Oil spills pose a growing threat to marine environments, causing severe harm to marine life, ecosystems, and coastal habitats. Effective and efficient oil spill detection methodology is crucial for mitigating these impacts. Hyperspectral imaging collects and processes information from across the electromagnetic spectrum. Unlike traditional imaging techniques which collect only three bands of light, hyperspectral imaging divides the spectrum into numerous narrow bands. Detection of oil spills mostly depends on the traditional SAR (Synthetic Aperture Radar) data which may not always provide sufficient discrimination between oil spills and other features in the marine environment. The current work focuses on improving the detection of oil spills by using hyperspectral images and analyzing them with UNet segmentation along with Explainable AI (XAI) methods. To make the model’s predictions easier to understand, a visual explanation of the deep learning results is provided, based on the SHapley Additive exPlanation (SHAP) method tailored for segmentation tasks. The combined approach helps increase the accuracy of oil spill detection in marine areas.