S6. Carbon and energy intensity reductions in climate policy cases

Will economic growth and fossil fuel scarcity help or
hinder climate stabilization? Overview of the RoSE
multi-model study
Electronic Supplementary Material 1 –
Extended Analysis, Tables and Figures
Elmar Kriegler1,*, Ioanna Mouratiadou1, Gunnar Luderer1, Nico Bauer1, Robert J. Brecha1,2,
Katherine Calvin3, Enrica De Cian4, Jae Edmonds3, Jiang Kejun5, Massimo Tavoni4, Ottmar
Edenhofer1,6,7
1
Potsdam Institute for Climate Impact Research, Potsdam, Germany
2
University of Dayton, Dayton, USA
3
Pacific Northwest National Laboratory, Joint Global Change Research Institute at the University of
Maryland, College Park, Maryland, United States
4
Fondazione Eni Enrico Mattei and Euro-Mediterranean Center on Climate Change, Milan,
Italy
5
Energy Research Institute, National Development and Reform Commission, Beijing, China
6
Technische Universität Berlin, Berlin, Germany
7
Mercator Research Institute on Global Commons and Climate Change, Berlin, Germany
* Corresponding author: [email protected]
S1. Population and GDP as emissions drivers in the baselines
Figure S1: (a) Global population
projections adopted by the models in the
HI Pop (upper solid line) and all other
scenarios (lower solid line), (b) global
average GDP per capita (in market
exchange rates, MER) over time, and
(c) per capita CO2 emissions from fossil
fuels and industry as a function of GDP
per capita under four different
assumptions of economic growth and
population (see Panel c for legend). The
FS Gr SL Conv is not shown here
because it was not calculated by all
models. The funnels are spanned by
results from GCAM, REMIND and
WITCH (identical in the case of
population), while IPAC results are
shown separately by dotted lines because
of larger differences in final energy and
emissions projections. All results are
given for the baseline scenarios. Dashed
black lines show the 5th to 95th percentile
range of the full century baseline
scenarios in the AR5 database.
Extended analysis: Gross world
product and GDP per capita
estimates in IPAC differ slightly
from the other models in MER
terms due to coarser regional
resolution affecting the aggregate
effect of regional PPP to MER
conversion (ESM-2 Section 3.3).
IPAC foresees stronger trend
breaks in energy and carbon
intensity improving technological
progress between the first and
second half of the century.
S2. Final energy intensity as a function of economic growth
Figure S2: Final energy intensity over GDP per capita (in MER terms) for different assumptions about
economic growth (see legend for scenario color codes). Results are given for the baseline scenario. Model
trajectories are shown in dashed lines and assigned to individual models by the use of their first letter as marker
for the points in 2010, 2050 and 2100. Trajectories run from top left to bottom right over time, i.e. from high
energy intensity and low GDP per capita to low energy intensity and high GDP per capita. Dotted isolines
indicate final energy demand levels of 20 to 140 GJ per capita and year (in steps of 40 GJ).
Extended analysis: It can be seen that increases in GDP per capita levels overcompensate
energy intensity improvements in almost all cases. The exception is IPAC in the slow growth
cases, because IPAC assumes strong energy intensity improving technological progress
particularly in the second half of the century. A constant income elasticity of final energy
demand, ɛ, would show as a straight line with slope ɛ-1 in the loglog plot format chosen
above. Models show some deviations from a straight line, but in the large majority of cases
project an income elasticity of 0 < ɛ < 1 throughout the century.
S3. Energy and carbon intensities in the baselines
Figure S3: Carbon vs. energy intensity for all models and various assumptions about economic growth and
population (Panel a) and fossil fuel availability (Panel b; indicated by marker colors). Model trajectories are
shown in dashed lines and assigned to individual models by use of their first letter as marker for the starting
point in 2010 and endpoint in 2100. Trajectories run from right to left, i.e. from high to low final energy
intensity over time. Dotted curves in the carbon vs. energy intensity graph show isolines of 100 to 800 gCO 2
emissions per unit GDP (in steps of 100 gCO2). All results are given for the baseline scenarios.
S4. Cumulative fossil fuel use and fossil fuel prices
Model
GCAM
IPAC
REMIND
WITCH
Policy Case
Baseline
550 ppm
450 ppm
Baseline
550 ppm
450 ppm
Baseline
550 ppm
450 ppm
Baseline
550 ppm
450 ppm
Cumulative Extraction of Fossil Fuels (2010-2100) [ZJ]
Coal
Oil
Gas
Total
26-38
12-21
18-23
59-72
12-16
12-17
17-19
42-46
8-11
10-13
14-15
34-37
20-30
15-16
11-13
46-58
7-9
11
10-11
29-31
4
8
7-8
19-21
19-41
14-31
19-35
54-84
3-11
14-22
17-22
41-44
2-5
12-16
12-14
28-30
18-23
22-36
21-23
61-79
8-9
18-20
10-12
38
7
10-11
8-9
26
Table S1: Ranges of cumulative fossil fuel use across different assumptions about fossil fuel availability as
estimated by the four global models in the RoSE study for the baseline, 550 ppm and 450 ppm CO2e scenarios.
Extended analysis: Oil and gas use are predominantly supply driven, with lowest use
observed in the LO Fos scenario (12 ZJ over 2011-2100 for oil and 18 ZJ for gas in GCAM,
excluding IPAC) and highest use in the HI Fos scenario for oil (36 ZJ in WITCH) and the LO
Oil scenario for gas (where gas and coal are as plentiful as in HI Fos, but oil limited to LO
Fos; 35 ZJ in REMIND). IPAC constitutes an exception as it projects a demand driven
constraint on future cumulative oil use due to large technological progress in the transport
sector leading to a partial replacement of oil-based transportation fuels even in the baseline.
In contrast, the use of coal as the lowest grade fossil fuel can be strongly affected by the
availability of oil and gas, with only slightly higher use in the HI Fos (21-29 ZJ) than in the
LO Fos scenario (18-26 ZJ; Fig. S4c). Significantly higher coal use is observed in a situation
with plentiful coal but limited oil and gas resources (HI Coal scenario: up to 41 ZJ in
REMIND). IPAC and WITCH are exceptions as they do not project coal use to vary
significantly between the HI Fos and HI Coal scenarios. For WITCH, this is mostly due to a
limited representation of coal use in non-electric sectors. For IPAC, this can be traced back to
the favorable economics of coal vs. gas use even in the HI Fos case (see Fig. 4b,c).
Concerning the dependence of fossil fuel prices on cumulative extraction, model differences
are sufficiently constrained for oil, where the low (LO Fos), medium (DEF) and high
availability (HI Fos) of oil resources dominates the increase in global oil prices as a function
of cumulative oil extraction (Fig. S4a). For gas (Fig. S4b) and coal (Fig. S4b), this is only
true for the cumulative extraction-price relationships in GCAM and REMIND. WITCH
shows consistently lower prices, lower price variations, and lower variations of coal and gas
extraction, which can be explained by the limited uses of these fuels outside the power sector
in the model version used in the RoSE study. IPAC shows consistently higher price increases,
particularly for gas, and as a result significantly lower gas use than the other models. It
apparently assumed lower gas supply in the economically more accessible resource grades.
Figure S4: Fossil fuel prices (indexed to
2010) as a function of cumulative
extraction for (a) oil, (b) gas, and
(c) coal for all models and various
assumptions about fossil fuel availability
(indicated by marker colors). All panels
include the LO Fos, DEF, and HI Fos
assumptions, while the panel on coal
prices also includes the HI Coal case
(favoring coal over oil and gas), and the
panel on gas prices the LO Oil case
(favoring gas over oil). Model
trajectories are shown in dashed lines
and assigned to individual models by use
of their first letter as marker for the
endpoint in 2100. All results are given
for the baseline scenario.
S5. Emissions, climate forcing and global mean warming
Figure S5: (a) Global Kyoto gas
emissions (dashed line at 2010 emissions
level), (b) total anthropogenic radiative
forcing (dashed lines show 450 and 550
ppm CO2 equivalent forcing), and (c)
global mean temperature increase since
preindustrial (dashed line at 2 degrees
Celsius) for the baseline, 550 ppm and
450 ppm CO2e climate policy cases.
Darker colors show the funnel spanned
by REMIND, GCAM and WITCH (solid
lines) under default assumptions (DEF)
while lighter colors the full range across
all economic growth and fossil fuel
availability assumptions (for the baseline
only, the two dimensions are separated
into light grey (economic growth) and
light blue (fossil resource) funnels).
IPAC did not model the climate
response in the RoSE study.
Kyoto gas emissions include the longlived GHGs controlled under the Kyoto
protocol (CO2, CH4, N2O, HFCs, and
SF6) and are calculated in CO2equivalent terms using 100 year global
warming potentials.
S6. Carbon and energy intensity reductions in climate policy cases
Figure S6: Carbon intensity vs energy intensity reductions relative to the baseline in the 550 ppm (Panel a) and
450 ppm CO2e (Panel b) climate policy cases for all models, economic growth and fossil resource assumptions.
Model trajectories are shown as dotted lines with markers at the end point in 2100, starting in the upper right
corner at 2005. Markers indicate model (by letter) and scenario (by color; see legend). Isolines show
hypothetical 0%, 20%, …, 100%, 120% emissions reductions relative to baseline assuming that GDP stays
roughly constant.
S7. Primary energy supply
Climate policy induces a clear transformation away from coal and towards non-fossil fuel
sources. In the climate policy cases, the structure of transformation remains largely
unchanged by economic growth (below) and fossil resource variations (next page).
Figure S7a: Sensitivity of primary energy projections to socio-economic assumptions. Shown is world
primary energy supply by source in 2050 (left column) and 2100 (right column) in the baseline (top row) and the
550 ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of
electricity projections across models and different assumptions about global economic growth and population.
Figure S7b: Sensitivity of primary energy projections to assumptions about fossil fuel availability. Shown
is world primary energy supply by source in 2050 (left column) and 2100 (right column) in the baseline (top
row) and the 550 ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the
variation of electricity mix projections across models and different assumptions about fossil fuel availability.
S8. Electricity generation
The electricity mix is characterized by strong structural transformations with distinct model
patterns, in both baseline and policy cases. In the climate policy cases, the effect of economic
growth (below) and fossil resource variations (next page) are small.
Figure S8a: Sensitivity of electricity use projections to socio-economic assumptions. Shown is world
electricity generation by source in 2050 (left column) and 2100 (right column) in the baseline (top row) and the
550 ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of
electricity projections across models and different assumptions about global economic growth and population.
Figure S8b: Sensitivity of electricity use projections to assumptions about fossil fuel availability. Shown is
world electricity generation by source in 2050 (left column) and 2100 (right column) in the baseline (top row)
and the 550 ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of
electricity mix projections across models and different assumptions about fossil fuel availability.
S9. Final energy demand
The electricity mix is affected by model differences and economic growth assumptions
(below). It is largely unaffected by fossil resource variations (next page).
Figure S9a: Sensitivity of final energy projections to socio-economic assumptions. Shown is world final
energy use by energy type in 2050 (left column) and 2100 (right column) in the baseline (top row) and the 550
ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of electricity
projections across models and different assumptions about global economic growth and population.
Figure S9b: Sensitivity of final energy projections to assumptions about fossil fuel availability. Shown is
world final energy use by type in 2050 (left column) and 2100 (right column) in the baseline (top row) and the
550 ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of
electricity mix projections across models and different assumptions about fossil fuel availability.
S10. Greenhouse gas emissions
Figure S10a: Sensitivity of emissions projections to socio-economic assumptions. Shown are global Kyoto
Gas emissions split into CO2 emissions from fossil fuel combustion and industry and the remaining emissions
(in CO2equivalent) in 2050 (left column) and 2100 (right column) in the baseline (top row) and the 550 ppm
(middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of emissions
projections across models and different assumptions about global economic growth and population.
Figure S10b: Sensitivity of emissions projections to assumptions about fossil fuel availability. Shown are
global Kyoto Gas emissions split into CO2 emissions from fossil fuel combustion and industry and the
remaining emissions (in CO2equivalent) in 2050 (left column) and 2100 (right column) in the baseline (top row)
and the 550 ppm (middle row) and 450 ppm CO2e (bottom row) policy cases. Each panel shows the variation of
emissions projections across models and different assumptions about fossil fuel availability.
S11. Carbon prices and mitigation costs
Carbon prices and mitigation costs for the 550 ppm CO2e target are signficiantly lower than
for the 450 ppm CO2e target (cmp. Fig. 4 in the main paper).
Figure S11: : (a+b) Net present value mitigation costs (discounted at 5% per year) over the period 2010-2100
(consumption losses in percentage net present consumption for ReMIND and WITCH, area under MAC in
percentage net present output for GCAM) and (c+d) carbon prices in 2050 for the 550 ppm CO2e target. IPAC
did not report carbon prices and mitigation costs.
Mitigation cost estimates vary with time and their net present value over the 21st century is
sensitive to the choice of discount rate. Fig. S12 shows that the mitigation cost patterns
discussed in the main paper are robust against such choices.
Figure S12: Sensitivity of mitigation cost estimates to the choice of metric. Shown are net present value
mitigation costs over the period 2010-2100 discounted at 5% (Panel a+b; as in Figure 4 in the main text) and
discounted at 1% (Panel c+d) as well as actual mitigation costs in the year 2050 (Panel e+f) for the 450 ppm
CO2e target. Mitigation costs are calculated as consumption losses in percentage of consumption for ReMIND
and WITCH, and area under MAC in percentage of output for GCAM. The variation of mitigation costs across
economic growth (left panels a, c, e) and fossil resource assumptions (right panels b, d, f) are shown separately.