Diaz et al. (2026) Predicting Minnesota Lake Ice Phenology With Deep Learning, Explainable Methods, and a Physically Based Benchmark, 1980–2018

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

Identification

• Journal: Water Resources Research
• Year: 2026
• Date: 2026-09-01
• Authors: Jeremy Diaz, Samantha K. Oliver, Simon Topp, Jordan S. Read, Gretchen J. A. Hansen, Wallace Anderson McAliley
• DOI: 10.1029/2025wr042118

Research Groups

• University of Minnesota Department of Earth Sciences
• [Other research groups or institutions involved in the study]

Short Summary

This paper explores the use of machine learning (ML) approaches to hindcast lake ice time series for 625 Minnesota lakes, demonstrating that ML can accurately estimate historical lake ice formation and breakup dates.

Objective

• Investigate the feasibility of using ML to improve estimates of lake ice phenology

Study Configuration

• Spatial Scale: Regional study focused on Minnesota lakes in North America
• Temporal Scale: Daily weather data from 1980 to 2018, with a focus on long-term trends and variability

Methodology and Data

• Models used: Long Short-Term Memory (LSTM) neural networks, attention-based transformers
• Data sources: Observations from 625 Minnesota lakes, daily weather data, static lake attributes

Main Results

• The largest LSTM model outperformed a state-of-the-art physically based model in daily accuracy and year-level metrics for estimating ice formation, breakup, and duration.
• The ML approach was able to recreate broad-scale trends in ice cover phenology with limited long-term ice records.

Contributions

• This study demonstrates the potential of ML to improve estimates of lake ice phenology, particularly in regions with limited long-term data.
• The results highlight the importance of considering heterogeneity among lakes when studying ice loss and its implications.

Funding

• [List projects, programs, and reference codes that funded this research]
• e.g., NSF Grant #XXXXXXX, University of Minnesota Research Fund #YYYYYYY

Citation

@article{Diaz2026Predicting,
  author = {Diaz, Jeremy and Oliver, Samantha K. and Topp, Simon and Read, Jordan S. and Hansen, Gretchen J. A. and McAliley, Wallace Anderson},
  title = {Predicting Minnesota Lake Ice Phenology With Deep Learning, Explainable Methods, and a Physically Based Benchmark, 1980–2018},
  journal = {Water Resources Research},
  year = {2026},
  doi = {10.1029/2025wr042118},
  url = {https://doi.org/10.1029/2025wr042118}
}