Intraseasonal Variability in Coupled GCMs: The roles of ocean feedbacks, high-frequency SST variability, and cumulus parameterization

It has long been recognized that tropical intraseasonal variability (ISV) in atmospheric general circulation models (AGCMs) is improved when those models are coupled to a dynamic ocean model, yielding a coupled GCM (CGCM). Several modeling studies have addressed the role of ocean coupling, and the resulting high-frequency SST variability, on ISV by running an AGCM driven with CGCM-generated SSTs. In general, ISV and the associated eastward and/or northward propagation (for the Asian summer monsoon) improves, sometimes dramatically, over that which is present in the AGCM simulations driven with monthly mean SSTs. Consequently, most studies conclude that coupling and/or high-frequency SST variability is critical for or beneficial to ISV simulation. Absent from the above studies is discussion of the role of cumulus parameterization in ISV response to SST variability. For example, a CAPE-based cumulus parameterization might respond differently to SST variability than a parameterization that employs a moisture convergence closure. Similarly, AGCMs with inherently realistic ISV (as would be evident in that model's AMIP-style simulation) may not exhibit much improvement when forced with high-frequency SSTs. We explore the effects of cumulus parameterization on the AGCM response to high-frequency SST variability through a suite of AGCM experiments driven with SSTs derived from the Super-parameterized Community Climate System Model (SP-CCSM). SP-CCSM produces realistic ISV in both boreal summer and winter seasons. We derive two SST datasets from SP-CCSM: 1) daily SST, with a 5-day running mean applied, and 2) monthly mean SSTs. Each of these SST time series is applied to three versions of the Community Atmospheric Model (CAM, the atmospheric component of CCSM): CAM3.0, CAM4.0, and SP-CAM. These three model configurations represent cumulus parameterizations that are progressively less sensitive to surface temperature, and have progressively better ISV, yet will all have a mean state very similar to that of SP-CCSM. Through this comparison, we hope to discern the role of ocean feedbacks, high-frequency SST variability, and cumulus parameterization on ISV simulation.