Nelvinebi/Niger-Delta-Inundation-Mapping

A machine learning system mapping surface water extent in Nigeria's Niger Delta using Sentinel-1 SAR, CHIRPS rainfall, and SRTM elevation. Phase 1 detects current inundation; Phase 2 distinguishes true floods from permanent wetlands via temporal change detection honestly scoped, spatially validated, and dashboard-ready. # 🌊 Niger Delta Inundation Mapping System (NDIMS) - Nigeria > A machine learning pipeline that fuses **Sentinel-1 SAR backscatter**, **CHIRPS rainfall**, and **SRTM elevation** data to map surface water inundation across the **Niger Delta, Nigeria** one of Africa's most flood-vulnerable regions delivering GIS-ready outputs and an interactive dashboard for land use planning, wetland monitoring, and early-stage flood response. --- ## 📌 Problem The Niger Delta is among the world's most ecologically complex and flood-prone river deltas, spanning approximately 70,000 km² across Bayelsa, Delta, and Rivers States. Annual flooding displaces hundreds of thousands of residents, degrades agricultural land, and disrupts oil infrastructure yet reliable, high-resolution inundation maps remain scarce for emergency planners and local government agencies. Traditional flood monitoring in the region is constrained by cloud cover (optical satellites fail during peak rainy seasons), limited ground-based gauge networks, and the absence of automated pipelines capable of processing SAR imagery at operational scale. Existing global flood products are too coarse (250 m–1 km) to delineate Local Government Area (LGA)-level flood extents needed by NEMA and SEMA for resource allocation. There is a critical need for a **reproducible, open-source pipeline** that leverages Synthetic Aperture Radar unaffected by cloud cover fused with rainfall and terrain data, to produce actionable, stakeholder-ready inundation maps of the Niger Delta. --- ## 🎯 Objective - Acquire and process **Sentinel-1 SAR backscatter (VV polarisation)** for the Niger Delta wet season using **Google Earth Engine** - Integrate multi-source environmental rasters **CHIRPS rainfall** and **SRTM digital elevation model** into a unified geospatial feature stack - Train a **Random Forest classifier** to distinguish inundated from non-inundated surfaces using SAR + rainfall + elevation features - Generate a * …