Interaction of nocturnal low-level jets with urban geometries as seen in joint urban 2003 data

Because accurate modeling of atmospheric flows in urban environments requires sophisticated representation of complex urban geometries, much work has been devoted to treatment of the urban surface. However, the importance of the larger-scale flow impinging upon the urban complex to the flow, transport, and dispersion within it and downwind has received less attention. Building-resolving computational fluid dynamics (CFD) models are commonly employed to investigate interactions between the flow and three-dimensional structures that make up the urban environment; however, such models are typically forced with simplified boundary conditions that fail to include important regional-scale phenomena that can strongly influence the flow within the urban complex and downwind. This paper investigates the interaction of an important and frequently occurring regional-scale phenomenon, the nocturnal low-level jet (LLJ), with urban-scale turbulence and dispersion in Oklahoma City, Oklahoma, using data from the Joint Urban 2003 (JU2003) field experiment. Two simulations of nocturnal tracer release experiments from JU2003 using Lawrence Livermore National Laboratory's Finite-Element Model in 3 Dimensions and Massively Paral-lelized (FEM3MP) CFD model yield differing levels of agreement with the observations in wind speed, turbulence kinetic energy (TKE), and concentration profiles in the urban wake, approximately 750 m downwind of the central business district. Profiles of several observed turbulence parameters at this location indicate characteristics of both bottom-up and top-down boundary layers during each of the experiments. These data are consistent with turbulence production due to at least two sources, the complex flow structures of the urban area and the region of strong vertical wind shear occurring beneath the LLJs present each night. Strong LLJs occurred each night, but their structures varied considerably, resulting in significant differences in the magnitudes of the turbulence parameters observed during the two experiments. Because FEM3MP was forced only with an upwind velocity profile that did not adequately represent the LLJ, the downward propagation of TKE observed during the experiments was absent from the simulations. As such, the differing levels of agreement between the simulations and observations during the two experiments can, in part, be explained by their exclusion of this important larger-scale influence. The ability of the Weather Research and Forecast Model (WRF) to simulate accurate velocity fields during each night was demonstrated, and the use of regional-scale simulation data was identified as a promising approach for representing the effects of important regional-scale phenomena such as the LLJ on urban-scale simulations.