Invited Talk
Hybrid AI for High-Resolution Peatland Mapping
Fahimeh Farahnakian · Invited Speaker
Abstract
Accurate mapping of boreal peatlands is essential for environmental monitoring, greenhouse gas inventories, and sustainable ecosystem management. This talk presents how recent advances in deep learning and data fusion techniques can improve peatland classification using open-access radar and optical satellite imagery.
The talk introduces a framework that integrates convolutional neural networks (CNNs) and Vision Transformers (ViTs) to capture both local spatial–spectral patterns and long-range contextual dependencies within complex peatland ecosystems. By combining spectral, textural, and structural information, the framework enables robust pixel-wise classification of peatland characteristics.
The presentation highlights the potential of hybrid AI architectures and multi-modal remote sensing for advancing large-scale environmental mapping and peatland intelligence applications.
Speaker Bio
Fahimeh Farahnakian is an AI specialist in the Geoinformatics Team at Geological Survey of Finland and a Docent of Computer Science at University of Turku. She is also the PI of an EU-funded project, Peat-Genie, coordinated by GTK, which focuses on developing AI-powered solutions for peatland monitoring and environmental intelligence. Her recent research focuses on machine learning, geospatial AI and explainable AI for environmental and geoscience applications. She has authored more than 60 scientific publications in international journals and conference proceedings.
Materials and Images
- Slides: To be added.
- Related links: To be added.
- Images: To be added.