Stochastic Detection Time Concept and its Economic Implications T. Ermolieva, M. Jonas, Y. Ermoliev, M. Makowski IIASA, 2 nd International Workshop on Uncertainty in Greenhouse Gas Inventories, IIASA, Laxenburg, 27-28th of September, 2007.
Outline 1. Kyoto protocol and detection of emissions 2. Uncertainty (Variability) matters 3. Practical Example: Long time data series 4. Variability of emissions: “fast” and “slow” systems 5. Emission signal detectability: stochastic detection techniques 6. Economic value of stochastic detection: a. carbon trading markets b. insurance of carbon credits transactions
Kyoto protocol and detection of emissions Kyoto – binding commitments to limit or reduce the emissions of six GHGs or groups of gases (CO2, CH4, N2O, HFCs, PFCs, and SF6). Each Party of the protocol calculates how much of gases its country emits by adding together estimates/reported emissions from individual sources. Often estimated/reported emissions are inaccurate: M. Gillenwater & F. Sussman & J. Cohen: Practical Policy Applications of Uncertainty Analysis for National Greenhouse Gas Inventories. In many countries, agreed emission changes are smaller than their underlying uncertainty. In IPCC practice, emission/emission changes are reported, but without rigorous signal detection
S The KP requires that net emission changes be “verified” on the spatial scale of countries by the time of commitment, relative to a specified base year. The key questions: 1. Whether reported emissions outstrip uncertainty and can be “verified/detected” ? 2. What percentage of all possible emissions can be detected within a given time ?
Uncertainty (variability) matters Net Emissions Below the target, larger uncertainty Below the target, smaller uncertainty Net Emissions Below the target, larger uncertainty Above the target, smaller uncertainty Source: M. Jonas et. al.
Variability matters Net Emissions ± 2 δ a 95 th confidence =
Variability matters Net Emissions A C B Base Commitment year/period year 95 th confidence 95 th confidence B C
Practical examples Longer data time series on FF , LUC and OU taken from global carbon budget: http://lgmacweb.env.uea.ac.uk/lequere/co2/carbon_budget.htm Fossil Fuel Emissions (FF) are estimated from data on the global consumption of coal, oil, and natural gas. The Land Use Change (LUC) are estimated using a bookkeeping model updated in August 2006 using revised data from the FAO of the United Nations. The mean Ocean Uptake (OU) for 1959-2005 is estimated using an ocean general model forced by observed atmospheric conditions of weather and CO2 concentration. The terrestrial uptake is estimated as a residual of all the sources minus the ocean uptake and atmosphere increase (Assessment Report 4, WG 1, Ch. 7, 2007, p. 519) .
Variability of emissions: “fast” and “slow” emissions dynamics Net Terrestrial Balance Fossil Fuels: strong dynamics, 8.0 small variability 6.0 Fluxes to Balance [Pg C/yr] y = 0.0544x + 1.5985 y = 0.1019x + 2.7158 ff_plus R 2 = 0.9132 R 2 = 0.9845 4.0 ffplus_smooth y = 0.0236x + 1.3588 atm_growth R 2 = 0.9015 atm_smooth 2.0 ocean_sink ocean_smooth net_terr 0.0 1959.5 1964.5 1969.5 1974.5 1979.5 1984.5 1989.5 1994.5 1999.5 2004.5 net_terr_smooth -2.0 y = 0.0239x - 0.2415 R 2 = 0.6931 Net terrestrial: slow dynamics, -4.0 large variability Time http://lgmacweb.env.uea.ac.uk/lequere/co2/carbon_budget.htm
Data series Net Emissions 0.20 0.15 0.10 0.05 1961 1963 1965 1967 1969 0.00 -0.05 y = 0.0239x - 0.2415 -0.10 R2 = 0.6931 -0.15 -0.20 -0.25 -0.30 Data De-trended Trend
Slow dynamics vs large variability: Net terrestrial uptake, 1960-1970 12 1.0 0.9 95th percentile value: 0.06 10 0.8 Average: -0.13 0.7 8 0.6 Frequency 6 0.5 0.4 4 0.3 0.2 2 0.1 0 0.0 Net Emissions 3 1 9 6 4 2 9 7 5 2 0 2 5 7 9 1 4 2 2 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 . . . . . . . . . . . . . . . . . 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - - - - - - - - Pg C/yr Frequency Cumulative % More emissions below average ! ± 2 δ a
Slow dynamics vs large variability: Net terrestrial uptake, 1985-1995 18 1.0 95th percentile value: 0.79 0.9 16 0.8 14 0.7 12 Frequency 0.6 10 0.5 Average: 0.6 8 0.4 6 0.3 4 0.2 2 0.1 0 0.0 9 4 9 4 8 3 8 3 7 2 7 2 6 1 6 1 5 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 Net Emissions . . . . . . . . . . . . . . . . . 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Pg C/yr Frequency Cumulative % More emissions above average ! 2 ± δ a
Fast dynamics vs small variability: Fossil fuel emissions, 1960-1970 16 1.0 0.9 95th percentile 14 Value: 0.3 0.8 Average: 0.3 12 0.7 10 Frequency 0.6 8 0.5 0.4 6 0.3 4 0.2 2 0.1 0 0.0 2.80 2.83 2.87 2.90 2.93 2.96 2.99 3.02 3.05 3.09 3.12 3.15 3.18 Net Emissions Pg C/yr Frequency Cumulative % More emissions below average ! ± 2 δ a
Fast dynamics vs small variability: Fossil fuel emissions, 1985-1995 16 1.0 95th percentile Value: 6 0.9 14 0.8 12 0.7 10 0.6 Frequency Average: 5.8 8 0.5 0.4 6 0.3 4 0.2 2 0.1 0 0.0 5.621 5.650 5.679 5.708 5.737 5.766 5.796 5.825 5.854 5.883 5.912 5.942 5.971 Net Emissions Frequency Cumulative % Pg C/yr More emissions above average ! 2 ± δ a
The need for “detection” of emission shapes 1. In 1960 to 1970, the terrestrial system was mostly a sink. 2. Average flow -0.13. Higher likelihood of flows larger than average. 3. More of probability mass below average 4. In 1985 to 1995, it turned to source. Average flow 0.6. 5. More of probability mass above average.
Emission signal: detectability Detect time when emission outstrips the uncertainty represented by a symmetrical interval Net Emissions 2ε ε = const time t 1 t 2 � F> � at t*=DT 2ε � � F ε = const time t 1 t 2 t*=DT
Emission signal: detectability Net Emissions ε 2 Net Emissions ε 2 ε 1 ε 1 a) b) time t 1 t 2 DT < t 2 time t 1 t 2 Net Emissions DT = t 2 ε 2 ε 1 ( 1 ) ε t ∆ > t ε dF d � � net 2 − � � c) dt dt � � t t 1 1 time DT > t 2 t 1 t 2
Stochastic detection of emission signal Net Emissions F max � 2 � 1 F min time DT < t 2 DT > t 2 t 1 t 2
E-sided vs stochastic detection, slow dynamics and large variability: Net terrestrial uptake, 1965 – 1985 1.0 0.9 95th percentile 0.8 Value: 54.7 0.7 Average: 17.7 Mean 0.6 0.5 Median: 15 0.4 0.3 E-sided approach: 7.9 0.2 E-sided approach: 7.9 0.1 0.0 Years 67.34 0.00 4.21 8.42 12.63 16.83 21.04 25.25 29.46 33.67 37.88 42.09 46.30 50.50 54.71 58.92 63.13
E-sided vs stochastic detection, slow dynamics and large variability: Net terrestrial uptake, 1965 – 1985 Example 1. 1965–1985 F(t 1 ) -0.13 Pg C yr-1 F(t 2 ) 0.18 Pg C yr-1 � (t 1 ) 0.16 Pg C yr-1 � (t 2 ) 0.39 Pg C yr-1 dt 20 Pg C yr-1 d � = � (t 2 )- � (t 1 ) 0.23 Pg C yr-1 |dF net |=|F(t 2 )-F(t 1 )| 0.32 Pg C yr-1 7.9 yr VT
E-sided vs stochastic detection, fast dynamics vs small variability: Fossil fuel emissions, 1965 – 1985 1.0 0.9 95th percentile 0.8 Value: 3 0.7 Average: 1.1 Mean 0.6 0.5 Median: 0.9 0.4 E-sided approach: 0.9 0.3 0.2 0.1 0.0 Years 0 3 5 8 1 4 6 9 2 5 7 0 3 . . . . . . . . . . . . . 0 0 0 0 1 1 1 1 2 2 2 3 3
Economic implications of stochastic detection techniques 1. Emissions are tradable commodities. 2. Variability of emissions is a key element for pricing commodities 2.a. Inclusion of various systems (forestry and land use CDMs) in carbon trading market: Carbon Market Europe, 21, 2006. (Available on request) 2.b. Slow dynamic systems (forestry, land use) long response times. 2.c. The “must” for an appropriate emission detection technique - affects prices. 3. Stochastic detection: what share of possible emissions is detectable within a given time interval. 4. Clean Development Mechanisms (CDM), Joint Implementation (JI) projects. 5. SwissRe and emission trading insurance: emissions uncertainties. http://www.swissre.com/pws/transactions.html 6. Insurance of “skewed” risks of emissions trading is similar to Catastrophic risks insurance.
Recommend
More recommend
