Jeremy Sousa and Shixuan Zhang1
In recent decades, increased spatial resolution has advanced the resolvability of physical features in earth system models; however, time step sensitivity has received less focus. This study investigates convergence of simulations at different time steps in the Department of Energy’s climate model, the Energy Exascale Earth System Model (E3SM), where the simulation with the smallest time step is treated as the reference solution for all other simulations. A few basic surface variables in the model such as pressure and temperature are selected as the medium for assessing the time step sensitivity both regionally and globally. A ten-day simulation of the coupled atmosphere-land model in E3SM is performed at different time steps, and the solutions are compared using root mean square error (RMSE). In regions where RMSE is maximized, further statistical analysis is performed to determine the forcing for the differences. In-situ high-frequency data are used as a reference for model output accuracy. Results demonstrate that while some features are correlated in time over different time steps, amplitude is not preserved.
1Pacific Northwest National Laboratory, Richland, WA