Harmonizing Legacy Eddy Covariance Data within the ICOS and FLUXNET Networks: Methodological Insights from Long-Term Observations

Long-term observations of carbon, water, and energy exchanges, together with meteorological measurements, are essential for quantifying climate variability and change and for supporting ecosystem and climate model development. Eddy covariance networks provide unique multi-decadal datasets. However, their legacy data, historical observations collected prior to or during the evolution of standardized protocols, represent a critical resource. These datasets are frequently fragmented and may suffer from inconsistent formats and incomplete or missing metadata describing: how, when, and where data were collected, processed, and analyzed. Without systematic curation and harmonization, such data remain difficult to interpret, compare, and reuse. 

A metadata-driven approach was applied to long-term datasets from nineteen ICOS Network stations in order to integrate legacy eddy covariance data into standardized data infrastructures. These long-term datasets are released through the FLUXNET Data System, a continuously updated, open-access platform that provides harmonized flux and meteorological observations, complemented by comprehensive metadata. In this contribution, we focus on a representative subset of these datasets to examine key methodological and practical aspects that are critical for effective long-term data integration. We demonstrate how detailed and structured metadata enable the identification and resolution of inconsistencies arising from changes in instrumentation, sensor characteristics, spatial representativeness, and data processing methodologies, over multi-decadal periods.

A systematic metadata cross-walking procedure is used to document and reconcile historical site-specific changes, ensuring temporal continuity, data comparability, and transparency. This case study highlights the central role of metadata in bridging legacy datasets with contemporary standards, supporting FAIR data principles, and enabling the construction of interoperable long-term observational datasets. The proposed approach enhances data quality, interpretability, and reusability, thereby maximizing the scientific value of long-term eddy covariance observations for climate and ecosystem research.