Identifying the Essentials: Distributed Resilience for Safeguarding Scientific Data in Times of Uncertainty

The long-term sustainability of data repositories in the Earth and environmental sciences is increasingly influenced by evolving institutional priorities, fluctuating funding, and shifting governance frameworks. Recent events have highlighted the vulnerabilities in the continued accessibility of major US-based climate and environmental datasets, particularly in the context of political shifts, underscoring the fragility of even well-established infrastructures. Against this backdrop, we propose a multi-level, network-oriented model for strengthening the resilience of Earth and environmental data infrastructures. This model, in addition to enhancing the self-healing capabilities of individual repositories, aims to establish a common framework for cooperative stewardship.

Until recently, frameworks like Core Trust Seal and the Repository Crisis Scorecards, developed by the ESIP Repositories Resilience Project, have focused on risk assessment and mitigation and some resilience, but less on recovery. Resilience is conceptualised in our approach as the coordinated interaction of several layers that collectively enhance the “rescuability” of essential scientific data. It includes networks of mutual support, in which repositories proactively coordinate to prepare for and respond to operational crises, sharing responsibilities to reduce the risk of isolated failures; harmonized technologies and standards, common protocols, and training, enabling efficient creation of rescue-ready data packages; a structured validation and stress-testing framework to assess vulnerabilities using transparent, scenario-based criteria; and a contingency layer providing shared resources, such as temporary storage or hosting, deliberately reserved to support other repositories, and enabling distributed, peer-to-peer style replication workflows that allow data to remain accessible even when local systems cannot operate fully. A further component of this approach is the prioritisation of critical or at-risk datasets, ensuring that limited rescue capacity is directed toward collections whose loss would most severely affect research continuity and societal monitoring needs.

We illustrate this approach with examples from existing Earth and environmental science repositories, and argue that even small and mid-sized infrastructures can benefit from strategies that preserve core data and metadata, even if complete restoration of complex interfaces or ingestion pipelines might be impractical. Given the heterogeneity, scale, and long-term relevance of environmental data, developing tiered, distributed resilience strategies is essential for maintaining scientific continuity in an era of increasing systemic uncertainty.