Scholarlink

<p>This study evaluates flood susceptibility in the Cuddalore district of Tamil Nadu, India, utilizing advanced machine learning techniques integrated with Geographic Information System (GIS) tools to support sustainable urban planning. Ten key factors curvature, digital elevation model (DEM), distance from the coast, distance from rivers, drainage density, slope, Stream Power Index (SPI), Topographic Wetness Index (TWI), rainfall, and Land Use Land Cover (LULC) were analyzed. A dataset of 270 flood inventory points, comprising both flood and non-flood locations identified from historical records and satellite data, was divided into 75% for training and 25% for testing the Artificial Neural Network (ANN) model. The study area was categorized into five flood susceptibility zones: very low, low, medium, high, and very high. The results revealed that 24.42% of the area is highly susceptible to flooding, while 39.98% falls within the very high susceptibility zone. Conversely, low and medium susceptibility zones account for only 10.83% and 4.15%, respectively. Model validation through a receiver operating characteristic (ROC) curve demonstrated a strong predictive performance with an Area Under the Curve (AUC) of 0.88. This study supports Sustainable Development Goal 11 (SDG 11) by promoting safe, resilient, and sustainable communities. Identifying high-risk flood zones enables data-driven strategies for urban resilience, disaster risk reduction, and sustainable land use planning, aiding urban planners, disaster managers, and policymakers in climate-adaptive infrastructure development.</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

Integrating artificial neural networks and earth observation data for flood susceptibility mapping: a case study of Cuddalore District, Tamil Nadu, India

  • 期刊论文

  • 发表日期：

    2026年4月22日

查看全文
  • M. Subbulakshmi,
  • Sachikanta Nanda

摘要

This study evaluates flood susceptibility in the Cuddalore district of Tamil Nadu, India, utilizing advanced machine learning techniques integrated with Geographic Information System (GIS) tools to support sustainable urban planning. Ten key factors curvature, digital elevation model (DEM), distance from the coast, distance from rivers, drainage density, slope, Stream Power Index (SPI), Topographic Wetness Index (TWI), rainfall, and Land Use Land Cover (LULC) were analyzed. A dataset of 270 flood inventory points, comprising both flood and non-flood locations identified from historical records and satellite data, was divided into 75% for training and 25% for testing the Artificial Neural Network (ANN) model. The study area was categorized into five flood susceptibility zones: very low, low, medium, high, and very high. The results revealed that 24.42% of the area is highly susceptible to flooding, while 39.98% falls within the very high susceptibility zone. Conversely, low and medium susceptibility zones account for only 10.83% and 4.15%, respectively. Model validation through a receiver operating characteristic (ROC) curve demonstrated a strong predictive performance with an Area Under the Curve (AUC) of 0.88. This study supports Sustainable Development Goal 11 (SDG 11) by promoting safe, resilient, and sustainable communities. Identifying high-risk flood zones enables data-driven strategies for urban resilience, disaster risk reduction, and sustainable land use planning, aiding urban planners, disaster managers, and policymakers in climate-adaptive infrastructure development.