Scholarlink

Urban flooding is a global issue triggered by factors such as low-lying zones, blocked drainage, and insufficient infrastructure. Rapid urbanization aggravates the problem, leaving cities vulnerable to flood impacts. Also, climate change amplifies the issue with more intense rainfall. Urban centers in Assam, particularly the Guwahati Metropolitan Area (GMA), are facing a growing surge in frequent floods annually, impacting diverse socio-economic groups and posing multifaceted challenges. Therefore, the present study uses an innovative combination of machine learning, the random forest (RF) algorithm, and a metaheuristic optimization algorithm, the ant colony optimization (ACO) model, to comprehensively model vulnerability to flooding in the dynamic urban catchment of Guwahati. The study has a dual objective of integrating a combination of ACO and RF algorithms to determine the current urban flood susceptibility and predict the future flood susceptibility for 2050 under different representative concentration pathways (RCPs). The analysis reveals flood susceptibility areas of 64.92 km2 (very high) and 66.28 km2 (high), concentrated in low-lying plains and near the main river. Moreover, the RF-ACO model excels in prediction with very low RMSE of 0.167 and an excellent MAE of 0.1447, showcasing its high accuracy in predicting flood susceptibility. Additionally, future flood susceptibility predictions for different RCPs indicate increasing trends in flood-prone areas compared to current scenario, emphasizing the need for specific flood management plans. By incorporating the findings from the flood hazard model and considering the impacts of climate change, policymakers can develop individual flood management strategies to improve urban resilience and protect the well-being of local population.

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

Quantifying Future Flood Risk in Guwahati Urban Watershed Using a Metaheuristic-Based Random Forest Model Under Climate Change Scenario

  • 章节

  • 发表日期：

    2025年

查看全文
  • Ishita Afreen Ahmed,
  • Swapan Talukdar,
  • Shahfahad,
  • Mirza Razi Imam Baig,
  • Mohd Rihan,
  • Asif,
  • Atiqur Rahman

摘要

Urban flooding is a global issue triggered by factors such as low-lying zones, blocked drainage, and insufficient infrastructure. Rapid urbanization aggravates the problem, leaving cities vulnerable to flood impacts. Also, climate change amplifies the issue with more intense rainfall. Urban centers in Assam, particularly the Guwahati Metropolitan Area (GMA), are facing a growing surge in frequent floods annually, impacting diverse socio-economic groups and posing multifaceted challenges. Therefore, the present study uses an innovative combination of machine learning, the random forest (RF) algorithm, and a metaheuristic optimization algorithm, the ant colony optimization (ACO) model, to comprehensively model vulnerability to flooding in the dynamic urban catchment of Guwahati. The study has a dual objective of integrating a combination of ACO and RF algorithms to determine the current urban flood susceptibility and predict the future flood susceptibility for 2050 under different representative concentration pathways (RCPs). The analysis reveals flood susceptibility areas of 64.92 km2 (very high) and 66.28 km2 (high), concentrated in low-lying plains and near the main river. Moreover, the RF-ACO model excels in prediction with very low RMSE of 0.167 and an excellent MAE of 0.1447, showcasing its high accuracy in predicting flood susceptibility. Additionally, future flood susceptibility predictions for different RCPs indicate increasing trends in flood-prone areas compared to current scenario, emphasizing the need for specific flood management plans. By incorporating the findings from the flood hazard model and considering the impacts of climate change, policymakers can develop individual flood management strategies to improve urban resilience and protect the well-being of local population.