This panel will discuss the current state of global, coupled climate modeling, the next phases in identifying and reducing model uncertainty, our ability to leverage an ever-increasing flow of environmental data, and the roles of government and industry in post-processing innovation.

The success of climate model projections has been scrutinized extensively over the past decade. Even as past model ensembles begin to show signs of verification as we approach the mid-21st century, the standard of deployable information for decision-makers continually becomes more stringent. The amount of available data through climate simulations and in situ and remote observation infrastructure has exploded. This has given rise to a variety of methods, including machine and deep learning, used to ingest, digest, and investigate the implications of modeling output as it pertains to global down to regional climatic trends and extremes. Across spatial and temporal scales, the predictability of various phenomena is subject to uncertainty introduced by observation gaps and assimilation, ensemble design, model complexity, and computational limitations. Discussion will center around the following key questions:

How do we assess and reduce the uncertainty in the near-term or coming decades? Research gaps in traditional modeling? What role will machine learning and deep learning serve in model initialization, execution, and post-processing? What techniques are available concerning convergent and divergent modeling solutions?

- Indicates paper has been withdrawn from meeting

- Indicates an Award Winner