Jørgen Brandt and Jesper H. Christensen

National Environmental Research Institute

Frederiksborgvej 399, P.O.Box 358

DK-4000, Roskilde, Denmark.

An integrated weather and air pollution forecast and scenario model system, THOR, for air pollution applications has been developed. The model system includes several meteorological and air pollution models capable of operating for different applications – ranging from hemispheric scale over European and urban background scales to urban street scale. Coupling models over different scales, makes it possible to account for contributions from local, near-local as well as remote emission sources in order to describe the air quality at a specific location - e.g. in a street canyon. The main purposes of the system are forecasting, nowcasting, emission reduction scenarios, retrospective analysis and air pollution assessments. Furthermore, the system is used in connection with the urban and background monitoring programs in Denmark.

This system of models is used to produce 3 days (72 hours) weather and air pollution forecasts, 4 times every day (starting at 00 UTC, 06 UTC, 12 UTC and 18 UTC). Visualizations and animations of these forecasts can be chosen for every 6 hours; from 00 to 72 hours after the start. The forecasts are normally finished at around 8 am, 2 pm, 8 pm and 2 am (CET).

The air pollution models described in the following, have been developed at DMU-ATMI:

• For the regional daily air pollution forecasts, the Danish Eulerian Operational Model (DEOM) is used. This model covers the whole of Europe with a grid resolution of 50 km x 50 km, and the model has three vertical layers. DEOM is based on the DEM model, which has been validated over a period of 9 years, since 1989.
• A new comprehensive 3-D air pollution model DEHM / REGINA (a nested version of DEHM) is running in test mode. This model will replace the DEOM model during the autumn 2002.
• For the urban background and urban street level air pollution forecasts, BUM (Background Urban Model) and OSPM (Operational Street Pollution Model) is used.
• The system is continuously validated against measurements from the rural and urban air quality monitoring networks including real time air quality measurements carried out at DMU-ATMI. Ozone measurements from seven Danish sites are used for real time validation of the system.
• In case of accidental releases, the Danish Rimpuff and Eulerian Accidental release Model (DREAM) is used. DREAM has a grid resolution of 25 km x 25 km and with a finer resolution of 5 km x 5 km near the source.
• The OML-multi model is a modern Gaussian plume model intended to be used for distances up to about 20 km from the multiple sources. The source is typically one or more stacks, and possibly also area sources. Typically, the OML model is applied for regulatory purposes. In particular, it is the recommended model to be used for environmental impact assessments when new industrial sources are planned in Denmark. The model can be used for both high and low sources.
• The Atmospheric Chemistry and DEPosition model, ACDEP is a semi-operational Lagrangian long-range transport model, using four days back trajectories. Within the Danish national background monitoring program, the model is run at the beginning of a year for the previous year and used e.g. to estimate the total nitrogen deposition to the inner Danish waters. ACDEP receives initial concentrations from the DEOM model and meteorological data from the Eta model. The ACDEP model is furthermore used operationally to calculate forecasts of the nitrogen depositions to the inner Danish water.

The used weather forecast model at the THOR-DMU system:

• The weather forecast model, which is used is the Eta model also applied in the SKIRON Project in Greece and in several institutes in USA as the National Weather Service at NCEP. Meteorological analyzed input data is obtained from the National Centers for Environmental Prediction (NCEP), USA which is the starting point for nearly all weather forecasts in the United States.
• The weather forecast model MM5 is also used in the THOR system. Currently it is not running in forecast model, but are using archived meteorological analyzed input from European Centre for Medium-range Weather Forecasts (ECMWF) in a reanalyze mode. The MM5 model has been run for a 12 year period (1990-2001).

Global meteorological data is obtained from NCEP. These data are used as input to the Eta model, which calculates 3 days high-resolution regional weather forecasts. The weather forecast are then used in a series of regional and hemispheric air pollution models (DEOM, DREAM and DEHM). It will be possible in the future to run both the Eta (or MM5) and DEHM/REGINA as nested models with a finer resolution over a limited area as e.g. Denmark. The European air pollution background values calculated by DEOM are used as input to the urban background model (BUM) (currently for the cities of Copenhagen, Aalborg and Berlin), and the Operational Street Pollution Model (OSPM) for calculation of air pollution forecasts at street level in major cities (currently 11 streets in Copenhagen and 10 streets in Aalborg). Different visualization and animation tools in 1, 2 and 3 dimensions are applied for presentation of the output from all the models.

The whole system has been optimized to run operational and fully automized on a powerfull local compaq alpha server ES45 with two 1Ghz processors with Tru64 UNIX. The models are parallelized by using OPENMP statements. The computing time for a three days forecast with the different models are: 45 minutes for the weather forecast, total 15 minutes for DEOM, BUM and OSPM, while the computing time for the 3-D air pollution model DEHM is 45 minutes and 15 minutes for the accidental release model DREAM.