Monte carlo estimates of uncertainties in predictions by a photochemical grid model (UAM-IV) due to uncertainties in input variables

Abstract

Because photochemical grid models such as UAM-IV are being used to make policy decisions concerning emissions controls, it is important to know (1) the uncertainties in the model predictions due to the combined effects of uncertainties in the full set of input variables, and (2) the individual input parameters whose variations have the greatest effect on variations in model predictions. A preliminary Monte Carloun certainty analysis system has been developed and the methodology has been demonstrated using anapplication of the standard U.S. regulatory model, UAM-IV, to the 230 km by 290 km New York City domain for the 6–8 July 1988 ozone episode. As a first step, ten modeling experts were asked to estimate the typical uncertainties of 109 UAM-IV input parameters, including 23 variables related to emissions, boundary conditions, and meteorological conditions; and 86 variables related to chemical rate constants. For many of the model inputs, the assumed range of uncertainty was about plus or minus 30% of a normal mid-range value, and, in most cases, the distributions were assumed to have a log-normal shape. The regulatory agency’s “base run” application of UAM-IV to this ozone episode was used to define the mid-range or median values of all input parameters. 50 Monte Carlo UAM-IV runs were then carried out by simple random sampling of each of the 109 input parameters from the assumed distributions. The 50 predicted values of peak hourly averaged ozone concentrations anywhere on the geographic domain for the episode were found to follow a log-normal distribution and exhibit a variability from 176 to 331 ppb. The locations of the 50 predicted ozone peaks varied from 100 km upwind (southwest) of New York City to 150 km down wind (northeast) of the city. Variability in the input parameter known as the anthropogenic volatile organic compound (VOC) area source emissions had the most influence on the variations in the 50 predicted peak ozone concentrations.