a b s t r a c tWater, food and energy are at the core of human needs and there is a boundless complex cycle among these three basic human needs. Ecosystems are in the center of this nexus, since they contribute to the provision of each component, making it imperative to understand the role of ecosystems in securing food, water and energy for human well-being. In this study we aimed to map and assess water provisioning services and associated benefits to support the ecosystem-water-food-energy nexus by taking into account environmental flow requirements for riverine ecosystems using the hydrological model Soil and Water Assessment Tool (SWAT). We developed a framework that includes indicators of renewable water (capacity of ecosystem to provide water) and water use (service flow) and we applied it in the Danube river basin over the period 1995-2004. Water scarcity indicators were used to map the possible water scarcity in the subbasins, and analyze the spatial match of water availability and water use. The results show that modelling is instrumental to perform the integrated analysis of the ecosystem-waterfood-energy nexus; and that spatial mapping is a powerful tool to display environmental availability of water provisioning and regulatory services delivered by ecosystems, and can support the nexus analysis.

The Water Framework Directive of the European Union requires member states to achieve good ecological status of all water bodies. A harmonized pan-European assessment of water resources availability and quality, as affected by various management options, is necessary for a successful implementation of European environmental legislation. In this context, we developed a methodology to predict surface water flow at the pan-European scale using available datasets. Among the hydrological models available, the Soil Water Assessment Tool was selected because its characteristics make it suitable for large-scale applications with limited data requirements. This paper presents the results for the Danube pilot basin. The Danube Basin is one of the largest European watersheds, covering approximately 803,000 km 2 and portions of 14 countries. The modeling data used included land use and management information, a detailed soil parameters map, and high-resolution climate data. The Danube Basin was divided into 4663 subwatersheds of an average size of 179 km 2 . A modeling protocol is proposed to cope with the problems of hydrological regionalization from gauged to ungauged watersheds and overparameterization and identifiability, which are usually present during calibration. The protocol involves a cluster analysis for the determination of hydrological regions and multiobjective calibration using a combination of manual and automated calibration. The proposed protocol was successfully implemented, with the modeled discharges capturing well the overall hydrological behavior of the basin.

Different SWAT models have been set-up to predict water discharge at the European scale, applying an innovative modelling protocol that involves sensitivity analysis, multi-variable calibration and regionalization of the calibrated parameters. In this application, two large regions have been considered: the Scandinavian and the Iberian peninsulas, with the main objectives: (a) to study the spatial variation of calibrated parameter sets obtained for selected sub-basins, and (b) identification of the most relevant hydrological processes in each region. The results of the analysis highlight that snow processes are dominant in Scandinavia; groundwater processes are significant in both the Scandinavian and Iberian peninsulas, while lateral flow is not significant in either region. Calibrated soil hydraulic parameters have different ranges of values in each region, reflecting a difference in runoff-generating mechanisms between the two studied regions. The contribution of this analysis is the assessment of the main differences between hydrological processes across Europe and understanding of the major transport pathways for pollutants.Key words large-scale model; multi-variable calibration; sensitivity analysis; spatial variability; SWAT model; parameter regionalization; Scandinavia; Iberian Peninsula Comparaison de jeux de paramètres calés de modèles SWAT pour les péninsules ibérique et ScandinaveRésumé Différentes modèles SWAT (Soil and Water Assessment Tool) ont été mises en place afin de prévoir les débits à l'échelle européenne selon un protocole de modélisation innovant comprenant une analyse de sensibilité, une calage multi-variable ainsi que la régionalisation des paramètres calés. Nous avons considéré dans cette application deux grandes régions: les péninsules scandinave et ibérique. Les objectifs principaux de cette recherche sont (a) d'étudier la variation spatiale des jeux de paramètres calés obtenus dans les sous-bassins sélectionnés et (b) d'identifier les processus hydrologiques dominants dans chaque région. Les résultats de l'analyse indiquent que les processus affectant le manteau neigeux sont dominants en Scandinavie, que les processus d'écoulements souterrains sont importants dans les deux péninsules, alors que l'écoulement latéral dans la zone non saturée n'est significatif dans aucune des deux régions. Les paramètres hydrauliques calés des sols varient sensiblement d'une région à l'autre, indiquant une différence des mécanismes de génération des écoulements. Notre analyse a contribué à une meilleure connaissance des processus hydrologiques à travers l'Europe, permettant ainsi une détermination plus précise des principales voies de transport des polluants.

Highlights Alternative regionalization techniques in large-scale model applications were investigated. Regionalization techniques are model-independent and are based on the similarity approach. Adapted measures of physiographic similarity meaningful for hydrological similarity were used. Regionalization involving unsupervised and supervised clustering were studied. These techniques proved to be helpful in large-scale model calibration.