There are several approaches on parameterizing turbulent-scale motions in cloud-resolving models (CRMs). This type of models explicitly resolves cloud-scale dynamics but covers a large horizontal domain so that an entire cloud system can be simulated. The most popular approach is the one- and a half-order prognostic turbulent kinetic energy (TKE) closure (1.5-order closure, for short). Third-order turbulence closure is only used in the CRM originally developed at the University of California-Los Angeles (UCLA; Krueger 1988) and currently used in several institutions. It has been generally accepted that 3rd-order closure is superior to 1.5-order closure through one-dimensional (1-D) modeling of boundary-layer (BL) clouds. This has never be examined in 2-D or 3-D model framework, where cloud dynamics plays an important role, except in an intercomparison mode. Therefore, the present study is aimed at examining the role of cloud dynamics and its impact on the performance of different turbulence closures in simulating a range of BL cloud regimes using 2-D CRMs.

We use three 2-D CRMs for this study. Model A is the beta.5 version of the Advanced Regional Prediction System (ARPS) with 1.5-order closure developed at the University of Oklahoma (Xue et al. 2000). Model C is the UCLA CRM that was extensively used at the Colorado State University by the second author and is currently used at NASA Langley Research Center (LaRC). Model AC is the same as Model A except for being implemented with the 3rd-order closure from Model C, instead of the 1.5-order closure in Model A. We use a horizontal resolution of 1000 m and a vertical resolution 100 m, which are much coarser than those in large-eddy simulation (LES) models and those in CRMs participated in the GCSS (GEWEX Cloud System Study) Working Group (WG) 1 intercomparison studies. The same cases are simulated using Models A, C and AC and the exact model setting is followed. These BL cloud regimes includes ASTEX (Atlantic Stratocumulus Transition EXperiment), BOMEX (Barbados Oceanographic and Meteorological Experiment), ARM (Atmospheric Radiation Measurement) and FIRE (First ISCCP Regional Experiment).

Preliminary results from BOMEX simulations shows that Model C performs reasonably well, compared to LES model results, for example, the cloud location (around 1200 m) and the TKE budget. The buoyant production is, however, higher near the lower boundary (below 300 m) than LES results. Clouds are also produced from the simulations of Models A and AC but at lower locations. The TKE budgets from these two models are also less comparable to LES results. We are still examining the computational aspects of the model dynamics in Models A and AC in order to improve the simulations and compare the performance of 1.5-order against 3rd-order closures. We will simulate the remaining three regimes and compare the detailed results with LES results to explore the roles of cloud dynamics and its impact on the performance of different turbulence closures.