Due to the complex nature of estimating inventories, a simple quantitative treatment cannot be given for the accuracy of emission estimates. However, an indication can be made of the systematic errors in a qualitative way for each pollutant. For individual years, there are considerable uncertainties associated with the emission estimates. Trends over time are likely to be more reliable.

The SO₂ emissions are likely to be the most accurate as they depend largely on the level of sulphur in fuels. Hence a comprehensive analysis of coal and fuel oil consumption by power stations and the agriculture, industry and domestic sectors will give accurate emission estimates. The emission totals are likely to be within an accuracy of ± 10-15% of the true figure. Margins of error result mainly from the uncertainties about sulphur retention in ash.

NO_x emissions are less accurate than SO₂ because they are based on relatively few measurements of emission factors and depend on combustion conditions that can vary widely. For example, the amount of NO_x emitted from vehicles varies as a function of speed as well as being dependent on the type of vehicle (this particular example is taken into account when estimating emissions from the road transport sector). In addition, many of the data used have been derived from other small samples which contain margins of error e.g. vehicle speed distributions. It is estimated that NO_x emissions are accurate to ± 30%.

The development of an accurate emission inventory for NMVOCs is hindered by a number of factors. First, the number of species covered by the term NMVOCs is very large and they arise from a diverse range of processes. Within a single industry sector such as printing, the variation in the quantity and type of organic solvents used, the different printing processes used, and the varying extent and types of abatement used on different presses make it difficult to apply a single, generally valid, emission factor across the industry. The cost of measuring emission factors for all of the different processes would be very high and not entirely satisfactory since the measurement of emission factors is also problematic. Many commonly employed measurement techniques such as flame ionisation detection do not respond with uniform sensitivity to all compounds. Large errors can therefore occur if emitted compounds are poorly detected by the measurement technique used. Methane is also associated with NMVOC emissions from many sources, for instance combustion processes, transport, and the oil and gas industries. Emission factors derived from measurements of these sources will often include a methane component. Secondly, activity data for many solvent using and other processes are not readily available leading to difficulties in producing a good time series. Finally, due to emission controls and other factors, emissions from many sectors are showing substantial reductions from year to year. Accordingly, emission estimation methods must be revised frequently in order to maintain the accuracy of the inventory.

Despite these issues, significant improvements have been made to the NMVOC inventory over the last several years, in particular regarding emissions from solvent use and production processes, as well as speciation and spatial disaggregation of emissions. As a result the uncertainty associated with the NMVOC emission estimates is considered to have significantly reduced. It is estimated that UK emissions are probably accurate to ± 30%.

Ammonia emissions are the most uncertain due largely to the nature of the major agricultural sources. Emissions depend on animal species, age, weight, diet, housing systems, waste management and storage techniques. Hence emissions are affected by a large number of factors which make the interpretation of experimental data difficult and emission estimates uncertain (DOE, 1994).