A Multi-Model Regional Decomposition of CO2 Emissions ... · lnCO 2 t;i;j lnCO 2 0;i;j ln 0 @ CO2...

Objectives and Motivation Data Methodology SSP Scenarios Results-Regional ∆CO2 Results-Model Uncertainty Appendix

A Multi-Model Regional Decomposition of CO2

Emissions: Socio-Economic Developments vsEnergy Efficiency and Carbon Intensity

Improvements

Michele Maurizio Malpede∗

*Fondazione Eni Enrico Mattei (FEEM)email: [email protected]

International Energy WorkshopAbu-Dhabi June 3, 2015

1 / 31


Overview

Objectives and Motivation

Data

Methodology

SSP Scenarios

Results-Regional ∆CO2

Results-Model Uncertainty

Appendix

2 / 31


Objectives of the Study

The study has 2 goals:

1. To explore regional distribution of CO2 emission variationsand their drivers from 2010 to 2100 and across differentsocio-economic scenarios

2. To examine how different integrated assessment modelspredict regional distributions of CO2 emissions

Motivation

• To determine the implications of each of the 4 drivers inevaluating the impacts of socio-economic scenarios on globaland regional economic systems and exploring the differencesbetween short-term (2030 vs 2010), medium-term (2050 vs2010) and long term (2100 vs 2010) effects

3 / 31


Data

Models

• AIM-CGE

• GCAM

• MESSAGE

• IMAGE

• REMIND

• WITCH

Time

2010-2100

Regions of the world

Advanced Economies: EUROPE, USADeveloping Economies: R5ASIA, R5MAF, R5LAM

Socioeconomic Scenarios

5 SSPs developed by the National Center for AtmosphericResearch (NCAR)

4 / 31


Methodology• Kaya and Yokoburi (1997) decompose carbon dioxide

emissions into 4 components, namely Population, GDP percapita, Energy Efficiency and Carbon Intensity:

CO2 = POP ∗ GDP

POP∗ PE

GDP∗ CO2

PE(1)

• As in Ang (2000) a difference in an aggregate indicator can bedecomposed into the sum of the effects by differences inexplaining factors and residual terms. Hence defining:∆CO2 = CO2t − CO20

∆CO2 = ∆POP+∆

(GDP

POP

)+∆

(PE

GDP

)+∆

(CO2

PE

)+∆res

(2)

• Each driver contributes in the following form (where t is theyear (i.e 2030, 2050, 2100), 0 is the starting year (i.e. 2010), istands for the region and j refers to the scenario):

5 / 31


Log Mean Divisia Index Method (LMDI)

∆POP =CO2t,i,j − CO20,i,j

lnCO2t,i,j − lnCO20,i,j

ln

(POPt,i ,j

POP0,i ,j

)(3a)

∆GDP

POP=

CO2t,i,j − CO20,i,j


ln

GDPt,i,j

POPt,i,j

GDP0,i,j

POP0,i,j

(3b)

∆PE

GDP=



ln

PEt,1,j

GDPt,i,j

PE0,i,j

GDP0,1,j

(3c)

∆CO2

PE=



ln

CO2t,i,j

GDPt,i,j

CO20,i,j

PE0,i,j

(3d)

∆res = 0 (3e)

6 / 31


The Shared Socio-economic Pathways (SSPs)

Figure: Conceptual framework and names of new global Shared Socio-economic

Pathways (SSPs) for climate change research as defined in O’Neill et al.(2012).

7 / 31


Results of the decomposition of ∆CO2 - SSP2

Characterizations of the SSP2 scenario:

• Middle of the Road. This is the most adopted SSP foranalysis.

• Medium Technology Growth

• Environmental Awareness

• Medium Energy Demand

• Medium Economic Growth

• Medium Population Growth

8 / 31


Results of the decomposition of ∆CO2 - SSP2

0

.25

.5

.75

1

1.25

1.5

1.75

2

2.25

2.5

2.75

3

3.25

3.5

3.75

4

CO

2

EUROPE R5ASIA R5LAM R5MAF USA

LMDI - CO2 - 2030vs2010 - SSP20

.51

1.5

22.

53

3.5

4C

O2


LMDI - CO2 - 2050vs2010 - SSP2

0

.5

1

1.5

2

2.5

3

3.5

4C

O2


LMDI - CO2 - 2100vs2010 - SSP2

9 / 31


Contribution of each driver in explaining ∆CO2 - SSP2

10 / 31


What We Deduce from the LMDI Decomposition for SSP2

R5MAF and R5ASIA ∆CO2 emissions Over Time

In such framework, R5ASIA and R5MAF countries will experiencethe same ∆CO2 for the period 2010-2050. After 2050 a divergenceis shown, largely due to a greater impact of Population growth andGDP per capita on R5MAF carbon dioxide emissions.

Europe, USA and R5LAM ∆CO2 emissions Over Time

These 3 regions will once again manage, to keep CO2 emissionsconstant over time. This partially due to a slow population growthand GDP per capita.

Role played by each factor

Population and GDP per capita will affect positively CO2

emissions, while Energy Efficiency and Carbon Intensity will play anegative role in terms of carbon dioxide pollution.

11 / 31


General Conclusion I

• Over the period 2010 to 2050 R5ASIA and R5MAF willpollute more than EUROPE, R5LAM and USA. While from2050 to 2100 R5MAF will experience a divergence in CO2

Emissions, R5ASIA will keep CO2 emissions at the same levelsof 2050, this is mainly due to a decrease of population and aminor increase of GDP per capita.

• This can be also seen by taking the Average yearly GrowthRates of CO2

log(1+gCO2 ) = log(1+gPOP)+log(1+g GDPPOP

)+log(1+g PEGDP

)+log(1+g CO2PE

)

(4)

where gx =(

xt+n

xt

) 1n − 1

For small values of x we obtain

gCO2 = gPOP + g GDPPOP

+ g PEGDP

+ g CO2PE

(5)

12 / 31


Results Average Yearly Growth Rates

-1

-.75

-.5

-.25

0

.25

.5

.75

1

1.25

1.5

1.75

2

2.25

2.5

2.75

3

Ave

rage

Yea

rly C

O2

Var

iatio

ns %


Average Yearly Growth Rates - 2010-2050 - CO2 - SSP2

-1

-.75

-.5

-.25

0

.25

.5

.75

1

1.25

1.5

1.75

2

2.25

2.5

2.75

3

Ave

rage

Yea

rly C

O2

Var

iatio

n %


Average Yearly Growth Rates - 2050-2100 - CO2 - SSP2

13 / 31


General Conclusion II: Uncertainty Among Models

• The source of uncertainty among the 6 models considered wasassessed by using Principal Component Analysis

• Considerable degree of homogeneity among models. Howeverthe PCs associated to IMAGE and MESSAGE differ fromthose associated to WITCH, REMIND, GCAM and AIM

• Results of the PCA show a greater difference among modelsin assessing Energy Efficiency in 2020 with respect to 2050,suggesting that the assumptions that models make on PrimaryEnergy consumption are different, while converging over time.

14 / 31


Example: R5ASIA - SSP2 - Primary EnergyComponent/Total Primary Energy

15 / 31


Fossil Components

16 / 31


Thanks for your attention

Corresponding author

email: [email protected]

AcknowledgementThe research leading to these results has received funding from theEuropean Union Seventh Framework Programme (FP7/2007-2013)under grant agreement n◦ 308481 (ENTRACTE).

17 / 31


Appendix:

Results of Regional ∆CO2 for the rest of SSPs

18 / 31


Results of the Decomposition SSP1

-1

-.75

-.5

-.25

0

.25

.5

.75

1

CO

2


LMDI - CO2 - 2030vs2010 - SSP1

-.5

0

.5

1

1.5

2

2.5

3

CO

2


LMDI - CO2 - 2050vs2010 - SSP1

-.5

0

.5

1

1.5

2

2.5

3

CO

2


LMDI - CO2 - 2100vs2010 - SSP1

19 / 31


In depth: The contribution of each driver in explaining∆CO2 - SSP1

20 / 31



• Most Optimistic Scenario in terms of Challenges to Mitigationand Adaptation

• Rapid Technology

• High Environmental Awareness

• Low Energy Demand

• Medium-High Economic Growth

• Low Population

21 / 31


What We Deduce from the LMDI Decomposition for SSP1

Middle East and Africa ∆CO2 emissions Over Time

R5MAF countries will experience a greater increase in CO2

emissions over time than the other regions. This is largely due toan increase in GDP per capita.

Europe, USA and R5LAM ∆CO2 emissions Over Time

These 3 regions will manage, in this Scenario, to keep CO2

emissions constant over time. This partially due to a slowpopulation growth and GDP per capita. Low primary Energydemand will also play a large impact on pushing down CO2 forR5LAM.

Role played by each factor

Energy Efficiency will play the largest role in pushing CO2

emissions down over time, while carbon intensity effect will bemore slightly consistent over time

22 / 31


LMDI - ∆CO2 - Regional Comparison - Over Time - SSP3

0

.25

.5

.75

1

1.25

1.5

1.75

2

2.25

2.5

2.75

3

3.25

3.5

3.75

4

CO

2


LMDI - CO2 - 2030vs2010 - SSP3

0

.5

1

1.5

2

2.5

3

3.5

4

CO

2


LMDI - CO2 - 2050vs2010 - SSP3

0

.5

1

1.5

2

2.5

3

3.5

4C

O2


LMDI - CO2 - 2100vs2010 - SSP3

23 / 31



24 / 31



• Worst Scenario in terms of Economic Growth

• Slow Technology

• Little Environmental Awareness

• High Energy Demand

• Low/Negative Economic Growth

• Reduced Trade

• Very High Population

25 / 31



-1-.

75-.

5-.

250

.25

.5.7

51

CO

2


LMDI - CO2 - 2030vs2010 - SSP4

-.5

0

.5

1

1.5

2

CO

2


LMDI - CO2 - 2050vs2010 - SSP4

-.5

0.5

11.

52

CO

2


LMDI - CO2 - 2100vs2010 - SSP4

26 / 31



27 / 31



Notes: For SSP4, No Data Available for IMAGE, MESSAGE andREMIND.

• Low Level of Challenges to Mitigation, but High Level ofChallenges to Adaptation

• Slow Technology

• High Inequality

• Low Energy Demand

• Low Economic Growth

• High Population

28 / 31



0.25.5

.751

1.251.5

1.752

2.252.5

2.753

3.253.5

3.754

4.254.5

4.755

5.255.5

5.756

CO

2


LMDI - CO2 - 2030vs2010 - SSP50

.51

1.5

22.

53

3.5

44.

55

5.5

6C

O2


LMDI - CO2 - 2050vs2010 - SSP5

0

.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6

CO

2


LMDI - CO2 - 2100vs2010 - SSP5

29 / 31



30 / 31



Notes: For SSP5, No Data Available for IMAGE, MESSAGE andREMIND.

• Worst Scenario in terms of CO2 Emissions

• High Level of Challenges to Mitigation, but Low Level ofChallenges to Adaptation

• Rapid Technology for Fossil

• High Energy Demand

• High Economic Growth

• Low Population

31 / 31

A Multi-Model Regional Decomposition of CO2 Emissions ... · lnCO 2 t;i;j lnCO 2 0;i;j ln 0 @ CO2...

Documents

Transcript of A Multi-Model Regional Decomposition of CO2 Emissions ... · lnCO 2 t;i;j lnCO 2 0;i;j ln 0 @ CO2...