Photochemical Models: Assessing Uncertainty and Sensitivity to Model Input (Invited Presentation)

Observations and photochemical models are often compared to test the understanding of atmospheric photochemistry. However, until the model and observation uncertainties are well understood, it is not possible to know if differences between observed and modeled reactive chemical species are insignificant or if they come from missing or incorrect model photochemistry. In addition, the possible causes of meaningful differences can be found by examining the sensitivity of the model-calculated reactive species to model inputs. One approach to achieving these goals uses global sensitivity analysis. For much of our work we have used the Random Sampling – High-Dimensional Model Representation (HD-HDMR) method. In this presentation, we will show how this method provides insight into differences between measured and modeled reactive chemical species in field and laboratory studies. We will also discuss how uncertainty analysis can be used to assess the potential improvement that new measurement strategies can have on air quality prediction.