Scholarlink

<p>Drought is one of the most persistent and complex hydroclimatic hazards, exerting severe pressure on agricultural systems, water resources, ecosystems, and socio-economic stability across diverse climatic regions. This review provides a comprehensive synthesis of advances in drought characterization, monitoring, and prediction, highlighting the evolution from traditional statistical drought indices to contemporary data-driven and hybrid modeling approaches. Conventional indices, including the Standardized Precipitation Index, Standardized Precipitation Evapotranspiration Index, and Palmer Drought Severity Index, have been widely applied due to their simplicity and interpretability; however, their reliance on historical stationarity assumptions and limited representation of land–atmosphere–hydrology interactions constrain their effectiveness under changing environmental conditions. Recent developments in machine learning and deep learning have improved the capacity to model nonlinear relationships among meteorological, hydrological, and environmental variables, enabling enhanced spatiotemporal representation of drought processes using ground observations and satellite-derived datasets. Nevertheless, challenges related to data availability, model transferability, computational complexity, and limited interpretability continue to restrict their operational applicability. To address these limitations, this review emphasizes the growing role of hybrid modeling frameworks that integrate physical process understanding with data-driven learning, improving robustness, transparency, and environmental relevance. Emerging advances in process-informed learning, explainable modeling, and multi-source data integration are critically discussed in the context of drought early-warning systems and climate-change adaptation. The review concludes by identifying key research priorities for developing integrated, physically consistent, and decision-orientated drought assessment frameworks to support sustainable water and environmental management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine and Deep Learning Approaches for Drought Characterization and Prediction: a Comprehensive Review

  • 期刊论文

  • 发表日期：

    2026年5月22日

查看全文
  • Chinthu Naresh,
  • Aneesh Mathew

摘要

Drought is one of the most persistent and complex hydroclimatic hazards, exerting severe pressure on agricultural systems, water resources, ecosystems, and socio-economic stability across diverse climatic regions. This review provides a comprehensive synthesis of advances in drought characterization, monitoring, and prediction, highlighting the evolution from traditional statistical drought indices to contemporary data-driven and hybrid modeling approaches. Conventional indices, including the Standardized Precipitation Index, Standardized Precipitation Evapotranspiration Index, and Palmer Drought Severity Index, have been widely applied due to their simplicity and interpretability; however, their reliance on historical stationarity assumptions and limited representation of land–atmosphere–hydrology interactions constrain their effectiveness under changing environmental conditions. Recent developments in machine learning and deep learning have improved the capacity to model nonlinear relationships among meteorological, hydrological, and environmental variables, enabling enhanced spatiotemporal representation of drought processes using ground observations and satellite-derived datasets. Nevertheless, challenges related to data availability, model transferability, computational complexity, and limited interpretability continue to restrict their operational applicability. To address these limitations, this review emphasizes the growing role of hybrid modeling frameworks that integrate physical process understanding with data-driven learning, improving robustness, transparency, and environmental relevance. Emerging advances in process-informed learning, explainable modeling, and multi-source data integration are critically discussed in the context of drought early-warning systems and climate-change adaptation. The review concludes by identifying key research priorities for developing integrated, physically consistent, and decision-orientated drought assessment frameworks to support sustainable water and environmental management.