An Evaluation of the Large-Eddy Simulation Option of the Regional Atmospheric Modeling System in Simulating a Convective Boundary Layer: A FIFE Case Study

Abstract

A large-eddy simulation (LES) model was used to simulate the convective boundary layer (CBL) that developed on 1 July 1987, over the domain of the First International Satellite Land Surface Climatology Project Field Experiment (FIFE). Three simulations were produced using different boundary conditions at the ground surface, namely, (i) spatial distribution of topography and spatial distribution of surface heat fluxes; (ii) spatial distribution of topography but mean surface heat fluxes; and (iii) no topography and mean surface heat fluxes. The diurnal variation of mean surface fluxes and their spatial distribution were derived from the FIFE network of observations. In all cases, the model was initialized with the atmospheric sounding observed in this domain at 0700, and run until 1500 local time. The resulting mean profiles of temperature and specific humidity were compared to those observed with atmospheric soundings at 0900, 1030, and 1230 local time. The simulated structure of turbulence was qualitatively compared with that obtained from a volume-imaging lidar (VIL) scanning the CBL over the simulated domain during that day. Power spectra and autocorrelations of mixing ratio were calculated from the model outputs and were compared to those obtained from the VIL. Overall, the model performed quite well. Observed atmospheric soundings were within 1 K and 1 g kg?1 of the simulated mean profiles of temperature and specific humidity, respectively, and indicated that the model correctly predicts the CBL height. Similarities in the structure of the eddies obtained from the model and the VIL were clearly identified. Spectral analysis indicated that resolved eddies (i.e., eddies larger than 200 m) are relatively well simulated with the model, but that the energy cascade is not well represented by the Deardorff 1.5-order-of-closure subgrid-scale parameterization. Autocorrelation analysis indicated that the model correctly simulates the characteristic size of the eddies, but that their mean lifetime is longer than that observed with the VIL, indicating a too weak dissipation of the eddies by the subgrid-scale scheme. Thus, this study emphasized the need to develop better subgrid-scale parameterizations for LES models. The different simulations also indicated that topographical features of the order of 100 m and micro?-scale heterogeneity of surface heat fluxes had only a minor to modest impact on the CBL developing over a relatively humid surface.