Scholarlink

<p>Precipitation is one of the most critical and complex natural phenomena, often leading to floods, which is the most devastating natural hazard worldwide. Pakistan experiences significant climate variability, particularly in precipitation patterns. This research aimed to assess precipitation characteristics across Pakistan’s subregions, with a focus on drought monitoring, prediction, and the identification of key drought drivers. Precipitation characteristics, including mean, median, first quartile, third quartile, standard deviation, and kurtosis, are interpolated across Pakistan using kriging techniques, providing a detailed picture of precipitation characteristics over the domain. For drought analysis, the Standardized Precipitation Index (SPI) is employed. Various machine learning (ML) algorithms are used to predict drought occurrences and identify significant drought-related factors. Model performance is evaluated using accuracy, Kappa value, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Results revealed notable regional variations in precipitation. Monsoon-dominated areas received the highest rainfall, while northern regions showed the highest monthly variation. Khyber Pakhtunkhwa (KP) experienced the highest overall rainfall, except during the monsoon season, when Punjab recorded the most. In terms of drought frequency, the northwest had the highest frequency (23.82) during 1981–1990, while the southern regions experienced more frequent droughts during 2011–2022. Among ML models, random forest (RF) outperformed others in identifying key drought factors. Lagged SPI-3 was the most influential predictor, followed by precipitation and surface pressure. These findings support future planning in water, agriculture, and health sectors.</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

Spatiotemporal variability of precipitation and meteorological drought in Pakistan: trends, patterns, and machine learning prediction

  • 期刊论文

  • 发表日期：

    2026年7月21日

查看全文
  • Anwar Hussain,
  • Firdos Khan

摘要

Precipitation is one of the most critical and complex natural phenomena, often leading to floods, which is the most devastating natural hazard worldwide. Pakistan experiences significant climate variability, particularly in precipitation patterns. This research aimed to assess precipitation characteristics across Pakistan’s subregions, with a focus on drought monitoring, prediction, and the identification of key drought drivers. Precipitation characteristics, including mean, median, first quartile, third quartile, standard deviation, and kurtosis, are interpolated across Pakistan using kriging techniques, providing a detailed picture of precipitation characteristics over the domain. For drought analysis, the Standardized Precipitation Index (SPI) is employed. Various machine learning (ML) algorithms are used to predict drought occurrences and identify significant drought-related factors. Model performance is evaluated using accuracy, Kappa value, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Results revealed notable regional variations in precipitation. Monsoon-dominated areas received the highest rainfall, while northern regions showed the highest monthly variation. Khyber Pakhtunkhwa (KP) experienced the highest overall rainfall, except during the monsoon season, when Punjab recorded the most. In terms of drought frequency, the northwest had the highest frequency (23.82) during 1981–1990, while the southern regions experienced more frequent droughts during 2011–2022. Among ML models, random forest (RF) outperformed others in identifying key drought factors. Lagged SPI-3 was the most influential predictor, followed by precipitation and surface pressure. These findings support future planning in water, agriculture, and health sectors.