Damages and costs of air pollution: An analysis of uncertainties

This paper evaluates the uncertainties of an impact pathway analysis which traces the fate of each pollutant or other burden, from the source to the receptors, using dose-response functions to evaluate the damage. The expression for the total damage is shown to be largely multiplicative, even though it involves a sum over receptors at different sites. This follows from conservation of matter which implies that overprediction of the dispersion model at one site is compensated by underprediction at another; the net error of the total damage arises mostly from uncertainties in the rate at which the pollutant disappears from the environment. Since the central limit theorem implies that the error distribution for multiplicative processes is likely to be approximately lognormal, one may be able to bypass the need for a detailed and tedious Monte Carlo calculation. Typical error distributions are discussed for the factors in the expression for the total damage, in particular those of two key parameters: the deposition velocity of atmospheric dispersion models, and the value of statistical life; they are close to lognormal. A lognormal distribution for the total damage appears plausible whenever the dose-response function is positive everywhere. As an illustration, results for several types of air pollution damage are shown (health damage due to particles and carcinogens, damage to buildings due to SO2, and crop losses due to O-3): the geometric standard deviation is in the range of 3 to 5. To the extent that the distribution of the result is lognormal, the geometric mean equals the median and the geometric standard deviation has a simple interpretation in terms of multiplicative confidence intervals around the median. (C) 1999 Elsevier Science Ltd.