Modelling Potential Fishing Zones in Kenya's Exclusive Economic Zone Using Machine Learning and Satellite-derived Oceanographic Variables

Kenya’s industrial fishing zones remain poorly characterised using spatially explicit predictive approaches, limiting the use of environmental information in fisheries management and marine spatial planning. This study develops a machine learning analytical workflow to predict potential fishing zones (PFZ) within Kenya’s ~100,204 km² industrial fishing zone using open-access satellite-derived oceanographic data and Global Fishing Watch (GFW) effort data for 2019, 2021 and 2022. The study develops a model that correctly identifies fishing locations based on oceanographic variables such as Sea Surface Height (SSH), chlorophyll-a concentration, salinity, and Sea Surface Temperature (SST). This model will be validated and applied across the full EEZ to produce spatially explicit fishing suitability maps (both binary and three-tier classifications), representing the first predictive fishing zone product developed specifically for Kenyan waters. These spatial outputs offer a practical, low-cost, and replicable decision-support tool to support fisheries management, marine spatial planning, and design of an effective marine protected area network. Similarly, by demonstrating that freely available remote sensing data can reliably predict industrial fishing activity, this study provides a scalable framework applicable across the western Indian Ocean region, directly supporting evidence-based ocean governance aligned with both national, regional and international policy priorities.