A meteorological preprocessor takes observational meteorological data from a given time period and a given spatial domain and computes out of this sparse data set all the meteorological information necessary to calculate pollutant concentrations in the air later on.
At the present moment, the preprocessor in use is a modified version of the MESOPAC II meteorological preprocessor [33, 39], developed for the Environmental Protection Agency of the U.S.A. This preprocessor interpolates given meteorological data in time and space and produces a gridded field of data in two vertical layers of the atmosphere. The grid dimensions of the rectangular domain of interest have to be specified by the user of the preprocessor by way of a control file. This grid is the computational grid, whose grid points represent all the receptor locations of the pollutants. Each of these receptor locations contains an ecosystem and some individuals whose response on the pollution is modeled in other modules later on.
The following data is necessary for a prespecified number of meteorological surface stations within a prespecified time interval for which the preprocessor has to compute data: hourly data with respect to cloud ceiling height, amount and type of precipitation in CD-144 coding, wind direction (in degree), wind speed, surface pressure, dry bulb temperature, relative humidity, and opaque cloud cover (in %). Of course, the latitude and longitude of the station providing the data has to be specified, too. Moreover, additional data from upper air stations providing rawinsonde data can be added to the data set: twice daily data with respect to soundings at different levels. At each level, data with respect to height, temperature, wind direction (in degree), and wind speed has to be given. Additionally, hourly precipitation data from surface stations can be added to the input. In any case, so called "land use data" has to be provided to allow for the computation of the friction velocity and the wind speed at the top of the surface layer. This land use data takes the form of a categorization of each surface grid point of the computational field specified according to twelve different categories.
The output of the MESOPAC II preprocessor consists of a binary file with the following data: at each grid point of the computational field, hourly data with respect to the lower-level wind field, the upper level wind field, mixing height, friction velocity, convective velocity scale, Monin-Obukhov length, PGT stability class, precipitation rate, surface air density, air temperature, solar radiation, relative humidity, and type of precipitation.
More details with respect to the preprocessor can be found in the user's manual [39].
Since the meteorological data set provided by the preprocessor is independent of the parameters of the pollutant and its source, the preprocessor needs to be called only once during an optimization run. Afterwards, the necessary data will be available in a file. This strategy has been followed for the design of the code. Subsequent computational tests showed significant savings in computation time.
Note that the preprocessor in use is diagnostic instead of prognostic, i. e. the input data consists of measurements, which are then interpolated according to a technique based on Voronoi-diagrams. In contrast to this, prognostic models are used, e. g., in forecasting and simulate the variables under consideration according to the underlying physical laws. It has to be expected that prognostic preprocessors are able to deliver more accurate data. On the other hand, diagnostic codes seem to be more efficient with respect to computation time. In any case, the diagnostic model used here can be easily replaced by any other kind of code, as long as the output data format is the same. The choice of the specific preprocessor used here was made due to the simplicity and the robustness of the approach.