Evaluation of CAM Single‐Column Models in Simulating Convectively Coupled Rossby Gravity Waves During TRMM‐KWAJEX

Abstract Leveraging observational data from the TRMM‐KWAJEX field campaign, which captures the evolution of mixed Rossby‐gravity (MRG) waves, this study evaluates how parameterization schemes influence convectively coupled MRG waves in the Community Atmospheric Model single‐column framework (SCAM). Four configurations are examined: the traditional approach that uses separate deep (ZM) and shallow (UW) convection schemes (CAM_ZMUW), unified shallow‐deep convection (CAM_UNICON), unified cloud‐turbulence (CAM_CLUBB), and super‐parameterization (CAM_SP). While precipitation is relatively insensitive to parameterization choice, diabatic heating fields differ markedly, which shape temperature, moisture and MRG wave characteristics. All simulations capture the general eastward tilt of MRG waves, but with varying fidelity. CAM_ZMUW and CAM_UNICON produce overly strong convection, while CAM_CLUBB underestimates low‐level early‐phase moistening. Only CAM_CLUBB and CAM_SP reproduce the successive vertical subgrid‐scale transport of moist static energy convergence, though the transport is weaker in CAM_CLUBB, whereas mass‐flux‐based schemes lack coherent vertical organization. Experiments combining Cloud Layers Unified by Binormals (CLUBB) with ZM deep convection behave similarly to CAM_ZMUW, indicating strong sensitivity to deep‐convection scheme. Applying stricter convective triggering in ZM convection alleviates the lower‐tropospheric dry bias but does not lead to an improved simulation of the convectively coupled MRG wave. The stochastic Subgrid Importance Latin Hypercube Sampler, a Monte Carlo technique designed to integrate over subgrid variability and improve estimates of its effects on microphysical process rates, has a relatively minor impact on MRG modeling in CLUBB. These results underscore the need to improve the representation of shallow‐to‐deep convective transitions in mass‐flux schemes and to enhance the vertical subgrid‐scale transport within the CLUBB framework.