Metafor: Managing Metadata For Climate Models

Climate model outputs are increasingly important, not only to climate scientists, but also to those concerned with the impact of climate change on disease, water resources, forest fires and fish stocks, including researchers, policymakers and the enlightened public.
Identifying, accessing and using the climate data stored in digital repositories requires accurate and complete metadata, or ‘data which describes the data’. Unfortunately, metadata is not always complete or understandable, and hence there is a need to standardise the description of climate data and the models that produce it.

Easy as CIM
The Metafor project is funded by the EU’s 7th Framework Programme and aims to develop a Common Information Model (CIM) to document and compare climate models. As modellers know, discriminating between two simulations is not easy, even if you were responsible for creating and running them both. Documentation currently revolves around runtime, but not the scientific detail or relevance of the model components. To counter this, the CIM exists to document the ‘simulation context and models’, otherwise known as the whys, wherefores and issues associated with any particular simulation.
The core Metafor team consists of eleven international partners from Europe and the USA, with additional collaborators from the US-led Curator  project. As Metafor progresses, more climate modelling groups will become involved in the development and governance of the CIM, and via interaction with the Coupled Model Intercomparison Project Phase 5 (CMIP5)  metadata questionnaire.

The CMIP5 Metadata Questionnaire
The Working Group on Coupled Modelling (WGCM), via the Coupled Model Intercomparison Project  (CMIP) panel, gave Metafor the task of developing a questionnaire to define and collect the model and experiment  metadata for CMIP5, to be used for the next Intergovernmental Panel on Climate Change assessment, due in 2013. CMIP5 will have a globally distributed data archive using technology primarily developed by the U.S. Earth System Grid (ESG) team, with chief responsibility for the archive held by the Program for Climate Model Diagnosis and Intercomparison  (PCMDI). The core archives are expected to comprise around 750 Terabytes of model-run data, hence the need to capture all the metadata to allow users to differentiate between and compare the experiments, and better understand the climate models used.
To develop the questionnaire, the Metafor team has been collecting controlled vocabulary from domain experts; this collection will also be used to populate the CIM with a standard structure. The questionnaire is web-based and will collect information from the CMIP5 modelling groups, including details of the models used, how the simulations were carried out, how the models conform to CMIP5 experiment requirements, and what hardware or platforms were used.

Spinning off the results
The questionnaire results will be made available to anyone interested in the metadata, whether or not they plan to use the CIM. The controlled vocabulary generated during the development of the questionnaire will also be spun-off and governed independently from the CIM, providing a valuable resource for the climate modelling community. The metadata collected by the questionnaire will be used by ESG and stored in the Metafor CIM archive.
Climate modelling is a complex process, so the CIM has been designed to be highly generic and flexible. The information provided in each CIM instance will vary according to the model and parameters described, but the CIM will store this information in a standard way and thus enable the development of tools to search for and compare CIM documents, and hence different climate models.

Integration and collaboration
Metafor does not exist in isolation. A key aim of the project is to integrate with other larger projects such as GEOSS  and INSPIRE , and to collaborate with other complementary projects alongside ESG, such as Curator, IS-ENES , GENESI-DR  and EuroVO-AIDA . Partnership with technology providers such as OGC GML  projects and UK NERC DataGrid  will ensure the development of new technology that will use the CIM to benefit users.
Metafor is halfway through its three years of operation and its project team has developed into a dedicated and tightly organised group of experts. This team have proposed a CIM development strategy, including a conceptual level and meta-model, and released CIM v1.0, which is freely available for comment and criticism at http://metaforclimate.eu/trac/browser/CIM. Excellent international collaborations and links have been established with US colleagues and there is strong buy-in by members of the climate modelling community, as demonstrated by Metafor’s key contribution to the CMIP5 metadata collection.
Further information about the Metafor project and the CMIP5 questionnaire can be found at http://metaforclimate.eu