Masten et al. (2025) An enhanced python framework for hydrological modeling in alpine catchments: Snow hysteresis and glacier ice melt

Identification

• Journal: Environmental Modelling & Software
• Year: 2025
• Date: 2025-12-19
• Authors: Martin Masten, Simon Seelig, Matevž Vremec, Magdalena Seelig, Gerfried Winkler
• DOI: 10.1016/j.envsoft.2025.106842

Research Groups

• Department of Earth Sciences, NAWI Graz Geocenter, University of Graz, Graz, Austria
• Alma Mater Europaea University, Maribor, Slovenia

Short Summary

This study introduces a new Python extension for the Rainfall-Runoff Modeling Playground (RRMPG) that incorporates snow cover hysteresis and glacier ice melt, crucial for alpine hydrology. The enhanced framework, tested in two Ötztal Alps catchments, significantly improves the accuracy and robustness of runoff and snow cover dynamics simulations compared to simpler models.

Objective

• To develop and implement a Python extension for the Rainfall-Runoff Modeling Playground (RRMPG) that integrates a modified linear snow cover hysteresis model and a degree-day glacier ice melt model.
• To enhance the representation of hydrological processes in high-alpine catchments, improving model robustness and accuracy, particularly under changing environmental conditions.

Study Configuration

• Spatial Scale: Two high-alpine catchments in the Ötztal Alps, Austria: Radurschltal (2.44 x 10^7 m²) and Kaunertal (6.00 x 10^7 m²). Catchments are divided into multiple elevation bands of equal area.
• Temporal Scale: Modeling period from January 1, 1961, to December 31, 2022, with specific calibration and validation periods within this range. Daily time steps are used for simulations.

Methodology and Data

• Models used:
  • Rainfall-Runoff Modeling Playground (RRMPG) Python package (extended with new modules).
  • CemaNeige snow model (basic and with modified linear hysteresis).
  • Degree-day approach for glacier ice melt.
  • Conceptual rainfall-runoff model GR4J (used in examples, compatible with ABC, HBV-EDU).
  • PyET v1.3.1 for potential evapotranspiration estimation (e.g., Hargreaves method).
  • SPOTPY for model calibration (e.g., Shuffled Complex Evolution Algorithm - SCE-UA).
• Data sources:
  • Observed runoff (discharge) from gauging stations (TIWAG for Radurschlbach, Hydrographical Service of Tyrol for Fagge).
  • MODIS snow cover data (Collection 6.1 products, Normalized Difference Snow Index - NDSI) from NASA National Snow and Ice Data Center.
  • Air temperature and precipitation data from nearby weather stations or gridded datasets (e.g., SPARTACUS) from GeoSphere Austria.
  • Glacier outlines (2015 data).

Main Results

• The new Python extension successfully integrates a modified linear snow hysteresis model and a degree-day glacier ice melt model into the RRMPG framework.
• In the Radurschltal catchment (ice-free), the hysteresis snow model significantly improved runoff simulation accuracy and robustness (higher Kling-Gupta Efficiency, KGE) compared to the basic CemaNeige model, especially during validation.
• Simulated Snow-Covered Area (SCA) from the hysteresis model showed strong agreement with MODIS data (KGE of 0.8) in Radurschltal, whereas the basic CemaNeige model substantially underestimated SCA.
• For the highly glaciated Kaunertal catchment, the combined snow hysteresis and ice melt model outperformed the basic CemaNeige model in simulating discharge, particularly during low flow periods and demonstrating improved robustness in validation (better KGE).
• The ice melt component accurately simulated melt during warm, snowless periods, with peak ice melt occurring in August, following the snowmelt peak in July, consistent with glaciological expectations.
• The framework demonstrated seamless integration with external Python packages like PyET (for potential evapotranspiration) and SPOTPY (for calibration), further enhancing model performance (e.g., achieving superior KGE for the Fagge catchment).

Contributions

• First Python-based implementation of the modified linear snow cover hysteresis within the CemaNeige framework, integrated into the RRMPG package.
• Integration of a degree-day glacier ice melt model into RRMPG, specifically designed to enhance rainfall-runoff simulations in high-alpine environments.
• Development and implementation of a multi-objective calibration strategy in Python, leveraging both observed discharge and satellite-derived snow cover data (MODIS SCA) to improve model robustness and reduce parameter uncertainty.
• Creation of a modular and interoperable framework that facilitates straightforward coupling with various rainfall-runoff models, input estimation methods, and calibration approaches, surpassing the functionality of existing Python tools.
• Provision of an open-source extension with a detailed step-by-step Python tutorial and example input data, promoting user adoption, flexibility, and compatibility with Python-based workflows.

Funding

• Austrian Academy of Sciences (ESS project ECOSPRING)
• University of Graz

Citation

@article{Masten2025enhanced,
  author = {Masten, Martin and Seelig, Simon and Vremec, Matevž and Seelig, Magdalena and Winkler, Gerfried},
  title = {An enhanced python framework for hydrological modeling in alpine catchments: Snow hysteresis and glacier ice melt},
  journal = {Environmental Modelling & Software},
  year = {2025},
  doi = {10.1016/j.envsoft.2025.106842},
  url = {https://doi.org/10.1016/j.envsoft.2025.106842}
}