Premier et al. (2026) Assessing the impact of Earth Observation data-driven calibration of the melting coefficient on the LISFLOOD snow module

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-03-03
• Authors: Valentina Premier, Francesca Moschini, Jesús Casado-Rodríguez, Davide Bavera, Carlo Marin, Alberto Pistocchi
• DOI: 10.5194/hess-30-1189-2026

Research Groups

• Institute for Earth Observation, Eurac Research, Bolzano, Italy
• European Commission, Joint Research Centre (JRC), Ispra, Italy
• Arcadia SIT, Milano, Italy
• Rey Juan Carlos University, Madrid, Spain

Short Summary

This study evaluates the LISFLOOD hydrological model's snow module and the impact of calibrating its snowmelt coefficient using Earth Observation (EO) snow cover fraction (SCF) data across nine European basins. It demonstrates that while EO-based calibration significantly improves snow cover representation, its impact on basin-level discharge simulations is minimal, suggesting that standard discharge-based calibration adequately captures snow dynamics for streamflow.

Objective

• Evaluate LISFLOOD’s current snow module and investigate the effects of a post-replacement of the snowmelt coefficient (Cm) across nine European basins with varying snow influence.
• Assess the ability of the current LISFLOOD setup to reproduce snow and streamflow.
• Determine the potential of EO-Cm to improve snow process realism.
• Analyze the impact of the new, distributed EO-Cm on hydrological performance.
• Test whether modifying only the snowmelt coefficient affects discharge performance, and whether the standard calibration adequately represents both snow dynamics and streamflow.

Study Configuration

• Spatial Scale: Nine snow-dominated hydrological basins across Europe (Italy, Switzerland, Austria, Germany, France, Spain, Slovakia, Sweden), ranging from 1200 km² to 12100 km². Model grid resolution of 1 arcminute (approximately 0.75 km to 1.48 km). EO data at 50 m resolution, aggregated to LISFLOOD grid.
• Temporal Scale: EO dataset spans six hydrological years (1 October 2017 to 30 September 2023), with the 2022/23 season used for independent assessment. LISFLOOD model runs from 1990 to 2023 (including two warm-up years). Daily time steps for simulations and data.

Methodology and Data

• Models used:
  • LISFLOOD (v5.0): Open-source, spatially distributed hydrological model, utilizing a temperature-index (degree-day) approach for snowmelt.
  • Snow cover parameterization: Swenson and Lawrence (2012) (primary), Zaitchik and Rodell (2009) (alternative).
  • Optimization algorithm: L-BFGS-B (from SciPy library) for snowmelt coefficient estimation.
• Data sources:
  • Earth Observation (EO) snow cover fraction (SCF): Daily, gap-filled, high-resolution (50 m) binary snow dataset, integrating Sentinel-2 and MODIS data (Premier et al., 2021).
  • Meteorological forcings: EMO-1 dataset (precipitation and temperature fields).
  • Observed river discharge data: From CEMS Hydrological Data Collection Centre, Andalucian water agency (Hidrosur), and Agenzia Regionale per la Prevenzione e Protezione Ambientale (ARPAV).
  • Digital Elevation Model (DEM): MERIT DEM (90 m resolution).
  • Snow Water Equivalent (SWE) reanalysis: IT-SNOW (for Adige), Swiss Operational Snow-Hydrological (OSHD) model system (for Dischma Valley/Alpenrhein).

Main Results

• The standard LISFLOOD calibration (L-Cm) showed biases in SCF from -1 % to 22 % and RMSE values from 20 % to 55 % compared to EO-SCF.
• The EO-based calibration (EO-Cm) improved both bias and RMSE by up to 8 % for SCF, particularly during the snow depletion phase.
• EO-Cm exhibited high values in basins or areas with ephemeral snow (e.g., Laborec, Mörrumsån, lower Adige), indicating a wider parameter range than standard LISFLOOD calibration.
• No strong correlations were found between EO-Cm and topographic, geographic, or land cover features, though elevation showed a negative correlation in some alpine basins.
• SWE intercomparison with IT-SNOW and OSHD showed LISFLOOD's tendency to underestimate SWE, likely due to precipitation input discrepancies, but comparable or slightly improved metrics with EO-Cm.
• At the basin level, the optimized EO-Cm generally did not significantly change simulated discharge in terms of Kling-Gupta Efficiency (KGE), with KGE differences being minimal or showing slight improvements/degradations depending on the basin.
• The EO-Cm application led to noticeable divergences in discharge within smaller upstream catchments (Normalized Euclidean Distance up to 4, locally higher), but these effects were smoothed downstream.
• The EO-Cm primarily influenced the timing of snow accumulation and melt phases, which in turn affected the timing and magnitude of runoff contributions, especially during and immediately following the snowmelt period.

Contributions

• Developed and applied a parsimonious calibration strategy for the LISFLOOD snowmelt coefficient using high-resolution, gap-filled Earth Observation snow cover fraction data.
• Demonstrated that EO-driven calibration significantly improves the spatial realism and temporal representation of snow cover dynamics (especially depletion) in a large-scale hydrological model without requiring a full model recalibration.
• Quantified the impact of improved snowmelt parameterization on hydrological outputs, showing that while snow metrics improve, basin-level discharge performance remains largely stable, highlighting the equifinality of discharge-based calibration.
• Provided a foundation for a future two-step calibration approach in LISFLOOD, aiming to reduce equifinality and enhance process realism.
• Showed that standard calibration procedures already provide an acceptable representation of snow dynamics for streamflow, but local divergences in upstream catchments are revealed by EO-based calibration.

Funding

• Joint Research Centre (JRC) through Tender no. JRC/IPR/2023/VLVP/2678 – “Support to LISFLOOD model development: testing of the snow module”.
• European Space Agency (ESA) through the ESA Snow CCI project (contract no. 4000124098/18/I-NB) and the ESA EXPRO+ AlpSnow - Alps Regional Initiative project (contract no. 4000132770/20/I-NB).
• EU Interreg Alpine Space project Alpine Drought Observatory (ADO).

Citation

@article{Premier2026Assessing,
  author = {Premier, Valentina and Moschini, Francesca and Casado-Rodríguez, Jesús and Bavera, Davide and Marin, Carlo and Pistocchi, Alberto},
  title = {Assessing the impact of Earth Observation data-driven calibration of the melting coefficient on the LISFLOOD snow module},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-1189-2026},
  url = {https://doi.org/10.5194/hess-30-1189-2026}
}