No abstract

No abstract

As a data intensive field that unites researchers from many disciplines, Earth System Science (ESS) is an ideal site for examining evolving cross‐disciplinary data practices. This paper reports on results from a survey examining data sharing, data reuse, and research reproducibility practices of ESS researchers, aimed at informing improvements in data services for interdisciplinary sciences. Data reuse was found to be very high for new and comparative analyses but very limited for reproducing research. Data sharing was also strong, mostly through supplements to published papers, with moderate use of open access repositories. At the same time, there was interesting variability in both data sharing and reuse among ESS disciplines. The most pronounced challenges to reuse and reproducibility stem from limited documentation on how data are collected and managed, practices that are poorly supported by institutions, funders, and publishers. A more refined approach to “reproducibility” is needed that aligns with priorities and practices within the research community. Just as importantly, advances in data service models for ESS and other interdisciplinary fields need to account for the diverse and distributed system of repositories and build a workforce with deeper knowledge of the complex data and methods that drive integrative systems science.

The DCC Curation Lifecycle Model has played a vital role in the field of data curation for over a decade. During that time, the scale and complexity of data have changed dramatically, along with the contexts of data production and use. This paper reports on a study examining factors impacting data curation practices and presents recommendations for updating the DCC Curation Lifecycle Model. The study was grounded in a review of other lifecycle models and informed by a site visit to the Digital Curation Centre and consultation with expert practitioners and researchers. Framed by contemporary conditions impacting the conduct of research and provision of data services, the analysis and proposed recommendations account for the prominence of machine-actionable data, the importance of machine learning for data processing and analytics, growth of integrated research workflows, and escalating concerns with fairness, accountability, and transparency of data and algorithms.

No abstract