Scholarlink

<p>Drought, a complex natural hazard exacerbated by climate change, poses significant challenges to water resources, agriculture, livestock, and socio-economic stability globally. This study introduces a novel, multi-phase framework for regional drought monitoring and prediction. First, precipitation data quality is enhanced using auxiliary information, and the standardized drought indices (SDI) are computed through a 12-component Gaussian mixture distribution (12-CGMD), validated by Bayesian information criterion (BIC) values. Second, hierarchical clustering based on the Davies-Bouldin index (DBI = 1.7700) identifies six homogeneous clusters of meteorological stations, enabling the development of a multi-regional aggregated standardized drought index (MRASDI) through spatiotemporal bootstrapping. Distinct drought patterns were observed, including significant shifts from 1983 to 1996 in clusters 3, 4, and 5 and increased drought frequency in clusters 1 and 2 post- 1996. Lastly, the Boruta algorithm assesses meteorological station relevance within clusters, validating MRASDI. Machine learning models—random forest (RF) and support vector machines (SVM)—are applied to predict cluster-specific drought severity, with RF demonstrating superior accuracy, particularly in Cluster 5. Validation across 52 stations in Pakistan spanning 1968–2016 confirms the framework’s robustness. This study provides a scientifically robust tool for improving drought monitoring and prediction, offering insights for drought mitigation and climate resilience strategies tailored to regional needs.</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

Innovative drought monitoring: development and application of the multi-regional aggregated standardized drought index (MRASDI)

  • 期刊论文

  • 发表日期：

    2025年4月8日

查看全文
  • Asad Ellahi,
  • Ibrahim A. Nafisah,
  • Mohammed M. A. Almazah,
  • Nafisa A. Abasheir,
  • Ijaz Hussain,
  • Muhammad Mubashar Dogar

摘要

Drought, a complex natural hazard exacerbated by climate change, poses significant challenges to water resources, agriculture, livestock, and socio-economic stability globally. This study introduces a novel, multi-phase framework for regional drought monitoring and prediction. First, precipitation data quality is enhanced using auxiliary information, and the standardized drought indices (SDI) are computed through a 12-component Gaussian mixture distribution (12-CGMD), validated by Bayesian information criterion (BIC) values. Second, hierarchical clustering based on the Davies-Bouldin index (DBI = 1.7700) identifies six homogeneous clusters of meteorological stations, enabling the development of a multi-regional aggregated standardized drought index (MRASDI) through spatiotemporal bootstrapping. Distinct drought patterns were observed, including significant shifts from 1983 to 1996 in clusters 3, 4, and 5 and increased drought frequency in clusters 1 and 2 post- 1996. Lastly, the Boruta algorithm assesses meteorological station relevance within clusters, validating MRASDI. Machine learning models—random forest (RF) and support vector machines (SVM)—are applied to predict cluster-specific drought severity, with RF demonstrating superior accuracy, particularly in Cluster 5. Validation across 52 stations in Pakistan spanning 1968–2016 confirms the framework’s robustness. This study provides a scientifically robust tool for improving drought monitoring and prediction, offering insights for drought mitigation and climate resilience strategies tailored to regional needs.