Scholarlink

<p>Urbanization-induced land use and emission changes are increasingly influencing regional precipitation dynamics in mountainous terrains, yet their impact remains underexplored in the Indian Himalayan foothills. The study investigates rainfall variability and hydroclimatic extremes across urban and non-urban districts of Uttarakhand using integrated statistical and machine learning approaches. Forty years (1984–2023) of MERRA-2 reanalysis data were analyzed, employing the Mann-Kendall trend test, Sen’s slope estimator, and ETCCDI-based indices (Consecutive Dry Days, Consecutive Wet Days). Machine learning models (Support Vector Machine and Random Forests) were developed to classify extreme precipitation events using a 70:30 train-test validation. Results showed much higher rainfall in urban areas (<i>Haridwar</i>: 377.64&#xa0;mm) than in non-urban regions (<i>Tehri Garhwal</i>: 116.18&#xa0;mm), with a stronger rising trend in <i>Dehradun</i> (slope: 9.06 × 10⁻⁵). Extreme event analysis showed prolonged dry days (81 days in 2022) and extended wet days (58 days in 2023), with urban districts exhibiting greater intensities. Random Forest outperformed Support Vector Machine with a slightly higher accuracy (0.786-0.799 vs. 0.740-0.767). The study shows that urbanization increases vulnerability to extreme rainfall, highlighting the need for disaster risk reduction, climate-resilient urban planning, and better water management. It offers a scientific basis for localized climate adaptation in the Himalayas and supports India’s National Action Plan on Climate Change by improving extreme weather prediction.</p> Graphical abstract <p>The graphical abstract visually encapsulates a data-driven framework for assessing rainfall variability and hydroclimatic extremes in the urbanizing Himalayan foothills of Uttarakhand, India. Utilizing NASA’s MERRA-2 reanalysis dataset, the study integrates multi-stage analyses, beginning with Mann-Kendall trend detection and extending to extreme event characterization using Consecutive Wet Days (CWD) and Consecutive Dry Days (CDD) indices. The workflow highlights a robust correlation analysis identifying relative humidity, dew point temperature, and surface pressure as primary climatic drivers of precipitation variability across varying topographies. Spatial patterns reveal escalating rainfall trends and intensifying extremes, with districts like <i>Nainital</i>, <i>Almora</i>, and <i>Pithoragarh</i> showing heightened susceptibility to hydroclimatic hazards such as flash floods and landslides. A machine learning component compares predictive capacities of Random Forest and Support Vector Machine models, with Random Forest achieving up to 79.93% classification accuracy in <i>Udham Singh Nagar</i>, as validated by ROC curve analysis and spatial accuracy maps. Overall, the graphical abstract offers valuable insights that can inform climate-sensitive urban planning and disaster preparedness strategies in this rapidly transforming Himalayan state, emphasizing the importance of understanding localized climate patterns for effective resource management and risk reduction.</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

Rainfall Variability and Rising Extremes in Urbanizing Himalayan Foothills: A Machine Learning and data-driven Exploration of Hydroclimatic Shifts in Uttarakhand, India

  • 期刊论文

  • 发表日期：

    2025年10月17日

查看全文
  • Aayushi Tandon,
  • Amit Awasthi,
  • Kanhu Charan Pattnayak,
  • Sumanta Das

摘要

Urbanization-induced land use and emission changes are increasingly influencing regional precipitation dynamics in mountainous terrains, yet their impact remains underexplored in the Indian Himalayan foothills. The study investigates rainfall variability and hydroclimatic extremes across urban and non-urban districts of Uttarakhand using integrated statistical and machine learning approaches. Forty years (1984–2023) of MERRA-2 reanalysis data were analyzed, employing the Mann-Kendall trend test, Sen’s slope estimator, and ETCCDI-based indices (Consecutive Dry Days, Consecutive Wet Days). Machine learning models (Support Vector Machine and Random Forests) were developed to classify extreme precipitation events using a 70:30 train-test validation. Results showed much higher rainfall in urban areas (Haridwar: 377.64 mm) than in non-urban regions (Tehri Garhwal: 116.18 mm), with a stronger rising trend in Dehradun (slope: 9.06 × 10⁻⁵). Extreme event analysis showed prolonged dry days (81 days in 2022) and extended wet days (58 days in 2023), with urban districts exhibiting greater intensities. Random Forest outperformed Support Vector Machine with a slightly higher accuracy (0.786-0.799 vs. 0.740-0.767). The study shows that urbanization increases vulnerability to extreme rainfall, highlighting the need for disaster risk reduction, climate-resilient urban planning, and better water management. It offers a scientific basis for localized climate adaptation in the Himalayas and supports India’s National Action Plan on Climate Change by improving extreme weather prediction.

Graphical abstract

The graphical abstract visually encapsulates a data-driven framework for assessing rainfall variability and hydroclimatic extremes in the urbanizing Himalayan foothills of Uttarakhand, India. Utilizing NASA’s MERRA-2 reanalysis dataset, the study integrates multi-stage analyses, beginning with Mann-Kendall trend detection and extending to extreme event characterization using Consecutive Wet Days (CWD) and Consecutive Dry Days (CDD) indices. The workflow highlights a robust correlation analysis identifying relative humidity, dew point temperature, and surface pressure as primary climatic drivers of precipitation variability across varying topographies. Spatial patterns reveal escalating rainfall trends and intensifying extremes, with districts like Nainital, Almora, and Pithoragarh showing heightened susceptibility to hydroclimatic hazards such as flash floods and landslides. A machine learning component compares predictive capacities of Random Forest and Support Vector Machine models, with Random Forest achieving up to 79.93% classification accuracy in Udham Singh Nagar, as validated by ROC curve analysis and spatial accuracy maps. Overall, the graphical abstract offers valuable insights that can inform climate-sensitive urban planning and disaster preparedness strategies in this rapidly transforming Himalayan state, emphasizing the importance of understanding localized climate patterns for effective resource management and risk reduction.