Ground Level Ozone Peak Forecast using Neural Networks and Kalman Filter

0404081 - UIVT-O 20000202 RIV SK eng J - Journal Article
Pelikán, Emil - Eben, Kryštof - Vondráček, Jiří - Krejčíř, Pavel - Keder, J.
Ground Level Ozone Peak Forecast using Neural Networks and Kalman Filter.
Meteorologický časopis. Roč. 3, č. 2 (2000), s. 3-8. ISSN 1335-339X
Grant - others:APPETISE(XE) IST-99-11764; MŽP ČR(CZ) ZZ520/2/97; MŠMT ČR(CZ) VS96008
Institutional research plan: AV0Z1030915
Keywords : ozone forecast * neural classifications * Kalman filter * genetic algorithms * Kohonen maps * Czech Republic
Subject RIV: BB - Applied Statistics, Operational Research

A forecasting system for the maximum ozone concent. for the following day based on statistical methods has been developed and tested within the framework of cooperation between the Czech Hydromet. Inst. and Inst. of Computer Science. Predictors based on neural networks and Kalman filter have been developed for urban, rural and mountain types of measurement stations.The stations were clustered with the help of Kohonen maps. Input variables were selected by statistical tests and a genetic algorithm.
Permanent Link: http://hdl.handle.net/11104/0124352