Modelling Spatial Uncertainties associated with Emission Disaggregation in an integrated Energy Air Quality Assessment Model

This paper presents a methodology to model spatial uncertainties associated to the disaggregation of sectoral emission allocation in Luxembourg. We apply it to the integrated Luxembourg Energy Air Quality assessment model for the urban and regional scale. National aggregated sectoral emissions of primary air pollutants are computed by an energy model which minimises the cost of the reference energy system. The model describes five sectors, i.e. agriculture, transport, industry, residential and commercial at five-year periods. The sectoral emissions are spatially allocated to obtain hourly to daily emission maps. The air quality model simulates the dispersion of the emitted pollutants and their chemical reactions to produce ozone for typical episodes for each five-year period. Both models are coupled by an Oracle Based Optimisation Engine (OBOE) to find the optimal energy system with a constraint on ozone concentrations. When disaggregating emissions from the national to the urban scale, small scale variation and its associated uncertainty within and across land-use boundaries have to be modelled to obtain realistic emission maps. We propose to decompose the emission value for each sector into its mean, standard deviation and spatially correlated Gaussian distributed error with zero mean and unit variance. The spatial error is modelled by spatial stochastic simulation assuming a semivariogram with local and regional scale variability to account for within-boundary and across-boundary variation. A Monte Carlo approach is used to simulate the spatially correlated error and its distribution at each grid cell of the emission map. Finally, the average disaggregated emission value and its corresponding local uncertainty can be computed for each grid cell accounting for spatial correlation at different scales and across land-use boundaries. Once uncertainties are assessed, the disaggregated emissions and its associated uncertainties can be propagated through the air-quality model for policy assessment under uncertainties.