Olaleye et al. (2026) A Decomposition-Based Hybrid Prophet–LSTM Framework for SPEI-12 Drought Forecasting in Kano State, Nigeria

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

Identification

• Journal: AgriEngineering
• Year: 2026
• Date: 2026-07-27
• Authors: Oluwatobi Solomon Olaleye, Oluwaseun Temitope Faloye, Oluwafemi E. Adeyeri, Olayiwola Akin Akintola, Akinwale T. Ogunrinde, Bolaji Adelanke Adabembe, Toju Babalola, John Omodara Akinremi
• DOI: 10.3390/agriengineering8080307

Research Groups

Not specified

Short Summary

The study develops a hybrid Prophet–LSTM architecture to predict drought in Northern Nigeria, demonstrating that decoupling deterministic trends from stochastic residuals significantly improves the accuracy of hydroclimatic forecasting.

Objective

• To enhance the near-term predictive capacity for non-stationary hydroclimatic time series in the Sudan–Sahel transition zone to better manage drought risks.

Study Configuration

• Spatial Scale: Kano State, Northern Nigeria (Sudan–Sahel transition zone).
• Temporal Scale: 1980–2024 (analysis period) and 2025–2030 (projection period).

Methodology and Data

• Models used: Hybrid Prophet–Long Short-Term Memory (LSTM) architecture.
• Data sources: Standardized Precipitation Evapotranspiration Index (SPEI-12) derived from CRU TS v4.09.

Main Results

• Identified a statistically significant trend toward moisture recovery (p < 0.0001).
• The hybrid model outperformed standalone baselines with a Nash–Sutcliffe Efficiency (NSE) > 0.87 and a 67.2% reduction in Root Mean Square Error (RMSE).
• Achieved directional accuracy exceeding 87% in simulating hydroclimatic transitions.
• Projected a positive moisture shift of approximately 0.9 SPEI units for the 2025–2030 period.

Contributions

• Establishes a technical framework for improving drought prediction by decoupling non-linear noise from deterministic signals in non-stationary time series.
• Provides a robust tool for near-term climate prediction applicable to national drought early warning systems.

Funding

Not specified

Citation

@article{Olaleye2026DecompositionBased,
  author = {Olaleye, Oluwatobi Solomon and Faloye, Oluwaseun Temitope and Adeyeri, Oluwafemi E. and Akintola, Olayiwola Akin and Ogunrinde, Akinwale T. and Adabembe, Bolaji Adelanke and Babalola, Toju and Akinremi, John Omodara},
  title = {A Decomposition-Based Hybrid Prophet–LSTM Framework for SPEI-12 Drought Forecasting in Kano State, Nigeria},
  journal = {AgriEngineering},
  year = {2026},
  doi = {10.3390/agriengineering8080307},
  url = {https://doi.org/10.3390/agriengineering8080307}
}