Multi-dimensional assessment of flood susceptibility drivers in the urban watershed of Guwahati

Floods are a critical natural disaster causing widespread global devastation, exacerbated by the interaction of natural hydrological phenomena and human activities, particularly in rapidly urbanizing areas like Guwahati, India. The present study systematically examines the influence of sixteen envir...

Full description

Saved in:
Bibliographic Details
Published in:Acta geophysica Vol. 73; no. 6; pp. 5815 - 5837
Main Authors: Ahmed, Ishita Afreen, Talukdar, Swapan, Naikoo, Mohd Waseem, Shahfahad, Rahman, Atiqur
Format: Journal Article
Language:English
Published: Cham Springer International Publishing 01.12.2025
Springer Nature B.V
Subjects:
Accuracy
Agricultural land
Algorithms
Artificial intelligence
Artificial neural networks
Climate adaptation
Deep learning
Disaster management
Drainage
Earth and Environmental Science
Earth Sciences
Emergency preparedness
Environmental risk
Explainable artificial intelligence
Feature selection
Flood risk
Flooding
Floods
Genetic algorithms
Geophysics/Geodesy
Geotechnical Engineering & Applied Earth Sciences
Hydrology
Infrastructure
Land cover
Land use
Machine learning
Natural disasters
Neural networks
Parameters
Population density
Proximity
Rainfall
Research Article - Hydrology and Hydraulics
Resilience
Sensitivity analysis
Spatial analysis
Statistical analysis
Storm damage
Structural Geology
Support vector machines
Urban areas
Urban climates
Urban planning
Urban watersheds
Watersheds
Wetness index
Zoning ordinances
India
Flood susceptibility modeling
Machine learning algorithms
Guwahati urban watershed
Urban flood resilience
Feature selection methods
Human and natural impacts
ISSN:1895-7455, 1895-6572, 1895-7455
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Floods are a critical natural disaster causing widespread global devastation, exacerbated by the interaction of natural hydrological phenomena and human activities, particularly in rapidly urbanizing areas like Guwahati, India. The present study systematically examines the influence of sixteen environmental and socioeconomic parameters on flood dynamics using a comprehensive modeling framework that combines statistical analysis, advanced machine learning (ML) algorithms, and four feature selection methods: genetic algorithm (GA), Boruta, recursive feature elimination (RFE), and recursive partitioning (Rpart). The application of ML models—support vector machine (SVM) and random forest (RF)—with GA-derived parameters resulted in high predictive performance (accuracy: 0.892; precision: 0.887; recall: 0.868), while deep neural networks (DNNs) integrated with explainable artificial intelligence (XAI) enabled robust sensitivity analysis. The results show a strong positive correlation between the topographic wetness index (TWI) and land use and land cover (LULC) with flooding events. Conversely, factors such as proximity to agricultural land and built-up areas show a negative correlation. LULC, precipitation, distance to built-up areas, and population density were identified as important factors influencing inundation patterns. Furthermore, rainfall, LULC, and proximity to built-up areas emerged as key predictors, with SHAP values highlighting rainfall as the most influential driver of flood risk. Spatial analysis delineated approximately 773 km 2 as very low and 38.47 km 2 as very high susceptibility zones. These findings of the study offer actionable insights for urban planners, policymakers, and disaster management authorities to design targeted mitigation strategies, inform zoning regulations, and prioritize resilient infrastructure development. The approach also contributes to broader urban climate resilience strategies, aligning with SDG 11 and SDG 13 objectives.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1895-7455
1895-6572
1895-7455
DOI:10.1007/s11600-025-01666-7