Improving snow water equivalent modelling: a comparative study of hybrid machine learning techniques

Pomarol Moya, Oriol; Nussbaum, Madlene; Mehrkanoon, Siamak; Kraaijenbrink, Philip D. A.; Gouttevin, Isabelle; Karssenberg, Derek; Immerzeel, Walter W.

doi:10.5194/tc-20-1427-2026

Articles | Volume 20, issue 2

https://doi.org/10.5194/tc-20-1427-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/tc-20-1427-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 20, issue 2

Research article

|

03 Mar 2026

Research article |

| 03 Mar 2026

Improving snow water equivalent modelling: a comparative study of hybrid machine learning techniques

Oriol Pomarol Moya, Madlene Nussbaum, Siamak Mehrkanoon, Philip D. A. Kraaijenbrink, Isabelle Gouttevin, Derek Karssenberg, and Walter W. Immerzeel

Data sets

ESM-SnowMIP meteorological and evaluation datasets at ten reference sites (in situ and bias corrected reanalysis data) C. Menard and R. Essery https://doi.org/10.1594/PANGAEA.897575

Model code and software

oriol-pomarol/snow_project: Initial release - zenodo integration O. Pomarol Moya https://doi.org/10.5281/zenodo.17434422

Crocus simulations for ESM-SnowMIP exercise M. Lafaysse https://doi.org/10.5281/zenodo.15197745

Short summary

Two hybrid Machine Learning (ML) approaches predicting daily Snow Water Equivalent (SWE) were evaluated across ten Northern Hemisphere sites. By integrating meteorological data with Crocus snow model simulations, these hybrid models outperformed both standalone Crocus and traditional ML models. Notably, augmenting measured SWE data with Crocus simulations significantly improved performance at unseen locations, offering a promising new approach to long-term SWE prediction.