BaySenseAI - A scalable biodiversity observation and prediction platfom based on novel community sensors and artificial intelligence

About

“You only treasure what you measure”. Anthropogenic biodiversity loss is a global concern and a national and federal priority, in particular in the context of climate change. Nevertheless, our information on current biodiversity status and change is still very incomplete. Technological advances in environmental and species sensing, coupled with appropriate AI systems, could make it possible, for the first time, to measure biodiversity at scale – i.e. over large spatial extents and at high resolution and frequency. Realizing this vision, however, requires combining specialized skills from multiple disciplines. This proposal brings together a interdisciplinary team of experts in environmental sensing, biodiversity research, and artificial intelligence. Using a case study in the Bavarian alps (Berchtesgaden National Park), we will develop a scalable next-generation biodiversity observation and prediction platform, synergistically combining cutting-edge technologies from AI/deep learning, remote and proximal sensing as well as modern biodiversity sensors such as metabarcoding, visual image analysis and bioacoustics. Our goal is to develop a foundational AI model that can incorporate multimodal inputs (images, tabular data, time series), as well as incomplete local species observations from rapid assessment tools, to predict species compositions at scale and with high accuracy. The case study will be used to train this model in automatically selecting new sampling locations, so that it could later be scaled to act as an observation platform for the entire Bavarian state. Our research program will deliver multiple scientific and societal benefits, including better understanding of the drivers of species distribution and abundance; efficient data generation pipelines; a scalable biodiversity model; and low-marginal-cost biodiversity maps to improve policy planning.