The role of AI in tackling climate change and its impacts: from science to early warning

Go back to programme

The role of AI in tackling climate change and its impacts: from science to early warning

  • Watch

    * Register (or log in) to the AI4G Neural Network to add this session to your agenda or watch the replay

  • Earth observation and climate data tend to be extremely large and detailed. Sometimes they are noisy but often also hard to interpret.

    Using data driven algorithms to complement traditional processing schemes holds great promise to speed-up the creation of actionable insights, speed-up the development of novel applications and improve the quality of  the output compared to existing algorithms.

     

    However, accounting for the data-gravity of earth-observation data, data-access and training schemes like distributed computing, federated-learning and generally filtering and sharing of data across borders must be employed.

     

    Furthermore, compared to the volume of available data itself, high quality annotations are either expensive or not available in abundance.

     

    The latter has led to the adoption of concepts used by large language models, but adapted to geospatial data, resulting in geospatial foundation models. Trained on very large volumes of earth observation data, these models can be adapted to various downstream tasks, e.g. for floods or wildfires, with a minimal number of labels, while generalizing across continents.

     

    In this presentation, we will discuss above topics, with some emphasis on their applicability to early warning systems.

    Speaker(s)
    • i

      Speaker abstract

      Title: Large scale AI models to create earth observation actionable insights - Abstract: Earth observation and climate data tend to be extremely large and detailed. Sometimes they are noisy but often also hard to interpret. Using data driven algorithms to complement traditional processing schemes holds great promise to speed-up the creation of actionable insights, speed-up the development of novel applications and improve the quality of the output compared to existing algorithms. However, accounting for the data-gravity of earth-observation data, data-access and training schemes like distributed computing, federated-learning and generally filtering and sharing of data across borders must be employed. Furthermore, compared to the volume of available data itself, high quality annotations are either expensive or not available in abundance. The latter has led to the adoption of successful concepts used by large language models, but adapted to geospatial data, resulting in geospatial foundation models. Trained on very large volumes of earth observation data, these models can be adapted to different various downstream tasks (e.g. flood or wildfire segmentation) with minimal amount of labels, while generalising across continents. In this presentation, we will discuss above topics, with some emphasis on their applicability to early warning systems.
      Jonas Weiss
      Senior Research Scientist
      IBM
    Speaker(s)
    • i

      Speaker abstract

      Title: Causal and explainable machine-learning models for disaster-induced displacement / Abstract: Disaster-induced displacement poses a significant threat to communities worldwide, demanding proactive measures for effective preparation. However, understanding the drivers of such displacement is a challenging task. Developing robust early warning systems require dealing with numerous complex and interacting factors in space and time, and not only predicting but also understanding the underlying reasons. New data collection initiatives allow us to address this challenge using machine learning. In this session, we will explore how recent AI developments can shed light on the preparedness of human mobility under extreme weather hazards. Specifically, we will showcase two case studies: (1) explainable AI (xAI) to examine displacement driven by sudden-onset disasters, such as floods and storms, at a global scale, and (2) the use of causal machine learning to understand the dynamics of displacement drivers during droughts in Somalia. Our research demonstrates the potential of explainable AI and causal machine learning algorithms in anticipating displacement, guiding policy-making, and supporting humanitarian efforts.
      José M. Tárraga
      Image Processing Lab
      Universitat de València
    Share this session
    In partnership with

    Are you sure you want to remove this speaker?