WHOI PO

Title: Emulators and Parameterizations: Indirect Tools for Ocean Science
Abstract: 
As Earth system models grow in complexity and computational cost, two powerful approaches offer opportunities to accelerate progress and deepen understanding: parameterizations and emulators. In this talk, I will clarify the distinction between these two strategies—parameterizations as embedded simplifications within models to represent subgrid processes like ocean mixing, and emulators as stand-alone surrogates or companions to such models that reproduce outputs efficiently and accurately under various forcings.

Focusing on the ocean, I’ll discuss recent results showing how regional mixed layer depth (MLD) serves as an emergent constraint on climate sensitivity, quantifying how surface mixing processes predict the spread in model projections of climate change. By combining observational MLD climatologies from Argo with energy balance emulators, we attribute about 40% of uncertainty in projected climate sensitivity to upper ocean processes.

I’ll also show how an emulator (the energy balance model–Kalman filter (EBM–KF)) can do many things we struggle to do with climate models.  It assimilates global surface temperature and ocean heat content to generate rapid, probabilistic projections and allows efficient exploration of policy thresholds, internal variability, and the impact of external forcings like volcanic eruptions.

Together, these new ocean-focused tools—emulators optimized for interpretability, accuracy, and speed—provide new ways to study the oceans' role in climate.