Evaluation of diagnostic and prognostic flow fields over prealpine complex terrain by comparison of the Lagrangian prediction of concentrations with tracer measurements

Abstract

A short range tracer experiment was carried out under fair-weather conditions over prealpine complex terrain in Switzerland. The prevailing wind direction in the troposphere over the Swiss Plateau was characterized by a northeasterly flow condition that is called bise in Switzerland. This wind system is oriented along the Swiss Plateau due to channelling effects between the Alps in the south and the Jura mountains in the north. Over the prealpine experiment area, however, the SODAR measurements gave evidence of the presence of a thermally driven wind system with a diurnal cycle. The air pollution dispersion scenario of the tracer experiment was simulated by a Lagrangian air quality model for passive air constituents called PARTRAC. The calculated data were compared to the measured tracer concentrations. PARTRAC requires 3D-flow fields which were either generated diagnostically by the mass consistent wind model CONDOR, or prognostically by the non-hydrostatic mesoscale meteorological model MEMO. The sensitivity of the CONDOR wind fields to varying input of measured wind data was investigated by comparing the calculated concentration fields to observational data. The reliability of the concentration predictions was assessed on the basis of statistical performance measures by comparing the predicted half-hourly concentrations to the corresponding measured data paired at equal locations and times. This sensitivity study was aimed at providing PARTRAC with the most representative diagnostic flow fields in order to numerically reproduce the observations as reliably as possible and to establish a benchmark to which the prognostic modelling results were compared in a second step. The prognostic flow fields were generated by a passive model nesting technique in order to reproduce the thermally driven diurnal flow circulation between the prealpine part of the Swiss Plateau and the Alps. The occurrence of this wind regime was suggested by the SODAR measurements and by the dispersion model results relying on the diagnostic wind fields. The initial conditions for the prognostic model MEMO were delivered by a regional weather forecast model that is in operation both at the weather services in Germany (DWD) and Switzerland (SMA). As will be demonstrated in the first part of this paper, the diagnostic transport and dispersion modelling does not necessarily produce a reliable description of an air pollution episode over complex terrain, such as the Swiss Plateau. Therefore, the density of the permanently operating network of weather stations in Switzerland—ANETZ and ENET—definitely appears to be too small to be considered as a sufficient data base for reliable diagnostic concentration predictions in the case of an emergency situation. On the other hand, the prognostic mesoscale modelling of flow fields in complex terrain suffers from the same deficiency as the diagnostic approach inasmuch as reliable results often require a sensitivity study with the nesting of flow fields originating from different meteorological processes at different scales. Therefore, the data selection for such a grid nesting procedure requires a phenomenological knowledge of the different meteorological processes involved. In the case of MEMO at the time of this study, this data selection could be accomplished in different ways, which did not necessarily produce always identical results.