New forecasting contest aims to advance AI-driven species distribution modelling in the UK
Accurate forecasting of species distributions can help improve our understanding of how species respond to environmental change. These forecasts can subsequently support more effective environmental monitoring, inform conservation practice, and help identify emerging trends in the spread of invasive species.
Species distribution modelling typically combines species occurrences with environmental data to predict where species are likely to be found across space and time. Advances in modelling approaches, combined with increasingly comprehensive datasets such as the NBN Atlas, offer new opportunities to generate more accurate forecasts.
As part of the SPHERE-PPL project, and in collaboration with the National Biodiversity Network (NBN) Trust, researchers from Imperial College London, Dr Will Pearse and Dr Alex Rabeau, have launched a new forecasting contest focused on improving species distribution modelling for nine key UK species. Using data from the NBN Atlas, participants are invited to develop models that forecast species occurrences within a given UK region for 2026.
Participants will need to consider different ecological contexts when forecasting the distributions of distinct species as part of this contest. These include rarer species with narrower ranges that are likely to be under-sampled (e.g., the Osprey), which may require different forecasting approaches than those used for more common and widespread species (e.g., the Otter or Common pipistrelle).
The NBN Atlas is one of the UK’s most comprehensive sources of biodiversity data, bringing together more than 387 million species occurrence records from over 190 data partners. Participants are encouraged to incorporate additional relevant predictor datasets alongside NBN Atlas data to develop and improve their forecasting models.
The contest launches on 1 October and runs until 14 December. Anyone with a background in data science, statistics or machine learning and an interest in ecology is encouraged to participate.
For more details and instructions on how to enter the contest please visit the contest GitHub repository.
About SPHERE-PPL
SPHERE-PPL is in an initiative focused on advancing the application of AI forecasting models and Probabilistic Programming Languages across health and environmental sciences. This is principally achieved by building networks that connect organisations operating in these domains – such as NGOs, government departments, and private companies – with analytical experts. A key mechanism for this is the development of open source forecasting contests aimed at developing accurate models to tackle real-world challenges.