Creating the framework for the next generation Energy Exascale Earth System Model (E3SM) at PROCEED (Perturbed physics ensemble Regression Optimization Center for ESM Evaluation and Development)

Predicting the magnitude of future environmental change necessitates modelling the complex and multi-scale Earth system and coupling it to future economic, energy, and infrastructure decisions. Earth System Models

earth system model (ESM)

A model that simulates how chemistry, biology, and physical forces work together. These models are similar to, but much more comprehensive than, global climate models.

Source: U.S. Department of Energy Source

(ESMs), which are the foundation of climate predictions, must parameterize unresolved processes. This leads to substantial parametric uncertainty in future climate predictions. A great deal of the parametric uncertainty in future climate predictions come from processes occurring at the scale of cloud droplets, aerosols,

aerosols

Small particles or droplets, such as wildfire smoke, volcanic gases, or salty sea spray, that float in the air. They are emitted by both natural events and human activities.

Source: NASA Global Climate Change Source

and chemistry. These processes must also be parameterized in high-resolution convection-permitting global models and Large Eddy Simulation (LES). We will need to use highly-parameterized models to make climate predictions for the foreseeable future.

The Perturbed parameter ensemble (PPE) Regression Optimization Center for ESM Evaluation and Development (PROCEED) is a university-lab partnership. The University of Wyoming (UW), together with the University of Hawaii at Manoa (UHM), will develop new approaches to improve understanding of how uncertainty in the parameterization of microscale processes imprints on to global climate through aerosol-cloud interactions.

aerosol-cloud interaction

The interaction of airborne particles (aerosols) with clouds. Scientists study how aerosols affect clouds and how these interactions can impact the Earth's climate.

Source: DOE Source

  Collaborators include the DOE Pacific Northwest National Lab (PNNL) and Lawrence Livermore National Lab (LLNL). The long-term goal of PROCEED is to create a unique, observationally-driven model evaluation and development framework in the context of the Energy Exascale Earth System Model (E3SM).

Energy Exascale Earth System Model (E3SM)

The Energy Exascale Earth System Model (E3SM) project includes Earth System Models, simulations, and predictions to investigate energy-relevant science using code optimized for DOE's advanced supercomputers.

Source: NVCL Web Page Source

PPEs comprise hundreds of integrations of an Earth system model randomly varying uncertain model parameters within their probable ranges. Machine learning

machine learning (ML)

Machine Learning (ML) refers to the field and practice of using algorithms that are able to “learn” by extracting patterns from a large body of data.

Source: Center of Excellence Source

is used to train surrogate models on these integrations to systematically explore parametric uncertainty in Earth system models. PROCEED will apply these techniques to E3SM to develop a PPE.

Once developed, evaluating what parts of parameter space are consistent with observations is important to understanding what PPEs are telling us about physical processes in the Earth system. PROCEED will develop methodologies that allow comparison between Earth system models and in situ measurements from the surface and from airborne platforms. In situ observations are suited to constraining the small, fast processes governing aerosols, clouds, and atmospheric chemistry. Techniques will be implemented for sampling E3SM and nudging it to observed weather to facilitate comparison to observations. There is a substantial scale gap between the global model resolution of E3SM and airborne and surface observations. To bridge this scale gap and account for sampling uncertainty PROCEED will utilize simulations conducted as part of the Large-Eddy Simulation (LES) ARM Symbiotic Simulation and Observation (LASSO) exercise to produce statistically-robust observational constraints for use with the E3SM PPE. Developing these techniques has the potential to produce higher confidence climate predictions through its use of new observational data sources.

The PROCEED framework will deliver three key results: (i) a framework for systematically confronting parametric uncertainty; (ii) a methodology for utilizing airborne and surface observations that are well-suited for constraining parameterized, microscale processes; and (iii) physical understanding of how parameterized processes limit the predictive ability of global models in the context of radiative forcing due to aerosol-cloud interactions.