I'm pleased to share that I've successfully completed an 8-week full-time internship at the Christian Doppler Laboratory for Geospatial and EO-based Humanitarian Technologies (GEOHUM) at the University of Salzburg's Department of Geoinformatics.
During this intensive research position, I had the opportunity to work on developing and testing a deep learning model specifically designed for flood identification using Sentinel-1 radar imagery. The project encompassed the complete geospatial data workflow:
- Acquiring and processing Sentinel-1 radar data using QGIS and Python
- Developing training and testing datasets for model development
- Training a Convolutional Neural Network (CNN) for automated flood classification
- Conducting comprehensive qualitative and quantitative accuracy assessments
This experience allowed me to bridge theoretical knowledge from my studies with practical applications in humanitarian technologies. Working with radar imagery presented unique challenges compared to optical satellite data, particularly in developing robust classification methods for flood detection under various environmental conditions.
I'm grateful to the team at Z_GIS for their guidance and support throughout this internship. The skills I've developed in deep learning applications for geospatial analysis will be invaluable as I continue my research in the Copernicus Master in Digital Earth program.