Session 1M.6 Statistical convergence in simulated moist absolutely unstable layers

Monday, 24 October 2005: 11:45 AM
Alvarado GH (Hotel Albuquerque at Old Town)
George H. Bryan, NCAR, Boulder, CO; and R. Rotunno

Presentation PDF (565.7 kB)

Previous work has shown that simulated deep moist convection becomes turbulent once the grid spacing (dx) is of O(100 m). However, our simulations of mesoscale convective systems are usually not statistically converged, even with dx = 125 m, despite having a grid spacing well into the inertial subrange. To investigate the issue, we simulate a simple case of a moist unstable layer in a 2-km cube. The layer is perturbed by 3d, small amplitude, white noise perturbations, and the turbulent transition is studied using grid spacing as small as 3.9 m. The results confirm a theoretical result that grid spacing should be approximately two orders of magnitude smaller than the largest turbulent eddies. In this case, this means that ~20 m grid spacing is required to achieve statistical convergence for the simulation of 2 km large eddies (analgous to cumulonimbus clouds). More relevant to the mesoscale processes community, we find that convection simulated with inadequate resolution tends to be too intense, leading to cloud tops that are too high and vertical mass transport that is overestimated. A short (3 h) simulation of a squall line with 62.5 m grid spacing supports these conclusions.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner