Water scarcity in Nigeria has worsened because of climate change, especially in peri-urban agricultural areas that experience environmental shocks and poor infrastructure. Bayelsa State is the focus of this study because its sustainable agriculture is under threat from erratic rainfall, tidal flooding and rapid urban expansion. The research employs artificial intelligence (AI) to build predictive models which include long short-term memory networks for temporal forecasting and random forest classifiers for spatial risk assessment with feature importance analysis. The research draws data from satellite records and IPCC scenario models (RCP 4.5 and RCP 8.5) and local hydrological inputs for the period from 1996 to 2024. The preliminary modeling results indicate that water stress during dry seasons will become more severe because of climate change and changes in land use. The research uses its findings to recommend adaptive planning strategies which include smart irrigation and flood-resilient water storage systems and community-based early warning protocols (Rasmussen et al., 2020). The research integrates predictive AI models with contextual adaptation pathways to support proactive policy development and sustainable planning for vulnerable agricultural populations in Bayelsa. The framework presented offers a scalable solution for other flood-prone agricultural regions in Sub-Saharan Africa.