DEEP LEARNING APPLIED TO REMOTE SENSING FOR THE ANALYSIS AND MAPPING OF LANDSCAPE UNITS OF THE COASTAL PLAIN OF MARANHÃO STATE/ BRAZIL

• Grant number: 24/15740-4
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor
• Start date: December 26, 2024
• End date: December 25, 2025
• Field of knowledge: Humanities - Geography - Human Geography
• Principal Investigator: Archimedes Perez Filho
• Grantee: Francisco Afonso Cavalcanti Júnior
• Supervisor: Fabian Faßnacht
• Host Institution: Instituto de Geociências (IG). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Institution abroad: Freie Universität Berlin, Germany
• Associated to the scholarship: 22/15889-2 - Environmental transformations in coastal plains of the State of Maranhão/Brazil resulting from holocene climatic pulsations and the use and occupation of land, BP.PD

Abstract

This project aims to apply deep learning techniques in remote sensing for the detailed mapping and analysis of the landscape units of the coastal plain of the state of Maranhão/Brazil, covering the eastern and western coasts. Deep learning convolutional neural networks (CNNs) will be employed to identify, classify, and monitor different landforms and land uses, such as dunes, paleodunes, marine terraces, and other geomorphological features. The research aims to identify, classify and monitor the different landforms and land uses in the region. The integration of these advanced techniques with geographic information systems (GIS) will make it possible to create detailed maps and spatial analysis of changes in land use and geomorphological features. The methodology involves the collection of high-resolution data, rigorous pre-processing, analysis with artificial intelligence, and validation of the results with field data. The study seeks not only to contribute to the literature on coastal geomorphology, but also to meet the need to identify, more accurately, the differentiations and ruptures in the landscape that may go unnoticed in a conventional analysis in remote sensing.