Generation of reliable validation data for microscale dispersion modeling in urban areas

Predicting local air pollution is one of the most challenging problems applied meteorological research is facing. Most of the air pollution sources are located in the urban roughness layer and turbulent transport phenomena are dominated by several complex factors. The quality of microscale modeling has a direct impact on the quality of life and human health. Furthermore, the results of microscale modeling serve as input for large scale models affecting the quality of dispersion modeling up to the global scale. Quality assurance and model validation play a crucial role in the development of reliable microscale dispersion models.

A problem of model validation is the lack of qualified reference data. Field data and the results of experiments in boundary layer wind tunnels can be used as source of validation data but both sources of data have specific limitations. Field data represent unique situations with a rather complex set of boundary conditions. A proper recording of all boundary conditions during field measurements is often impractical because of limitations in instrumentation. In a strict physical sense, it is impossible to define exactly what kind of dispersion situation has been captured in the field and further assumptions are required to integrate field data into model validation. Changing boundary conditions because of the diurnal cycle and constantly changing weather cause large scatter even in time-averaged field results. The limited representativeness of field data with respect to steady-state results of a microscale dispersion model restricts a direct comparison of numerical results with field data. Laboratory results are model results which already incorporate simplification and abstraction from the physical reality. Still, wind tunnel data have major advantages over field data regarding model validation. Extensive datasets with different levels of complexity can be generated for constant boundary conditions. It is possible to simulate steady-state dispersion the same way like it is modeled by most of the microscale dispersion models. In addition, the boundary conditions can be controlled and measured with high accuracy. Complete laboratory data sets can provide all information required to identify the physical state of an experiment and model tests can be set up without further assumptions. If a microscale dispersion model is simulating simplified laboratory situations properly it is likely that it will predict complex field situations properly as well.

To overcome main problems related to validation data, the CEDVAL project (Compilation of Experimental Data for VALidation purposes) has been carried out at the Meteorological Institute of Hamburg University. Sponsored by the German Federal Environmental Agency, the main goal of the project was to design, set up and maintain a WWW-database of wind tunnel data that meet all essential requirements on validation data with respect to completeness of documentation, documented quality and accuracy of reference data and easy access. The contribution will introduce the basic concept of the database and discuss the quality and reliability of the datasets in CEDVAL. An overview of the datasets available so far will be given and typical validation data sets with different complexity will be presented.