BaySenseAI

A Scalable Platform for Monitoring and Predicting Biodiversity Based on Novel Community Sensors and Artificial Intelligence

„You only treasure what you measure.“

The loss of biological diversity is one of the greatest challenges of our time. What is discussed globally affects Bavaria very concretely: species are disappearing, habitats are changing, and climatic extremes are increasing. Yet, although biodiversity enjoys high political priority, we often lack a precise, comprehensive picture of its current state and dynamics.

This is where BaySenseAI comes in. The project combines cutting-edge technologies from Earth observation, biodiversity research, and artificial intelligence into a shared vision: a scalable, AI-based platform that not only documents biodiversity but can also provide forecasts regarding its future changes.

At its core are foundation models—powerful AI models capable of integrating multimodal data sources. Satellite imagery, climate time series, field surveys, and biodiversity assessments are analyzed together to precisely predict local species compositions.

The platform is being developed and tested in the Berchtesgaden National Park—a region of extraordinary ecological diversity. In the long term, its application is planned for all of Bavaria. BaySenseAI thus bridges a central monitoring gap and lays the foundation for data-driven, sustainable nature conservation strategies.

Background

Global biodiversity loss has long been felt in Bavaria. Species react sensitively to climate change, land-use changes, and human interference. However, effective conservation measures require reliable information: Where are species communities changing? Which habitats are particularly affected? And how are rare or elusive species developing?

Until now, many answers have relied on laborious fieldwork—precise, but time-consuming and costly. At the same time, technologies are available today that were unimaginable just a few years ago.

BaySenseAI utilizes this technological turning point. State-of-the-art biodiversity monitoring methods—such as audio recorders to capture bird and insect calls, camera traps for mammals, or DNA metabarcoding to analyze diversity at the genetic level—are combined with Earth observation data. Satellites provide continuous information on vegetation status, temperature, and moisture, supplemented by LiDAR and hyperspectral aerial surveys that capture the structure of habitats in three dimensions.

This data diversity is merged using artificial intelligence. Foundation models can process images, time series, and text information simultaneously—and derive ecological patterns from them. For the first time, this creates the possibility to map biodiversity changes on a large scale, systematically, and scalably.

Methods and Objectives

The development of the platform begins in a model region with high informative value: the Berchtesgaden National Park. Stretching between alpine peaks, forests, and valley landscapes, the region offers a distinct ecological gradient—a natural „stress test“ for the model. The motto: If it works in Berchtesgaden, it will work elsewhere.

Implementation follows three closely integrated steps:

1. Collection of high-resolution biodiversity data: Species and communities are recorded on over 200 sample plots using advanced in-situ sensors.
2. Integration of diverse Earth observation data: Satellite and aircraft-based systems provide complementary environmental information that is spatially and temporally processed.
3. Linkage via AI-supported foundation models: Both data streams are integrated into a common model.

A special added value lies in the ability of foundation models to independently recognize ecological connections. Species are not viewed in isolation, but as part of a community. If the model identifies one species with high confidence, it can infer that certain accompanying species are also likely to be present. The model thus operates at the level of entire species communities—a paradigm shift compared to classical, single-species-based approaches.

The long-term goal is to scale the project to all of Bavaria. This will enable comprehensive, locally differentiated biodiversity forecasts as well as robust conclusions regarding changes in all Bavarian ecosystems.

Immediate Added Value for the Free State of Bavaria

BaySenseAI creates an innovative platform for biodiversity monitoring in the Free State. The resulting forecasts open up diverse application possibilities:

• Support in the designation and prioritization of protected areas
• Foundation for regional planning processes
• Early identification of sensitive or threatened habitats
• Assessment of climate-induced changes

Furthermore, the project bridges practical research in Berchtesgaden National Park with university-level cutting-edge research. Bavaria thus positions itself as a pioneer for AI-supported biodiversity analysis in Europe.

Potential Synergies within bayklif2

BaySenseAI works closely with Berchtesgaden National Park. The data collected there will be available to other research projects in a timely manner, thereby strengthening the scientific network in Bavaria.

The methods developed in the project are modular and expandable in the long term. They can be transferred to other regions, additional species groups, or new sensor technologies.

In this way, not only is a platform created for current issues, but also a foundation for the biodiversity research of tomorrow.

BaySenseAI tells a story of technological innovation, interdisciplinary collaboration, and the shared goal of making Bavaria’s biological diversity visible, measurable, and therefore protectable.