EC Tier 1 uncertainty analysis

1
EC Tier 1 uncertainty analysis

• Uncertainty workshop
• Bernd Gugele, ETC-ACC
• 5 September 2005

2
Availability of Table 6.1 of the Tier 1
uncertainty analysis as of 15 April 2005
Member State Year Coverage Member State Year Coverage
Austria 2003 96 Ireland 2003 100
Belgium 2001 99 Italy 2002 100
Denmark 2002 92 Netherlands 2002 100
Finland 2002 97 Spain 2002 100
France 2002 100 Sweden 2003 100
Germany 2003 100 United Kingdom 2002 100
Greece 2002 100
For 13 MS Tier 1 uncertainty estimates
available Four MS 2003, eight MS 2002, one MS
2001 Most MS cover all source categories (except
for LUCF) 95 of total EU-15 GHG emissions in
2003 are covered
3
Uncertainty analysis approach (1)

• MS uncertainty estimates were grouped by source
  categories
• Level of detail varies between MS
• 1A (stationary), 1A1 or 1A1a
• split by fuels or not
• 2001/2002 used because more detailed fuel split

4
Uncertainty analysis approach (2)
5
Uncertainty analysis approach (3)

• Then for each source category a range of
  uncertainty estimates was calculated
• lower bound assuming no correlation among MS
  uncertainty estimates
• upper bound assuming full correlation among MS
  uncertainty estimates
• Basic equations on summing up of variances with
  correlation
• VAR(XY)VAR(X)VAR(Y)2COVAR(X,Y)
• COVAR(X,Y)rsqrtVAR(X)VAR(Y)
• VAR variance
• COVAR covariance
• R correlation coefficient (assumed to be 1)

6
Uncertainty analysis approach (4)

• Then single uncertainty estimate calculated for
  each source category assuming that
• MS uncertainty estimates fully correlate if
  default EF are used and/or Tier 1 methods are
  used
• The fact that AD are most likely uncorrelated was
  neglected
• Then uncertainty estimates for total GHG
  emissions and for the sectors were calculated
• Stationary fuel combustion
• Transport
• Fugitive emissions
• Industrial processes
• Agriculture
• Waste

Also plausibility checks were made with Monte
Carlo tool
7
Uncertainty analysis example

• CO2 emissions from 2 A 2 Lime Production
• MS uncertainties range from 5 30
• Four MS use the same, default emission factors
  (DE, GR, SE, UK)

Emissions of four member states are considered
to be correlated, because they use the same,
default EF
Uncertainty EU-15 13.3
8
EU-15 uncertainty estimates

• Lowest uncertainty estimates for stationary fuel
  combustion (1 ) and transport (3 )
• Highest estimates for agriculture (44 - 83 ).
• For agriculture the range depends on assumption
  on N2O emissions from soils
• lower bound not correlated
• upper bound fully correlated

9
Uncertainty estimates and recalculations
10
Uncertainty contribution
15 sources contribute 99.7 to EC uncertainties
11
Uncertainty estimates 4D N2O
EC range 89 - 233
12
Uncertainty estimates 1A3b CO2
EC range 3.4 - 7.4 EC estimate 3.4
13
Uncertainty estimates 4B N2O
EC range 50 - 133 EC estimate 109
14
Uncertainty estimates 4B CH4
EC range 33 - 69 EC estimate 33
15
Uncertainty estimates 1A1a CO2
EC range 3.1 - 5.0 EC estimate 3.1
16
Issues for Working group

• Level of detail of EC uncertainty analysis
• Currently detailed level depending on MS
  availability
• Combination of different levels (e.g. 1A1-1A1a
  1A) appropriate?
• Methods/assumptions to combine uncertainty
  estimates at EC level
• Criteria to identify correlations?
• Currently focus on EF and methods neglect AD
• Is assumption on full correlation appropriate?
• Improvement of EC uncertainty analysis
• Combination of various years alternative?
• Not for all MS/sectors (i.p. LULUCF) available
  gap filling?
• How to estimate trend uncertainty? No AD, EF
  uncertainty for EC
• Correction for large uncertainties? Tier 2
  approach?
• Timing of EC uncertainty analysis
• How often and when
• Use of EC uncertainty estimates e.g. Tier 2 Key
  source analysis

17
(No Transcript)
18
Uncertainty 1A4b CO2
EC range 3.2 - 7.0 EC estimate 3.3