Biological and Environmental Research - Earth and Environmental System Sciences
Earth and Environmental System Modeling

Diagnosing and Addressing Numerical Noise by Improving Spatial Discretizations in CLUBB

Monday, December 7, 2020 - 04:00
Add to Calendar

The Cloud Layers Unified By Binormals (CLUBB) package provides the turbulence and cloud macrophysics parameterizations in both the Community Earth System Model (CESM) and the Energy Exascale Earth System Model (E3SM), via the Community Atmosphere Model (CAM) and the E3SM Atmosphere Model (EAM), respectively. CLUBB treats each column of a discrete atmospheric grid separately, allowing for testing in a single-column configuration where the forcing terms normally provided by the dynamical core and other physics parameterizations are specified as cases. In the Rain in Cumulus over the Ocean (RICO) case, noisy artifacts were observed that persisted at higher temporal accuracy (smaller timestep size). Such artifacts must be diagnosed and addressed before the next generation of Earth system models (ESMs) can use higher resolution to produce more accurate solutions.

Diagnosing the noisy artifacts involved determining whether they originated from the highly nonlinear system of PDEs used in CLUBB (the continuum equations) or from the numerical discretizations used in CLUBB (the discrete equations). A series of self-convergence and timestep sensitivity tests determined the artifacts were indeed numerical. Replacing the responsible spatial discretizations with numerically consistent and stable schemes removed the noisy numerical artifacts for a range of resolutions. This work will support the next generation of ESMs by removing the corresponding numerical artifacts in CAM and EAM once the changes to the standalone CLUBB code are integrated, as well as providing a diagnostic approach that may be applicable to other physics parameterizations.

 

This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 and Pacific Northwest National Laboratory under Contract DE-AC06-76RLO 1830.

LLNL-ABS-812973

Link for More Information: