Matthew B. Jones

The American Geophysical Union (AGU) announces the new living collection “Impactful Datasets in Earth, Space, and Environmental Sciences” which recognizes the creation and curation of datasets enabling research, analysis, and decision making beyond what is captured by traditional academic impact metrics. The collection highlights datasets through the dimensions of People, Planet, and Prosperity, and includes a curated set of nominated datasets and accompanying special collection of data papers. Developed with the Earth Science Information Partners (ESIP) community, the nomination and review process initially invites contributors from across research, operations, and public service, to nominate datasets important to their work, and then submit a data paper that expands upon the nomination text for inclusion in the special collection. This commentary highlights the rationale for the living collection, and the process of nomination. The collection serves as a living reference that strengthens data culture, supports reproducible geoscience, and provides enduring access to critical information for science and society.

Despite substantial research and conservation efforts, many salmon populations are in decline. Globally, salmon research is not delivering effective decision support products to help managers apply research insights as informed management actions. Data Mobilization (DM) is a key step towards building the wider evidence base required to deliver accountable, reliable, and usable scientific advice to managers. Best practices for DM are being adopted throughout the scientific community but have not permeated deeply into the culture of salmon research and conservation. To address this, we present a strategy for Salmon Data Mobilization (SDM). This strategy defines three spheres of agencies and practitioners that must interact to advance SDM: (1) authoritative bodies that can create policies to support SDM uptake; (2) agencies that can promote, fund, and implement those policies; and (3) the broad salmon community of practice that can support uptake of SDM within focused interest groups. We sketch a future for SDM and propose functional changes required to improve it throughout the community.

This paper assesses trending AI foundation models, especially emerging computer vision foundation models and their performance in natural landscape feature segmentation. While the term foundation model has quickly garnered interest from the geospatial domain, its definition remains vague. Hence, this paper will first introduce AI foundation models and their defining characteristics. Built upon the tremendous success achieved by Large Language Models (LLMs) as the foundation models for language tasks, this paper discusses the challenges of building foundation models for geospatial artificial intelligence (GeoAI) vision tasks. To evaluate the performance of large AI vision models, especially Meta’s Segment Anything Model (SAM), we implemented different instance segmentation pipelines that minimize the changes to SAM to leverage its power as a foundation model. A series of prompt strategies were developed to test SAM’s performance regarding its theoretical upper bound of predictive accuracy, zero-shot performance, and domain adaptability through fine-tuning. The analysis used two permafrost feature datasets, ice-wedge polygons and retrogressive thaw slumps because (1) these landform features are more challenging to segment than man-made features due to their complicated formation mechanisms, diverse forms, and vague boundaries; (2) their presence and changes are important indicators for Arctic warming and climate change. The results show that although promising, SAM still has room for improvement to support AI-augmented terrain mapping. The spatial and domain generalizability of this finding is further validated using a more general dataset EuroCrops for agricultural field mapping. Finally, we discuss future research directions that strengthen SAM’s applicability in challenging geospatial domains.

In this review, we adopt the definition that 'Data citation is a reference to data for the purpose of credit attribution and facilitation of access to the data' (TGDCSP 2013: CIDCR6). Furthermore, access should be enabled for both humans and machines (DCSG 2014). We use this to discuss how data citation has evolved over the last couple of decades and to highlight issues that need more research and attention. Data citation is not a new concept, but it has changed and evolved considerably since the beginning of the digital age. Basic practice is now established and slowly but increasingly being implemented. Nonetheless, critical issues remain. These issues are primarily because we try to address multiple human and computational concerns with a system originally designed in a non-digital world for more limited use cases. The community is beginning to challenge past assumptions, separate the multiple concerns (credit, access, reference, provenance, impact, etc.), and apply different approaches for different use cases.

Many data packaging standards are available to researchers and data repository operators and the choice to use an existing standard or create a new one is challenging. We introduce the DataONE Data Package standard which is based on the existing OAI-ORE Resource Map standard. We describe the functionality Data Package provides, implementation considerations, compare it to existing standards, and discuss future extensions to the standard including the ability to describe execution environments via WholeTale "Tales"" and alternate serialization formats.

GeoLink has leveraged linked data principles to create a dataset that allows users to seamlessly query and reason over some of the most prominent geoscience metadata repositories in the United States. The GeoLink dataset includes such diverse information as port calls made by oceanographic cruises, physical sample metadata, research project funding and staffing, and authorship of technical reports. The data has been published according to best practices for linked data and is publicly available via a SPARQL Protocol and RDF Query Language (SPARQL) end point that at present contains more than 45 million Resource Description Framework (RDF) triples together with a collection of ontologies and geo-visualization tools. This article describes the geoscience datasets, the modeling and publication process, and current uses of the dataset. The focus is on providing enough detail to enable researchers, application developers and others who wish to leverage the GeoLink data in their own work to do so. The dataset is available at http://hdl.handle.net/1912/9524.

The Code of Practice for Research Data Usage Metrics standardizes the generation and distribution of usage metrics for research data, enabling for the first time the consistent and credible reporting of research data usage. This is the first release of the Code of Practice and the recommendations are aligned as much as possible with the COUNTER Code of Practice Release 5 that standardizes usage metrics for many scholarly resources, including journals and books. With the Code of Practice for Research Data Usage Metrics data repositories and platform providers can report usage metrics following common best practices and using a standard report format. This is an essential step towards realizing usage metrics as a critical component in our understanding of how publicly available research data are being reused. This complements ongoing work on establishing best practices and services for data citation.