Technological Challenges

GENIEfy will extend our Earth System Modelling capability in several dimensions, including model size (spatial resolution), duration of integrations, simulation multiplicity (number of replicates), and process complexity.

The science will challenge grid technology in the following ways:

• Increased model size, requires the deployment of massive memory and the fastest processing speeds attainable, probably in a parallel mode.
• Increased duration and multiplicity of simulations demand faster model components and maximisation of the pool of processors accessible for model experiments.
• Increased process complexity demands a more general coupling approach for components, and a more user-friendly system to engage new users.
• Furthermore, we pose the fundamental technical challenge of how data, from models and observations, and newly developed code, can be retained, shared and 'recycled' into the model development process.

Grid technologies will advance the science by:

• Easing the construction of new instances of Earth system model,
• automating the process of model tuning,
• speeding up the execution of individual long integrations,
• enabling larger ensembles to be run,
• easing their execution,
• and feeding and recycling data back into model development.

We propose to extend the existing GENIE modelling framework to enable the coupling of:

• full complexity components using community standards from PRISM including the OASIS coupler,
• integrated assessment components using the Bespoke Framework Generator (BFG) approach (as in CIAS/softIAM),
• components from the proposed SIAMESE and e-JULES projects. We thus aim to provide a generic modelling framework that is better integrated with existing Earth system modelling and Integrated Assessment activities and comprises well documented and well structured code incorporating existing standards where appropriate. Innovative data assimilation techniques will be incorporated as an integral part of the model building and tuning process.

We rely on the following development/initiatives within the e-science core programme: middleware provided by the Geodise and BioSimGrid projects provide for scripting environment usage of the Grid-enabled resources (computing, data handling etc). We will also look to exploit the web services based middleware stack from the Open Middleware Infrastructure Institute as it becomes available. We envisage ongoing usage of the National Grid Service resources (both data and compute).

Scientific Research Challenges

Traceability