Scholarlink

<p>Landslides in the Himalayan region, particularly in Joginder Nagar hills, pose recurring risks due to heavy rainfall and fragile geology. Traditional warning systems, dependent on fixed rainfall thresholds, often fail to capture real-time slope behavior. This study presents a real-time landslide early warning system integrating IoT sensors and AI models. A multi-sensor probe was installed near active slopes in Joginder Nagar, with secondary arms embedded horizontally to monitor soil moisture, tilt, and rainfall intensity. Using historical data from the India Meteorological Department (IMD), we simulated environmental conditions and calibrated thresholds. The AI models—decision trees and support vector machines—processed live sensor input to assess risk levels. Our results show that when rainfall exceeded 200 mm and moisture crossed 38%, accompanied by a tilt change over 1.2°, the system triggered alerts with over 90% accuracy. This proves that the deployed device, combined with AI analytics, can reliably predict landslide-prone conditions in real time.</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

Real time environmental sensing for landslide risk reduction using IoT and AI in Joginder Nagar hills

  • 期刊论文

  • 发表日期：

    2025年10月6日

查看全文
  • Parul Arora,
  • Ritika Wason,
  • Ujjwal Singh Thakur,
  • Devansh Arora,
  • M. N. Hoda

摘要

Landslides in the Himalayan region, particularly in Joginder Nagar hills, pose recurring risks due to heavy rainfall and fragile geology. Traditional warning systems, dependent on fixed rainfall thresholds, often fail to capture real-time slope behavior. This study presents a real-time landslide early warning system integrating IoT sensors and AI models. A multi-sensor probe was installed near active slopes in Joginder Nagar, with secondary arms embedded horizontally to monitor soil moisture, tilt, and rainfall intensity. Using historical data from the India Meteorological Department (IMD), we simulated environmental conditions and calibrated thresholds. The AI models—decision trees and support vector machines—processed live sensor input to assess risk levels. Our results show that when rainfall exceeded 200 mm and moisture crossed 38%, accompanied by a tilt change over 1.2°, the system triggered alerts with over 90% accuracy. This proves that the deployed device, combined with AI analytics, can reliably predict landslide-prone conditions in real time.