Monitoring and managing oil spillage and environmental degradation through geoinformation

The Niger Delta region of Nigeria is a major oil-producing area which experiences frequent oil spills which severely impacts the local environment and communities. Effective environmental monitoring and management remain inadequate in this area due to negligence, slow response times following oil spills, and difficulties regarding access and safety. This thesis investigates the pervasive issue of oil spills in the Niger Delta region, employing a multidisciplinary approach to provide insight into their spatiotemporal patterns, environmental impacts, and socio-economic consequences on local communities. The research integrates advanced geospatial techniques, remote sensing analysis, and community based participatory methods to provide a framework for oil spill impact assessment and quantification. Utilising the spNetwork package in R, Network Kernel Density Estimates (NKDE) and Temporal Network Kernel Density Estimates (TNKDE) were employed to analyse oil spill patterns along the pipeline network. By transforming pipeline data into 500-metre lixels (linear pixels), this network-based approach uncovered chronically high-risk oil spill hotspots zones and tracked their temporal evolution, thereby offering a vital evidence base for targeted intervention and remediation. This method surpasses traditional spatial analyses by incorporating network constraints and revealing critical spatiotemporal patterns where spills recur over time. Furthermore, a remote sensing approach was developed, leveraging geospatial cloud computing and machine learning to evaluate vegetation health indices (SR1, SR2, NDVI, EVI2, GRNDVI, GNDVI). These indices were analysed using Slow Moving Average (SMA) regression, which revealed significant declines in vegetation health following oil spill events. The contaminated landcovers exhibit a Spearman’s correlation coefficient (ρ) ranging from -0.68 to -0.82, p < 0.005 and P-values below 0.05 in most landcover categories, suggesting a clear and consistent downward trend in the indices’ values, reflecting a decrease in vegetation health in contaminated areas between 2016 and 2023. A Random Forest (RF) classifier further quantified the extent of contaminated land cover, demonstrating the effectiveness of this method for monitoring environmental damage in this challenging terrain. The classifier revealed that landcovers, including contaminated vegetation, wetland, farmland, and grassland cover approximately 4% (1,180 hectares) of the total area. Conversely, the non-contaminated prioritised landcovers (non-contaminated vegetation, wetland, farmland, and grassland) account for 96% (32,215 hectares) of the total landcover area.