Developing a reusable infrastructure for machine learning on diverse earth observation data for sustainable agriculture and forestry
2025
McClelland, Jennifer | Dominic, Anto Raja | Beyer, Florian | de Kock, Arno | Golla, Burkhard
The integration of machine learning (ML) into environmental protection, particularly in sustainable agriculture and forestry, is increasingly vital given the spatio-temporal scale of the data and analysis. Earth observation data from Sentinel-1 (S1), Sentinel-2 (S2), weather, and LiDAR provide valuable insights, but applying ML algorithms to these diverse datasets presents challenges due to their differences in data structure and formats as well as spatial, spectral and temporal resolutions. This research develops a multi-purpose, extensible infrastructure using open-source technologies, implemented within the cloud platform CODE-DE at Julius Kühn-Institut (JKI), to streamline ML applications for geo-located earth observation data. The infrastructure supports diverse data types, including satellite, weather, and LiDAR records, and is adaptable to future ML models. It has been rigorously tested for detecting plant growth stages (BBCH), demonstrating its potential in agricultural analysis. Future work will extend this to detecting tree and shrub growth events. This research contributes to sustainable agriculture by advancing reusable ML solutions for environmental monitoring.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Editeur Gesellschaft für Informatik e. V. (GI)
ISSN 2944-7682Cette notice bibliographique a été fournie par Julius Kühn-Institut
Découvrez la collection de ce fournisseur de données dans AGRIS