Characterization of turbulence at Oklahoma City inflow conditions during JU2003

Overview: Validation of computational models for the prediction of urban dispersion flows in the event of a chemical, biological, or radioactive (CBR) incident is in progress nationwide. The numerical synthesis of turbulence, required at inflow boundaries for urban dispersion modeling, is built upon an understanding of atmospheric flow conditions in the urban boundary layer. The frequency or wave number content of the turbulent flow drives dispersion rates, and also determines the numerical grid spacing required for computational modeling. To obtain site specific data for validation studies, researchers at NRL performed spectral analysis of atmospheric flow data taken from tower locations at Oklahoma City during the field studies commonly denoted JU2003. Sonic anemometer data taken at 10 Hz and provided by field researchers was analyzed to obtain Eulerian correlation curves, integral scales, and spectral or frequency content, thus characterizing wind conditions upstream of the urban center. Findings from the conditions studied indicate that 90% of the turbulent energy is resolved, while approximately 10% is subgrid in LES numerical simulations. Turbulence amplitudes in the flow and amplitude spectra will be presented, as will a plot of energy spectra and cumulative energy content vs. wave number.

Background: Numerical modeling and prediction of urban dispersion flows has been the subject of intense effort in the past several years due to concern over the effects of CBR weapons. Numerical simulation has been employed by several groups to study various scenarios at low cost, and provide critically important tools to emergency responders. Efforts towards validation of time accurate large eddy simulation (LES) models are currently underway at the Naval Research Laboratory. Important to modeling the inflow to the model, it is noted that both mean flow and turbulent values vary from city to city location, as a function of weather conditions, surrounding topography, and local suburban structure around the city core. This variation is not just with wind velocity, but because of heating effects, it may vary as a function of time during the day, even under similar wind conditions. Mean flow and turbulent boundary conditions have been tuned on a city to city basis to drive the numerical boundary conditions, and in this present effort, a further refinement of turbulence structure and intermittency using site specific turbulence wind flow data is used to enhance model accuracy. Accurate prediction of urban dispersion requires that conditions present at the site be set appropriately at city model boundary conditions, and this has been recognized as a pacing technical concern for several years.

Boundary Conditions Required for Urban Dispersion Modeling: Both fine-scale and large-scale eddy structures of importance exist in urban scale atmospheric flows. Large scale eddies will drive overall urban turbulent flow patterns. In this highly non-isotopic urban case, the buildings themselves are "chopping up" the turbulence, and vortices formed around structures by large eddies will cause both significant dispersion and control finer turbulence generation. There is direct coupling of large scale to fine scale turbulence forced by the urban structures, and a need to accurately specify all turbulence at the inflow across a wide range of scales.

Understanding the limit of the frequency range that must be generated synthetically for turbulence modeling is based upon a detail examination of the energy content with frequency in urban measurements. An estimate is that eddies in the range of 20 meters to 500 meters will have a significant impact on urban dispersion, and drive turbulence generation amongst building structures. Turbulent scales above this size can likely be modeled as mean flow changes, or better, coupled meso-scale modeling can provide input of these conditions. Turbulence below the 20 meter range can be modeled by urban scale LES. The accuracy of this estimate of required turbulence scales at the inflow boundary is a main point in the presentation.

The boundary surfaces for the urban dispersion model may be viewed as a thin or low rectangular box containing the urban city domain. Turbulence varies as a function of altitude, and it is clear from the work of other researchers that an assumption of the law of the wall from classic boundary layer theory would be totally erroneous. Even the mean boundary layer profile changes as a function of heating and time of day, and the presence of suburban structures greatly increase turbulence near the ground. Characterizing and specifying the inflow conditions are valuable to reduce code start up times and the number of upstream model grid points.

Selected References

Allwine, K.J., Leach, M.L. Stockham, L.W., Shinn, J.S., Hosker, R.P., Bowers, J.F., Pace, J.C., "Overview of Joint Urban 2003 – An atmospheric Dispersion Study in Oklahoma City", PNNL-SA-40036, Presented at the Symposium on Planning, Nowcasting and Forecasting in the Urban Zone, January 11-15, 2004, Seattle, Washington.

De Wekker, Stephan F. J., Berg, Larry K., Allwine, Jerry, Doran, J. Christopher, Shaw, Willam J. "Boundary Layer Structure Upwind and Downwind of Oklahoma City During the Joint Urban 2003 Field Study", American Meteorological Society, 5^th Conference on Urban Environment, 23-27 August 2004, Vancouver, B.C., Canada.