Vesuvius in A.D. 79. Journal of Volcanology and Geothermal Research 49, 285-311.
Dufek, J., Bergantz, G.W. (2007) Suspended load and bed-load transport of particle laden gravity currents: the role of particle-bed interaction. Theoretical and Computational Fluid Dynamics 21, p. 119-145.
Dufek, J., Bergantz, G.W. (2008) Dynamics and deposits generated by the Kos Plateau Tuff eruption: Controls of basal particle loss on pyroclastic flow transport. Geochemistry Geophysics Geosystems 8, doi:10.1029/2007GC001741
Dunlap, R., Mark, L., Rugaber, S., Balaji, V., Chastang, J., Cinquini, L., DeLuca, C., Middleton, D., Murphy, S. (2008) Earth system curator: metadata infrastructure for climate modeling. Earth Science Informatics 1, 131-149, doi:10.1007/s12145-008-0016-1
Espositi-Ongaro, T., Clarke, A.B., Neri, A.,Voight, B., Widiwijayanti, C. (2008) Fluid dynamics of the 1997 Boxing Day volcanic blast on Montserrat, West Indies. Journal of Geophysical Research 113, B03211, doi:10.1029/2006JB004898.
Goodhue, D. L., Thompson, R. L, (1995) Task-technology fit and individual performance. MIS Quarterly 19, 213 - 236.
Hale, A. J., Calder E. S., Loughlin, S., Wadge, G. (submitted, a) Modelling the Lava Dome Extruded at Soufriere Hills Volcano, Montserrat, August 2005 - May 2006. Part I: Dome shape and internal structure. Journal of Volcanology and Geothermal Research (Submitted Jan ‘09)
Hale, A. J., Calder E. S., Loughlin, S., Wadge, G. (submitted, b) Modelling the Lava Dome Extruded at Soufriere Hills Volcano, Montserrat, August 2005 - May 2006. Part II: Rockfall Activity and Talus Deformation. Journal of Volcanology and Geothermal Research (Submitted Jan ‘09)
Head, E.M., Bluth, G.J.S., Carn, S.A. (2007) Eruptions of Nyamuragira volcano, D.R. Congo (1980-2006). Presented at the 26th ECGS Workshop, Active Volcanism and Continental Rifting with special focus on the Virunga (North Kivu, DRC), European Center for Geodynamics and Seismology, Luxembourg, November 19-21.
Head, E.M., Wallace, P.J., Carn, S.A., Bluth, G.J.S., Sims, K.W.W., Shaw, A. (2008) Interpreting SO2 degassing with melt inclusions: Nyamuragira volcano, D.R. Congo (1980-2006). Presented at the IAVCEI General Assembly, Reykjavik, Iceland, Aug 18-22.
Huppert, H.E., Sparks, R.S.J. (2006) Extreme natural hazards: population growth, globalization and environmental change. Philosophical Transactions of the Royal Society Series A 364, 1875-1888.
Ishimine, Y. (2006) Sensitivity of the dynamics of volcanic eruption columns to their shape. Bulletin of Volcanology 68, 516-537.
Johnson, C.M. (2001) A survey of current research on online communities of practice. The Internet and Higher Education 4, 45-60
LaFemina, P.C., Connor, C.B., Hill, B.E., Saballos, J.A., Strauch, W. (2004) Exploring the link between tectonism and magmatism: the 1999 seismic swarm and eruption of Cerro Negro, Nicaragua. Journal of Volcanology and Geothermal Research 137, 187-199.
Nachmias, R., Mioduser, D., Oren, A., Ram, J. (2000) Web-supported emergent-collaboration in higher education courses. Educational Technology 3, ISSN 1436-4522.
Neri, A., Dobran, F. (1994) Influence of eruption parameters on the thermofluid dynamics of collapsing volcanic columns. Journal of Geophysical Research 99, 11833-11857.
Oberhuber, J.M., Herzog, M., Graf, H.-F., Schwanke, K. (1998) Volcanic plume simulation on large scales. Journal of Volcanology and Geothermal Research 87, 29-53.
Poeter, E.P., Hill, M.C. (2007) MMA, a computer code for multi-model analysis. Techniques and Methods Report 6-E3, U.S. Geological Survey.
Rajasekar A., Vernon, F., Moore, R. (2007) Rule-Oriented Data Systems: A Grid-Based Cyberinfrastructure for Large-Scale and Long-Term Data Management. Fall meeting of the American Geophysical Union, San Francisco, December 2007.
Renschler, C.S. (2003) Designing geo-spatial interfaces to scale process models: The GeoWEPP approach. Hydrological Processes 17, 1005-1017.
Renschler, C.S. (2005) Scales and Uncertainties in volcano hazard prediction. Journal of Volcanology and Geothermal Research 139, 73-87.
Rogova, G., Bursik, M., Hanson-Hedgecock, S. (2007) Interpreting the pattern of volcano eruptions: intelligent system for tephra layer correlation. In The 10th International Conference on Information Fusion 10, 1-7, http://ieeexplore.ieee.org/Xplore.
Rogova, G.L., Bursik, M.I., Hanson-Hedgecock, S. (2009) Intelligent system for interpreting the pattern of volcano eruption. ISIF Journal of Advances in Information Fusion, in press.
Sawyer, G.M., Carn, S.A., Oppenheimer, C., Tsanev, V.I., Burton, M. (2008) Investigation into magma degassing at Nyiragongo volcano, Democratic Republic of Congo. Geochemistry Geophysics Geosystems 9, Q02017, doi:10.1029/2007GC001829.
Sparks, R.S.J., Wilson, L., Hulme, G. (1978) Theoretical modeling of the generation, movement, and emplacement of pyroclastic flows by column collapse. Journal of Geophysical Research 83, 1727-1739.
Suzuki, T. (1983) A theoretical model for dispersion of tephra. In: Volcanisn: Physics and Tectonics (Shimozura, D., and Yokoyama, I., eds.), Tokyo, 95-113.
Suzuki, Y. J., Koyaguchi, T. (2009) A three-dimensional numerical simulation of spreading umbrella clouds. Journal of Geophysical Research 114, B03209, doi:10.1029/2007JB005369
Suzuki Y. J., T. Koyaguchi, M. Ogawa, I. Hachisu (2005), A numerical study of turbulent mixing in eruption clouds using a three-dimensional fluid dynamics model. Journal of Geophysical Research 110, B08201, doi:10.1029/2004JB003460.
Vacher, H.L. (2001) Better math, better geology. Geotimes 46, n. 3, pp. 13, 31.
Valentine, G.A. (2003) Towards integrated natural hazard reduction in urban areas. In: Geosciences in the Cities (Heiken, G., Fakundiny, R., Sutter, J., eds.), American Geophysical Union Special Publication Series 56, 63-74.
Valentine, G.A., Perry, F.V. (2009) Volcanic risk assessment at Yucca Mountain, NV, USA: integration of geophysics, geology, and modeling. In: Volcanism, Tectonism, and Siting Nuclear Facilities (Connor, C., Connor, L., Chapman, N., eds.), Cambridge University Press, in press.
Valentine, G.A., Wohletz, K.H. (1989) Numerical models of Plinian eruption columns and pyroclastic flows. Journal of Geophysical Research 94, 1867-1887.
Wilson, L. (1976) Explosive eruptions, III. Plinian eruption columns. Geophysical Journal of the Royal Astronomical Society 45, 543-556.
Wohletz, K.H., McGetchin, T.R., Sandford II, M.T., Jones, E.M. (1984) Hydrodynamic aspects of caldera-forming eruptions: Numerical models. Journal of Geophysical Research 89, 8269-8286.
Woods, A.W. (1988) The fluid dynamics and thermodynamics of plinian eruption columns. Bulletin of Volcanology 50, 169-193.
Zhang, D.Z., VanderHeyden, W.B. (2001) High-resolution three-dimensional numerical simulation of a circulating fluidized bed. Powder Technology 116, 133-141.
Zigurs, I., Buckland, B. K. (1998) A theory of task/technology fit and group support systems effectiveness, MIS Quarterly 22, 313-334.