In recent work, Tulich and Mapes (2007) studied multi-scale propagating waves using a two-dimensional cloud resolving model (CRM) which was forced by uniform cooling. Their model spontaneously generated horizontally-propagating wave packets with phase speeds in the range 16-18 m/s. The modeled domain was 8192 km in the horizontal, with horizontal grid spacing of 2km.
Other work by Bretherton et. al. (2005) used a three-dimensional CRM with a 576 km X 576 km horizontal domain and 3 km X 3km grid spacing. Rather than propagating modes, their model demonstrated convective self-aggregation.
Though many details differentiate these experiments, it is imperative to understand whether modeled results capture true physics or if they are manifestations of the model itself. To probe this question in the context of large-scale propagating waves, we perform a series of experiments using a CRM. Three of the experiments use a two-dimensional domain spanning 7000 km in the horizontal, and one uses a three-dimensional "bowling alley" domain spanning 7000 km X 20 km in the horizontal. The 2D runs had horizontal grid spacings of 1 km, 2 km, and 4 km; the 3D run used grid a 2 km X 2 km horizontal grid size. All simulations were run for 15 days. After 15 days, no obvious large-scale waves developed in the run with a 4 km horizontal grid size; All other runs spontaneously generated propagating modes with phase speeds in the range 12-15 m/s. In addition to these runs, a 2D simulation on a 2000 km horizontal domain with a 1 km horizontal grid spacing produced no large-scale waves. These results indicate that both domain and grid size are important for generating large-scale wave propagation within our CRM. The important question of whether these waves are purely manifestations of 2D vertical momentum transport cannot completely be addressed with a 3D bowling alley domain, though our results suggest that these modes can exist in 3D models.