A Method for Interpreting the Role of Parameterized Turbulence on Global Metrics in the Community Earth System Model

The parameterization of subgrid-scale processes such as boundary layer (PBL) turbulence introduces uncertainty in Earth System Model (ESM) results. This uncertainty can contribute to or exacerbate existing biases in representing key physical processes. This study analyzes the influence of tunable parameters in an experimental version of the Cloud Layers Unified by Binormals (CLUBBX) scheme. CLUBB is the operational PBL parameterization in the Community Atmosphere Model version 6 (CAM6), the atmospheric component of the Community ESM version 2 (CESM2). We perform the Morris one-at-a-time (MOAT) parameter sensitivity analysis using short-term (3-day), initialized hindcasts of CAM6-CLUBBX with 24 unique initial conditions. Several input parameters modulating vertical momentum flux appear most influential for various regionally-averaged quantities, namely surface stress and shortwave cloud forcing (SWCF). These parameter sensitivities have a spatial dependence, with parameters governing momentum flux most influential in regions of high vertical wind shear (e.g., the mid-latitude storm tracks). We next evaluate several experimental 20-year simulations of CAM6-CLUBBX with targeted parameter perturbations. We find that parameter perturbations produce similar physical mechanisms in both short-term and long-term simulations, but these physical responses can be muted due to nonlinear feedbacks manifesting over time scales longer than 3 days, thus causing differences in how output metrics respond in the long-term simulations. Analysis of turbulent fluxes in CLUBBX indicates that the influential parameters affect vertical fluxes of heat, moisture, and momentum, providing physical pathways for the sensitivities identified in this study.