Building and disseminating local hydrological narratives under regional warming levels

Integrating information from multimodel ensemble hydroclimate projections in adaptation strategies poses significant challenges from an operational point of view. Three of these challenges are addressed here to provide information relevant to local water managers in France: (i) the uncertainty from ensembles derived from multiple Global Climate Models (GCMs), Regional Climate Models (RCMs), and Hydrological Models (HMs), (ii) the conceptual differences from scenario-based projections to Global/Regional Warming Level approach, and (iii) the semantic complexity of hydroclimate modelling chains.

This work builds on a large multimodel ensemble of hydrological projections from the Explore2 French national project (Sauquet et al., 2025), and the National Reference Warming Trajectory for Adaptation (TRACC, Soubeyroux et al., 2024) which defines three Regional Warming Levels (RWLs) : +2°C, +2.7°C, +4°C. The Explore2 hydrological projections consist in up to 153 transient daily streamflow series for over 4,000 locations in France. The approach taken is to select through a dedicated clustering algorithm four individual projections that adequately sample possible changes for time slices defined by the RWLs. Selection is made separately for large basins striking a balance between spatial consistency across local studies, and regional diversity in the hydrological responses ro RWLs. Selection is made based on low-, average- and high-flow indicators to capture contrasting changes across the hydrological regime.

The selected individual projections are called Narratraccs -- short for "TRACC narratives" -- and organised around four families of overall changes represented by one letter (X, E, C, M): eXtrêmes (extremes), Étiages (Low-flows), Crues (floods) and Modérés (Moderate). Each Narratracc belongs to one or more of these families for a given basin and a given RWL. The individual projections originally defined through their GCM/RCM/HM modelling chain (e.g. HadGEM2-ES / CCLM4-8-17 / MORDOR-TS) are renamed and articulated as, for example, "E1: decreased annual streamflow, slightly less severe floods, and much more severe low-flows".

Narratraccs are disseminated via the MEANDRE-TRACC dedicated portal (https://meandre-tracc.explore2.inrae.fr/) which extends the widely-used MEANDRE portal (https://meandre.explore2.inrae.fr/). The latter aimed to synthesise messages from the Explore2 project. MEANDRE-TRACC enables users to visualise and download changes associated with each Narratracc. It also redirects to one-page summary sheets for individual subbasins (Héraut et al., 2025) which are hosted by the Explore2 dataverse (https://entrepot.recherche.data.gouv.fr/dataverse/explore2 .

This work has been funded by the Agence de l'Eau Loire-Bretagne through the EHCLO R&D project.

References

Héraut, L. et al., 2025, Analyse des débits et de la recharge potentielle des aquifères par niveau de réchauffement et par secteur hydrographique — Fiche de synthèse, https://doi.org/10.57745/QDCSBZ, Recherche Data Gouv, V3

Sauquet, E. et al. A large transient multi-scenario multi-model ensemble of future streamflow and groundwater projections in France, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2025-1788, 2025.

Soubeyroux, J.-M. et al. (2024) À quel Climat s’adapter en France selon la TRACC ? Partie 1 —
Concepts et données de base pour les températures et précipitations. Météo-France. https://hal.science/hal-04797481v3