Abstract
Two alternative parameterizations for nonlocal turbulence mixing are tested in a 1D boundary-layer model against a dataset from the 1983 Boundary-Layer Experiment (BLX83) in Oklahoma. One method, proposed previously by Stull and Driedonks, is based on a nonlocal approximation to the turbulence kinetic energy (TKE) equation. An alternate method, based on a nonlocal approximation to the Richardson number, is simplified here from earlier parameterizations for transilient turbulence theory. Convective mixed-layer simulations of the vertical profiles of mean variables and fluxes using both methods are compared to the BLX83 observations and to simulations using a traditional slab model.
The TKE method develops a surface layer that is too thick compared to BLX83 data, particularly in the early morning. It also lacks the subadiabatic lapse rate that is observed in the top of the mixed layer. The Richardson number approach produces more accurate mixed-layer profiles, but lacks the general physical interpretation of the TKE method. Nonlocal spectral decompositions of the flux and intensity of mixing confirm that large-size eddies dominate within the middle of the mixed layer. Based on this limited validation, the Richardson number method is recommended for convective boundary layers, but the TKE approach should be used for modeling more general boundary layers that can include clouds and stable and/or windy conditions.
