Mixture of linear regression models for short term PM10 forecasting in Haute Normandie (France)

Abstract

Forecasting PM10 concentrations accurately will all for improved early warning procedures,useful for safety reasons and opens for example the possibility to restrict circulation or to decidefree public transportation. So the need of a statistical pollution forecasting tool from particulatematter is an important issue for the public authorities.Hourly concentrations of PM10 have been measured in three cities of Haute-Normandie(France): Rouen, Le Havre and Dieppe. The Haute-Normandie region is located at northwestof Paris, near the south side of Manche sea and is heavily industrialized. We consider sixmonitoring stations reflecting the diversity of situations. We have focused our attention onrecent data from 2007 to 2011.We forecast the daily mean PM10 concentration by modeling it as a mixture of linearregression models involving meteorological predictors and the average concentration measuredon the previous day. The values of observed meteorological variables are used for fitting themodels while the corresponding predictions are considered for the test data, leading to realisticevaluations of forecasting performances, which are calculated through a leave-one-out schemeon the four years.We discuss in this paper several methodological issues including estimation schemes,introduction of the deterministic predictions of meteorological models and how to handle theforecasting at various horizons from some hours to one day ahead.