Treatment of Uncertainties for National Estimates of Greenhouse Gas Emissions

  1. Overview and Results
4.1 Uncertainties in Emissions in 1990 and 2010

4.1.1 Carbon Dioxide

Carbon dioxide is produced primarily from combustion. In this study, processes that result in the removal from the atmosphere, such as certain land management practices, are considered separately, in accordance with IPCC Guidelines [6]. Emissions in 1990 were taken in the main from the most recent UK Greenhouse Gas Inventory, whereas those in 2010 were inferred from DTI forecasts. Data on carbon dioxide sinks and sources in land use change and forestry were provided by ITE

Uncertainties in emissions were estimated in the main by examining the likely uncertainties in emission factors and activity rates. In some cases where data do not exist, or the quality of data is insufficient, the uncertainties in emissions from particular sources were estimated directly by judgement. Table 1 displays the data on uncertainties used for the different sources of carbon dioxide in terms of the uncertainties in the main input parameters, or in the total estimated emission (e.g. land use change) as appropriate.

The calculated uncertainties in total emissions of carbon dioxide in 1990 and 2010 are displayed in Figure 1. The corresponding information for sinks is given in Figure 2. Probability density functions of emissions and sinks for 1990 and 2010 are given in Figures 3 and 4, respectively.

4.1.2 Methane

Methane is produced mainly from waste disposal (landfilling), agricultural practices and fugitive emissions from fuels. Emissions from combustion are relatively low. Emissions in 1990 were taken from a range of available sources of information, whereas those in 2010 were based on DETR forecasts.

Owing to the sparsity of data, uncertainties in total emissions (largely estimated in previous studies) were derived in the main by expert judgement. Uncertainties in the largest single emission source, from landfills, were based on a previously conducted comprehensive study. This involved constructing an empirical distribution of emissions on the basis of a range of landfill site types in the UK and their methane management regimes. In this previous study, density functions were formulated for a range of model input parameters. These govern the rate of landfill methane production and subsequent oxidation.

Table 3 displays data on the uncertainties for the individual sources within the methane inventory. The empirical distribution used for the emissions from landfills was skewed positively (that is, a longer tail for the higher emissions than for the lower).

The calculated uncertainties in total emissions of methane in 1990 and 2010 are displayed in Figure 5. The probability density functions of emissions are given in Figure 6.

4.1.3 Nitrous Oxide

Nitrous oxide is produced mainly from industrial sources (manufacture of adipic and nitric acids), agricultural practices (such as the application of artificial fertilisers), road traffic and power generation.

Emissions of nitrous oxide in 1990 and 2010 were obtained from previously published information and from the DETR database. Emissions were represented as products of emission factors and activity rates. For most sources (notably agricultural soils), overall uncertainties were dominated by uncertainties in the emission factors. The uncertainties in the input parameter values were obtained from both available information and from expert judgement.

Table 4 displays the uncertainties applied to the emission factors and activity rates used for the various sources of nitrous oxide included in the analysis.

The calculated uncertainties in total emissions of nitrous oxide in 1990 and 2010 are displayed in Figure 7. The probability distribution functions of emissions are given in Figure 8. In the latter case, the positive skew of the distribution is particularly noticeable 11. This is caused by the uncertainty in emissions from agricultural soils which as it is so wide (and assumed to be positive) has been modelled by a log-normal distribution.

4.1.4 Halocarbons and Sulphur Hexafluoride

Emissions of these gases arise from a large number of sources, including refrigerants, fire-fighting agents, solvents and during the production of magnesium and aluminium. The emissions themselves arise by leakage during manufacture, use and disposal. Information on the rates of production of the gases, and on estimated losses during their life cycles, was obtained from the MCG spreadsheet [29]. The estimation of the emissions of the various gases is itself complex, being dependent on assessments of the specific inventories in any year, together with additions and losses on a year by year basis. Emission factors correspond to the percentage losses during manufacture, disposal and product life, and were applied to activity data on the consumption or production of fluid or the current existing in the 'bank' of products. Table 5 displays the uncertainties in these three factors for the three categories of gases of concern.

The calculated uncertainties in total emissions of HFCs in 1990 and 2010 are displayed in Figure 9. The corresponding probability density functions of emissions are given in Figure 10.

Figures 11-14 display the corresponding information for PFCs and SF

4.1.5 Aggregated Total Emissions (GWPs)

In order to provide an overall measure of uncertainty in the UK Inventory of greenhouse gases, the respective 100 year-GWPs (Table 6) were used in the analysis to weight the relative contributions of each gas. These contributions for 1990 and 2010 are shown in Figure 15. As can be seen, in 1990 the total GWP is dominated by carbon dioxide (78%), followed by methane and nitrous oxide (at about the 10% level). In 2010, the relative contribution of carbon dioxide increases (to 85%), with corresponding reductions in methane and nitrous oxide. The contribution of emissions of sulphur hexafluoride rise by about a factor of two, from 0.07% to 0.14%, whereas those of the other industrial gases fall substantially, reflecting changes in management practices partly necessitated by concerns over ozone depletion.

The calculated uncertainties in the total aggregate emissions in 1990 and 2010 are displayed in Figure 16. The related probability density functions are given in Figure 17.

4.2 Changes in Emissions Between 1990 and 2010

In addition to displaying the uncertainties in emissions in the specific years 1990 and 2010, Figures 1-14, 16 and 17 also show the distributions of percentage changes in emissions over these two decades, in terms of the standard widths of the distributions themselves, and as probability density functions. As in the case of emissions, the distributions of possible changes are presented in terms of the standard measure of uncertainty (i.e. as 2´Övariance ¸ mean value, and as 1st and 99th percentiles), and as probability density functions themselves.

4.3 Main Sources of Uncertainties

Rank order correlation coefficients for all input parameters were evaluated for the calculations of the uncertainties in emissions in 1990 and 2010, and for the predicted change in emissions between these two times, and are given in Table 7. Values are calculated for each gas separately, and for the aggregated (GWP-weighted) total.

In order to keep the table to a manageable size, only coefficients whose magnitudes exceed 0.3 are given in the Table (i.e.
|coefficient|>0.3). It should be noted in particular that no single (model) parameter uncertainty was found to dominate the uncertainties in the emissions of carbon dioxide, methane, nitrous oxide or the aggregate total.

11 A consequence of the adoption of a log-normal distribution for the specific emission from agricultural soils.