Selasa, 31 Maret 2015

Baseline time accounting: Considering global land use dynamics when estimating the climate impact of indirect land use change caused by biofuels

Jesper Hedal Kløverpris & Steffen Mueller
Received: 7 February 2012 / Accepted: 10 August 2012 / Published online: 11 September 2012
# Springer-Verlag 2012

Abstract
Purpose Current estimations of the climate impact from
indirect land use change (ILUC) caused by biofuels are
heavily influenced by assumptions regarding the biofuel
production period. The purpose of this paper is to propose
a new method (baseline time accounting) that takes global
land use dynamics into account that is consistent with the
global warming potential, that is applicable to any phenomenon
causing land use change, and that is independent of
production period assumptions.
Methods We consider ILUC in two forms. The first is called
“accelerated expansion” and concerns ILUC in regions with
an expanding agricultural area. The second is called
“delayed reversion” and concerns ILUC in regions with a
decreasing agricultural area. We use recent trends in international
land use and projections of future land use change
to assess how ILUC from biofuels will alter the development
in global agricultural land use dynamics compared to
the existing trend (i.e., the baseline development). We then
use the definition of the global warming potential to determine
the CO2 equivalence of the change in land use
dynamics.
Results and discussion We apply baseline time accounting
to two existing ILUC studies in the literature. With current
trends in global agricultural land use, the method significantly
reduces the estimated climate impact in the previous
ILUC studies (by more than half). Sensitivity analyses show
that results are somewhat sensitive to assumptions regarding
carbon sequestration and assumptions regarding postreversion
ecosystems.
Conclusions The global dynamic development in land use
has important implications for the time accounting step
when estimating the climate impact of ILUC caused by
biofuel production or other issues affecting land use.
Ignoring this may lead to erroneous conclusions about the
actual climate impact of ILUC. Several land use projections
indicate that the global agricultural area will keep expanding
up to and beyond 2050. We therefore recommend to apply
the baseline time accounting concept as an integrated part of
future ILUC studies and to update the results on a regular
basis.
Keywords Biofuels . Bioethanol . Corn ethanol . ILUC .
Indirect land use change . Time accounting
1 Introduction
Liquid biofuels provide a renewable alternative to fossil
fuels in the transportation sector. For instance, cornbased
ethanol can replace conventional gasoline, which
causes average greenhouse gas (GHG) emissions of
roughly 95 g CO2 equivalents per megajoule of fuel
(g CO2e/MJ) when considering the entire life cycle from
crude oil extraction through refining and combustion
(Argonne National Laboratory 2010a). For comparison,
Liska et al. (2009) estimated GHG emissions from US
corn-based ethanol to be 38–48 g CO2e/MJ. The
Responsible editor: Llorenc Milà i Canals
Electronic supplementary material The online version of this article
(doi:10.1007/s11367-012-0488-6) contains supplementary material,
which is available to authorized users.
J. H. Kløverpris (*)
Novozymes A/S,
Krogshøjvej 36,
2880 Bagsværd, Denmark
e-mail: jklp@novozymes.com
S. Mueller
Energy Resources Center (MC 156),
University of Illinois at Chicago,
1309 South Halsted Street, Room 208,
Chicago, IL 60607, USA
Int J Life Cycle Assess (2013) 18:319–330
DOI 10.1007/s11367-012-0488-6
corresponding number in the GREET Model version
1.8d (Argonne National Laboratory 2010b) is 53 g
CO2e/MJ (Argonne National Laboratory 2010a). This
includes the energy efficiency improvements in ethanol
production documented by Mueller (2010). Based on
these numbers, substituting conventional gasoline with
corn-based ethanol provides a significant potential for
reduction of GHG emissions. However, emissions from
so-called indirect land use change (ILUC) have not been
considered in this comparison. ILUC may occur when
existing agricultural land previously used for food or
feed production is devoted to the production of biofuel
feedstocks. Simply speaking, the resulting drop in the
supply of feed or food can cause a relative increase in
agricultural prices, which could provide incentives to
increase production elsewhere (Kløverpris et al. 2008).
To some extent, this production increase may come
from conversion of new land to agricultural land, and
this may result in GHG emissions, e.g. from forest
clearing. To estimate the GHG emissions from ILUC
and relate it to one unit of biofuels, it is necessary to
estimate the amount of land indirectly affected, the
types of land indirectly affected (grassland, forest,
etc.), and the greenhouse gas emissions from these land
types resulting from their conversion to agriculture.
Finally, the GHG emissions from ILUC must be allocated
to the volume of biofuels produced. This step is
most often referred to as “time accounting” because the
initial ILUC emissions must be ascribed to the subsequent
biofuel production, which may take place during
several decades. Searchinger et al. (2008) were the first
to come up with a full analysis of ILUC emissions from
biofuels. Since then, several other researchers have refined
and improved ILUC modeling, e.g. Hertel et al.
(2010). However, the time accounting approach applied
by Searchinger et al. has somewhat set the standard for
many other ILUC studies. Searchinger et al. simply distributed
the ILUC emissions over 30 years of corn ethanol production.
This is also known as the annualization method. The choice of
a 30-year accounting period is clearly somewhat arbitrary and,
at the same time, has significant implications for the results: If
the accounting period is doubled, the ILUC emissions are
halved and vice versa. Furthermore, the annualization method
does not consider what would have happened to the land
indirectly affected by biofuel production in a scenario without
the biofuels. Instead, it is implicitly assumed that the global
agricultural area would remain constant without biofuels and
global land use dynamics are thereby ignored.1 A more
sophisticated approach to the time accounting issue is
clearly a much needed addition to the scientific debate
about ILUC. The purpose of this paper is to present a time
accounting method that takes the shortcomings of the
annualization method into consideration and to illustrate
how this method would affect ILUC emissions calculated
in other studies. We will refer to the concept as “baseline
time accounting.”
This paper will mainly focus on biofuels, but the philosophy
behind baseline time accounting is applicable to any
phenomenon resulting in ILUC such as highways and buildings
on agricultural land, shifting from conventional to
organic farming (in case of changed yields), political setaside
programs, and everyday dietary choices made by the
consumer in the supermarket.
We will use the term “land use baseline” (or just baseline)
to describe the constant development in land use taking
place as a result of all other drivers than the specific subject
under study.
2 Methods
2.1 Current land use baseline trends
We begin by looking at the most recent trends in the land
use baseline. From 1998 to 2007,2 the cropland area in the
developed world (loosely defined as Europe, North
America, and Oceania) decreased steadily by a total of
4 % or roughly 2.2 million ha (Mha) per year on average
(FAOSTAT 2010). When comparing to data in Roques et al.
(2011), it appears these cropland changes were inversely
correlated with changes in idle cropland, but Europe and
the USA also saw an increase in forest cover during the
period described (FAO 2010; Smith et al. 2010).
The recent trends in cropland areas in the developed
world indicate that neither direct nor indirect effects from
biofuels have caused any (gross) expansion of the cropland
area in this part of the world from 1998 to 2007.
Instead, biofuel production has likely slowed down the
rate at which land has gone out of production due to
idling or other causes. We will later refer to this effect as
“delayed reversion”.
From 1998 to 2007, the developing world (loosely defined
as Africa, Asia, and Latin America) has seen a 5 %
increase in cropland area or roughly 4.9 Mha/year on average.
This has mainly been driven by changes in Africa and
Latin America (FAOSTAT 2010). See Electronic
Supplementary Material 1 (Section 1) for further details.
1 Interestingly, the baseline dynamics are often considered in the estimation
of the amount of land indirectly affected, which is explicitly
modeled as a net change, i.e., a change in relation to a baseline (see
e.g., Hertel et al. 2010).
2 The latest 10-year period for which global land use data were available
in the FAOSTAT database (FAOSTAT 2010) at the time of
writing.
320 Int J Life Cycle Assess (2013) 18:319–330
2.2 Baseline implications for time accounting
As shown in the previous section, the land use baseline is
not static as implicitly assumed in the annualization method
(used for time accounting in most existing studies of ILUC
emissions from biofuels). The fact that the baseline is dynamic
is actually what allows us to discard the annualization
method (and its dependency on a fixed biofuel production
period) and instead estimate global warming from ILUC
emissions specific to the year in which a given volume of
biofuels is produced. The issue of a dynamic baseline has
previously been addressed by Kløverpris et al. (2010).
However, this was in relation to land quality, not GHG
emissions (graphically illustrated in Fig. 3 in the previous
paper). The baseline time accounting concept builds on the
approach discussed by Kløverpris et al. (2010) and the
implications for the modeling of GHG emissions from
ILUC are discussed below. Whereas Kløverpris et al.
(2010) only dealt with marginal changes in agricultural land
use (typically adequate for product life cycle assessment),
we will show how baseline time accounting can also cover
larger changes in land use, which may occur as result of
biofuel policies or other policies affecting agricultural land
use, e.g. the EU set-aside policy.
2.2.1 Accelerated expansion
We begin by explaining ILUC in the form of accelerated
expansion. When biofuels cause ILUC in a region where
arable land use is already increasing, the additional
(indirect) land use effect will be to speed up the expansion
of the agricultural area. New land will thereby be taken into
production earlier than it otherwise would have been in the
baseline. If an existing crop field is used for biofuel feedstock
production during 1 year, another area of new land
will come into production somewhere else 1 year earlier
compared to the baseline. If the production of feedstock is
maintained in the next year (on the same existing agricultural
land), yet another area of remote new land will come
into production 1 year earlier than it otherwise would have
had and so forth. This is illustrated in Fig. 1, which shows
how two consecutive years of biofuel production each has
the same ILUC impact, namely to bring a given area (indicated
by a star) into production 1 year sooner than in the
baseline. Note that this yearly impact is independent of the
assumed production period as long as the baseline is
expanding.
Figure 1 illustrates a situation in which the subject under
study (in this case biofuel production) causes indirect land
use change, which does not exceed the annual baseline
expansion of the agricultural area. In that sense, Fig. 1 is
relevant to all marginal changes but also changes of a
significant size (seen in relation to baseline changes).
In case of a larger biofuel program, the accelerated expansion
(the area coming into production sooner than in the
baseline) may exceed the next year's baseline expansion
whereby some land comes into production not just 1 year
but two or more years earlier than in the baseline. We will
later show how the baseline time accounting methodology
can address this situation. Furthermore, ILUC could potentially
go beyond the expansion that would anyway have
occurred in the baseline, at some point. This situation is
discussed in Section 4.3.
2.2.2 Delayed reversion
The other form of ILUC is delayed reversion. In regions
where agricultural land use is declining, ILUC will slow this
process down. If an existing crop field is occupied for
biofuel feedstock production during 1 year, it will indirectly
postpone (by 1 year) the time by which a corresponding area
of agricultural land is going out of production (analogous to
“accelerated expansion”). Another year of biofuel feedstock
production (on the same existing agricultural land) would
have the same effect although it would now be another area
of agricultural land, which was delayed in its “retirement”
Agricultural
land
ILUC
B
Year 1
Biofuels feedstock production
ILUC
Year 2
Year 3
(No biofuels
production)
Legend
Agricultural frontier in baseline
New agricultural frontier due to ILUC
B
B
Agricultural frontier in previous year
*
*
* Area brought into production one year
earlier than in the baseline
Agricultural
land
Agricultural
land
Fig. 1 Accelerated expansion. In a region with an expanding agricultural
area, the indirect land use effect of biofuels produced elsewhere
will bring new land into production earlier (in this case 1 year earlier)
than in a baseline situation without the biofuels. This is illustrated with
2 years of biofuel production in the figure above. Note that each year
has the same ILUC impact (asterisk)
Int J Life Cycle Assess (2013) 18:319–330 321
from agriculture. This is graphically illustrated in Fig. 2.
Based on Section 2.1, we find it likely that the “retired” land
will go idle and start the early reversion to a natural state.
We therefore designate this type of ILUC delayed reversion.
Also in this case, the ILUC impact of several years of
biofuel production would be the same and assumptions
about the production period are unnecessary (assuming
yields and other parameters stay the same). Note that under
the assumption of a static baseline, delayed reversion would
not be a possibility because there would be no reversion (in
the baseline) to delay. This again shows why the baseline is
important.
2.3 Estimating an ILUC factor
As discussed above, ILUC will change the timing of land
use changes in the baseline. This means that these changes
would occur anyway, but they must still be accounted for.
We now move on to the global warming implications of the
time shift in GHG emissions caused by ILUC seen in
relation to a dynamic baseline. The earlier GHG emissions
in the case of accelerated expansion and the postponed
carbon sequestration in the case of delayed reversion cause
GHGs to be present in the atmosphere for a longer time than
in the baseline. During this additional time, they will cause
warming, which must be considered.
The GHG emissions from indirect land use change attributed
to one unit of biofuels is often referred to as “the
ILUC factor” (although it is in fact an addend, not a factor).
Since the ILUC factor is added to the direct emissions from
biofuel production, the unit of the ILUC factor must be
consistent with the direct emissions. These are typically
measured by their global warming potential seen over
100 years (here referred to as GWP100). The GWP100
expresses the CO2 emission, which would cause the same
cumulative radiative forcing (CRF) as a given radiative
effect during the accounting period of 100 years. There
has been some debate over the GWP concept (see e.g.,
Fuglestvedt et al. 2003; Shine 2009; Levasseur et al. 2010).
For instance, while a variety of arguments support 100 years
as a reasonable accounting period (Fearnside 2002), this time
period is basically arbitrary. It is, however, beyond the scope
of this paper to enter these discussions.We will simply accept
that the GWP100 is the common choice of metric for global
warming impacts and describe how an ILUC factor for accelerated
expansion and delayed reversion can be derived accordingly.
Note that we are not exchanging one arbitrary
choice (the assumed biofuel production period) for another
(the accounting period in the GWP concept). We are simply
ensuring consistency between the ILUC factor estimation and
the GWP concept.
The GWP100 is defined as the CRF of a radiative effect
during 100 years divided by the CRF of a pulse emission of
one unit of CO2 during the same period of time
(Ramaswamy et al. 2001). The GWP can be derived for
GHG emissions but also for other radiative effects, e.g.,
changes in albedo (Muñoz et al. 2010). In accordance with
the GWP100 definition, we will define the ILUC factor
under dynamic baseline conditions as the difference between
the CRF of the ILUC emissions and the CRF of the
baseline emissions3 during the same 100-year period seen
relative to the CRF of a pulse emission of one unit of CO2.
This can also be expressed as in the equation below:
ILUC factor ¼ CRFILUC CRFBaseline
CRFCO2
where CRF designates the cumulative radiative forcing over
the same period of 100 years. For a conceptual illustration of
Agricultural
land
B
Year 3
ILUC
Year 1
Year 2
(No biofuels
production)
B
*
Biofuels feedstock production
Legend
Agricultural frontier in baseline
New agricultural frontier due to ILUC
B
Agricultural frontier in previous year
* Area brought out of production one year
later than in the baseline
Agricultural
land
Agricultural
land
Year 0
(No biofuels
production)
Agricultural
land
*
ILUC
Fig. 2 Delayed reversion. In a region with a decreasing agricultural
area, the indirect land use effect of biofuels produced elsewhere will
delay the reversion of land going out of production (in this case by
1 year) compared to a baseline situation without the biofuels. This is
illustrated with 2 years of biofuel production in the figure above. Note
that each year of production has the same ILUC impact (asterisk)
3 The baseline emissions being equal to the ILUC emissions but
occurring 1 year later or more (see Table 1).
322 Int J Life Cycle Assess (2013) 18:319–330
CRFILUC and CRFBaseline, see Fig. 3. The method proposed
here for handling temporal shifts in GHG emissions is not
novel. A similar approach (the Lashof method) has previously
been published by Fearnside et al. (2000). For further
discussion, see Electronic Supplementary Material 1
(Section 2).
In order to calculate the CRF values for the land use
emissions in the ILUC and baseline scenario, we rely on a
spreadsheet-based climate model based on Forster et al.
(2007). The model provides yearly increased radiative forcing
based on emission data determined by the user. The
GHG emissions can be inserted in different years so a
temporal emissions profile for land use change can be established.
The model takes into account the atmospheric GHG
residence time.4 The CRF of ILUC emissions seen over a
100-year period can thereby be established as well as the
CRF of baseline emissions seen over the same period of
time. It is thus possible to calculate the ILUC factor relevant
for dynamic baseline conditions. In the following, we explain
how the GHG emissions from ILUC are inserted into
the spreadsheet model. The model itself is provided as an
electronic appendix in two versions (see Electronic
Supplementary Materials 2 and 3) illustrating the two case
studies described in Section 3.1.
ILUC from biofuel production (or other phenomena affecting
the demand for land) will typically be seen in both
the developed and the developing world (Searchinger et al.
2008; Hertel et al. 2010; Kløverpris et al. 2010). When the
ILUC occurring in the developing world is smaller than the
annual baseline expansion (E) and ILUC in the developed
world is smaller than the annual baseline reversion (R), the
emission profiles to be inserted in the spreadsheet climate
model can be expressed as in Table 1. The ILUC emissions
are assumed to take place instantaneously after the production
of the triggering biofuel production (year 1), whereas in
the baseline, the same emissions (from the same land) would
not have taken place until 1 year later (year 2), following the
logic explained in Section 2.2.1. For the developing world,
ILUC emissions are modeled as accelerated expansion with
GHG emissions from conversion of above- and belowground
biomass (a) and the same emissions in the baseline,
only happening 1 year later (see Section 2.2.2). For simplicity,
above- and below-ground carbon emissions are assumed
to occur instantaneously. Distributing these emissions more
realistically over several years would only have a minor
influence on results since the time shift in the emissions
profile would still be the same. Carbon sequestration could
be included during the first year of the baseline emissions
profile (see Table 1), but, as the carbon sequestered during
this year is released as part of the subsequent years' emissions
from above- and below-ground biomass, the baseline
carbon sequestration in year 1 has been omitted for the
developing world.5 For the developed world, ILUC emissions
are modeled as delayed reversion, which means that
the land in the baseline will sequester carbon for an additional
year compared to the biofuel (ILUC) scenario (see
Table 1). In principle, carbon sequestration in both the ILUC
and the baseline situation could be included from year 2 and
onwards. However, these emissions would simply cancel
out each other and thereby not influence the result.
So far, we have only considered a situation in which the
studied change (biofuel production) does not cause ILUC,
which exceeds the annual rate of land use change in the
baseline (cf. Fig. 1). This will always be the case when
studying marginal changes in land use. When looking at
larger changes where ILUC in the developing world does
exceed annual baseline expansion (E), some of the accelerated
expansion will happen more than 1 year in advance of
the baseline expansion. In that case, the emission profiles in
Table 2 should be used in the climate model (see Electronic
Supplementary Material 1, Section 3.1 for details).
If ILUC exceeds the annual baseline reversion in the
developing world, 1 year of reversion will be delayed, and
the net land use change exceeding the annual baseline
reversion will consist of expansion into an area that recently
“reverted.” However, that is only in the first year of biofuel
production. In the next year, biofuel production will delay
the reversion of the area that was “reconverted” due to the
first year of biofuel production. The emissions profiles for
the developed world in Table 1 are therefore relevant for all
4 The significance of this was also discussed by O'Hare et al. (2009).
5 An alternative approach could be to include baseline carbon sequestration
in year 1. and then subtract it from the above- and below-ground
emissions in year 2. However, this would only have a negligible
influence on results and therefore we chose the other simpler approach.
CRFILUC
Legend
Radiative forcing (RF) from ILUC
RF from baseline land use change
CRFBaseline
Radiative forcing (W/m2)
1 100
Time (y)
Fig. 3 Radiative forcing from land use emissions. Conceptual illustration
(not to scale) of radiative forcing from land use change in a
situation where ILUC causes land to come into production 1 year
earlier than in the baseline. The radiative forcing decreases after the
land use change because of GHG removal from the atmosphere caused
by different mechanisms
Int J Life Cycle Assess (2013) 18:319–330 323
years of production except the first, which has been given
further consideration in Electronic Supplementary Material
1 (Section 3.2).
2.4 Additional methodological elaboration
The method proposed in this paper has been discussed
extensively prior to publication (at conferences, in working
groups, during the review process, and in other academic
forums). Particularly one key question has always been at
the center of the debate: What if biofuel production continues
or, in other words, will biofuels not, at some point, lead
to land use expansion on top of maximum baseline expansion?
Since this question is key to understanding the baseline
time accounting concept, it will receive special attention
in this section. First of all, we stress that the baseline time
accounting concept is not directly applicable when/if the
object of study (in this case biofuels) leads to agricultural
expansion beyond what would have occurred anyway in the
baseline. We will later refer to such expansion as “additional
expansion” (as opposed to accelerated expansion). Secondly,
whether biofuels lead to additional expansion depends on the
duration of their production and the baseline. Thirdly, if biofuel
production should not stop, it is certainly wrong to
assume a production period of 30 years. Interestingly,
Searchinger et al. (2008) chose a 30-year production period
partly because they found that “ethanol is typically viewed as
a bridge to more transformative energy technologies.” If
Searchinger and colleagues are right, all (or at least a substantial
part) of the agricultural expansion possibly caused
(indirectly) by biofuel production would most likely have
occurred at some point anyway. This illustrates the shortcomings
of the annualization method and the need for consideration
of the baseline.
The fundamental basis for the thinking behind the baseline
time accounting concept is the typical product-oriented
approach in consequential life cycle assessment. We ask the
question: What is the environmental impact, in this case the
climate impact specifically, of producing and using a given
product at a given time under the given circumstances
relevant at that time. It is well known that the circumstances
surrounding a system under study, e.g., market trends, may
have important implications for the LCA of a given product.
For a general discussion, see e.g., Ekvall and Weidema
(2004). The present paper basically considers how trends
in agricultural land markets influence the climate effect of
ILUC caused by a given product. When land use baseline
trends change, this will affect the estimation of the ILUC
factor, especially when looking at larger land use changes
(cf. Table 2). Note that the baseline time accounting method
does not rely on an a priori assumption about biofuel production
stopping sometime in the future (if biofuels is the
object of study). What the method does is to allow for an
estimation of the GWP100 from ILUC caused by the production
of a “land-consuming” product at a given point in
time−as long as the estimated ILUC does not lead to a
greater agricultural area than the baseline peak. For further
discussion of “additional expansion”, see Section 4.3.
Table 1 Land use emission profiles
Region Year → 1 2 3 4 5 …
Developing ILUC a − − − − −
Baseline −* a − − − −
Developed ILUC − − − − − −
Baseline −c − − − − −
The general form of ILUC and baseline emission profiles for the
developing world (accelerated expansion) and the developed world
(delayed reversion) when ILUC does not exceed annual baseline expansion
and reversion, respectively. The unit of a and c is mass of
CO2e per functional unit, for biofuels typically grams of CO2e/
megajoule
a total above- and below- ground C emissions, c average annual carbon
sequestration
* See discussion in text about carbon sequestration
Table 2 Land use emission profiles
Year 1 2 3 n
ILUC a − − −
Baseline − If Ia≤E: a If Ia≤E: 0 If Ia≤(n−1)E: 0
If Ia>E: E/Ia×a If E<Ia≤2E: (Ia−E)/Ia×a If (n−2)E<Ia≤(n−1)E: (Ia−(n−2)E)/Ia×a
If Ia>2E: E/Ia×a If Ia>(n−1)E: E/Ia×a
The general form of ILUC and baseline emission profiles for the developing world (accelerated expansion) when ILUC exceeds annual baseline
expansion
a total above- and below-ground carbon emissions (mass of CO2e per functional unit, for biofuels typically ‘g CO2e/MJ’); Ia ILUC in the form of
accelerated expansion in the developing world (in million hectares); E annual baseline expansion in the developing world (assumed to currently be
4.9 Mha, based on Section 2.1)
324 Int J Life Cycle Assess (2013) 18:319–330
3 Results
3.1 Baseline time accounting applied to existing studies
To illustrate the influence of baseline time accounting on
existing ILUC factor results, we apply the concept on the
studies by Searchinger et al. (2008) and Hertel et al. (2010),
both relating to corn-based ethanol. We derive emissions
data directly from these studies. Above- and below-ground
emissions in the developing world (a) are estimated at 2,240
and 110 g CO2e/MJ for the Searchinger and Hertel study,
respectively (see detailed calculations in Electronic
Supplementary Material 1, Section 4). As for the carbon
sequestration values to be used for the delayed reversion
calculations, it is less straight forward to use the data in the
original studies. The reason is that the authors use foregone
sequestration values for mature ecosystems, which only
sequester relatively small amounts of carbon. To avoid
underestimation of the carbon sequestration delayed by
ILUC in the developed world, and, in order to consistently
use the data from the original studies, we exploit that these
studies estimate above- and below-ground carbon emissions
(a) from land conversion in the developed world. Based on
the assumption that the carbon lost upon conversion of land
will be resequestered within 100 years, we estimate the
annual average carbon sequestration delayed in the developed
world by dividing the above- and below-ground emissions
(a) with 100 (i.e., c 0 a/100, for the developed world).
This gives delayed carbon sequestration of 5.9 and 10.1 g
CO2e/MJ for the Searchinger and Hertel study, respectively.
There are two important reasons for dividing by 100 years.
First, it is consistent with GWP100. Second, it is generally
considered reasonable to assume that land reverting to a
natural state will have reached maximum carbon stock or
close-to-maximum carbon stock within 100 years. If this is
not the case, the baseline time accounting concept will
slightly underestimate the climate impact of delayed reversion.
On the other hand, if carbon sequestration is continuously
offset by management activities (e.g., plowing of idle
land to avoid full renaturalization) or not taking place at all
(e.g., due to development on the land), the concept will, to
some extent, overestimate the climate impact of delayed
reversion. Note that only the average sequestration rate for
1 year is required (c in Table 1), but, to estimate this number,
assumptions about future carbon sequestration over the
GWP100 accounting period are necessary. Further discussion
and detailed calculations are available in Electronic
Supplementary Material 1 (Section 4). See also sensitivity
analyses in Section 3.2.
The emission figures described in the previous paragraph
(above- and below-ground emissions in the developing
world and carbon sequestration emissions for the developed
world) are to be used in the climate model. In order to do so,
it is necessary to consider whether the changes in cropland
modeled by Searchinger et al. (2008) and Hertel et al. (2010)
exceed the changes in the baseline (cf. Section 2.3). We do
so by comparing to the ongoing changes in cropland area
described in Section 2.1 (−2.2 Mha/year in the developed
world and +4.9 Mha/year in the developing world). For the
Searchinger study, we find that ILUC in the developing
world exceeds ongoing expansion by roughly 70 %, and
we therefore apply Table 2 for establishing of emissions
profiles. These are shown in Table 3. For the Hertel study,
we find that ILUC in the developed world exceeds baseline
reversion by roughly 15 %, and we therefore consider the
special case of the first year of production. Further discussion
and detailed calculations are available in Electronic
Supplementary Material 1 (Section 4).
Table 4 summarizes the results from applying the baseline
time accounting concept to existing ILUC studies (spreadsheet
calculations available in Electronic Supplementary
Materials 2 and 3). For the special case of the first year of
biofuel production (see last part of Section 2.3), an ILUC
factor of 31 g CO2e/MJ is estimated for the Hertel study
(because delayed reversion in the developed world exceeds
annual baseline reversion).
As shown in Table 4, the ILUC factors based on the
baseline time accounting concept (under current baseline
development in global agricultural land use) are significantly
lower (60–70 %) than the ILUC factors based on the 30-
year annualization method. The main reason is that the
baseline time accounting method is based on the additional
radiative forcing from a temporal shift in land conversion
caused by the production of a given quantity of biofuels
under the baseline conditions relevant at the time of production.
The background situation (baseline) is important because
it predicts the “alternative fate” of the land that is
allegedly brought into production as an indirect result of
biofuel production in a given year. By comparing ILUC
emissions from conversion of land at the agricultural frontier
with the emissions resulting from the baseline conversion
of the same land, the climate impact (ILUC factor) is
estimated. As implied by the equations in Table 2, the rate of
baseline expansion in the developing world (E) can have a
substantial impact on results. If baseline expansion is slow, a
large indirect land conversion (Ia in Table 2) may push land
into production more than 1 year in advance of the baseline
conversion (cf. Searchinger case in Table 3), which will
increase the ILUC factor.
3.2 Sensitivity analyses
Two main aspects of uncertainty could potentially influence
the results derived with the baseline time accounting concept,
both relating to delayed reversion. One is the uncertainty
relating to the average carbon sequestration per year,
Int J Life Cycle Assess (2013) 18:319–330 325
the other is the uncertainty relating to postreversion ecosystems
(whether retired cropland turns into forest, grassland,
or parking lots).
For the Searchinger study, we used the spreadsheet climate
model to investigate the sensitivity to the emissions
profile for carbon sequestration. We assumed that the same
amount of carbon was sequestered within the GWP accounting
period but at different speeds (within 1, 20, and
99 years). We found that this could change the result (the
contribution to the ILUC factor from delayed reversion)
within a range of +2 to −20 %. Thus, the shape of the
emissions profile for carbon sequestration has a significant
influence on the result but does not change it by orders of
magnitude. See Electronic Supplementary Material 1 for
details (Section 4).
For the Hertel study, our average annual (delayed) carbon
sequestration was based on an average of conversion to
forest and grassland (see Electronic Supplementary
Material 1, Section 4.2.2). Assuming reversion only to forest
gave a total ILUC factor of 18 g CO2e/MJ, whereas
reversion only to grassland gave a total ILUC factor of 4 g
CO2e/MJ (i.e. plus/minus two-thirds compared to the result
in Table 4). These numbers are both below the original
ILUC factor estimate (27 g CO2e/MJ) from Hertel et al.
(2010) but show that results are sensitive to postreversion
ecosystem assumptions. More details are available in
Electronic Supplementary Material 1 (Section 4.2.4).
4 Discussion
4.1 Future changes in the land use baseline
As implied by the baseline time accounting concept, it is the
dynamics of the land use baseline at the time of the biofuel
production, which determines the climate impact of indirect
land use change. The reason is that the baseline indicates
what would have happened to the land indirectly affected by
biofuels if the fuels had not been produced. The analysis of
current baseline trends in Section 2.1 justifies our approach
for current biofuel production but does not describe the
future conditions for baseline time accounting. We therefore
looked at several projections of future changes in global
agricultural land use (Fischer et al. 2002; Bruinsma 2003;
Alder et al. 2005; Bakkes and Bosch 2008; Stehfest et al.
2009; Bruinsma 2009; Fischer 2009) to assess the future
validity of the baseline time accounting concept.
These projections are highly driven by changes in world
population and consumers' dietary choices, but they differ in
several aspects such as yield assumptions, modeling framework,
considered land use types, etc. (see Electronic
Supplementary Material 1, Section 5). Some of them include
future biofuel production, whereas none of them consider
future production of biomaterials, which may increase
considerably in the future (King et al. 2010).
In some of the land use projections, the future trend in
agricultural land use in the developed world is negative
while positive in others. However, all of the studies agree
that the total agricultural area will increase up to 2030.
Furthermore, all of the studies, except one (Stehfest et al.
2009), agree that total agricultural land use will increase up
to and beyond 2050. This shows that baseline time accounting
can be applied for many years to come. It will, however,
be necessary to update ILUC factor calculations in the future
as baseline conditions and other aspects change. This is not
Table 3 Land use emissions (in
grams of CO2e/megajoule) as
modeled with the baseline time
accounting concept
Original study Region Year → 1 2 3 4 …
Searchinger et al. (2008) Developing ILUC 2,240 − − − −
Baseline − 1,322 918 − −
Developed ILUC − − − − −
Baseline −5.9 − − − −
Hertel et al. (2010) Developing ILUC 110 − − − −
Baseline − 110 − − −
Developed ILUC − − − − −
Baseline −10.1 − − − −
Table 4 Estimated ILUC factors
30 years annualization
(g CO2e/MJ)
Baseline time accounting
(g CO2e/MJ)
Searchinger et al. (2008)
Developing world 78 24
Developed world 26 6
Total 104 30
Hertel et al. (2010)
Developing world 3 1
Developed world 24 10
Total 27 11
326 Int J Life Cycle Assess (2013) 18:319–330
unique to the ILUC issue. Many carbon intensity values
(e.g., for fossil fuels) change over time.
4.2 Isolating the subject of study from the baseline
The purpose of the baseline time accounting concept is to
describe the climate impact of a change seen relative to a
dynamic baseline or, in other words, a change on top of all
other changes. In order to do so, it is necessary to develop a
reasonable estimate of the rate of change in the baseline (E
in Table 2 and C in Electronic Supplementary Material 1,
Section 3). If studying larger changes (e.g., US corn ethanol
production or the EU set-aside program), it is necessary to
make sure that the change itself is not counted when estimating
the underlying baseline changes.
This introduces some challenges in the case studies used
in the present paper. The baseline trends are estimated based
on changes in the global agricultural area from 1998 to
2007, but, during this period, the production of corn ethanol
had already begun. In 1998, US annual ethanol production
had slowly reached 5.3 gigaliters (Gl) or 9 % of the total
mandate of 56 Gl (RFA 2012). Potentially, this production
had already had an indirect effect on the global agricultural
area. In 2001, annual corn ethanol production in the US
reached 6.7 Gl and then increased more swiftly to 24.6 Gl in
2007 (RFA 2012). Due to the inertia of the global economy
and especially land markets, it is unlikely that the increased
production of ethanol in the last part of the period from 1998
to 2007 had any influence on the global agricultural area by
the end of the period. We did, however, consider the implications
of a “worst-case scenario” assuming an instantaneous
(indirect) effect on the global agricultural area from
the increase in annual US corn ethanol production up to
2007. This adjustment increased the ILUC factor for the
Searchinger study (estimated by use of the baseline time
accounting concept) from 30 to 50 g CO2e/MJ. Despite the
substantial increase in the result, it is still more than 50 %
lower than the ILUC factor estimated by Searchinger et al.
(2008) with 30-year annualization. For the Hertel study
(estimating a much lower cropland increase in the developing
world), the adjustment of annual baseline expansion in
the developing world had no influence on results. For
details, see Electronic Supplementary Material 1, Sections
4.1.5 and 4.2.5.
4.3 Additional expansion
At some point, the global agricultural area will stop expanding
and potentially start to gradually contract. This is, however,
not foreseen within the temporal scope of the land use
projections mentioned in Section 4.1, with the exception of
Stehfest et al. (2009) who indicate that global agricultural
land use contraction might occur already around 2030.
When approaching such a tipping point, biofuel production
could (indirectly) cause expansion that otherwise would not
have occurred. We designate this additional expansion (as
opposed to accelerated expansion). When considering a
marginal change, additional expansion would only occur
in the exact moment when the baseline shifts from expansion
to contraction (the tipping point) and therefore additional
expansion would not be of relevance in practice
(consider a marginal change in relation to the dotted line
in Fig. 4). Additional expansion might, however, become
relevant when studying larger changes extending many
years into the future. This is because a large change (maintained
continuously) will cause additional expansion when
approaching the time of the tipping point (see full line in
Fig. 4).
Note that only at the exact time of the tipping point the
indirect land use change would consist fully of additional
expansion. The annualization method implicitly assumes
this stage as being permanent (by implicitly assuming a flat,
nondynamic baseline). As indicated in Fig. 4, the indirect
land use change will, in fact, consist of a mix of accelerated
and additional expansion when approaching the baseline
tipping point from the left. On the immediate right of the
tipping point, indirect land use change will consist of a mix
of additional expansion and delayed reversion.
In the case of biofuels causing additional expansion,
there is no longer any baseline land use conversion to
compare to, and baseline time accounting can no longer be
applied directly as described above. Based on the review of
land use projections in Section 4.1, additional expansion
seems unlikely in the near future. We will, however, add
some additional considerations to how the baseline time
accounting concept could potentially be extended to also
address additional expansion (in combination with accelerated
expansion and delayed reversion). The climate model
in Electronic Supplementary Materials 1 and 2 could be
used to model a multiyear production scenario for biofuels.
This obviously means that production period assumptions
become unavoidable (as in the annualization method), but
this approach would allow for a rigorous and consistent
Accelerated
expansion
Area
Time
Delayed
reversion
Additional expansion
Tipping point
Fig. 4 Additional expansion. Conceptual illustration of how a larger
change in land use (full curve) may cause “additional expansion”
around the tipping point of the land use baseline (dotted line)
Int J Life Cycle Assess (2013) 18:319–330 327
incorporation of baseline implications (as opposed to the
annualization method). Note that in case of additional expansion,
land indirectly affected by biofuels (or another
subject of study) would not have come into production in
the baseline and must therefore be assumed to go out of
production and start reversion after termination of its
(indirect) cause of conversion. Note also that the approach
outlined above would, conceptually, be in consistence with
the method proposed by Levasseur et al. (2010) in which a
dynamic inventory of GHG emissions is computed (further
info given in Section 4.5). We will leave further considerations
for future research.
4.4 Flat baselines
In some parts of the world, little change in the agricultural
area is observed, and little change is projected despite the
increasing pressure on global land resources caused by
population growth, changing diets, etc. If biofuel production
indirectly leads to expansion of agricultural land use in a
region with a flat baseline, it could be considered additional
expansion to which the baseline time accounting does not
apply (see previous section). Meanwhile, it is unlikely that
biofuels will cause ILUC in a region which is otherwise
unaffected by the increasing pressure from the global market.
It is therefore important to be critical towards ILUC
modeling predicting land use changes in regions with flat
land use baselines. For instance, the GTAP Model applied
by Hertel et al. (2010) predicts a 0.4-Mha increase in the
Canadian croplands as a result of US corn ethanol
production (more than 10 % of the estimated ILUC),
but the Canadian cropland area has been relatively constant
since 1980.
4.5 Comparison to other time accounting methodologies
The present paper has compared the baseline time accounting
concept to the annualization method as applied by
Searchinger et al. (2008) and Hertel et al. (2010). Other
studies have also addressed the challenges of properly accounting
for time in GHG analyses of systems with carbon
stock changes. In this section, key studies in the literature
are briefly compared to the baseline time accounting
concept.
O'Hare et al. (2009) looked at ILUC emissions from US
corn ethanol production. They also developed a spreadsheet
model, which could take the atmospheric residence time of
carbon emissions into account. This allowed them to consider
the warming impact of early emissions (from land use
change) compared to that of later emissions (from biofuel
production and avoided fossil fuel combustion). They concluded
that this would lead to a higher relative impact of the
ILUC emissions. Just as the annualization method, the
method presented by O'Hare et al. (2009) relies on an
assumed biofuel production period with significant influence
on results.
Kendall et al. (2009) also criticized the time accounting
method applied by Searchinger et al. (2008) for some of the
same reasons mentioned by O'Hare et al. (2009) and in the
present paper. Kendall et al. (2009) developed a time correcting
factor (TCF) to characterize the impact of a pulse
CO2 emission annualized over a given time horizon. The
TCF was also based on the residence time of GHG emissions,
and results were highly dependent on assumptions
regarding the production period of biofuels.
Levasseur et al. (2010) took a different approach and
developed time-dependent characterization factors to take
into account the timing of GHG emissions in LCA in order
to ensure consistent time frames. By using a biofuel case,
Levasseur et al. (2010) demonstrated how different analytic
time horizons had a big impact on the results. The authors
did not explore the impact of the biofuel production period,
but, if this had been changed, it would also have affected
results considerably with the method suggested by
Levasseur et al. (2010).
Cherubini et al. (2011) also studied emissions from bioenergy.
However, their focus was not on ILUC but on the
temporary release of biogenic carbon between feedstock
harvesting/combustion and feedstock regrowth. They argued
that the warming effect of the temporary release should
be accounted for, and, just as in the present study, the
authors took their point of departure in the GWP methodology
and developed global warming potentials for
biogenic GHG emissions from different crop and forestry
rotation systems.
Müller-Wenk and Brandão (2010) also studied the
release of biogenic carbon emissions suggesting a
slightly different approach than Cherubini et al. (2011).
The authors derived “fossil combustion-equivalent”
amounts of biogenic carbon transferred to the air from
different biomes as a result of land transformation and
land occupation. As in the present paper and several
other studies mentioned herein, Müller-Wenk and
Brandão (2010) build on the atmospheric residence time
of CO2, but they recommend an accounting period of
500 years to be used in their analysis.
All of the studies mentioned above build on radiative
forcing of GHG emissions, just as the concept presented in
the present paper. When converting GHG emissions at different
points in time to impacts, the different studies (including
the present study) have a lot of similarities. Seen in
this perspective, the suggested approach in the present paper
for estimating an ILUC factor is not the real novelty of
the paper. It is the findings in Section 2.2 on the
implications of a dynamic baseline that add new
insights to the analysis of ILUC.
328 Int J Life Cycle Assess (2013) 18:319–330
5 Conclusions
The global dynamic development in land use has important
implications for the time accounting step when
estimating the climate impact of ILUC caused by biofuel
production or other issues affecting land use. The
key element of the baseline time accounting concept is
to consider what would happen to the land indirectly
affected by biofuel production in a baseline situation
without the specific subject under study. If the land
indirectly affected would have been affected anyway at
a later stage, the omission of this aspect may lead to
erroneous conclusions about the actual climate impact of
ILUC. When applying the baseline time accounting
concept to existing biofuel studies, we find that the
estimated ILUC impact is significantly reduced under
current baseline conditions. The baseline time accounting
concept does not rely on an arbitrary choice of
biofuel production period. On the other hand, it requires
an understanding of the trend in baseline land use at the
time during which the biofuels are produced. Several
land use projections indicate that the global agricultural
area will keep expanding up to and beyond 2050. Our
recommendation is therefore to apply the baseline time
accounting concept as an integrated part of future ILUC
studies and to update the results on a regular basis just
as other estimations of carbon intensities are being
updated.
Acknowledgments We thank R. Plevin (University of California,
Berkeley) for providing background data for the study by Hertel
et al. (2010) and Elke Stehfest (Netherlands Environmental Assessment
Agency) for providing background data for the Bakkes
and Bosch (2008) and Stehfest et al. (2009) land use projections.
A special thanks to M. Persson (Chalmers University of Technology,
Sweden) for helping with the applied climate model and for
constructive critique. Finally, we are thankful to the many peers
who have engaged actively in the discussion of the baseline time
accounting concept, including the time accounting subgroup in the
expert workgroup on indirect land use change established by the
California Air Resources Board in 2010.

References
Alder J, Bennett E, Carpenter S, Christensen V, Foley J, Maerker M,
Masui T, Morita T, O’Neill B, Peterson G, Ringler C, Rosegrant
M, Schulze K (2005) Changes in ecosystem services and their
drivers across the scenarios. In: Carpenter SR, Pingali PL, Bennett
EM, Zurek MB (eds) Ecosystems and human well-being: scenarios,
Vol 2. Millennium ecosystem assessment, Island Press, Washington,
Covelo and London
Argonne National Laboratory (2010a) GREET (Greenhouse gases,
regulated emissions, and energy use in transportation) Model
1.8d.1, http://greet.es.anl.gov. Accessed September 2010
Argonne National Laboratory (2010b) Summary of Expansions and
Revisions in GREET1.8d Version, Systems Assessment Section,
Center for Transportation Research, Argonne National Laboratory,
memo available at http://greet.es.anl.gov
Bakkes JA, Bosch PR (eds) (2008) Background report to the
OECD environmental outlook to 2030. Overviews, details,
and methodology of model-based analysis (MNP Report
500113001, 2008, www.pbl.nl/en/publications/2008/Back
groundreporttotheOECDEnvironmentalOutlookto2030.html).
Accessed June 2010
Bruinsma (ed) (2003) World agriculture: Towards 2015/2030. An FAO
Perspective. Earthscan, London
Bruinsma J (2009) The resource outlook to 2050. Paper presented at
FAO Expert meeting on how to feed the world in 2050, Rome,
24–26 June 2009
Cherubini F, Peters GP, Berntsen T, Strømman AH, Hertwich E (2011)
CO2 emissions from biomass combustion for bioenergy: atmospheric
decay and contribution to global warming. GCB Bioenergy
3:413–426
Ekvall T, Weidema BP (2004) System boundaries and input data in
consequential life cycle inventory analysis. Int J Life Cycle Assess
9:161–171
FAO (2010) Global forest resources assessment 2010—key findings.
Food and Agriculture Organization of the United
Nations, Rome
FAOSTAT (2010) United Nations Food and Agricultural Organisation,
http://faostat.fao.org. Accessed October 2010
Fearnside PM (2002) Why a 100-year time horizon should be used for
global warming mitigation calculations. Mitig Adapt Strateg Glob
Chang 7:19–30
Fearnside PM, Lashof DA, Moura-Costa P (2000) Accounting for time
in mitigating global warming through land-use change and forestry.
Mitig Adapt Strateg Glob Chang 5:239–270
Fischer G (2009) World food and agriculture to 2030/50. Paper presented
at FAO Expert meeting on how to feed the world in 2050,
Rome, 24–26 June 2009
Fischer G, Shah M, van Velthuizen H (2002) Climate change and
agricultural vulnerability. Remaprint, Vienna
Forster P, Ramaswamy V, Artaxo P, Berntsen T, Betts R, Fahey DW,
Haywood J, Lean J, Lowe DC, Myhre G, Nganga J, Prinn R, Raga
G, Schulz M, van Dorland R (2007) Changes in atmospheric
constituents and in radiative forcing. In: Solomon S, Qin D,
Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller
HL (eds) Climate Change 2007: the physical science basis. Contribution
of Working Group I to the Fourth Assessment Report of
the Intergovernmental Panel on Climate Change. Cambridge University
Press, Cambridge and New York
Fuglestvedt JS et al (2003)Metrics of climate change: assessing radiative
forcing and emission indices. Clim Chang 58:267–331
Hertel TW et al (2010) Global land use and greenhouse gas emissions
impacts of U.S. maize ethanol: estimating market-mediated
responses. Bioscience 60:223–231
Kendall A, Chang B, Sharpe B (2009) Accounting for time-dependent
effects in biofuel life cycle greenhouse gas emissions calculations.
Environ Sci Technol 43:7142–7147
King DA, Inderwildi OR, Williams A, Hagan A (eds) (2010) The
future of industrial biorefineries. World Economic Forum,
Cologny/Geneva
Kløverpris J, Wenzel H, Nielsen PH (2008) Life cycle inventory
modeling of land use induced by crop consumption. Part 1:
Conceptual analysis and methodological proposal. Int J Life Cycle
Assess 13:13–21
Kløverpris JH, Baltzer K, Nielsen PH (2010) Life cycle inventory
modelling of land use induced by crop consumption. Part 2:
Example of wheat consumption in Brazil, China, Denmark and
the USA. Int J Life Cycle Assess 15:90–103
Levasseur A, Lesage P, Margni M, Deschênes L, Samson R (2010)
Considering time in LCA: dynamic LCA and its application to
Int J Life Cycle Assess (2013) 18:319–330 329
global warming impact assessments. Environ Sci Technol
44:3169–3174
Liska AJ et al (2009) Improvements in life cycle energy efficiency and
greenhouse gas emissions of corn-ethanol. J Ind Ecol 13:58–74
Mueller S (2010) 2008 National dry mill corn ethanol survey. Biotechnol
Lett 32:1261–1264
Müller-Wenk R, Brandão M (2010) Climatic impact of land use in
LCA—carbon transfers between vegetation/soil and air. Int J Life
Cycle Assess 15:172–182
Muñoz I, Campra P, Fernández-Alba AR (2010) Including CO2-
emission equivalence of changes in land surface albedo in
life cycle assessment. Methodology and case study on greenhouse
agriculture. Int J Life Cycle Assess 15:672–681
O'Hare M et al (2009) Proper accounting for time increases crop-based
biofuels' greenhouse gas deficit versus petroleum. Environ Res
Lett 4:024001. doi:10.1088/1748-9326/4/2/024001
Ramaswamy V, Boucher O, Haigh J, Hauglustaine D, Haywood J,
Myhre G, Nakajima T, Shi GY, Solomon S (2001) Radiative
forcing of climate change. In: Houghton JT, Ding Y, Griggs DJ,
Noguer M, van der Linden PJ, Dai X, Maskell K, Johnson CA
(eds) Climate change 2001: The scientific basis, intergovernmental
panel on climate change. University Press, Cambridge
RFA (2012) Renewable Fuels Association, www.ethanolrfa.org/pages/
statistics. Accessed August 2012
Roques S, Garstang J, Kindred D, Sylvester-Bradley R, Wiltshire J
(2011) Idle cropland for future crop production. World Agric 2
(2):40–42
Searchinger T et al (2008) Use of U.S. croplands for biofuels increases
greenhouse gases through emissions from land use change. Science
319:1238–1240
Shine KP (2009) The global warming potential—the need of an interdisciplinary
retrial. Clim Chang 96:467–472
Smith P et al (2010) Competition for land. Philos Trans R Soc B
365:294–2957
Stehfest E et al (2009) Climate benefits of changing diet. Clim Chang
95:83–102
330 Int J Life Cycle Assess (2013) 18:319–330

Tidak ada komentar:

Posting Komentar