Treatment of Uncertainties for National Estimates of Greenhouse Gas Emissions



A4   References

Table A1. Example Output from a Qualitative Uncertainty Analysis of an Emissions Inventory

Inventory Component Basis of Uncertainty Description (examples only)
Survey Survey respondent expertise Incorrect or incomplete answers could result because of lack of understanding or incorrect interpretation
  Unknown answers Some equipment is very old, some has been modified, hence manufacturers' ratings would be no longer applicable
  Incorrect responses Some respondents did not read instructions. Confusion over units.
  Data entry Typographical and data omission errors possible. Transcription by support staff?
  Omitted sources x% of companies did not return survey results. Some returns incomplete.
Emissions methodology Emission factors Emission factors represent an average. Results could be biased by distribution of responses.
  Fugitive emissions Calculated using an empirical formula. Assumptions therein may not be valid in all cases.
  Applicability/usage A more applicable emissions methodology could exist. Methodology could have been applied incorrectly because of an incorrect assumption.



Table A2. AP-42 Rating System for Emissions Test Data


Rating Description
A Tests are performed by a sound methodology and are reported in sufficient detail for adequate validation
B Tests are performed by a sound methodology, but lacking sufficient detail for adequate validation
C Tests are based on an unproven or new methodology, or are lacking a significant amount of background information
D Tests are based on a generally unacceptable method, but the method may provide an order-of-magnitude value for the source


Table A3. AP-42 Rating System for Emission Factors


Rating Quality Rating Discussion
A Excellent Factor is developed from A and B-rated source test data taken from many randomly chosen facilities in the industry population. The source category population is sufficiently specific to minimise variability.
B Above average Factor is developed from A or B-rated test data from a 'reasonable number' of facilities. Although no specific bias is evident, it is not clear if the facilities tested represent a random sample of the industry. As with an A rating, the source category population is sufficiently specific to minimise variability.
C Average Factor is developed from A, B and/or C-rated test data from a reasonable number of facilities. Although no specific bias is evident, it is not clear if the facilities tested represent a random sample of the industry. As with the A-rating, the source category population is sufficiently specific to minimise variability.
D Below average Factor is developed from A, B and/or C-rated test data from a small number of facilities, and there may be reason to suspect that these facilities do not represent a random sample of the industry. There also may be evidence of variability within the source population.
E Poor Factor is developed from C or D-rated test data, and there may be reason to suspect that the facilities tested do not represent a random sample of the industry. There also may be evidence of variability within the source category population.



Table A4. Guidelines Proposed for Emissions Factors
Ratings in the UK

Rating Description
A An estimate based on a large number of measurements made at a large number of facilities that fully represent the sector
B An estimate based on a large number of measurements made at a large number of facilities that represent a large part of the sector
C An estimate based on a number of measurements made at a small number of representative facilities, or an engineering judgement based on a number of relevant facts
D An estimate based on a single measurement or an engineering calculation derived from a number of relevant facts and some assumptions
E An estimate based on an engineering calculation derived from assumptions only






Table A5. Protocol for Combining Quality Ratings for
Emission Factors and Activity Rates

Combination Factor Final Factor Combination Factor Final Factor
E - E E C - C C
E - D D D - A C
E - C D C - B B
D - D D C - A B
E - B D B - B B
E - A C B - A A
D - C C A - A A
D - B C    




Table A6. Data Quality Codes Recommended by the IPCC



  Estimates   Quality   Documentation   Disaggregation
Code Meaning Code Meaning Code Meaning Code Meaning
Part Partly estimated H High confidence in estimation H High (all background information included) 1 Total emissions estimated
All Full estimate of all possible sources M Medium confidence in estimation M Medium (some background information included) 2 Sectoral split
NE Not estimated L Low confidence in estimation L Low (only emission estimates included) 3 Sub-sectoral split
IE Estimated but included elsewhere            
NO Not occurring            
NA Not applicable            









Table A7. Data Attribute Rating System: How Scores are Evaluated
for an Emission Inventory

Attribute Emission factor Activity rate Emissions
Measurement/method e1| a1| e1| X a1|
Source specificity e2| a2| e2| X a2|
Spatial allocation e3| a3| e3| X a3|
Temporal Allocation e4| a4| e4| X a14
Composite Score S4i=1 ei
___
4 		S4i=1 ai
___
4 		S4i=1 ei x ai
______
4

Table A8. Summary of Methods for Assessing Uncertainties in Emissions Estimates or Evaluating Data Quality
Method Advantages Disadvantages Utility/Relevance
Qualitative
  • low cost
  • comprehensive
  • subjective
  • limited value of output
  • useful only for screening and prioritisation
Semi-quantitative/ Data Quality Ranking
  • moderate cost
  • comprehensive
  • documented and structured approaches available
  • useful in prioritising areas for further quantitative study
  • subjective (but less so than above)
  • output still not fully quantitative, hence intercomparisons, etc., difficult
  • select the most appropriate tool (e.g. DARS, IPCC scoring)
  • use as basis for a more quantitative analysis in particular areas of the UK emissions database
Quantitative:
1. Expert estimation
  • moderate cost
  • requires considerable active participation from a wide range of experts
  • owing to lack of information, cannot be used for large areas of emissions database
  • not suited to present study
2. Error propagation
  • not computationally complex
  • unless database structured appropriately, considerable cost unable to produce meaningful estimates if : parameter values are correlated or uncertainty exceeds a critical magnitude
  • owing to lack of information, cannot be used for large areas of emissions database
  • difficult to interface with less quantitative methods
  • in the context of this study, does not have any advantages over the following technique
3. Direct simulation
  • can accommodate large parameter uncertainties and correlations between parameter values
  • allows most important component sources of uncertainty to be identified
  • unless database structured appropriately, considerable cost
  • considerable care required to evaluate output distributions (emissions estimates)
  • owing to lack of information, cannot be used for large areas of emissions database
  • difficult to interface with less quantitative methods
  • set up probabilistic framework for areas of database amenable to quantitative analysis (i.e. those sections for which quantitative information is available)