Statistical dependence in input data of national greenhouse gas inventories: Effects on the overall inventory uncertainty

Winiwarter W & Muik B (2010). Statistical dependence in input data of national greenhouse gas inventories: Effects on the overall inventory uncertainty. Climatic Change 103 (1): 19-36. DOI:10.1007/s10584-010-9921-7.

Full text not available from this repository.

Abstract

An uncertainty assessment of the Austrian greenhouse gas inventory provided the basis for this analyss. We isolated the factors that were responsible for the uncertainty observed, and compared our results with those of other counries. Uncertainties of input parameters were used to derive the uncertainty of the emission estimate. Resulting uncertainty using a Monte Carlo approach was 5.2% for the emission levels of 2005 and 2.4 percentage points for the 1990-2005 emission trend. Systematic uncertainty was not assessed. This result is in the range expected from previous experience in Austria and other countries. The determining factor for the emission level uncertainty (not the trend uncertainty) is the uncertainty ssociated with soil nitrous oxide N2O emissions. Uncertainty of the soil N2O release rate is huge, and there is no agreement even on the magnitude of the uncertainty when country comparisons are made. In other words, reporting and use of N2O release uncerainty are also different between countries; this is important, as this single factor fully determines a country's national grenhouse gas inventory uncertainty. Inter-country comparisons of emission uncertainty are thus unable to reveal much about a country's inventory quality. For Austria, we also compared the results of the Monte Carlo approach to those obtained from a simpler error propagation approach, and find the latter to systematically provide lower uncertainty. The difference can be explained by the ability of the Monte Carlo approach to account for statistical dependency of input parameters, again regarding soil 2O emissions. This is in contrast to the results of other countries, which focus less on statistical dependency when performing Monte Carlo analysis. In addition, the error propagation results depend on treatment of skewed probability distributions, which need to be translated into normal distributions. The result indicates that more attention needs to be given to identifying statitically dependent input data in uncertainty assessment.

Item Type: Article
Research Programs: Atmospheric Pollution (APD)
Bibliographic Reference: Climatic Change; 103(1-2):19-36 (November 2010) (Published online 14 July 2010)
Depositing User: IIASA Import
Date Deposited: 15 Jan 2016 08:43
Last Modified: 18 Feb 2016 16:52
URI: http://pure.iiasa.ac.at/9217

Actions (login required)

View Item View Item

International Institute for Applied Systems Analysis (IIASA)
Schlossplatz 1, A-2361 Laxenburg, Austria
Phone: (+43 2236) 807 0 Fax:(+43 2236) 71 313