Text, Data and People : How to Represent Earth System Science

Abstract

Earth system science – a collective name for a number of disciplines, as these are applied to a specific topic - is among the most data intensive sciences. Another characteristic feature of this field of research is the international cooperative work, which is organized around expeditions and other coordinated observation campaigns. These campaigns make use of an extensive array of instruments mounted on ships, airplanes, satellites and trucks. To make the most out of costly datasets from observations and “in silico” experiments – that is: modeling results – data have to be published in a well documented form, with or without strong links to classical publications, in which these data are interpreted. Due to the special circumstances, “scholarly communication” within the community probably is more personal but also more difficult than in other areas. In many cases, even small teams are international with multiple institutional affiliations of their members. Since the object of their observation may comprise a substancial part of the globe, a single observation can last for years. In terms of e-science, all these factors clearly call for supporting “groupware” systems or even “knowledge management” systems. We will make a case for these systems in this session just in order to point out the importance of people, the groups they form and the projects and campaigns they perform together. These are important objects, to be described in their own right – not just as metadata to mark up datasets or publications. We will discuss the application of the open access paradigm as well as Open Archives protocols and common metadata schemes, as they are applicable to datasets as well as people and groups. One outcome of this discussion will be a critical evaluation whether complex metadata schemes – as the 1000 attribute ISO 19115 – are useful for open, loosely coordinated harvesting schemes or if these should be applied in closed information systems only. Regarding people and organizations, we strongly recommend using the eduPerson object classes and attributes from the Internet2 / Grid middleware standards. Finally, we will demonstrate a working OAI-PMH service provider which harvests information about (text-) publications, datasets and researchers, which are described in a way as outlined above.