Jian et al. (2026) Short-term drought forecasting in Iran using multi-source machine learning: an assessment of autoregressive, teleconnection-driven, and hybrid paradigms

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2026
• Date: 2026-09-14
• Authors: Jun Jian, Peyman Mahmoudi, Pouria Jafari, Alireza Ghaemi, Jing Yang, Fatemeh Firoozi
• DOI: 10.5194/nhess-26-4407-2026

Research Groups

• Navigation College, Dalian Maritime University
• Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai)
• Department of Physical Geography, Faculty of geography and environmental planning, University of Sistan and Baluchestan
• Department of Electronic and Electrical Engineering, Faculty of Electrical and Computer Engineering, University of Sistan and Baluchestan
• Faculty of Geographical Science, Key Laboratory of Environmental Change and Natural Disaster, Beijing Normal University
• Department of Humanities and Social Science, Farhangyan University

Short Summary

This study provides a comprehensive comparative assessment at the national scale of Iran to determine the relative superiority or synergy of three competing paradigms in short-term drought forecasting: autoregressive (based on temporal memory), teleconnection-driven (based on large-scale climate drivers), and hybrid. The results show that the optimal model structure is highly location-dependent, with the hybrid approach prevailing in arid and semi-arid regions.

Objective

• Evaluate the performance of three competing paradigms in short-term drought forecasting: autoregressive, teleconnection-driven, and hybrid.
• Identify the primary sources of predictability for drought conditions in Iran.

Study Configuration

• Spatial Scale: National scale of Iran, covering a diverse range of climates from hyper-arid to humid.
• Temporal Scale: 30-year period (January 1993 to December 2022) with monthly time steps.

Methodology and Data

• Models used: Nine machine learning algorithms: Multiple Linear Regression, K-Nearest Neighbors, Artificial Neural Network, Least Squares Support Vector Machine, Random Forest, Gradient Boosting Regression, Extreme Gradient Boosting, Bayesian Ridge Regression, and Long Short-Term Memory.
• Data sources: Station-based precipitation data from 96 synoptic stations, large-scale climate indices (teleconnections) from the National Centers for Environmental Prediction (NCEP) and the Climate Prediction Center (CPC).

Main Results

• The hybrid approach integrating temporal memory with large-scale climate drivers prevailed in arid and semi-arid regions.
• The Random Forest model demonstrated superior accuracy and lower error rates compared to other models.
• The geographical distribution of forecast skill highlights a strong dependency on regional hydroclimatology.

Contributions

• This study provides the first nationwide, multi-model, and multi-paradigm assessment of drought predictability in Iran.
• The findings highlight the importance of considering both temporal memory and large-scale climate drivers for accurate drought forecasting.

Funding

• Not specified.

Citation

@article{Jian2026Shortterm,
  author = {Jian, Jun and Mahmoudi, Peyman and Jafari, Pouria and Ghaemi, Alireza and Yang, Jing and Firoozi, Fatemeh},
  title = {Short-term drought forecasting in Iran using multi-source machine learning: an assessment of autoregressive, teleconnection-driven, and hybrid paradigms},
  journal = {Natural hazards and earth system sciences},
  year = {2026},
  doi = {10.5194/nhess-26-4407-2026},
  url = {https://doi.org/10.5194/nhess-26-4407-2026}
}