Item

Abstract

Reusability of research data is one of the four FAIR principles. Envisioning future data reuse scenarios early in the data life cycle requires anticipation, since data reuse is often not carried out by the data producers, and reuse scenarios are constantly evolving. Data reuse is especially challenging when the data reuser is active in a different domain than the data producer. The application of data science methods, for instance, poses a growing demand on the (meta-)data information quality. In Earth Sciences, the development of data-driven models or data-integrated predictive simulations often first requires to assemble a homogenous and sanity checked data compilation as training data, which is made up of individually heterogeneous and non-consistent data sets. In order to do that in an efficient way the data sets have to comply with the FAIR paradigms. Here, we share our experience from creating a data compilation from sea ice core data with focus on temperature and salinity measurements. First, we will report on the FAIRness of publicly available sea ice data. The heterogeneous character of the data morphology and metadata availability makes interoperability challenging and reuse laborious. To overcome these deficiencies, we developed a workflow to create data compilations. We will conclude with a descriptive analysis of the sea ice core data compilation.