HyperSpec-RDM An Experience Report on Implementing FAIR-complaint Research Data Infrastructure for Hyperspectral Imaging

Hyperspectral Imaging (HSI) is a technique used in various disciplines, including environmental monitoring, material sciences, and biomedical research. However, managing, preserving, and sharing hyperspectral data — often reaching several hundred terabytes per project — poses significant challenges in Research Data Infrastructure (RDI). At CeMOS Research- and Transfer Center, the largest volume of research data originates in the medical and biomedical sectors, with a strong focus on HSI. Individual projects frequently accumulate multiple terabytes of data, necessitating extensive preprocessing and analysis. Ensuring compliance with the FAIR Principles for data archiving is particularly challenging as metadata annotation is essential to make data findable, accessible, interoperable and reusable throughout its lifecycle [1]. The complexity increases further through the need for the collaboration with external research institutions and industrial partners. A key challenge within CeMOS is the lack of standardized guidelines, tools, and infrastructure for managing research data throughout its lifecycle — from planning and collection to archiving and reuse. Currently, researchers rely on fragmented approaches, limiting the overall benefit of Research Data Management (RDM) practices. Addressing this gap is critical not only to ensure research integrity but also to align with funding agency mandates and international best practices in RDM. To address the challenges associated with the storage, metadata annotation and access of hyperspectral data, the open-source Dataverse repository for research data was implemented [2]. A container-based installation was chosen to ensure maximum flexibility of the service and enable seamless migration. Simultaneously, the scalability of the applications was considered, which is why Kubernetes was selected as the deployment environment. Kubernetes facilitates seamless scalability and making the approach resilient to future changes, as it enables responding effectively to increasing requirements. The repository [3] formed the foundation for the development. In this repository Dataverse version 4.20 was provided. This repository was updated extensively and improved as the needs and requirements for higher versions changed. Today the project repository [4] supports the most recent versions of Dataverse and is equipped with several scripts providing easier deployment, customization, additional features and services such as an identity provider. This Dataverse instance integrates into the proposed RDI in [5] and was further adapted to the needs of HSI [6]. Additionally, the Research Data Management Organiser (RDMO) was integrated into the proposed infrastructure to provide researchers with the ability to create robust Research Data Management Plans (DMPs) for their research proposals and to comply with the latest requirements of the funding authorities. In addition, extensive training materials were created to prepare the research community for the use of RDMO, Dataverse and the general handling of research data. Several workshops were held to learn how to use the individual services and to develop metadata specifically tailored to HSI. The proposed RDI significantly improves the accessibility and usability of hyperspectral data. In the future, the RDI, training materials and offerings will be rolled out to the entire university to ensure comprehensive adoption and effective RDM.