Towards an Explanation for the Resolution Sensitivity of CAM
A set of idealized moist bubble experiments were developed to help understand the resolution sensitivity of CAM. Mimicking the behavior of more complex configurations, the horizontal scale of the perturbations was assumed proportional to the grid spacing, and resultant vertical velocities across hydrostatic resolutions agreed with that expected from a scale analysis of the (dry) anelastic equations. Through experimenting with different physics time-steps, it was found that the vertical motion in high-resolution simulations was severely damped in the experiments and in more complex, aqua-planet configurations, using more conventional (larger) physics time-steps.
Understanding how the resolved vertical velocity varies with resolution may help to inform model sensitivities. We have shown that using too large a physics time-step at higher-resolutions leads to time-truncation errors that should be considered when developing a high-resolution model.
The moist bubbles are developed in a non-rotating framework, in which the planets radius and horizontal scale of the perturbation, were reduced by the same factor, for a range of scale factors. The perturbations may be thought of as surrogate near-grid scale physics tendencies (think stratiform clouds), the horizontal scale of which are observed to decrease with resolution in more complex, aqua-planet configurations. The experiments were performed with moist physics routines of varying complexity, although convection schemes were not considered.