yuhang et al. (2025) A machine leaning model for hydrological drought prediction: Model development and application

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

Identification

• Journal: SHILAP Revista de lepidopterología
• Year: 2025
• Date: 2025-12-01
• Authors: Yao yuhang, OU Ming, LI Min, FENG Zilong
• DOI: 10.13522/j.cnki.ggps.2025144

Research Groups

Information not provided in the paper text.

Short Summary

This study developed a hybrid Boruta-CNN-LSTM model to accurately forecast hydrological drought at the catchment scale, demonstrating its superior performance in predicting spatiotemporal drought variations in the Huai River Basin.

Objective

• To develop and evaluate a robust model for accurately forecasting hydrological drought at the catchment scale.

Study Configuration

• Spatial Scale: Catchment scale, specifically the Huai River Basin and its sub-regions.
• Temporal Scale: Monthly data over a 61-year period (1960-2020).

Methodology and Data

• Models used: Hybrid Boruta-CNN-LSTM model, Boruta-random forest (feature selection algorithm), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Boruta-LSTM (for comparison).
• Data sources: Measured monthly precipitation, evapotranspiration, and soil moisture data.

Main Results

• The Boruta algorithm effectively reduced interference from redundant features, identifying monthly standardized precipitation index, evapotranspiration, and soil temperature as the most influential factors among 31 initial variables for drought prediction across all sub-regions.
• The hybrid Boruta-CNN-LSTM model demonstrated the highest accuracy for hydrological drought prediction, achieving a coefficient of determination (R²) of 0.8536, a mean absolute error (MAE) of 0.262, and a root mean square error (RMSE) of 0.352.

Contributions

• Development of a novel hybrid Boruta-CNN-LSTM model that significantly improves the accuracy of hydrological drought prediction at the catchment scale.
• Demonstration of the effectiveness of the Boruta algorithm in feature selection for enhancing drought forecasting model performance.
• Provision of a robust and accurate model for predicting spatiotemporal drought variations under various hydrological scenarios.

Funding

Information not provided in the paper text.

Citation

@article{yuhang2025machine,
  author = {yuhang, Yao and Ming, OU and Min, LI and Zilong, FENG},
  title = {A machine leaning model for hydrological drought prediction: Model development and application},
  journal = {SHILAP Revista de lepidopterología},
  year = {2025},
  doi = {10.13522/j.cnki.ggps.2025144},
  url = {https://doi.org/10.13522/j.cnki.ggps.2025144}
}