ADVANCING ANTARCTIC ICE-FREE ZONE MAPPING: INTEGRATING DEEP LEARNING AND IN SITU VALIDATION FOR MULTI-TEMPORAL SATELLITE IMAGERY

• Grant number: 25/13513-3
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: March 20, 2026
• End date: March 19, 2027
• Field of knowledge: Physical Sciences and Mathematics - Geosciences
• Principal Investigator: Erivaldo Antonio da Silva
• Grantee: Eduardo Soares Nascimento
• Supervisor: Pedro Miguel Berardo Duarte Pina
• Host Institution: Faculdade de Ciências e Tecnologia (FCT). Universidade Estadual Paulista (UNESP). Campus de Presidente Prudente. Presidente Prudente , SP, Brazil
• Institution abroad: Universidade de Coimbra (UC), Portugal
• Associated to the scholarship: 23/16874-1 - METHODOLOGY TO MAP ICE-FREE ZONES IN THE ANTARCTIC REGION USING SATELLITE IMAGES, BP.DR

Abstract

This Doctoral research internship aims to significantly advance the methodology for mapping Antarctic ice-free zones, crucial indicators of climate change impact. Building on foundational work in Brazil that uses multi-temporal Landsat and Sentinel imagery and initial Machine Learning (Random Forest) classifications, this 12-month mobility period at the University of Coimbra, Portugal, seeks to overcome persistent classification challenges, particularly concerning spectrally ambiguous features like clouds, vegetation, and ice-snow transitions. Under the supervision of Professor Pedro Pina, a leading expert in polar remote sensing and advanced image analysis, the internship will focus on three key objectives. Firstly, it will investigate and implement advanced classification algorithms, including Support Vector Machines (SVM) and Deep Learning architectures (e.g., Convolutional Neural Networks), to enhance discrimination accuracy. Secondly, a validation framework will be established, leveraging unique in situ ground truth data and very-high-resolution drone imagery provided by Professor Pina's research group, which is critical for ground-truthing and refining the methodology. Thirdly, the refined methodology will be expanded and applied to map ice-free zones across additional key Antarctic peninsulas, contributing to a robust multi-temporal analysis framework for assessing their evolution. This international collaboration is indispensable for integrating cutting-edge techniques and expert knowledge, promising substantial academic gains, including scientific publications and the development of a highly accurate, generalizable mapping solution vital for understanding Antarctic environmental dynamics. (AU)