Title: Development of short-term ozone forecasting for the smog early warning system. Client: UBA (German Federal Environment Agency) Dept. II, F&E-Vorhaben 104 028 17 carried out in co-operation with the Geosciences Dept., Institute of Meteorology, Freie Universität Berlin. Software: S+, Fortran Description: Ozone plays a substantial national and international role as a principal component of photochemical smog. Both the ozone itself, as well as its precursor substances, move beyond national borders without hindrance. This is why information on ozone pollution and its development is so important. The aim of the project described here "The statistical prediction of maximum daily ozone concentrations" was the development of a statistical prognosis system allowing the production of one- to three-day forecasts with details on the degree of forecast inaccuracy. This model development process resulted in a prognosis system consisting of a large number of variable, non-linear regression models. One individual model for each of the 325 ozone stations and optional for each of the 10 large-scale weather sites. The specification and estimation process runs fully automatically under the S+ program system by using stepwise regression. The resulting prognosis system is far superior to the statistical approach adopted by the Federal Environmental Agency up to now and is so constructed as to allow the system to automatically adapt to changes in data at any time. |