OZONE FORECASTING FROM METEOROLOGICAL VARIABLES PART II - DAILY MAXIMUM GROUND-LEVEL OZONE CONCENTRATION FROM LOCAL WEATHER FORECASTS

A ground-level ozone concentration prediction procedure is tested unde r real-world circumstances. Models are constructed for measured meteor ological variables to predict ozone concentrations. Subsequently, a we ather forecast is used as input to the prediction in order to obtain a n estimate of the next day's maximum ozone concentration. The work inv estigates whether the additional variance, introduced by the use of we ather forecasts, still allows accurate ozone concentration predictions . Two simulation studies were performed in order to check the predicti ve ability and the effect of the weather forecast accuracy. Good predi ction accuracy was obtained. Only 2.6% of the predictions lead to resi duals exceeding 40 mu g/m(3). In addition, the influence of the uncert ainty in the weather forecasts is generally low in these cases (3.39 t o 3.75 mu g/m(3)). Real predictions in the region of Grenland, Norway, for the summer of 1996 showed good agreement between measured and cal culated maximum ozone levels. Only 3.5% of the predictions yielded res iduals exceeding 40 mu g/m(3). (C) 1998 Elsevier Science B.V. All righ ts reserved.