Editors’ Highlights are summaries of recent papers by AGU’s journal editors.
Source: Journal of Advances in Modeling Earth Systems
Earth Systems Models are unable to resolve all the processes affecting climate, so important unresolved processes are represented by parameterizations. The traditional approach is to hand-tune these parameters using expert knowledge, a time-consuming process that involves matching climate model output to observational data as closely as possible.
Yarger et al. [2024] introduce an alternative “automated calibration” framework that chooses a subset of the uncertain parameters automatically and in a fraction of the time. The automated method builds a very fast machine learning-based substitute (surrogate) of the Earth System Model to predict the present-day climatology for 11 spatial fields over four seasons, and then uses the surrogate to search for the parameters that best match present-day climate observations. Applying the automated calibration to the atmospheric component of the U.S. Department of Energy’s Energy Exascale Earth System Model (E3SM) version 2 reduces the root mean square error by 2.7% on average across the different spatial fields and seasons, compared to the hand-tuned solution.
Citation: Yarger, D., Wagman, B. M., Chowdhary, K., & Shand, L. (2024). Autocalibration of the E3SM version 2 atmosphere model using a PCA-based surrogate for spatial fields. Journal of Advances in Modeling Earth Systems, 16, e2023MS003961. https://doi.org/10.1029/2023MS003961
—Jiwen Fan, Editor, JAMES