An evaluation concept for microscale numerical models of the obstacle layer

A contribution to subproject SATURN

H. Panskus, K. Heinke Schluenzen

Meteorologisches Institut, ZMK, Universität Hamburg, Bundesstr. 55, D-20146 Hamburg

1. Summary

An evaluation concept for micro-scale, obstacle resolving numerical transport and fluid models has been developed (Panskus, 1999). In a first step the concept has been specified for non-reactive pollutants. For this part details of the user oriented and scientific assessment as well as the validation test cases have been specified.

2. Aim of the research

This SATURN contribution is done in the frame of the development of the microscale model MITRAS. The MITRAS development is a cooperation of Alfred Wegener Institut fuer Polar und Meeresforschung, Bremerhaven (S. Lopez, C. Lüpkes) Fraunhofer Institut für Atmo-sphärische Umweltforschung, Garmisch-Partenkirchen (S. Trepte, T. Schöne-meyer), Institut für Tropospärenforschung, Leipzig (O. Knoth, M. Lambrecht, E. Renner) and Meterologi-sches Institut, Universität Hamburg (G. Bischof, B. Leitl, M. Schatzmann, +authors).

MITRAS will resolve obstacles with a horizontal resolution of some metres in an area of some hundreds to some thousands of metres. A forcing utility will be implemented in MITRAS so that it can be nested into models of the urban area scale which have a coarser horizontal resolution and don’t resolve the obstacles. MITRAS will include passive tracer transport as well as chemical reactions by directly solving the gas phase chemistry within the micro-scale model. The chemistry module considers all important chemical reactions close to traffic sources. Soot is transported as aerosol with its deposition dependent on size. Soot is treated as a sink for VOC compounds. For a realistic simulation of mixing effects a turbulence parameterization for obstacle resolving models is developed. A very important part of the model development is the evaluation of the model and its results. For this wind tunnel measurements will be used.

3. Activities during the year

Different turbulence schemes based on prognostic equations for turbulent kinetic energy and the energy dissipation rate have been tested by comparing model results with wind tunnel observations. A test version of MITRAS which includes a Prandtl-Kolmogorov turbulence closure (SATURN activity 19), a new numerical solver (SATURN activity 6) and on-line calculation of photolysis rates (SATURN activity 27) is available now and is tested by the model developer group. An evaluation concept applicable to urban transport and flow models has been developed (SATURN activity 29). The concept has been applied to the first version of MITRAS and detailed comparisons of model results with wind tunnel measurements were performed.

4. Principal results

Based on a validation concept for atmospheric mesoscale models (Schlünzen, 1996, 1997) and on the suggestions of the Model Evaluation Group (1994) an evaluation concept for obstacle resolving microscale numerical models has been developed (Panskus and Schlünzen, 1998; Panskus 1999). The objectives of the model evaluation are the checking of the applicability of microscale models for simulating flow fields within the obstacle layer. As a result of an evaluation the reliability of a model can be improved, model shortcomings and possible model improvements can be detected.


Figure 1: Outline of the evaluation strategy




The evaluation concept is outlined in Figure 1. The target parameter of the validation are concentrations of non-reactive pollutants within the obstacle layer. Concentration fields of passive tracers are strongly influenced by the flow fields. Therefore the model dynamics including stratification effects has to be validated. Chemical reactions are currently left out. For checking the passive tracer transport the user oriented and scientific assessment as well as the validation test cases have been specified.

For the validation a number of easy to run test cases are defined, which are specific for testing different model parts. The comparison data are wind tunnel measurements conducted by Schatzmann et al. at the University of Hamburg (SATURN activity 67). Evaluation measures are the classification of differences, correlation coefficient, skewness of the distributions, bias and differences of standard deviations. Ranges of validity have been defined ahead with defining the test cases.

The wind tunnel measurements were conducted for a steady state situation and thus they have to be compared to the steady state model solution. A criterion to define the steady state was derived from the scale analysis of the momentum equations and checked with the microscale, obstacle resolving model MITRAS (Panskus et al., 1997). For reducing the computational effort the minimum free flow velocity was determined and also checked with several MITRAS test runs. All comparisons of model results and wind tunnel data are conducted for non-dimensional values.

The evaluation concept has been applied to the second model update of MITRAS using the Prandtl-Kolmogorov turbulence closure. The comparison to wind tunnel measurements was done for a single quasi 2-dimensional building for two different boundary layer closures. One of the closures uses height above ground for specifying the maximum mixing length, whereas the other is using the minimum wall distance. The latter closure gives much better model results near the building as has been shown by comparison of model and wind tunnel data and by applying the evaluation measures.

5. Main conclusions

The evaluation concept developed is well applicable to obstacle resolving models. The number of test cases seems to be sufficient to detect major model shortcomings. The number of test cases is small enough to finish the tests within a reasonable amount of time (few months if all comparison data are well prepared). The evaluation measures are well chosen to evaluate the model. The evaluation of the current test version of MITRAS has shown that some improvements of the model are still necessary.

6. Aim for the coming year

In the coming year the current test version of MITRAS will be updated (turbulence closure, chemistry module, nesting procedure) and this version of MITRAS will be evaluated in total. Applications of MITRAS to the urban atmosphere are planned as well.

7. Acknowledgements

This work was funded by the German Ministry of Education, Science, Research and Technology within Tropospheric Research Program under grant number 07TFS10/LT1-B.1. The authors are responsible for the contents of this publication.

8. References

Model Evaluation Group; Model Evaluation Protocol. European Communities Directorate-General XII, Science Research and Development Version 5 (1994) pp 14.

Panskus, H.; Entwicklung einer Evaluationsvorschrift für mikroskalige Modelle mit Anwendung auf das Modell MITRAS. PhD thesis Fachbereich Geowissenschaften, Universität Hamburg, in preparation (1999).

Panskus, H., O. Knoth, M. Lambrecht, C. Lüpkes, M. Schatzmann, K.H. Schlünzen, T. Schönemeyer and S. Trepte; Konzeption eines mikroskaligen Modells in bodenfolgenden Koordinaten mit Chemie. Poster, Fachtagung METTOOLS III, Universität Freiburg (10.-12.3.97).

Panskus, H. and K.H. Schlünzen; Validierungskonzept für prognostische mikroskalige Modelle mit Anwendung in der Hindernisschicht. Annalen der Meteorologie, Deutscher Wetterdienst, 37 (1998) 175 –176.

Schlünzen, K.H.; Validierung hochauflösender Regionalmodelle. Berichte aus dem Zentrum für Meeres- und Klimaforschung, Universität Hamburg, Meteorologisches Institut, A23 .(1996): pp 184.

Schlünzen, K.H; On the validation of high-resolution atmospheric models. J. Wind Engineering and Industrial Aerodynamics 67&68 (1997) 479- 492.