An automated approach for updating land cover change maps using satellite imagery

Abstract

Land cover change is a critical aspect of global environmental dynamics, influencing ecosystems, biodiversity, and climate change. This study presents an automated approach for updating land cover maps across Europe, combining Sentinel-1 and Sentinel-2 satellite imagery within the Copernicus framework. The application utilises machine learning algorithms to categorise land cover changes into classes such as no change, retained/reclassified, deurbanisation, afforestation, and urbanisation. Case studies in Poland, Greece, and Italy demonstrate the application's effectiveness, revealing the impact of motorway construction, afforestation efforts, and rapid urbanisation. Overall accuracy rates ranged from 68% to 95%, emphasising the reliability of the methodology. The open-source application, implemented in Python Jupyter and Voila, provides a user-friendly platform for researchers and stakeholders to monitor and analyse land cover changes, supporting informed decision-making for sustainable land management and conservation efforts. This study contributes valuable insights to understanding and addressing the environmental consequences of land cover changes in diverse geographical contexts.