Configuring, Building and Running Models in GENIE

The following casestudy provides a workflow perspective on the Grid ENabled Integrated Earth systemmodelling (GENIE) framework, which was introduced in Vol. 1. The three aims in constructing the framework were; (i) the flexible coupling of Earth system components; (ii) to tune and run the resulting models on a wide variety of platforms, including the Grid; and (iii) to archive, query and retrieve the results. The component models incorporated into GENIE to date have been designed to be computationally inexpensive. This feature allows researchers using the framework to construct ensemble experiments to investigate long-term climatic trends or to conduct comprehensive parameter-space or climate sensitivity studies. GENIE also incorporates a full carbon cycle which makes it particularly suitable for recreating paleo-climates spanning the full ice-core record. The framework also has a role in simulating future climate change, including the response of slowly varying features of theEarth system, such as the deep ocean, sediment deposition andmineralweathering processes. The workflow description is split into four main areas. First is the task of gaining access to the model files, including source code, documentation, configuration descriptions and sample input and outputs. When considering access to files, it is natural to look at the facilities provided by version control systems (VCSs). Section7.2 looks inmore detail at the categories of information that must be recorded in amodel configuration. The section also reflects upon aspects such as the ease of use of model configurations as well as their provenance and evolution when harnessed to a growing model. The third area is the task of building an executable. Topics of particular relevance to the GENIE framework in this regard include code portability and dependencies as well as the incorporation of build-time model configurations. Lastly, we will outline some issues relevant to running and monitoring the model in the setting of a large ensemble or tuning experiment.