Ceballos-Tavares et al. (2026) Forecasting meteorological droughts in a hydrological basin using artificial neural networks

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

Identification

• Journal: Hydrology research
• Year: 2026
• Date: 2026-02-17
• Authors: Jesús Alberto Ceballos-Tavares, Israel Velasco
• DOI: 10.2166/nh.2026.025

Research Groups

[Not specified in the abstract]

Short Summary

This study developed and evaluated artificial neural network models for meteorological drought forecasting in Mexico's Conchos River Basin, demonstrating high predictive performance (Mean Squared Error < 0.1, Coefficient of Determination > 0.90) using complex architectures and specific optimizers.

Objective

• To develop and evaluate artificial neural network models for predicting meteorological droughts, specifically the three-month Standardized Precipitation-Evapotranspiration Index (SPEI-3), using climatic variables (precipitation and temperature) and drought indices (SPI, SPEI) as inputs in the Conchos River Basin, Mexico.

Study Configuration

• Spatial Scale: Conchos River Basin, Mexico.
• Temporal Scale: Prediction of the three-month Standardized Precipitation-Evapotranspiration Index (SPEI-3).

Methodology and Data

• Models used: Artificial Neural Networks (ANN), specifically Multilayer Perceptron (MLP) with backpropagation. Different architectures (e.g., two hidden layers: 256–128 neurons) and optimizers (Adam, Adamax, Stochastic Gradient Descent (SGD), RMSprop) were evaluated.
• Data sources: Climatic variables (precipitation and temperature) and meteorological drought indices (Standardized Precipitation Index (SPI) and Standardized Precipitation-Evapotranspiration Index (SPEI)).

Main Results

• The developed models achieved good performance with a Mean Squared Error (MSE) less than 0.1 and a Coefficient of Determination (R²) greater than 0.90.
• More complex architectures, particularly those with two hidden layers (256–128 neurons), demonstrated superior performance.
• Optimizers such as Adam and Adamax significantly outperformed SGD and RMSprop.
• Artificial neural network models showed strong potential to accurately capture relevant drought patterns within hydrological sub-basins.

Contributions

• Addresses a significant gap in the application of artificial intelligence and machine learning methods for drought forecasting within Mexico.
• Provides a robust methodology for meteorological drought prediction using ANNs, applicable to hydrological sub-basins.
• Strengthens the capabilities of early warning systems and supports decision-making processes in water resource management.
• Highlights the effectiveness of machine learning in drought research and advocates for its expanded implementation in the Mexican context.

Funding

[Not specified in the abstract]

Citation

@article{CeballosTavares2026Forecasting,
  author = {Ceballos-Tavares, Jesús Alberto and Velasco, Israel},
  title = {Forecasting meteorological droughts in a hydrological basin using artificial neural networks},
  journal = {Hydrology research},
  year = {2026},
  doi = {10.2166/nh.2026.025},
  url = {https://doi.org/10.2166/nh.2026.025}
}