Overview

In the first phase of the GENIE project we have successfully shown the power of the Grid and e-Science tools to perform fundamentally new kinds of experiment with modularised Earth system models (ESMs). GENIEfy will extend the modelling framework and tools developed in GENIE to include standards from the European "Program for Integrated Earth System Modelling" (PRISM, http://prism.enes.org) and coupling methods from PRISM and the Tyndall Centre's Community Integrated Assessment System (CIAS) to ensure compatibility with both. We will thus engage the Earth System Modelling (ESM) and Integrated Assessment (IA) communities and enable them to address hitherto intractable scientific problems, by utilising the power of e-Science technology. GENIEfy will provide key parts of the QUEST strategic activity in Earth System Modelling by supplying a spectrum of fast ESMs that are traceable to one another and have improved traceability to full complexity ESMs. Data assimilation techniques will be integrated in our framework and used both to improve the fast ESMs and train them to emulate full complexity ESMs. We will overlap with a second spectrum of full complexity Hadley ESMs by including fast components at equivalent resolution to FAMOUS and HadCM3. The framework we produce will be well documented, structured code that is easily maintainable, user-friendly and thus of maximum value beyond the lifespan of the project. Use of the Bespoke Framework Generator (BFG) and engagement in its development will aid future compatibility with the Met Office Flexible Unified Modelling Environment (FLUME), and we thus complement the SIAMESE proposal.

Achievements thus far: In phase 1 of the GENIE project we have successfully:

(1) Built and tested several instances of a new modular Earth System Model, at different levels of process complexity and spatial resolution, and used these to address important scientific problems, including the stability of the thermohaline circulation to multi-parameter freshwater inputs3 and simulations of the Last Glacial Maximum and mid-Holocene using PMIP boundary conditions.

(2) Demonstrated, for the first time:

• Extensive exploration of the parameter space of a 3D ESM, using ~1000 instantiations of the model, deployed using Grid technology, and involving ~40 million years of model integration.
• Automatic "tuning" of several 3D model parameters, by direct fitting of the model to real data, using Grid-enabled numerical optimisation software.
• Simultaneous determination of optimal values for a substantial number of 3D model parameters, with formal characterisation of the uncertainty of these, using the Ensemble Kalman Filter (EnKF), deployed on both cluster and Grid-based hardware. Development of the EnKF method relied on the existence of a fast GENIE ESM and is now being more widely adopted.

Fit to QUEST scientific goals and implementation strategy: We will provide a modelling framework with which users can couple together, tune and execute a spectrum of next generation fast Earth system models. The modelling framework will address the key requirements for traceability, inclusion of representations of ocean and land biosphere processes and their physical and chemical interactions with the ocean and atmosphere, and inclusion of interfacing with socio-economic models. Our framework will be able to incorporate modules generated in QUEST Themes 1 and 3, and will provide models that can address a number of the major research issues identified across QUEST. Separate proposals will be made to QUEST themes to do the necessary science, here we focus on providing the modelling framework and making the necessary investment in compatibility, modularity and structured programming to give a true community tool. In tune with the QUEST Implementation Plan we will build on our success in multi-institutional team building and the use of e-Science technologies to help remove cultural and organizational impediments to effective co-operation among institutions and disciplines.

GENIEfy

Scientific Research Challenges