Abstract

The air quality model ADMS-Urban is able to represent the fine scale variations in pollutant concentrations in an urban area by explicitly modelling urban sources including roads.  Causes of model errors arise principally from uncertainty in the emissions, in the underlying flow and turbulence fields and in the dispersion algorithms themselves. In this presentation the focus is on emission uncertainty: a Bayesian based inversion methodology, taking account of uncertainty in measured concentrations, is applied to model to optimise estimates of road traffic emissions using a network of sensors in Cambridge. This approach is compared with adjustments of standard emission factors using remote sensing, a methodology which has also been tested in application of ADMS-Urban to London.