Hindcast Simulations of the Madden–Julian Oscillation (MJO) in CESM: Assessing the Role of Regional Resolution Variations and Parameterized Physics

Various studies have linked the accurate simulation of the Madden Julian Oscillation (MJO) in climate models, to having either an adequate horizontal resolution to resolve large-scale/convection interactions, or the correct physical process parameterizations to enable the correct dynamics-physics coupling, sufficient to grow and propagate MJO-like disturbances.

The extent to which resolution and physical parameterizations matter in climate models will be presented in this talk. We use the NCAR Community Earth System Model (CESM) in HIndcast/CAPT-type configurations covering the 2009/10 period of the YOTC campaign. The period covers three active MJO events. Each simulation set constitutes a series of 20-day hindcasts, performed, in the control case, with a global, 1 degree resolution. Our default configuration uses version 5 of the Community Atmosphere Model (CAM5). This version has demonstrated significant skill in hindcasts out to 20-days for certain measures of the MJO (Klingaman et al., 2015). Paradoxically, the MJO performance of CAM5 AMIP-type and fully coupled simulations is poor in comparison.

With this model version we assess the role of resolution both globally and regionally. A global high-resolution (0.25 deg) set of simulations is compared to the 1 degree set for local MJO skill and more remote extra-tropical impacts in the Northern hemisphere. Additionally, we assess the potential role of resolution by isolating its importance in different regions. Using a regionally refined version of CESM we provide higher resolution (0.25 deg) within the 1 degree global simulations over a number of key tropical regions. These include over the Indian Ocean, where the MJO initiates; over the Maritime Continent, where barrier effects can damp the MJO; and over the West Pacific, where MJOs grow to maturity.

Finally, we investigate the role of improved parameterized physics in the hindcast of MJO activity. A development version of the CESM atmosphere model (CAM6), uses a radically different set of parameterized physics, including a high-order turbulence scheme (CLUBB), and is used to generate an additional set of simulations for the YOTC period. With this full experiment set we demonstrate the relative merits of resolution and parameterized physics on the simulation of the MJO in CESM.