Articles | Volume 19, issue 20
https://doi.org/10.5194/gmd-19-10309-2026
https://doi.org/10.5194/gmd-19-10309-2026
Development and technical paper
 | 
29 Oct 2026
Development and technical paper |  | 29 Oct 2026

Using a two-stage Rosenbrock solver to increase computational efficiency in the DOE's Exascale Earth System Model version 1 (E3SMv1)

Rong-You Chien and Joshua S. Fu

Data sets

Input files for ROS2-E3SM Box Model v1.0 for GMD submission R.-Y. Chien and J. S. Fu https://doi.org/10.5281/zenodo.19753716

chemUCI output of the E3SMv1 for GMD submission R.-Y. Chien and J. S. Fu https://doi.org/10.5281/zenodo.19703404

trop_strat_mozart output of the E3SMv1 for GMD submission R.-Y. Chien and J. S. Fu https://doi.org/10.5281/zenodo.19703406

Model code and software

E3SMv1 branch used for the GMD submission Q. Tang et al. https://doi.org/10.5281/zenodo.19745988

pshchien/E3SM_ROS2_BOX: ROS2-E3SM Box Model v1.0 for GMD submission (v1.0) R.-Y. Chien and J. S. Fu https://doi.org/10.5281/zenodo.19597033

pshchien/ROS2-E3SM: ROS2-E3SM v1.1 for GMD submission (v1.1) R.-Y. Chien and J. S. Fu https://doi.org/10.5281/zenodo.19124522

Download
Short summary
We implemented a two-stage Rosenbrock solver (ROS2) in E3SMv1 and evaluated it with two atmospheric chemistry mechanisms. At a 180 s timestep, ROS2 reduced chemistry integration time by ~33% versus the original implicit solver on Cori/KNL CPUs, while maintaining stable one-year simulations and small ozone differences (≤1.03 ppb). Species-dependent differences in ozone, NO, NO2, and OH were small. Results support ROS2 as a lower-cost option, with further convergence and CPU/GPU testing needed.
Share
Altmetrics
Final-revised paper
Preprint