Conclusion

Abstract

It is evident from this book that for planning and strategical purposes the air pollution modeller has a choice of a number of methods to predict long-term climatological averages of pollutant concentrations. We are not able to choose one method as the best because, although under a given set of conditions one method performs better than another, under another set of conditions it need not necessarily be the best method. This arises because of the inadequate observational data used for validation. It seems that we must be more careful in collecting observations and that we should broaden and deepen our data base; if this could be accomplished, then a decision as to the value of a model could be easily made. Unfortunately, we cannot achieve this, at least as far as yearly averages are concerned. The randomness of climatological means is also true for the mean concentrations. We are able to estimate the next year’s average pollutant concentrations as well as the yearly mean temperature or the amount of rainfall. We know a priori that it will (with a high probability) be within a given range, but that is all. There is always an element of chance that this year one model will give a better fit than another; however, the results obtained for the next year will probably favour another model. Only relatively long records can show a certain stability; only long time series will prove the actual usefulness of a given model.