Luo et al. (2026) A New Multiscale Deep Learning Model for Daily Runoff Prediction in Snow-Influenced Alpine Catchments

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-09-24
• Authors: Pingping Luo, Yajun Zhu, Chong‐Yu Xu, Maochuan Hu, Jing Wu, Jiachao Chen, Madhab Rijal, Fatima Fida
• DOI: 10.3390/rs18193309

Research Groups

• Institute of Mountain Hydrology (IMH), University of Natural Resources and Life Sciences, Vienna, Austria
• Department of Geoinformation and Cartography, University of Salzburg, Austria

Short Summary

This study proposes a multiscale deep learning model, IWOA-VCBA, for accurate daily runoff prediction in snow-influenced alpine catchments. The model achieved high predictive skill in the Lienz catchment and demonstrated useful direct cross-catchment application to three target catchments.

Objective

• Investigate the feasibility of a multiscale deep learning approach for predicting daily runoff in snow-influenced alpine catchments

Study Configuration

• Spatial Scale: Catchment scale, with focus on Lienz (Austria) and transfer experiments to Ziller, Möll, and Salzach catchments
• Temporal Scale: Daily time step, with a 30-year simulation period for the Lienz catchment and 10-year simulation periods for the target catchments

Methodology and Data

• Models used: IWOA-VCBA (multiscale deep learning model), incorporating VMD, CNN-BiLSTM-Attention, and IWOA
• Data sources: Meteorological variables from a regional climate model, antecedent runoff data from the Lienz catchment

Main Results

• High predictive skill of IWOA-VCBA in the Lienz catchment (KGE = 0.969, NSE = 0.947)
• Strong snowmelt-season performance (KGE = 0.961, NSE = 0.945)
• Useful direct cross-catchment application to three target catchments (mean KGE = 0.908, mean NSE = 0.841)

Contributions

• Development of a multiscale deep learning model for accurate daily runoff prediction in snow-influenced alpine catchments
• Demonstration of the potential for direct cross-catchment application without target-catchment retraining or hyperparameter re-optimization

Funding

• Austrian Science Fund (FWF) project P 32459-N32
• European Regional Development Fund (ERDF) through the Tyrolean Government

Citation

@article{Luo2026New,
  author = {Luo, Pingping and Zhu, Yajun and Xu, Chong‐Yu and Hu, Maochuan and Wu, Jing and Chen, Jiachao and Rijal, Madhab and Fida, Fatima},
  title = {A New Multiscale Deep Learning Model for Daily Runoff Prediction in Snow-Influenced Alpine Catchments},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18193309},
  url = {https://doi.org/10.3390/rs18193309}
}