Serpa-Usta et al. (2025) Hybrid Deep Learning Models for Predicting Meteorological Variables Associated with Santa Ana Wind Conditions in the Guadalupe Basin

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

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-11-14
• Authors: Yeraldin Serpa-Usta, Dora‐Luz Flores, Álvaro López-Ramos, Carlos Fuentes, Franklin Muñoz, Neila María González Tejada, Álvaro Alberto López-Lambraño
• DOI: 10.3390/atmos16111292

Research Groups

Not explicitly provided in the text.

Short Summary

This study explored the predictive capability of hybrid deep learning architectures to model the temporal evolution of key atmospheric variables during Santa Ana wind events in the U.S.-Mexico border region. The Bidirectional LSTM with Attention (BiLSTM–Attention) model achieved the best overall performance, demonstrating high accuracy for temperature and relative humidity.

Objective

• To explore the predictive capability of several hybrid deep learning architectures—Long Short-Term Memory (LSTM), Convolutional Neural Network combined with LSTM (CNN–LSTM), and Bidirectional LSTM with Attention (BiLSTM–Attention)—to model the temporal evolution of wind speed, wind direction, temperature, relative humidity, and atmospheric pressure during Santa Ana wind events.

Study Configuration

• Spatial Scale: U.S.–Mexico border region, specifically the Guadalupe Basin in northern Baja California.
• Temporal Scale: 1980 to 2020 (41 years).

Methodology and Data

• Models used: Long Short-Term Memory (LSTM), Convolutional Neural Network combined with LSTM (CNN–LSTM), Bidirectional LSTM with Attention (BiLSTM–Attention).
• Data sources: Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2) reanalysis data.

Main Results

• The BiLSTM–Attention model achieved the best overall performance among the tested architectures.
• It showed particularly high accuracy for temperature (R² = 0.95) and relative humidity (R² = 0.76).
• Angular errors for wind direction were maintained below 35 degrees.
• Model performance was evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R² metrics, and compared against persistence and climatology baselines.

Contributions

• Demonstrates the potential of hybrid deep learning models to capture nonlinear temporal dependencies in meteorological time series.
• Provides a methodological framework to enhance hydrometeorological understanding and water resource management in the Guadalupe Basin under Santa Ana wind conditions.

Funding

Not explicitly provided in the text.

Citation

@article{SerpaUsta2025Hybrid,
  author = {Serpa-Usta, Yeraldin and Flores, Dora‐Luz and López-Ramos, Álvaro and Fuentes, Carlos and Muñoz, Franklin and Tejada, Neila María González and López-Lambraño, Álvaro Alberto},
  title = {Hybrid Deep Learning Models for Predicting Meteorological Variables Associated with Santa Ana Wind Conditions in the Guadalupe Basin},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos16111292},
  url = {https://doi.org/10.3390/atmos16111292}
}