A method of sensivity analysis to describe the uncertainty in emission inventories

Emissions from natural sources are poorly described quantities. The NATAIR project has specifically been devised to improve the methods available for assessment of these quantities. Reliability and robustness of data produced within the project are therefore critical to any applicability of results. A semi-quantitative approach was chosen to provide this important property of any inventory. This approach took advantage of the project's sectoral design, where a specific expert is responsible for each source sector. We used expert elicitation to obtain the respective expert's view on the uncertainties encountered for each source sector. In a first round, each expert was asked to provide general information about parameters influencing the reliability for the specific sector. Then using this information and a detailed description on the respective algorithm to obtain emission estimates, we prepared specific questionnaires to be answered by the experts. The questionnaires were sent to the experts to allow for preparation before a telephone interview was performed. In these structured interviews, experts provided information about input data most critical for their specific source sector, and the sensitivity of the resulting emissions. We focused on the most relevant parameters (key parameters) to influence the specific sector for each of the emitted compounds. Using this information from each source we weighted the contribution of each key parameter and of the respective source to total emissions to arrive at the overall contribution of this key parameter to the total emission sensitivity for a given compound. Compiling uncertainty of the inputs and sensitivity, we were also able to assess their influence on the uncertainty of the resulting emissions. This allowed to rank the contributions of parameters, and to provide input information required for atmospheric model sensitivity analysis.