Scholarlink

<p>Floods pose a significant threat to Odisha, India, due to its diverse geography and climatic conditions, which result in considerable risks to communities, infrastructure, and the economy. This study analyzed flood dynamics in Odisha by incorporating hazard, vulnerability, and risk assessments through the application of various machine learning models (MLMs). A total of 15 and 13 conditioning factors were used for flood hazard and flood vulnerability analyses, respectively. Multicollinearity diagnostic tests, the Pearson correlation coefficient, and the Boruta model were employed to analyze the relationships between flood occurrences and the factors influencing them. The findings revealed that approximately 11.87% of the area falls into the high to very high flood susceptibility category, while 14.80% is classified as highly vulnerable and 6.85% of the region faces significant flood risk. Coastal districts, including Jagatsinghapur, Kendrapara, Puri, Baleswar, Bhadrak, and Cuttack, are particularly flood-prone, with the Baitarani, Brahmani, Mahanadi, Budhabalanga, and Rushikulya river basins being the most affected. The RF model demonstrated superior performance in both flood hazard (AUC = 0.963) and vulnerability (AUC = 0.956) assessments, followed by Bagging, SVM, KNN, and GLM.</p> Graphical Abstract <p></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 Machine Learning and Geospatial Analysis for Flood Hazard, Vulnerability, and Risk Assessment in Odisha, India

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月22日

查看全文
  • Manoranjan Mishra,
  • Rajkumar Guria,
  • Richarde Marques da Silva,
  • Somanath Sethi,
  • Minati Mishra,
  • Mansour Almazroui,
  • Rabin Chakrabortty,
  • Celso Augusto Guimarães Santos

摘要

Floods pose a significant threat to Odisha, India, due to its diverse geography and climatic conditions, which result in considerable risks to communities, infrastructure, and the economy. This study analyzed flood dynamics in Odisha by incorporating hazard, vulnerability, and risk assessments through the application of various machine learning models (MLMs). A total of 15 and 13 conditioning factors were used for flood hazard and flood vulnerability analyses, respectively. Multicollinearity diagnostic tests, the Pearson correlation coefficient, and the Boruta model were employed to analyze the relationships between flood occurrences and the factors influencing them. The findings revealed that approximately 11.87% of the area falls into the high to very high flood susceptibility category, while 14.80% is classified as highly vulnerable and 6.85% of the region faces significant flood risk. Coastal districts, including Jagatsinghapur, Kendrapara, Puri, Baleswar, Bhadrak, and Cuttack, are particularly flood-prone, with the Baitarani, Brahmani, Mahanadi, Budhabalanga, and Rushikulya river basins being the most affected. The RF model demonstrated superior performance in both flood hazard (AUC = 0.963) and vulnerability (AUC = 0.956) assessments, followed by Bagging, SVM, KNN, and GLM.

Graphical Abstract