1. Recent methodological changes in the GAINS model M. Amann, W. Asman, I. Bertok, J. Cofala, C. Heyes, Z. Klimont, W. Schöpp, F. Wagner Meeting of the Task Force on Integrated Assessment Modelling Prague, May 2-4, 2007

2. Recent methodological changes Update to the City-delta methodology RAINS cost curve-based optimization replaced by GAINS measured-based optimization 5-years meteorological conditions EC4MACS work plan

3. Changes to the City-delta methodology

4. Changes since December 2006 New population and city-domain data (“compact” city shapes including ~70% of population) Target metric: population-weighted PM2.5 concentration for health impact assessment Refined results from the three urban models Revised functional relationship Multi-year meteorology Modified assumptions on urban emissions

5. “Compact urban shape” for which the urban increment is computed – Prague

6. Compact urban shapes for which the urban increments are computed Paris London Lisbon Krakow Milan Berlin

7. Urban increments computed by the three models for the 5*5 km center grid cell and population-weighted

8. Urban increments computed by Chimere, CAMx, RCG, compared with the City-delta regression

9. Hypothesis of the City-delta functional relationship Δc … concentration increment computed with the 3 models α. β … regression coefficients D … city diameter U … wind speed Δq … change in emission fluxes d … number of winter days with low wind speed

10. Urban per-capita emissions by SNAP sector

11. Emission densities (red) and computed urban increments (blue)

12. Contribution of long-range transport (blue) and local primary PM emissions (red) to urban PM2.5 AT BE Bulgaria FI France

13. Contribution of long-range transport (blue) and local primary PM emissions (red) to urban PM2.5 Italy Netherlands NO Poland PT

14. Contribution of long-range transport (blue) and local primary PM emissions (red) to urban PM2.5 Germany GR HU

15. Contribution of long-range transport (blue) and local primary PM emissions (red) to urban PM2.5 United Kingdom

16. Sectoral contributions to background concentrations of primary PM2.5 components from urban sources AT BE Bulgaria FI France

17. Sectoral contributions to background concentrations of primary PM2.5 components from urban sources United Kingdom

18. Summary Substantial revisions of methodology and input data Health impact assessment based on population-weighted increments – conservative assumption? Largest uncertainties associated with quality of urban emission estimates. Large discrepancies cannot be readily explained More plausible on emissions assumptions improve estimates Validation hampered by lack of quality-controlled monitoring data Sensitivity analysis explored implications on optimization results

19. Mathematical formulation of the GAINS optimization

20. The RAINS optimization in brief Objective function: Minimization of total emission control costs C: Cost curves describe for each pollutant the relation between emission reductions x and costs: Emission reductions x must be sufficient to satisfy environmental constraints:

21. Constructing a cost curve with RAINS Compute for each emission control option: –(Further) reduction potential beyond baseline, –unit abatement costs, –marginal abatement costs in comparison to a less effective measure. Rank available measures according to their marginal costs

22. An example cost curve for SO 2

23. Objective function: Minimization of total emission control costs C for add-on measures (x) and structural changes (y): Application levels (x) and substitution potentials (y) are constrained: Emissions are calculated from activity levels and emission factors: Sectoral emissions of current legislation may not increase: Activity levels plus substitution levels must remain constant to satisfy demand: The GAINS optimization in brief

24. RAINS optimization: –Decides how far to move up the cost curve (keeping underlying activity levels fixed!) –Cost curves are independent from each other –Thus multi-pollutant measures are treated as single-pollutant GAINS optimization –Decides which measure to use (incl. multi-pollutant) –Each measure has one cost, but can have impacts on more than one pollutant –If cost-effective and possible, change the underlying activity (through, e.g., efficiency improvements) Differences between the RAINS and GAINS optimization

25. Comparison of cost curves Examples for Germany and Greece

26. Treatment of multi-pollutant measures

27. Multi-pollutant measures (1) Structural measures: –Energy savings, efficiency improvements, bans: all pollutants ↓ –Increased use of natural gas: CO 2, SO 2, VOC, NO x, PM ↓ CH 4 ↑ –Biomass: CO 2 ↓ VOC, PM, CH 4 ↑ Stationary sources: –SCR, SNCR: NO x, CO ↓, NH 3, N 2 O ↑ –Fluidized bed combustion: SO 2, NO x ↓, N 2 O ↑ –Advanced residential combustion: VOC, PM, CO, CH 4 ↓ –FGD: SO 2, PM ↓ CO 2 ↑ –IGCC: CO 2, SO 2, NO x, PM ↓ –CHP: all pollutants ↓ Mobile sources: –Euro-standards: NO x, VOC, PM, CO ↓ NH 3, N 2 O ↑ –Low sulfur fuels: SO 2, PM ↓ –Diesel: CO 2, VOC ↓, PM, NO x, SO 2 ↑

28. Multi-pollutant measures (2) Agricultural sources: –Low emission pig housing – NH 3, CH 4 ↓ N 2 O ↑ –Covered storage of slurry – NH 3 ↓ CH 4 ↑ –Injection of manure – NH 3 ↓ N 2 O ↑ –Anaerobic digestion (biogas) – CH 4, N 2 O ↓ CO 2 ↑ NH 3 ↓ ↑ Other sources –Gas recovery and flaring: CH 4 ↓ CO 2, PM, VOC, SO 2, NO x, CO ↑ –Gas recovery and re-use: CH 4 ↓ CO 2 ↑ –Improving flaring efficiency: PM, VOC, NO x, SO 2, CO ↓ –Waste incineration: CH 4 ↓ CO 2 ↑ –Gas recovery from wastewater treatment: CH 4 ↓ CO 2 ↑ In total approx 500 measures with multi-pollutant impacts considered in GAINS

29. Conclusions: ‘Technology approach’ vs. cost curve approach Advantages: –Adequate representation of multi-pollutant technologies –Dynamic interaction between activity data (e.g. energy system) and emission control –Simultaneous treatment of AP control and GHG abatement increases economic efficiency –No arbitrary allocation of costs to individual pollutants necessary Disadvantage –Simple single pollutant cost curves can be constructed but may be misleading, because arbitrary assumption about the valuation of multi-pollutant technologies must be included

30. Comparison of “GAINS-in-RAINS mode” and full GAINS optimizations for air pollutants

31. Treatment of multi-pollutant measures in RAINS Not possible to have them endogenously included in the optimization with single-pollutant cost curves Way out: –Ignore impacts on less import pollutant, or –Use pre-defined packages of measures (e.g., Euro-IV/V) –Two optimization runs for all other measures with different boundary conditions (with and without the multi-pollutant measures) –Comparison of resulting costs –For most important measures only

32. Reports on methodology and cost curve comparisons at www.iiasa.ac.at/rains

33. Multi-year meteorology

34. Atmospheric dispersion based on meteorological conditions of 1996, 1997, 1998, 2000, 2003 Sensitivity analysis with 2003

35. Loss in statistical life expectancy computed with different meteorological conditions (for 2000)

36. Estimates of mortality from ozone for year 2000 emissions for different meteorological conditions

37. Estimates of unprotected forest area for year 2000 emissions for different meteorological conditions

38. Estimates of ecosystem area with excess nitrogen deposition for year 2000 emissions for different meteorological conditions

39. Summary For EU-27, PM and ozone impacts from 5-yrs meteorology very similar to 1997. Acidification ~10% higher, eutrophication ~5% higher But different trends in different regions across Europe Implications on meaures 2003 produces higher health impacts for PM and ozone

40. EC4MACS European Consortium for Modelling of Air Pollution and Climate Strategies Overview of the 5 years work plan

41. The EC4MACS model system GAINSPOLESPRIMES CAPRI TM5EMEP CCE-IMPACTS TREMOVE BENEFITS Global/ hemispheric boundary conditions European policy drivers Energy Transport Atmosphere Agriculture Ecosystems GEM-E3 Cost- effectiveness Impacts FASOM

42. General work plan 2007: –Methodological improvements 2008: –Data collection –Feedbacks on methodological improvements 2009 –Interim assessment –Methodology workshop 2010 –Uncertainty assessment –Bilateral consultations on input data 2011 –Final assessment