Correlations between air pollution and meteorological conditions

Abstract

A variety of statistical methods are used with observational data in order to forecast total or average air pollution concentrations from multiple sources. Pollution emission rates and position of pollutant sources are assumed, by the majority of forecasting methods, to remain virtually unchanged during the observation and prediction periods. Hence, there will naturally be certain errors and/or restrictions on these forecasts, which are not present in those made by the numerical methods (Chapter 4) that allow for variations of emissions with time. The constant emission assumption is applicable to a degree for comparatively short-term forecasts i.e. periods from a few hours to a few days. In addition, for areas of multiple pollution sources with varying individual emission rates one can assume that an increase of emission from some of them may be compensated by lower emissions from others. Thus a higher average or total air pollution concentration is related, presumably, to changes in meteorological conditions or synoptic situation. The development of a prediction method starts by identifying periods of severe atmospheric pollution, which are then correlated with meteorological elements or combinations of weather conditions observed during those periods. These elements are regarded as predictors. Different forecasting rules are developed similarly. Use is also made of methods of statistical extrapolation in time of air pollution variations, as well as procedures for identifying autocorrelation effects and persistance factors.