Machine Learning-Based Flood Inundation Mapping and LULC Classification in Ahoada West, Rivers State, Nigeria Using Satellite Imagery

Abstract Flooding is one of the most recurrent and destructive hydro-environmental hazards in Nigeria Niger Delta, exacerbated by climate variability, rapid land use change, and geomorphological vulnerability. This study integrates multi-temporal Sentinel-1 Synthetic Aperture Radar (SAR), Sentinel-2 multispectral imagery, geospatial analysis, and machine learning (ML) techniques to map flood inundation and analyze land use/land cover (LULC) dynamics in Ahoada West Local Government Area, Rivers State, Nigeria, from 2018 to 2023. Flood inundation extent and severity were derived using DEM-based flood depth modeling, while LULC classification was performed using five supervised ML algorithms: Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Extreme Gradient Boosting (XGBoost), and Decision Tree (DT). Model performance was evaluated using overall accuracy, Kappa coefficient, and class-based metrics. Results show that flood extent increased non-linearly with water level, with the highest inundation recorded in 2022, affecting approximately 98.93 km² (22.5%) of the study area under very high flood severity. LULC analysis revealed a 12.6% increase in built-up areas between 2018 and 2023, largely at the expense of wetlands and vegetated land. Among the tested classifiers, RF achieved the highest accuracy (91.3%) and Kappa coefficient (0.89). Flood Hazard Index (FHI) analysis identified river/stream (0.91) and built-up areas (0.62) as the most flood-vulnerable classes. The findings demonstrate that integrating satellite data with ML provides a robust framework for flood risk zoning and land-use planning in data-scarce deltaic environments. The study recommends incorporating ML-derived flood risk maps into early warning systems, spatial planning, and climate adaptation strategies in the Niger Delta.