Scholarlink

<p>The Indian Sundarban delta is highly exposed to flood as a common consequences of storm surge and tropical cyclone originated from the Bay of Bengal. Saline water intrusion, loss of agricultural production, food shortage, economic losses, disruption of rural infrastructure and fatalities aggravate the challenges of coastal inhabitants and make their livelihood complex. In this regard, a site-specific strategic plan is required to mitigate the adverse impacts of flood. Flood susceptibility assessment is an urgent necessary to identify the most susceptible zone and identify the prime conditioning factors for formulating preventive measures. Naïve Bayes Tree with Frequency Ratio (NBT-FR), Logistic Regression (LR) and Random Forest (RF) models have been applied to assess flood susceptibility in this study. RF model achieves the highest prediction accuracy (90.59%) followed by LR (89.41%) and NBT-FR (86.47%). Two Optimum Combinations of conditioning factors have been identified comparing the accuracy of different combinations of conditioning factors using RF and Naïve Bayes (NB) machine learning models. Drainage density, elevation, drainage proximity, Topographic Wetness Index, Slope and Plan curvature are the common factors in both Optimum Combinations derived by RF and NB. Rainfall deviation and Land Use/Land Cover are also included under the Optimum Combinations derived by RF and NB models respectively. Such eight factors are considered as prime factors in this study, which is needed to be emphasised for flood management planning.</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

Flood susceptibility mapping in Indian Sundarban delta using multivariate statistics and machine learning algorithms in GIS

  • 期刊论文

  • 发表日期：

    2025年3月24日

查看全文
  • Souvik Kundu,
  • Tarun Kumar Mondal

摘要

The Indian Sundarban delta is highly exposed to flood as a common consequences of storm surge and tropical cyclone originated from the Bay of Bengal. Saline water intrusion, loss of agricultural production, food shortage, economic losses, disruption of rural infrastructure and fatalities aggravate the challenges of coastal inhabitants and make their livelihood complex. In this regard, a site-specific strategic plan is required to mitigate the adverse impacts of flood. Flood susceptibility assessment is an urgent necessary to identify the most susceptible zone and identify the prime conditioning factors for formulating preventive measures. Naïve Bayes Tree with Frequency Ratio (NBT-FR), Logistic Regression (LR) and Random Forest (RF) models have been applied to assess flood susceptibility in this study. RF model achieves the highest prediction accuracy (90.59%) followed by LR (89.41%) and NBT-FR (86.47%). Two Optimum Combinations of conditioning factors have been identified comparing the accuracy of different combinations of conditioning factors using RF and Naïve Bayes (NB) machine learning models. Drainage density, elevation, drainage proximity, Topographic Wetness Index, Slope and Plan curvature are the common factors in both Optimum Combinations derived by RF and NB. Rainfall deviation and Land Use/Land Cover are also included under the Optimum Combinations derived by RF and NB models respectively. Such eight factors are considered as prime factors in this study, which is needed to be emphasised for flood management planning.