Global atmosphere models, such as the one in E3SM (Energy Exascale Earth System Model), typically consider many physical processes that are partitioned into various modules such as a dynamical core, deep convection, radiation, etc. A standard approach to coupling the modules is where one module modifies the model state and then passes it the next module. This sequential splitting approach is just one of many potential process coupling approaches, begging the question of which approach to implement for a given collection of processes. The choice is typically made after some empirical investigation comparing numerical results from a few implemented coupling approaches. Such a brute force approach is both time consuming and typically does not inform as to what other approaches might be advantageous.
To address the question of how to choose a coupling approach, a semi-discrete error analysis framework was developed that distinguishes the component of the local truncation error due to the process coupling approach from the local truncation error due to the time integration methods for each process. By interpreting the results of the analysis of candidate coupling methods, possibly combined with some empirical measurement of the terms using existing simulations, it is possible to choose the advantageous method without needing to first implement the method. Furthermore, the error analysis results can inform research into new coupling approaches that provide additional benefit for a given set of processes. This work will demonstrate the utility of the error analysis framework both in the a priori choice of coupling approach and in the development of improved approaches using aerosol lifecycle in the E3SM Atmosphere Model (EAM).
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. LLNL-ABS-838172.