1. Long-term Greenhouse Gas Stabilization and the Risks of Dangerous Impacts M. Webster, C.E. Forest, H. Jacoby, S. Paltsev, J. Parsons, R. Prinn, J. Reilly, M. Sarofim, A. Schlosser, A. Sokolov, P. Stone, C. Wang Engineering Systems Division MIT Joint Program on the Science and Policy of Global Change Massachusetts Institute of Technology Society for Risk Analysis New England Chapter April 28, 2009

2. Calvin’s View on Risky Decisions

3. Outline Motivation MIT IGSM Model Framework Parametric Uncertainty Resulting Uncertainty in Projections Exploring Risk-Risk Tradeoffs

4. Climate Change Policy: Choosing a Long-Term Target UN Framework Convention on Climate Change –“…stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system.” How do we choose this stabilization level?

5. CCSP Product 2.1a Study of GHG Stabilization Scenarios Three Models: –MERGE (EPRI/Stanford) –MiniCAM (PNNL/UMD) –IGSM (MIT) Source: Clarke et al., 2007

6. Stabilization Scenarios (Source: U.S. CCSP Product 2.1a)

7. Global Mean Temperature Change (Deterministic)

8. Question for this Study How can we use information about risks of exceeding thresholds to guide our choice among long-run stabilization targets? Use the uncertainty in the CCSP results from one model (MIT IGSM)? Objective: Frame the choice of long-term stabilization target as a risk management decision –Consider risks of both climate impacts and abatement costs

9. MIT Integrated Global Systems Model

10. Uncertainty in Economics 110 Uncertain Parameters, including: –Productivity growth rates (historical data) –Energy efficiency growth rate (historical data) –Ease of substituting inputs (historical data) –Costs of new technologies (expert judgment) Main Uncertain Outputs: –Emissions (GHGs, urban pollutants) –Costs (consumption loss, carbon prices)

11. GDP Growth Uncertainty

12. Methodology Latin Hypercube Monte Carlo –400 random samples of all parameters –Impose correlation where justified by empirical data and/or theory Impose each CCSP scenario as an emissions cap over time –Not a fixed radiative forcing target –No banking/borrowing –DO allow GHG trading using GWPs –DO allow trading between nations each period

13. Uncertainty in CO 2 Emissions (No Policy)

14. Carbon Prices in 2020

15. Global Welfare Loss (%) in 2020

16. Uncertainty in Total Primary Energy Sources 2050

17. Relative Contribution to Variance Energy Supply Energy Demand Scale of Economy Other Uncertainties Predict which most affect cum. CO 2, carbon prices.

18. Top Ten Drivers of Uncertainty in Abatement Cost

19. Uncertainty in Climate Parameters Emissions Uncertainty from EPPA Climate Sensitivity Heat & Carbon Uptake by Deep Ocean Radiative Forcing Strength of Aerosols CO 2 Fertilization Effect on Ecosystem Trends in Precipitation Frequency

20. Results: Temperature Change Impacts of Stabilization Paths Global Mean Surface Temperature Increase ( o C) (1981-2000) to (2091-2100) No Policy Level 1 Level 3 Level 4 Level 2

21. Results: Sea Level Rise (Excluding Greenland and WAIS)

22. Communicating the Odds of Temperature Change

23. Communicating the Impact of Policy No Policy Stringent Policy (~550 ppm)

24. Δ T > 2 o C Δ T > 4 o C Δ T > 6 o C No Policy 400 in 40017 in 201 in 4 Stabilize at 750 400 in 4001 in 41 in 400 Stabilize at 650 97 in 1007 in 100<1 in 400 Stabilize at 550 8 in 101 in 400<1 in 400 Stabilize at 450 1 in 4<1 in 400 USING THE IGSM, WHAT IS THE PROBABILITY OF GLOBAL WARMING for 1980-2100, WITHOUT & WITH A 450, 550, 650 or 750 ppm CO 2 -equivalent STABILIZATION POLICY? (400 random samples for economics & climate assumptions)

25. Sea Level Rise > 0.2m Sea Level Rise > 0.4m Sea Level Rise > 0.6m No Policy 400 in 40013 in 209 in 100 Stabilize at 750 396 in 4001 in 5< 1 in 400 Stabilize at 650 97 in 1001 in 10< 1 in 400 Stabilize at 550 9 in 101 in 50< 1 in 400 Stabilize at 450 7 in 10<1 in 400 USING THE IGSM, WHAT IS THE PROBABILITY OF GLOBAL SEA LEVEL RISE for 2000-2100, WITHOUT & WITH A 450, 550, 650 or 750 ppm CO 2 -equivalent STABILIZATION POLICY? (400 random samples for economics & climate assumptions)

26. Δ WL>1% Δ WL>2% Δ WL>3% No Policy--- Stabilize at 750 1 in 1001 in 400<1 in 400 Stabilize at 650 3 in 1001 in 200<1 in 400 Stabilize at 550 1 in 41 in 501 in 200 Stabilize at 450 7 in 103 in 101 in 10 USING THE EPPA, WHAT IS THE PROBABILITY FOR WELFARE LOSS (% change in 2020), WITHOUT & WITH A 450, 550, 650 or 750 ppm CO 2 -equivalent STABILIZATION POLICY? (400 random samples for economics assumptions)

27. Marginal Reduction in Probability of Exceeding 5 o C Global Temperature Change Probability of exceeding target Reduction in Probability (percentage points) Cum. CO 2 Emissions 2000-2100 (GtC) Reduction in Cumulative CO 2  Prob/  Cum No Policy 54.0%1605.0 -- Stabilize at 750 2.5%51.5%1123.1481.9 0.107% Stabilize at 650 0.3%2.3%910.9212.2 0.011% Stabilize at 550 0.0%0.3%634.7276.2 0.001% Stabilize at 450 0.0% 381.1253.6 0.000%

28. Tradeoffs in Choosing Stabilization Targets: Expected Values

29. Risk-Risk Tradeoffs in Choosing Stabilization Targets

31. Key Insights Economics –GDP growth important, not biggest driver –Energy demand parameters critical –High returns on reducing uncertainties in AEEI, elasticities of substitution, etc. Climate Science –Uncertainty still wide –Mean and upper tails indicate likelihood of significant impacts without some GHG reductions

32. Key Insights (II) Decision-Making –Problem is one of risk management –Risk-risk tradeoffs give different insights than focusing on mean/reference values –Suggestive that for a 450ppm, cost risk may outweigh the reduction in temperature risk

34. Δ WL>1% Δ WL>2% Δ WL>3% No Policy--- Stabilize at 750 1 in 123 in 2003 in 400 Stabilize at 650 1 in 31 in 203 in 400 Stabilize at 550 9 in 103 in 51 in 4 Stabilize at 450 98 in 10096 in 10085 in 100 USING THE EPPA, WHAT IS THE PROBABILITY FOR WELFARE LOSS (% change in 2050), WITHOUT & WITH A 450, 550, 650 or 750 ppm CO 2 -equivalent STABILIZATION POLICY? (400 forecasts with equally probable economics assumptions)

35. Uncertainty in CO 2 Emissions (No Policy)

36. Why are the probabilities shifted to higher temperatures than in our previous calculations (Webster et al, 2003)? Radiative Forcing Increases? –Emissions (higher lower bound) –Reduced Ocean Carbon Uptake –Additional forcing such as Black Carbon & Tropospheric Ozone (additional forcing included but still calibrated by net aerosols in 1990s) Climate Model Response? –Climate Model Parameters show higher response Learning? –Distributions better defined –Distributions shifted higher

37. IPCC AR4 Temp Chg Uncertainty Relevant Comparison To IGSM No Policy

38. Typical Production Function in EPPA

39. Uncertainty in SO 2 Emissions (No Policy)

40. Uncertainty in SO 2 Emissions (No Policy vs. CCSP-550)

41. Uncertainty in Methane Emissions

42. Uncertainty in NO x Emissions

43. Uncertainty in BC Emissions

44. Zonal Temperature Change 2000-2100 (Median)

45. Zonal Temperature Change 2000-2100 (95 th Percentile)

46. PDFs of Global Mean Temp. Chg.

47. PDFs of Sea Level Rise (Excluding Greenland and WAIS)

48. Global Electricity Consumption by Technology and Fuel

52. Historical 1950-2000 (%) Projected Annual Average Growth Rate (%) 2000-2100 RegionMeanStd Dev0.050.50.95 USA2.22.3%1.72.12.5 CAN2.32.3%1.72.12.5 MEX2.25.2%1.22.12.9 JPN4.93.5%1.72.22.7 ANZ2.01.8%2.02.32.6 EUR2.81.6%1.92.12.4 EET1.13.9%2.12.83.3 FSU1.15.3%2.02.83.7 ASI4.34.7%1.82.63.3 CHN4.33.7%2.53.13.7 IND2.32.7%2.32.73.1 IDZ2.75.0%1.12.63.9 AFR1.01.8%2.02.32.6 MES2.33.3%1.52.12.6 LAM1.72.0%1.72.12.5 ROW2.23.5%1.72.32.8 GLOBAL2.22.42.6

61. Carbon Price Under Level 1 (450ppm)