From Sentinel‑2 to Geo‑Embeddings: Practical Vegetation and Land‑Cover Mapping

In this hands‑on session, students use Sentinel‑2 data to compute vegetation indices, explore spectral signatures, and build a simple land‑cover classification map of Göttingen (forest, crops, parks, other) directly in Python. They then revisit the same task with TESSERA geo‑embeddings, comparing how traditional spectral features and learned representations differ in accuracy, robustness, and interpretability for urban‑regional vegetation mapping.