A practical approach for the assessment and illustration of uncertainty in emissions modelling: a case study using GAINS Ireland.

Abstract: National outlooks of emission levels are important components of international environmental policymaking and associated national policy development. This is the case for both greenhouse gas emissions and transboundary air pollutants. However, there is uncertainty inherent in the production of forecasts. In the climate context, IPCC guidelines have been established to support national teams in quantifying uncertainty within national inventory reporting of historic emissions. These are presented to indicate the potential range of deviation from reported values and to offer added evidence for policy decisions. However, the method and practice of accounting for uncertainty amongst emission forecasts is both less clear and less common. This paper posits that the role of forecasts in setting international targets and planning policy action renders the management of ‘forecast’ uncertainty as important as addressing uncertainty in the context of inventory and compliance work. Failure to explicitly present uncertainty in forecasting delivers an implicit and misplaced confidence in a given future scenario, irrespective of parallel work on other scenarios and sensitivities. However, it is acknowledged that approaches to uncertainty analyses within the literature are often highly technical and the models used are both computationally demanding and time-intensive. This can limit broader adoption where national capacities are limited and scenario development is frequent. This paper describes an approach to presenting uncertainty, where the aim is to balance the technical and temporal demands of uncertainty estimation against a means of delivering regular and practical estimation and presentation of uncertainty for any given scenario. In turn this methodology should help formalise the recognition of the uncertainty dimension in emissions forecasts, for all stakeholders engaged. [Copyright &y& Elsevier]