A regional evapotranspiration network for climate-resilient water management in mountain agro-ecosystems

Accurate quantification of evapotranspiration (ET) is a key challenge for understanding land–atmosphere interactions and optimizing water use in agriculture. Under ongoing climate change, reliable estimates of actual ET are critical in mountain regions such as in the Alps, where changes in temperature, precipitation, and snow dynamics strongly influence water availability and ecosystem functioning. While distributed information on ET is commonly derived through models based on remote sensing, meteorological variables, and crop coefficients, this study highlights the added value of direct ET observations. Indeed, model-based approaches, while useful for large-scale estimates, often fail to capture the strong spatial variability associated with heterogeneous landscapes and the complex response of ET to climatic drivers. Direct ET measurements, obtained through eddy covariance, are more reliable under stress conditions because they capture actual increases in ET during drought, driven by elevated atmospheric demand and sustained water supply from deep soil reservoirs. These mechanisms are commonly underrepresented by models due to simplifications in plant, soil, and hydraulic parameterizations.Direct observations of actual ET are therefore essential for improving irrigation water requirement (IWR) assessments and supporting water management in agriculture. During 2025, within the Agile Arvier project (Next Generation EU), we implemented a regional network for direct ET measurements across the Aosta Valley region (north-western Italian Alps). Seven LI-710 evapotranspiration sensors (LI-COR) were installed across representative agricultural systems in the area. To further strengthen the observational network, these measurements have been integrated with long-term ET datasets from two ICOS-associated sites in the region (IT-Tor, IT-TrF), providing decadal-scale context, and with three additional LI-710 already present in the same territory. The final network totalizes twelve monitoring sites, including a vineyard, an apple orchard, six meadows and pastures, a European larch forest, an abandoned pasture, a wetland, and a high-altitude rocky environment, distributed along an altitudinal gradient from approximately 500 to 3100 m a.s.l. Results will include site-specific comparison between observed ET and estimates derived from regional IWR datasets, to refine irrigation requirement estimates based on observations. This comparison is expected to improve the reliability of irrigation planning tools and to support the Regional Agricultural Department in developing more efficient and adaptive water management strategies. Finally, the release of an online, open-access ET database will be presented, allowing to researchers, land owners, sector experts, and policymakers to access and download data, thereby contributing to transparent, evidence-based decision-making.