Accelerating the Usability of Robust Climate Information: Recent Advances in AI and ML Methods for Modeling Climate Impacts and Extreme Events (Invited Presentation)

For climate information to be actionable, it must be comprehensive, high-resolution, carefully designed for the intended application and have constrained and well-quantified uncertainty bounds. Importantly, the time horizon of the analytics must be aligned with the application and use case. Making climate risk-aware decisions necessitates robust and actionable climate intelligence, yet having intelligence that meets the varied needs of decision makers has been a computationally difficult goal at best and at worst, a roadblock to using climate information. Recent research in machine learning and artificial intelligence has already helped to address these issues – as well as other modeling challenges – by optimizing new sensor networks, filling in the forecast time horizons needed between weather and climate models, and enabling new climate risk tools, to name just a few. This talk will provide an overview of the applications as well as the limitations of recent work and reflect on how these advances have shifted the landscape of actionable insights for climate risk-aware decision-making.