Assessing the Evolution of Point Bars along the Niger River in the Niger Delta, Nigeria (1974-2024) using Remote Sensing and Machine learning

Abstract The evolution of point bars along the Niger River in the Niger Delta, Nigeria, from 1974 to 2024 using remote sensing and machine learning. The dataset includes DEM and Landsat series satellite imagery, supplemented by Sentinel-2 data for high-resolution analysis from USGS. Rainfall data from 1983 to 2023 was sourced from the Center for Hydrometeorology and Remote Sensing. Object-Based Image Analysis (OBIA) identified and mapped point bars. Support Vector Machines automated classification, ensuring accuracy through precision and recall metrics. ArcGIS 10.5 tracked changes over time, revealing significant fluctuations in river area and point bar dimensions. River area fluctuated significantly, decreasing from 46,376.54 km² in 1974 to 42,711.05 km² in 1984, rising to 49,601.2 km² in 2014, and falling to 44,796.47 km² in 2024. Contrarily, the total point bar area increased dramatically from 1,945.63 km² in 1974 to a peak of 7,026.33 km² in 2004, followed by a decline to 8,087.89 km² in 2024. This trend indicates complex interactions between sediment deposition, erosion, and river flow patterns. Notably, the maximum point bar area peaked in 1994 at 2,273.85 km², then fell to 783.84 km² in 2024, while the mean point bar area showed a general decrease over the study period. Spatial autocorrelation analysis revealed significant clustering of point bars and elevation (Global Moran’s Index = 1.059738), and a strong positive relationship between point bars and rainfall (R² = 0.9921). These findings align with global trends, underscoring the impact of climatic fluctuations and human activities on river morphology and sediment dynamics.