Niu et al. (2025) Tail-Aware Forecasting of Precipitation Extremes Using STL-GEV and LSTM Neural Networks

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

Identification

• Journal: Hydrology
• Year: 2025
• Date: 2025-10-30
• Authors: Haoyu Niu, Samantha Murray, Fouad Jaber, Bardia Heidari, Nick Duffield
• DOI: 10.3390/hydrology12110284

Research Groups

Not specified in the provided text.

Short Summary

This study introduces a hybrid modeling framework combining Generalized Extreme Value (GEV) distribution fitting with deep learning (LSTMs) to forecast monthly maximum precipitation extremes. The approach, which uses a tail-weighted loss function, demonstrates strong predictive performance in identifying anomalously high precipitation months.

Objective

• To develop and evaluate a hybrid modeling framework that combines Generalized Extreme Value (GEV) distribution fitting with deep learning models (specifically LSTMs) to forecast monthly maximum precipitation extremes.

Study Configuration

• Spatial Scale: Not specified in the provided text.
• Temporal Scale: Monthly (for maximum precipitation extremes).

Methodology and Data

• Models used: Generalized Extreme Value (GEV) distribution fitting, Long Short-term Memory (LSTM) deep learning models. A tail-weighted loss function was designed for training.
• Data sources: Not specified in the provided text.

Main Results

• The proposed hybrid GEV–deep learning model achieved strong predictive performance in both the cumulative distribution function (CDF) and residual domains.
• The model accurately identified anomalously high precipitation months.

Contributions

• Introduction of a novel hybrid GEV–deep learning framework for forecasting extreme precipitation events, transforming the problem into a bounded probabilistic learning task.
• Design of a crucial tail-weighted loss function to emphasize rare, high-impact events and address class imbalance in extreme precipitation predictions.
• Offers a promising solution for early warning systems and long-term climate resilience planning in hydrologically sensitive regions.

Funding

Not specified in the provided text.

Citation

@article{Niu2025TailAware,
  author = {Niu, Haoyu and Murray, Samantha and Jaber, Fouad and Heidari, Bardia and Duffield, Nick},
  title = {Tail-Aware Forecasting of Precipitation Extremes Using STL-GEV and LSTM Neural Networks},
  journal = {Hydrology},
  year = {2025},
  doi = {10.3390/hydrology12110284},
  url = {https://doi.org/10.3390/hydrology12110284}
}