Use of machine learning and AI to improve the utility of climate risk data (Invited Presentation)

As the impacts of our changing climate become more prevalent, scientists in academia, industry, government, and non-profit organizations grapple with the challenge of making climate risk data useful and accessible. Many scientists and organizations rely on machine learning and artificial intelligence to enhance the spatial and temporal resolution of climate data. Furthermore, the ability to scale our models is a major consideration and machine learning models can be used to augment dynamic models. Several applications are presented showcasing how machine learning can be used to downscale climate risk metrics, improve model biases, and adjust for local conditions that moderate climate risks like wildfire and extreme heat. We will highlight how these algorithms are trained, validated, and interpreted to counter perceptions of methodological “black boxes”. We also explore multiple approaches, including AI, to make our data more easily accessible and understandable for non-scientist decision makers using climate risk information.