This deliverable report describes the development of prior uncertainty dataset related to the European and global prior emission datasets (PED) for the year 2018. We provide an overview of the developed methodology to estimate prior uncertainties in the aggregated PED starting from a detailed set of uncertainties. This also includes the estimation of gridded uncertainties and spatial error correlation lengths. For the global emissions only CO2 is considered, whereas for the European emissions CO and NOx are also included. Since the data underlying the global and European datasets is different, a slightly different approach is used for the datasets, but where possible we kept our methods consistent.
Two methods are developed that are discussed and compared in this report. The first method is an extensive Monte Carlo simulation, which can take into account non-Gaussian error distributions and error correlations. However, since this method is computationally expensive, especially for a high-resolution dataset, we also used a relatively simple error propagation approach. We find that this simpler method gives comparable results, especially considering the uncertainties in the prior uncertainty data.
Abstract