16th International Conference on Harmonisation within Atmospheric Dispersion Modelling for Regulatory Purposes 8-11 September 2014, Varna, Bulgaria ______________________________________________________________________ “ŠOŠTANJ” DATA SET FOR VALIDATION OF MODELS OVER VERY COMPLEX TERRAIN Primož Mlakar 1 , Marija Zlata Božnar 1 , Boštjan Grašič 1 , Giuseppe Brusasca 2 , Gianni Tinarelli 2 , Maria Grazia Morselli 2 and Sandro Finardi 2 1 MEIS d.o.o., Mali Vrh pri Šmarju 78, SI-1293 Šmarje-Sap (Slovenia) 2 ARIANET Srl, Via Gilino n. 9, 20128 Milano (Italy) Abstract: In 1991 three institutions from Slovenia and Italy performed a measuring campaign to collect data for validation of models. The target was a local area around the Šoštanj thermal power plant (ŠTPP) in north-east of Slovenia. The measurements data set was intended for testing of the models capable of reproducing air pollution from point sources in complex meteorological and terrain situation that characterised Šoštanj area. At that time the database was elaborated and a written report was produced. The diskettes with data and the printed report have been delivered to several scientific institutions and libraries, but then after mostly forgotten. As one of the co-authors of the work we decided to renew the database and to produce nowadays needed information that will allow other researchers to use this experimental database for modern models testing. The measurements data set describes in details the following situation: Šoštanj thermal power plant at that time did not have desulphurisation pl ants therefore SO2 continuously emitted by its blocks was acting like a tracer in the environment (other sources of SO2 were negligible in comparison to ŠTPP). On-line automated measurements of SO2 were available at all stacks of the ŠTPP and at 6 ambient stations in the basin and on the surrounding hills that were measuring also ground level meteorology. The area is characterised by a very complex meteorological situations (characterized by low winds and thermal inversions). The air pollution patterns are even nowadays not so easy to reconstruct. Out of the three weeks of data there are two days very interesting for detail models performance testing (one in a neutral case and one in a very complex convective case). In the paper we will present a detailed description of the experiment (area description, site, terrain, measurements stations location, database and two most interesting cases suitable for a model validation (emission, ambient concentrations, meteorology). The available description and data offered will be sufficient for other modellers to use the validation set. In addition we will present examples of our validation of the AriaIndustry models Swift and Spray. Key words: air pollution dispersion modelling, database, validation, complex terrain, field dataset. INTRODUCTION It is essential to realise with regard to the use of dispersion models for regulatory purposes how well models can illustrate what is actually happening in nature (Božnar, M. Z. et al., 2012). This was already a question two decades ago, when an initiative was launched in Europe to unify these models. A group of researchers who worked at the Jožef Stefan Institute (IJS) (from Ljubljana) , CISE and ENEL (from Milan) institutes at that time realised that the greatest problems with modelling arise when we wish to simulate processes taking place over highly complex terrain. In the spring of 1991, the three institutes thus conducted a measurement campaign in the surroundi ngs of the Šoštanj Thermal Power Plant (ŠTPP) in north-eastern Slovenia (Elisei, G. et al., 1992, Božnar, M. Z. et al., 1994a, Božnar, M. Z. et al., 1994b). The goal of the measurement campaign was to obtain air pollution field dataset over complex terrain for the purposes of testing and validating dispersion models. A modern automated measuring system for the measurement of meteorological parameters and SO2 concentrations was already operating at the time in the direct vicinity of the ŠTPP and emission measurements were automatically carried out in the stacks of the ŠTPP. The key to the planning of this experiment was the fact that the ŠTPP was not yet equipped with a desulphurisation unit at the time and thus emitted very large quantities of SO2 into the atmosphere. The