A Stochastic Predictor of Air Pollution Based on Short-Term Meteorological Forecasts

Abstract

This paper illustrates a stochastic model of sulphur dioxide dispersion around a power plant. Precisely, the model describes the diurnal dynamics of a variable taken as representative of ground-level pollution [viz., the 2 h Dosage Area Product (DAP) in the sector of prevailing pollutant fallout]. Model exogenous inputs are power generated from the plant (regarded as an indirect measure of the emission), wind direction, and a properly defined atmospheric stability class, which depends on total radiation since sunrise, wind speed and Pasquill category at the end of the previous night. From the stochastic model, a real-time pollution predictor is derived, namely, a recursive relationship which, at the beginning of each time step, supplies forecasts of future DAP levels. The performance of such a predictor is tested on historical data and, in particular, the quality of the forecast in the presence of fumigation phenomena is pointed out. The basic difficulty in the actual implementation of the DAP predictor is due to the circumstance that it requires future radiation, wind direction and speed to be forecast separately at each time step. Such forecasts are supplied by three simple ?meteorological predictors? which are also illustrated in detail.