Environment

Objectives

We focus our expertise on crucial environmental issues, including downscaling global climate models for local insights, emulating complex ecological systems, modeling species distribution in changing environments, and cutting-edge applications of AI in environmental science.

Research Axes

Axis 1: Environment

Axis 1 aims to address the challenges of biodiversity loss and climate change by focusing on adaptation and mitigation, with an emphasis on practical applications.

Adaptation to biodiversity loss and climate change

AI models that provide locally relevant and accurate predictions on biodiversity loss and ecosystem degradation will be developed.

• Biodiversity modeling with finescale environmental data.
  Current biodiversity models often rely on large-scale factors such as macroclimate, which do not provide the fine-grained predictions needed by local managers. We are working on developing advanced AI models that downscale predictions to obtain more accurate and actionable information.
• Models of species/ecosystem responses.
  Understanding how species distributions evolve over time and space due to climate change is crucial for conservation planning. Quantifying uncertainty in these models and assessing the probability of extreme events is essential to develop robust decision-making tools for biodiversity protection.
• Scenario prediction.
  Integrated scenarios that combine ecological, environmental, social, and economic factors will be developed. These forecasts will guide decision-makers in planning a sustainable future that reconciles biodiversity conservation with human needs.

Mitigation – Nature-based climate solutions (NCS).

Nature-based climate solutions offer the advantage of reducing CO₂ emissions while enhancing biodiversity. We will develop models that leverage high-resolution remote sensing data to monitor plant biodiversity and carbon, and support effective environmental solutions.

Axis 2: Artificial Intelligence (AI)

Structured around the three principles of R³AI program – Robust, Responsible, and Reasoning AI – we aim to develop AI solutions that are not only innovative but also impactful.

• Robust AI: models that generalize across time, space and biological diversity.
• Responsible AI: models that represent the needs and constraints of diverse impacted communities.
• Reasoned AI: Models that perceive the ecological world, learn complex representations and cause-effect relationships, and offer perspectives that lead to more informed and sustainable decisions.

Challenges

AI algorithms are known to perform well when very large datasets are available and when deployed in in-distribution settings. However, the problems at issue in biodiversity monitoring, remote sensing, and climate science are often far from this paradigm. We are challenged by:

• limited labeled data due to lack of expert knowledge and on-the-ground observations validations
• generalization across geographies with often more abundant data present in the Global North
• nonstationary data – as the climate is changing, the data exhibits changes in distribution over time
• long-tailed data for the rare or hard-to-observe species poorly represented

Anticipated Impact

Artificial intelligence is emerging as a transformative force in addressing climate change and biodiversity loss, enabling unprecedented insights into complex ecological systems. Our cutting-edge research harnesses AI’s computational power to develop precise climate models and create tools that integrate biodiversity and environmental data. By providing policymakers with advanced, data-driven insights, we are empowering decision-makers to develop more effective strategies for sustainable resource management, conservation, and climate adaptation. As we continue to push the boundaries of AI technology, they’re not just observing environmental challenges — they’re actively developing innovative solutions to protect and restore our planet’s delicate ecosystems.

Researchers

• Marcel Babin – Université Laval
• Olivier Bahn – HEC Montréal
• Soumaya Cherkaoui – Polytechnique Montréal
• Sylvie Daniel – Université Laval
• Youssef Diouane – Polytechnique Montréal
• Andrew Gonzalez – McGill University
• Samira Keivanpour – Polytechnique Montréal
• Hugo Larochelle – Université de Montréal
• Tegan Maharaj – HEC Montréal
• Chris Pal – Polytechnique Montréal
• Sébastien Sauvé – Université de Montréal
• Oliver Sonnentag – Université de Montréal

Research Advisor

Dana F. Simon: dana.simon@ivado.ca