The programme and the presentations can be downloaded from this page.
The feedback from the attendants can be found in the minutes of the regular OATC meeting under Public Talk.
CSDMS) is a US initiative to enhance the international cooperation among earth surface scientists, to stimulate the development of interoperable simulation components, to increase the use of high-performance computing in surface dynamics modeling, and to provide a framework enabling integrated model coupling. It is funded by National Science Foundation and co-sponsored by various parties including the Office of Naval Research and NASA. The CSDMS community develops, supports, and disseminates integrated software modules that predict the movement of fluids, and the flux (production, erosion, transport, and deposition) of sediment and solutes in landscapes and their sedimentary basins. CSDMS nvolves the Earth surface - the dynamic interface between lithosphere, atmosphere, cryosphere, and hydrosphere.
CSDMS has chosen the OpenMI interface as a central element in the framework being built. The OpenMI standard will be combined with the Common Component Architecture (CCA) to provide a platform that runs on a single PC as well as on supercomputer infrastructure. It is an exciting new step in the development of the OpenMI standard to investigate together how well its concepts can be used to harness the computational power of the largest computers around. While raising very interesting technical issues, this choice for OpenMI underpins once more the international recognition for the standard and significantly broadens the range of models that can be coupled using the OpenMI interface.
Integrated Modeling for Integrated Environmental Decision Making White Paper. The approach advocated is open and outward looking with the development of an international community of practice forming an important part of the strategy. It recognises that to achieve acceptance, the value and cost effectiveness of integrated modelling has to be demonstrated. The help of research institutes and universities will be needed on two counts: first to solve the many scientific issues that will emerge as integrated modelling becomes widely applied, second to build a new body of skilled integrated modellers. A community of practice is needed because no single organisation has the resources, intellectual or financial, to meet all the challenges ahead. However, if the full power of the open source model can be harnessed, then there is a good chance that the rate of innovation can be driven up to the benefit of the whole modelling community.
From the 10th - 12th of December 2008, the Council for Environmental Regulatory Modelling convened a workshop for EPA staff and the agencies and industries with whom they work with. Its purpose was to establish support for integrated modelling and build a community of practice. An additional EPA objective was to establish a collaboration with the OpenMI Association.
The keynote speaker was Phillip Dibnerof the Open Geospatial Consortium. In extremely practical terms, he explained how the OGC raised the funds and resources that enabled it to do its work. He has also dealt with difficult issues, for example, achieving consensus on the direction of standards such as the OGC. To inform the meeting on the task ahead, a really lively bunch of external speakers was assembled. All were involved in developing enabling technologies for integrated modelling and building international communities to apply them to complex problems.Β Roger Moore of the Centre for Ecology and Hydrology and Chairman of the OpenMI Association talked on why openness and collaboration are essential for integrated modelling and are the key to the future of the OpenMI. Dan Ames of Idaho State University gave an entertaining account of waking up to discover that scattered around the world were 140,000 copies of his Map Windows software. However, the really exciting moment for him came when he realised that his users were beginning to sort each others problems out. Larry Murdoch of Clemson University outlined the plans for the CUAHSI Community Hydrologic Modelling Platform (CHyMP), the US counterpart of Wallingford Software's Openweb. Ken Rojas of the USDA and the new chairman of the Interagency Multi-Media Modelling Group explained how theΒ nine agency MOU for environmental modelling was being reinvigorated. Last but by no means least, Christopher Sherwood of the USGS and theΒ Community Sediment-Transport Modelling System described their experiences with an open-source approach for developing coupled models for waves, currents and sediment transport in coastal and estuarine environments.
A lively and thoughtful discussion followed, which considered different aspects of the need for a community of practice, its scope,Β objectives and how it might be organised. The discussions didn't always flow easily. It took some time for people to adjust to the idea of an international, open community as opposed to an EPA or US community. There was also concern that the new organisation should not replicate the Interagency Multi-Media Modelling Group or the international Environmental Modelling System society (iEMSs). However, these problems look solvable and the outcome was an agreement to form a 'Community for Integrated Environmental Modelling' (a working title until someone thinks of something better) - to join, contact Dan Ames.
One of the EPA's tasks in taking its integrated modelling forward is to upgrade its FRAMES system. The current version has reached the end of its life-cycle. While it comprises 80 models and a powerful set of tools, its major limitation is that process interaction can only be represented by running models sequentially, with data exchange being affected by file transfer. To explore how this situation can be improved, the EPA has been evaluating OpenMI. The outcome of the meeting is that the EPA will prototype a new design for FRAMES 3, where file transfer is replaced or supplemented by the OpenMI allowing data exchange on a time step by time step basis.
The JGrass team hosted the January 2009 OpenMI Association Technical Committee meeting at the University of Trento in Italy. Since 2007 JGrass implements in all its models and tools the OpenMI interfaces. A Console Engine (The JGrass Console4) has been developed to link together models the OpenMI way through a scripting language (Groovy or Beanshell, at the moment) that is then translated into OpenMI's linkable language for execution. This is a powerful tool that can be used by the scientific community to quickly create complex applications. Apart from the Horton Machine toolkit that contains more than 40 geomorphological and hydrological modules, users can find in JGrass a semi-distributed hydrological model for peak-flow analysis, and the Shalstab hillslope stability model. Besides, a full hydrological model is being implemented for the management of water resources, and river basins. The model is currently being tested on part of the second river of Italy, the Adige, financed by the Adige river basin authority. In another project, also the distributed hydrological model GEOtop4, is being ported to the system.
JGrass and all its models are released under the LGPL and GPL Free and Open Source license, and are freely downloadable at the project's homepage.
Actually, for its modeling purposes, the JGrass team extended the version of the 1.4 OPENMI development toolkit to include in memory the data types supported by the GIS community. These were taken from the GeoAPI project, that in these days is being proposed by Geomatys5 to the current OGC6 (Open Geospatial Consortium) meeting, as reference implementation for the OGC standards.
Over 60 participants from a wide range of organisations and industries attended the workshop, including representatives of the project partners and members of the OpenMI Association and its committees. Roger Moore of CEH Wallingford, President of the OpenMI Association, opened the day's proceedings with an introduction to the OpenMI and the OpenMI LIFE project. The LIFE project aims to demonstrate the applicability of the OpenMI through use on a number of real-world examples in the Scheldt and Pinios basins. The project aims to identify the issues that arise when models are joined together, widely disseminate those issues, and promote further research to investigate and resolve those issues.
Each of the four main use cases was then presented. Johan Van Assel of Aquafin detailed the work on the first Scheldt use case which is linking existing InfoWorks CS (sewer) and InfoWorks RS (river) models to improve the overall representation of flood processes in Leuven. He highlighted a key feature of this particular study which is the number of points where the river and sewer systems interact through both level and flow, and in total there are almost 200 points of interaction that have been represented. Most of the testing time has been spent ensuring the model linkages at these points are stable and give a reasonable representation of the flows through physical connections.
Neel Devroede of VMM-AWA then presented the second use case, which is linking together two river models to improve the representation of flooding in the Dijle river downstream of Leuven. OpenMI provides the mechanism to link a MIKE-11 model of the downstream river with an InfoWorks RS model upstream. Mr. Devroede identified some issues arising from linking the models, including the need to recalibrate the upstream models as a result of improved downstream boundary conditions. Initially, stability had been a concern, but by allowing an overlap in the linkage between the models, the resulting simulations were certainly stable.
Following a brief and informative demonstration of the OpenMI in action by Jan Gregerson of the OpenMI Association Technical Committee, Yves Ronse of VMM and Pol Magelmans from University of Liege introduced the third use case. This case is looking to assess the benefits of improved flow estimates for water quality modelling through linking InfoWorks RS and MIKE-11 into the PEGASE catchment wide water quality model. By introducing flow calculations from the river models into PEGASE at locations where observed flows are not available, it is hoped to improve the calculation of water quality in those areas. Pol Magelmans showed some encouraging results from the linked models that clearly show how the PEGASE results are being conditioned by those from the other models. However, he also emphasised the need for caution as any existing limitations in the river models have the potential to be reflected in the combined model results.
Finally Isabelle Neyskens of Flanders Hydraulics rounded off the use case presentations by covering the work being done to model the Lower Scheldt using a combination of the DELTARES WAQUA 3D model and MIKE-11. This shows that OpenMI can be comfortably used to integrate models of different dimensions: in this case a one-dimensional river model with a three-dimensional coastal model.
Throughout the day there were a number of lively discussion sessions, which ranged across a wide set of issues. Participants were interested in the institutional aspects of the use cases, for example how the different organisations involved in the use cases were developing procedures to work together effectively. The technical issues of linking particular model types and locations were debated, with many useful ideas for preventing or eliminating some of the stability issues that have been seen on occasions when running linked models. The need for ensuring correct calibration of integrated models was discussed, and the need to ensure that individual models of an integrated model are fit for purpose. Were they developed for the range of conditions to which they are now being applied?
Professor Willy Bauwens brought the workshop to a close with a summary of his thoughts on integrated modelling. In particular he stressed that integrated modelling does not remove the need to fully understand the individual models but makes that need even greater. The professor is new to OpenMI, but he was encouraged by the successes demonstrated at the workshop and will certainly be introducing OpenMI into his own field of work in the future.
Competent speakers from the Centre of Ecology and Hydrology (CEH), Deltares, the DHI, LicTek, the Ludwig-Maximilians-University and the BAW gave the presentations listed below. The workshop was with 35 participants fully booked.
It was one objective to build an OpenMI community. As a result a mailing list for Integrated Modelling in Germany and a mailing list for people with special interest in Integrated Modelling with 2D/3D models are created. Please send a mail to email@example.com in order to subscribe to these lists.
http://www.iemss.org/iemss2008). Amidst the wide variety of topics, the OpenMI Association (OA) convened a paper session on current practical OpenMI applications and a workshop on the strategy and future of the OpenMI.
Although scheduled at the never enviable last morning of the conference, both the paper session and the workshop attracted a good audience, including people from all around the world (America, Europe, Australia). The paper session presented user cases from within the ongoing EC funded OpenMI-Life project, but also an external contribution from Peter Schade (BAW, Germany), who was later awarded a prize for the best external (non OA related) OpenMI presentation of the past year.
During the workshop OA chairman Roger Moore introduced the audience in the current structure and strategy of the OA. Through an open discussion people were then invited to reflect on the most efficient and practical way to spread the OpenMI through a worldwide community of scientists and practitioners.
Overall, it was felt that the conference provided a good opportunity for the OA members to exchange ideas and to initiate further actions on dissemination.
The workshop was designed PEER members, NERC and NERC collaborative research centre directors, researchers and modellers with an interest in integrated modelling. It was particularly useful for those with science questions that they believe could be better answered using an integrated modelling approach. For those who are considering applying the OpenMI, a number of case studies were described. No prior knowledge of the OpenMI is assumed.
The workshop was held at CEH Wallingford, UK.
Front row: Jon Goodall, Jay Famiglietti, Peter Gijsbers, Rick Hooper, Michael Butts and Peter Sinding.
The purpose of the meeting was not only to exchange information but also to explore the opportunities for and, ideally, initiate collaboration. To this end, the meeting was organised around the headings of information exchange, identifying shared interests and initiating collaboration.
An interesting facet of integrated modelling is the number of levels on which people consider they have solved the model integration problem. Often, it takes a little while before the parties realise that they are actually talking at entirely different levels. To illustrate the point, the authors of a modelling framework, a model interface standard such as the OpenMI and a language such as C#, might all make such a claim. The reality is that C# is used to define the OpenMI interface standard and write its SDK; and the OpenMI might be used within a modelling framework system to enable the models to exchange data. The situation is analogous to the builder, architect and owner all claiming to have built a house.
Thus it was at the start of this meeting and another meeting with the US EPA. In these meetings, we usefully clarified the difference between CCA, a low level mechanism for linking models in the high performance computing environment, the OpenMI, a high level interface standard for the run time exchange of data between modelling components, and FRAMES, an entire integrated modelling system.
Coming back to the EU-NSF meeting, the discussion focussed primarily on the technical issues surrounding model linking. Topics of particular interest were linking models running on different platforms - .Net, Java, the high performance computing environment, etc. - and remote linking - linking models running on different machines. Also of interest was the development of modelling platforms, such as CUAHSI's CHyMP and OpenWeb. These provide the academic the community with the opportunity to link and run a wide range of models for teaching and research purposes. They also provide the academic community with a route to market for their models.
The purpose of linking models is to facilitate integrated modelling and in turn make integrated management feasible. While the benefits of integrated modelling are self-evident to some, this is far from being a universal view. In a separate meeting, part of the group discussed how more people could be persuaded of the benefits and a new cadre of scientists, engineers and IT people, skilled in the art, could be created.
The outcome of the meeting was the following series of initiatives: