Mixing height Determination for Dispersion Modelling

by

Petra Seibert, Frank Beyrich, Sven-Erik Gryning, Sylvain Joffre, Alix Rasmussen and P. Tercier

Report of Working group 2

of the COST 710 Action on

Pre-processing of Meteorological Data for Dispersion Modelling

Summary: Substances emitted into the atmospheric boundary layer (ABL,) are dispersed horizontally and vertically through the action of turbulence and eventually become well-mixed over this layer. Therefore, it has become customary to use the term "mixing layer" (ML). The ABL or ML height h is one of the fundamental parameters to characterise its structure and is required in dispersion models.

There are two basic possibilities for the practical determination of the h. It can be obtained from profile measurements, either in-situ (radiosonde, tethersonde, tower) or by remote sounding (sodar, clear-air radar, lidar). The other possibility is to use parameterizations or simple models with only a few measured parameters as input. Most of the relevant methods suggested in the literature are reviewed in this report. In any case, it is possible in principle, and also practised often, to substitute output from numerical weather prediction (or other, e.g. research) models for observed parameters.

Furthermore, the most important methods have been tested on data sets from two operational sites (Cabauw/NL, Payerne/CH) and a major field campaign (SADE, Germany). Parcel and Richardson number methods to analyse radiosoundings and mixing heights derived from sodar and wind profiler data have been investigated. Modules to determine h through parameterizations and models implemented in currently used meteorological pre-processors have been tested, too. These pre-processors were OML, HPDM, FMI, Servizi Territorio, and RODOS. For the stable and mechanically-dominated unstable ABL, they use similarity formulae based on the wind velocity, the Monin-Obukhov-length, and the Coriolis parameter while in the convective case simple slab models are integrated, based on an initial temperature profile and the surface heat flux.

A number of recommendations for operational mixing height determination have been formulated, including suggestions for the pre-processor development and for future research. The most important points are:

If suitable measured data are available, determination of h should be based on these data. In convective situations, the most reliable method is the parcel method applied to temperature profiles; bulk Richardson number methods can be used, too. MH determination in situations dominated by mechanical turbulence is much more difficult. If temperature and wind profiles are available, methods based on bulk Richardson numbers are considered to be the most appropriate. In many situations, data from remote sounding systems can give good results, but the available algorithms for their evaluation are not yet reliable enough for operational purposes.

All the pre-processors have problems in specific situations. In the stable and neutral ABL, they rely on similarity formulae involving surface layer parameters and the Coriolis parameter, which is not satisfactory from a physical point of view. Richardson methods appear to be better in this respect. However, the necessary input for these methods is often not available. Using one-dimensional numerical models with higher-order turbulence closure may become a solution in the future. If surface similarity methods are used, Nieuwstadts method appears to be superior to the u*/f approach for stable conditions. Further recommendations concern the initial profile in slab models, the ability of pre-processors to accept all sorts of measured data, and region-specific constants.

This report has been published as a part of the overall COST-710 Final Report.

Back to Homepage of  The COST 710 Action

This page is maintained by Helge Rørdam Olesen
It was last modified on April 9, 1999