Leveraging VAT for Spatio-Temporal Machine Learning on Environmental Data

VAT is one of the fundamental services offered in NFDI4Biodiversity. It is a web-based GIS application, powered by Geo Engine, that supports users to visualise, analyse and transform geospatial vector and raster data without requiring deep GIS knowledge. In this use case, VAT generates spatio-temporal training data for a species distribution modelling task. It easily handles the spatio-temporal alignment of different raster layers, while offering multiple curated data products as a solid basis for modelling. Specifically, in this use case, GBIF occurrences of Arnica montana are selected and combined with land use classification from the Ökosystematlas, weather data from CHELSA and topographic information. Weather data are also temporally aggregated annually using VAT.