An Open Science Framework For Data Set Versioning

The Data Stewardship Team at the NASA National Snow and Ice Data Center (NSIDC) Distributed Active Archive Center (DAAC) recently overhauled its data set versioning guidelines, aimed at improving the documentation of data set changes for better provenance and to better incorporate FAIR principles. What began as an effort to standardize practices specific to minor versioning—to address known pain-points and more clearly define the actions to be followed by the data management teams—turned into an improved framework for decision-making on which versioning pathway to take when faced with a request from data providers to make updates to published data sets. While practices around major-versioning were more or less consistent across data management teams at NSIDC, decision-making and actions in response to a variety of data set updates was handled in inconsistent ways for the types of updates where the need for major-versioning was not obvious. We approached this problem by synthesizing current recommendations and standards set by the data stewardship community – e.g., the Earth Science Information Partners (ESIP), the Core Trustworthy Data Repository Requirements, Research Data Alliance (RDA), Zenodo, DataCite – while considering the NASA Long Term Archive requirements and NASA Earth Science Data System (ESDS) standards set for data producers. Next we gathered a catalog of internal use cases which led to the categorization of triggers for data set updates. From there, a decision tree defined by questions related to changes in usability, interoperability, reproducibility, and provenance, was established to help the data management teams identify the most appropriate versioning pathway for a specific scenario. Once a pathway is identified, the actions to be taken for (1) major versioning, (2) minor versioning, or (3) updating without versioning are described in terms of the elements to be addressed (e.g. DOI record, collection metadata, User Guide, data file system, user announcements) and the staff members (subject matter experts) involved. These actions ultimately aim to ensure that the data set version history is well-documented and available to users, and that data files are preserved to the extent possible given contractual obligations and archiving constraints. This contribution is intended to foster discussion with other data repositories and NASA DAACs on how this framework could be improved, and to offer it as a specific example of open-science-based data stewardship strategy that could be adopted by other Earth Science data repositories.