Speaker
Description
As modern urban studies leverage high-cadence Earth Observation (EO) data for Smart Cities applications, the radiometric consistency of satellite imagery becomes a critical factor for automated analysis. However, solar activity cycles and long-term space climate variability significantly affect the ionosphere and upper atmosphere. These fluctuations introduce noise and geometric distortions into the datasets, which can degrade the precision of urban monitoring systems.
In the framework of my research internship at the German Aerospace Center (DLR) focused on Smart Cities, this study explores how Deep Learning could help mitigate these atmospheric effects. The focus is on investigating the potential integration of solar indices as auxiliary inputs for CNN or Transformer-based models. The goal is to see if a data-driven approach can improve automated correction in multi-temporal satellite series, particularly for urban footprint analysis. By connecting solar physics with remote sensing, this work aims to explore more reliable data processing methods for future smart cities infrastructures.