A biodiversity dashboard for Switzerland
2025 - 2026
Cooperation Financing
Rapidly growing biodiversity databases and innovations in deep learning gave rise to a new generation of versatile species distribution models with spatiotemporal resolution. Their complexity makes them interesting for many timely use cases, but it also challenges their deployment. We are building a web application that tailors outputs of such cutting-edge models directly to the needs of a broad user group.
Specifically, we collaborate with InfoFlora and info fauna designing an intuitive web interface combining two approaches:
- static maps of key biodiversity statistics optimized for large-scale overviews, and
- a dynamic dashboard for smaller-scale requests, allowing users to obtain predictions for targeted species at selected locations and timeframes.
Designed as a lasting infrastructure with a modular design, it can host a range of models across different taxa and spatiotemporal extents, while allowing for easy model updates and the addition of new ones.
The dashboard ¶
The Biodiversity Dashboard is built on deep species distribution models (deepSDMs) that jointly model the distribution and seasonal patterns of species (currently plants and butterflies) in Switzerland based on millions of citizen-science observations and high-resolution environmental data. The dashboard runs these computationally intensive models on demand and returns results tailored to your question.
The current capabilities hosted by the dashboard and introduced in the videos below are the following:
- finding the most probable species for a selected point or polygon,
- inspecting seasonal patterns in observation probabilities for species at a selected point,
- inspecting spatial patterns in observation probabilities for species within a polygon,
- simulating how environmental change (meteorology, land use and vegetation) affects observation probability, utilizing highly granular environmental and satellite data.