modeling climate change mitigation from alternative methods of...
TRANSCRIPT
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 2
Avai lab le a t www.sc iencedi rec t .com
ht tp : / /www.e lsev i er . com/ loca te /b iombioe
Modeling climate change mitigation from alternative methodsof charcoal production in Kenya
Rob Bailis*
Yale School of Forestry and Environmental Studies, 195 Prospect St, New Haven, CT 06511, USA
a r t i c l e i n f o
Article history:
Received 18 March 2008
Received in revised form
16 June 2009
Accepted 6 July 2009
Published online 22 August 2009
Keywords:
Charcoal
Land-use change
Sub-Saharan Africa
Greenhouse gas emissions
Carbon offsets
CO2Fix
* Corresponding author. Tel.: þ1 203 432 541E-mail address: [email protected]
0961-9534/$ – see front matter ª 2009 Elsevidoi:10.1016/j.biombioe.2009.07.001
a b s t r a c t
Current carbon accounting methodologies do not accommodate activities that involve
emissions reductions from both land-use change and energy production. This paper
analyzes the climate change mitigation potential of charcoal production in East Africa by
examining the impact of changing both land management and technology. Current
production in a major charcoal producing region of Kenya where charcoal is made as a
by-product of land clearance for commercial grain production is modeled as the ‘‘business-
as-usual’’ scenario. Alternative production systems are proposed based on coppice
management of native or exotic trees. Improved kilns are also considered. Changes in
aboveground, belowground, and soil carbon are modeled and two distinct baseline assess-
ments are analyzed: one is based on a fixed area of land and one is based on the quantity of
non-renewable fuel that is displaced by project activities. The magnitude of carbon emis-
sions reductions varies depending on land management as well as the choice of carbon-
ization technology. However, these variations are smaller than the variations arising from
the choice of baseline methodology. The fixed-land baseline yields annualized carbon
emission reductions equivalent to 0.5–2.8 tons per year (t y�1) with no change in production
technology and 0.7–3.5 t y�1 with improved kilns. In contrast, the baseline defined by the
quantity of displaced non-renewable fuel is 2–6 times larger, yielding carbon emissions
reductions of 1.4–12.9 t y�1 with no change in production technology and 3.2–20.4 t y�1 with
improved kilns. The results demonstrate the choice of baseline, often a political rather than
scientific decision, is critical in assessing carbon emissions reductions.
ª 2009 Elsevier Ltd. All rights reserved.
1. Introduction change mitigation depends on land use as well as energy
This paper models carbon dynamics of charcoal production
in order to quantify the greenhouse gas mitigation potential
of the charcoal trade in Kenya. In many African countries,
charcoal constitutes a major source of greenhouse gas
(GHG) emissions [1]. Like in any bioenergy system, carbon
dynamics from charcoal are linked to two distinct
processes: terrestrial carbon dynamics and emissions from
fuel processing and combustion. Consequently, the degree
to which the charcoal industry can contribute to climate
2; fax: þ1 203 436 9158.
er Ltd. All rights reserved
conversion technologies. The dual components of the
baseline complicate assessments of emissions reductions
(ERs) that result from changing production practices.
I analyze charcoal as it is current production in one part of
Kenya and model several alternative production methods
that have been proposed in order to promote sustainable
production [2–5].
Estimates of Kenya’s charcoal consumption vary widely
(Fig. 1), but it is likely one of the largest charcoal consumers
globally in absolute and per capita terms [1,6,7].
.
0
1
2
3
1993 1996
IEA
IEA
FAOSTAT*
Nyang
Kituyi
MoE
Millio
n m
etric to
ns
60% upward revision of IEA data based on 2000 data from the MoE.
200520021999
Fig. 1 – Point and time-series estimates of annual charcoal
consumption in Kenya between 1994 and 2004
[4,6,7,43,66,67]. * FAOSTAT includes charcoal consumption
data up to 2005, but the data abruptly drops from 650
thousand tons per year in 2000, which is already the
lowest of any published estimate, to just 16,500 t yL1 in
2001 and remains constant at that level until 2005 [6]. This
is an obvious error and has therefore been omitted from
the figure.
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 21492
The majority of the country’s charcoal originates in woody
savannah that constitutes over two-thirds of the country’s
area [5,8,9]. Woody savannah is a resilient ecosystem [10,11],
which can recover from charcoal production under some
management regimes [12–14]. When charcoal production
contributes to long-term deforestation, it is typically the result
of post-harvest management decisions. If trees cover is rees-
tablished, emissions are reduced relative to the deforestation
scenario. However, emissions are not entirely negated
because charcoal production releases 2–3% of the tree’s
carbon as CH4, which has nine times the warming impact of
a molar equivalent quantity of CO2, and a small quantity of
N2O, an even more potent GHG [15–19].
The degree of post-harvest tree regeneration depends on
environmental as well as socioeconomic factors, which are
mediated by environmental regulations that might be set at
a national or sub-national level. In Kenya, such regulations do
not favor sustainable charcoal production [20].
The Clean Development Mechanism (CDM) of the Kyoto
Protocol is the primary vehicle through which climate change
mitigation projects are developed. ‘‘Traditional’’ forms of
bioenergy are largely absent from the CDM. To date, two
methodologies involving charcoal production have been
approved by the CDM Executive Board (the approved method-
ologies are AM41 and AMS-III.K: see [21,22]). Each one calcu-
lates emission’s reductions based on reductions in CH4
induced by changing charcoal production technology. Neither
makes an attempt to assess carbon flux resulting from LUC [23].
However, methodological difficulties arise as a result of the
need to account for land-use change (LUC) and energy-related
emission reductions within a single project. To reduce
uncertainties, the CDM-EB defines biomass as either renew-
able or non-renewable. Their classification includes woody
and non-woody biomass from forested and non-forested land
as well as biomass wastes. Woody biomass originating from
forestland is the most appropriate for analysis of existing
charcoal production systems. In this context, biomass is
renewable if [24]:
1) the land remains a forest;
2) sustainable management practices are in place to ensure
that the level of carbon stocks does not systematically
decrease over time (stocks may temporarily decrease due to
harvesting); and
3) all national or regional forestry and nature conservation
regulations are complied with.
Biomass that fails to meet all three conditions is ‘‘non-
renewable biomass’’ (NRB) by default.
This paper proceeds as follows. Section 2 examines some of
the methodological challenges that arise when analyzing the
carbon implications of a shift from NRB to sustainable biomass.
Section 3 describes the charcoal production systems currently
used in Kenya and several alternatives to this system. Section 4
introduces the model used to estimate GHG emissions from
different charcoal production systems. Section 5 presents the
results and Section 6 discusses the policy implications and
suggests some possible paths for future research.
2. Methodological issues in LUC andbioenergy
The analysis relies on CO2Fix (version 3), a ‘‘carbon book-
keeping’’ model (additional details are provided below, but
also see [25–27]).
Greenhouse gas (GHG) emissions from charcoal result from
three distinct processes: LUC processes induced by wood
harvest, pyrolysis of the woody feedstock; and the combustion
of the charcoal itself. Several studies have quantified emis-
sions from both pyrolysis and end-use [18,28–31]. However,
there have been few analyses of the impact of the charcoal
economy on land cover, nor the effect of changes in charcoal
production technology. Others have assumed a fixed
percentage of biomass regenerates, but with little empirical
basis [1]. These approaches do not accurately assess carbon
dynamics at the stand level. In addition, none of the studies
conducted to date have considered soil carbon, which is likely
to change in response to LUC.
Additional approaches carbon dynamics in wood fuel
systems have utilized GIS models to identify imbalances
between wood fuel supply and demand [32,33]. At the stand
level, several empirical studies have measured biomass
regeneration in stands of trees harvested for charcoal
production [13,14,34,35]. Other researchers have modeled
stand-dynamics in fuel wood systems to estimate carbon flux,
including changes in carbon stored in the soil and in forest
products [26,36,37]. However, to date none have modeled
carbon dynamics in charcoal production systems.
In any bioenergy system, emission reductions are sensitive
to the choice of the ‘‘functional unit’’, a concept commonly
deployed in life-cycle assessment in order to systematically
define the performance of a good or service [38]. Characterizing
the functional unit in bioenergy systems in the context of GHG
abatement is challenging because LUC must be accounted for
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 2 1493
in parallel with emissions from processing and end-use. For
LUC activities, it is logical to assess changes in carbon stocks in
biomass and soil from a fixed area of land (e.g. 1 ha). Emissions
that result from processing harvested woodfrom that land area
are also accounted for. Thus, emissions from LUC and pro-
cessing attributable to the ‘‘functional unit’’ are summed and
compared to the ‘‘business-as-usual’’ (BAU) scenario,
accounting for ‘‘leakage’’ if applicable (see [39] and the
discussion in Section 6 below). However, when an activity
induces a shift away from NRB, it is logical to account for
emissions reductions based on a ‘‘functional unit’’ of energy
delivered to consumers. Indeed, the methodologies that the
CDM-EB approved for NRB substitution bases its assessment of
emission reductions on ‘‘the fuel consumption of the tech-
nologies that would have been used in the absence of the
project activity times an emission coefficient for the fossil fuel
displaced,’’ [39]. This raises challenging questions because
different emission reductions result from the choice of func-
tional unit as unit of land or energy.
EM kilns in Rwanda
Cassamance kilns in Rwanda
EM kiln in Zambia
Small EM kilns in Kenya
Medium EM kilns in Kenya
Round brick kiln in Brazil
Metal drum kiln in Thailand
Rice husk EM kiln in Thailand
EM kiln in Thailand
Mud beehive kiln in Thailand
Large EM kilns in Kenya EM kilns
Brick beehive kiln in Thailand
Hot tail kiln in Brazil
Rectangular metal kiln in Brazil
0
Fig. 2 – Charcoal and carbon yield from earthmound (EM) and im
is defined as the mass of charcoal to the mass of dry wood used
in the charcoal to the initial mass of carbon in the wood. * (* Sou
Medium kilns in Kenya [20]; Thailand [31]; Zambia [47]; and Rwa
there is only moderate correlation between the two variables (r
differed).
3. Charcoal production systems
3.1. Charcoal production technology
Charcoal can be made from many forms of biomass, including
agricultural residues and timber waste [40]. However, nearly
all charcoal in Kenya is made from native woody vegetation
[4,8]. Survey results show over 200 different tree species are
used [5,9,20]. Indigenous trees are favored, though in certain
areas exotic trees are also used [41–43].
Pyrolysis takes place in an enclosed space, ranging from
a simple pit or earthmound (EM) kiln to brick or metal kilns.
In industrialized settings, retorts may be used, which utilize
combustible compounds in the wood to generate heat for
pyrolysis [44,45], but retorts have not been used in Kenya.
EM kilns are typically thought to have low yields, although
empirical measurements show that yields, while variable,
can overlap with yields from improved kilns. For example,
40 80
Charc yield - EM kiln Charc yield - Imp kiln Carbon yield - EM kiln Carbon yield - Imp kiln St. Deviations
Yield (in percent)
proved kilns reported in published literature. Charcoal yield
as feedstock. Carbon yield is defined as the mass of carbon
rces: kilns in Brazil [17]; Large and small kilns in Kenya [17];
nda [49]. In studies reporting both mass and carbon yields,2 [ 0.64), indicating that the degree of pyrolyzation likely
300 600 900
EM kiln in Thailand
EM kiln in W. Africa
EM kiln in Kenya
EM kiln in Zambia
Rectangular metal kiln in Brazil
Brick beehive kiln in Thailand
Mud beehive kiln in Thailand
Beehive kiln in Brazil
Single drum kiln in Thailand
Round brick kiln in Brazil
EM
kiln
sIm
prov
ed k
ilns
Kilograms of carbon in CO2 equivalent units (using 100 yr GWP) per ton of charcoal produced
CO2
CH4
N2O
Fig. 3 – Emissions measured in traditional earthmound (EM) kilns and a variety of improved kilns reported in the literature
showing emissions of each GHG weighted by its 100-year global warming potential. Arrows indicate the emissions factors
used in this analysis. * (* Non-CO2 gases are expressed in CO2-equivalent units using 100-year global warming potential
(GWP), a dimensionless factor that expresses the warming impact of GHGs relative to the warming impact of CO2 [68].
Sources: emissions from Brazil and Kenya [17]; emissions from Thailand [31]; emissions from Zambia [28]; and emissions
from W. Africa [29]. Reference [28] found N2O levels very close to detection limits and do not report an emission factor).
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 21494
in Kenya, government literature cites yields of just 10%
(mass basis) [4]. However field work by this author and
others finds mass-based yields from EM kilns as high as
20–30% [30,46–49].
Yield is typically defined as the ratio of charcoal produced
to dry wood input. However, comparisons between different
studies should be made with caution as methodologies vary.
In addition, this definition of yield doesn’t account for the
degree of carbonization that the wood has undergone. High
temperatures and/or long residence times in the kiln yield
charcoal with a higher fraction of fixed carbon and lower
content of volatile matter, which lowers the mass of the end
product, but does not necessarily reduce the thermodynamic
efficiency of the process [50]. Thus, conversion efficiencies are
hard to interpret without information about the extent of
carbonization. Nevertheless, Fig. 2 is included here to show
Litter
Humification
BiomassIncrement
Litter fall
Tree biomass Stemwood- Foliage - Branches - Roots
Intermediatehumus
Harvestresidues and treemortality
Decomposition
Timberharvesting
Biomassmodule
Carbon in the atm
Fig. 4 – Land-use change model for estimating emiss
the range of efficiencies reported in the literature. Notably,
two studies measured carbon yield and both mass and carbon
yields from these studies are shown.
Traditional kilns are favored across sub-Saharan Africa
because they require very little capital investment, are flexible
in size and shape, and are well-matched to the dispersed
nature of the charcoal trade [5]. Measurements from charcoal
production show overlap in emissions from EM and improved
kilns. EM kiln emissions range from 0.514 to 0.840 t C (in CO2
equivalent units weighted by100 y global warming potentials)
per ton charcoal. Emissions from improved kiln technologies
vary from 0.400 to 0.820 t C per ton charcoal (Fig. 3).
This analysis relies on data from Kenya to estimate EM
emissions [30]. For the improved kiln, emissions from a Bra-
zilian metal kiln are used [30]. This is the cleanest production
technology reported and was chosen in order to show the
Stablehumus
Emissions from burning of wood and charcoal
Humification
Woodfuels- Firewood - Charcoal
Soil module
Productsmodule
osphere
ions from wood harvesting - adapted from [25].
Table 1 – Baseline and alternative models of charcoal production.
Model name Land management model Initial harvest Key assumptions
Baseline-EM Clearance of native vegetation (AG and BG)
via one-time charcoal production in a traditional
EM or improved kiln followed by wheat cultivation
All AG and 50% of
BG biomass is cleared
Initial soil carbon is lost due to annual
tillage so that 36% of initial soil-C is lost
after 20 years.aBaseline-IK
Tarch5-EM
Tarch10-EM
Tarch15-EM
Clearance of native vegetation (AG only) for
charcoal production in an EM kiln followed by
coppice management of native tree cover on
5, 10, and 15-year coppice cycles with harvested
wood processed in EM or improved kilns,
depending on the scenario.
All AG biomass.
BG biomass is left
intact
Coppiced stands regenerate without loss
for entire modeling period. Growth of AG
biomass during coppice cycles is boosted
by 5% relative to growth from seed due to
intact root structure.
Tarch5-IK
Tarch10-IK
Tarch15-IK
Euc10-EM Clearance of native vegetation (AG and BG) for
charcoal production in an EM kiln followed by
coppice management of E. grandis on a 10-year
cycle with harvested wood processed in EM
or improved kilns, depending on
the scenario.
All AG and 50% of
BG biomass is cleared
Coppiced stands regenerate without loss for
entire modeling period. Growth of AG biomass
during coppice cycles is boosted by 5% relative
to growth from seed germination due to large
intact root structure.
Euc10-IK
a This matches the IPCC’s default assumptions for conversion of woodlands to croplands [57, Ch. 3].
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 2 1495
largest possible contrast between technical options. In reality,
this option would likely be too costly for local production,
although it is possible that a scaled-down more locally
appropriate version could be developed.
3.2. Land management
There are two dominant modes of charcoal production in
Kenya and a third mode that is proposed, but not practiced: 1)
charcoal as a by-product of land conversion from woodland to
cropland, 2) charcoal produced on Kenya’s semiarid range-
lands without conversion to cropland and 3) management of
either natural woodlands or plantations for sustained char-
coal production.
Historically, much of Kenya’s charcoal production has
been linked to agricultural expansion [4]. When woodlands
are converted to croplands, there is a long-term loss of
terrestrial carbon and a reduction in soil carbon, as inputs of
carbon from leaf litter and decaying biomass are reduced and
soils are subject to annual tillage. This loss occurs whether or
not charcoal is produced because in the absence of charcoal
production trees are cleared and probably burned in situ.
However, the warming impact may be greater with charcoal
production because the GHG emission factors per ton of
standing biomass are greater when charcoal is produced than
when woodlands are burned [28].
In other modes of charcoal production, the loss of terres-
trial carbon may be temporary. In Kenya’s rangelands, trees
are harvested with little post-harvest management. Many
species can coppice and, left undisturbed, cleared woody
savannah typically regenerates [13,14]. However, regeneration
is contingent on natural and anthropogenic factors including
the presence of wildlife and livestock, precipitation patterns,
soil qualities, and natural or anthropogenic burning [11,51].
The net change in terrestrial carbon stocks on woodlands
composed of either exotic or indigenous tree species, when
managed for charcoal will depend on both the frequency of
harvest, and the management applied. This study analyzes
six charcoal production scenarios drawn from the three
modes of production. The business-as-usual (BAU) scenario
simulates current charcoal production in Narok District
(1�050South, 35�520East), where the author conducted exten-
sive field work in 2004 and 2005. At the time, Narok was
a major charcoal supply zone, providing roughly 30% of
Nairobi’s charcoal [20]. Charcoal in Narok originates in
woodlands dominated by Tarchonanthus camphoratus,
a common dioecious evergreen shrub. It is usually multi-
stemmed, growing up to 9 m, with a sparse crown and
a trunk up to 40 cm in diameter. It favors dry forest margins
or secondary woodlands and is often dominant or co-domi-
nant in association with Acacia spp. [52,53]. In the BAU
scenario, a full-grown stand of native vegetation is
completely cleared (including all aboveground (AG) and
a large portion of belowground (BG) biomass). Charcoal is
produced from the cleared biomass and wheat is planted on
the cleared land. A variation of BAU with an improved kiln is
included to demonstrate the effect of a technological shift
with no change in land management.
Additional scenarios are explored in which improved kilns
and coppice management of either native vegetation
(assumed to be pure stands of T. camphoratus) or a common
exotic species (Eucalyptus grandis) are introduced. Table 1
describes each scenario in more detail.
4. Estimating emissions from charcoalproduction
4.1. Modeling terrestrial carbon dynamics in charcoalproduction systems
CO2FIX (Version 3.1) was used to estimate terrestrial carbon
dynamics in each wood fuel production system. The total
stock of carbon at time t is the sum of carbon in three pools:
CTt ¼ Cbt þ Cst þ Cpt (1)
t defines an increment of time, CTt is the total stock of carbon,
Cbt is the carbon stored in AG and BG biomass, Cst is the carbon
stored in soil organic matter, and Cpt is the carbon stored in
Table 2 – Initial carbon stocks in stand of nativevegetation (T. camphoratus).
Component Carbon content (t ha�1)
Soil (carbon) 29.5
Biomass (carbon)
Stems 10.5
Foliage 0.3
Branches 9.6
Roots 9.1
Total 59.0
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 21496
forest products. In this case, charcoal is the only product, and
it is consumed in the same period as the harvest. All carbon
stocks are measured in tonne per hectare. The model treats
each component as a separate module that is parameterized
independently. A diagram of stocks and flows of carbon is
shown in Fig. 4; modules are distinguished by broken gray
lines.
From one time period to the next, carbon flows within and
between modules as depicted by the arrows in Fig. 4. After
a time period of n years, the net flux of carbon can be deter-
mined by summing the difference in each module between
t¼ n and t¼ 0:
DCT ¼ CTn � CT0 ¼ DCb þ DCs þ DCp
¼ ðCbn � CboÞ þ ðCsn � CsoÞ þ�Cpn � Cpo
�(2)
To simulate carbon dynamics for a given stand of trees,
each module must be parameterized according to the physical
properties of the stand. Parameterization is described below.
4.2. Biomass
The biomass module is parameterized with a biomass growth
function with branches, foliage and roots defined relative to
the stem. In addition, the turnover of each component must
be defined as well as wood density, carbon content, mortality,
and harvest-induced losses.
0
CAI
0
10
20
30
years
Bio
mass g
ro
wth
(d
ry to
ns · h
a-1 · y
-1) Euc (moderate)
Euc (slow) Tarch
252015105
Fig. 5 – Current and Mean Annual Increments (CAI and MAIs) f
slow-growth stands of E. grandis. Initial years of T. camphoratus
estimated in order to give maximum AG carbon of w25 t haL1. G
South Africa [55].
Each scenario has identical initial conditions, defined as
a mature stand of T. camphoratus, which is assumed to have
a total terrestrial carbon content of w60 t ha�1 apportioned
with roughly 35% in AG biomass, 20% in BG biomass, and 45%
in the soil. Table 2 shows estimated carbon allocation in the
initial plot. This allocation of soil and biomass carbon matches
the typical range for savannah woodlands where ‘‘root-to-
shoot ratios’’ of 0.5 are common [11,54]. Model parameters
also match empirical data for T. camphoratus measured by
Young and Francombe [35] in coppiced stands. That study
found a mean annual increment (MAI) of 2.3 dry t ha�1 after 6
years of regeneration. The growth function used in this
analysis simulates their findings: the MAI is 2.3 t ha�1 after six
years, increasing to 3 t ha�1 at 20 years and then declining to
simulate slowing growth of a mature stand.
To simulate growth of E. grandis, a growth function based
on in an ecological zone similar to the study area in Kenya was
used [55]. Two site quality (SQ) indices were chosen in order to
simulate two rates of tree growth (moderate and slow). Both
are managed on a 10-year rotation. The growth functions
(current annual increments (CAI) and mean annual incre-
ment) used for both species are shown in Fig. 5.
4.3. Soil
The soil module incorporates a soil carbon model that has
been shown to robustly estimate decomposition rates for
different types of litter across a range of climates [56]. The
model assumes carbon inputs to the soil via three litter
compartments: coarse consisting of stemwood, fine consisting
of branches and coarse roots, and non-woody consisting of
foliage and fine roots. The rate of carbon input into each
compartment is determined by the growth and turnover rates,
as well as inputs arising from mortality and slash (defined in
the biomass module). Depending on its chemical composition,
litter is partitioned into one of three subsequent compart-
ments: soluble compounds, holocellulose, and lignin-like
compounds. In addition to these compartments, there are two
0
10
20
30
MAI
years
Euc (moderate)
Euc (slow) Tarch
0 252015105
or a pure stand of T. camphoratus as well as moderate and
are based on empirical observations [35]; the later years are
rowth curves for E. grandis are based on empirical data from
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 2 1497
humus compartments, which receive inputs from the coarse
woody litter as well as the faster decomposing compounds.
The model applies temperature- and moisture-dependent
rates of partitioning and decomposition that have been cali-
brated across a range of temperatures [56].
For this analysis, the soil module was parameterized such
that initial soil carbon is roughly 30 t ha�1, which is the low
end for the range observed in savannah woodlands but is well-
matched to the steady state of carbon that would result from
leaf litter and root turnover of T. camphoratus as well as local
temperature and moisture conditions. The IPCC lists approx-
imate soil organic carbon under native vegetation in dry
tropical regions (rainfall less than 1000 mm y�1) between 35
and 60 t y�1 [54,57]. Ringius [58] notes soil carbon ranged
between 30 and 44 t ha�1 in several different soil types in
different locations in Kenya prior to crop cultivation [59].
4.4. Forest products
The product module tracks the carbon in the harvested wood. In
the case of stands that are managed strictly for woodfuels, there
are no long-lived products. All fuel-bound carbon extracted in
a given period is released to the atmosphere immediately.
Baseline
0
40
80
T. Camphoratus - 10 year coppice
0
40
80
t • h
a-1
E. Grandis (slow) - 10 year coppice
0
40
80
120
160
0 10 20 30
0 10 20 30
0 10 20 30Years
t • h
a-1
AG carbon
BG carbonSoil carbon
Total carbon
t • h
a-1
a
c
e
Fig. 6 – Carbon stocks in each stand o
4.5. Adjusting the model for coppice systems
CO2Fix is not designed to calculate carbon dynamics under
coppice management. The model assumes that biomass
either enters the pool of wood products or litter upon harvest
whereas under coppice management, BG biomass remains in
the pool of living biomass. Moreover, because there is already
a well-established root system, early stem growth in subse-
quent generations is typically faster than stem growth in trees
planted from seed or seedlings [60,61].
Coppicing was simulated by using BG biomass and soil-C at
end of the nth coppice cycle to define the initial BG and soil
conditions for the (nþ 1)th cycle. Lettens and colleagues
describe an alternate approach for dealing with coppicing
within CO2Fix [37].
5. Results
5.1. Emissions reductions from carbon storage
In each scenario, a mature stand of T. camphoratus is cleared
for charcoal production in order to reflect current practices in
T. Camphoratus - 5 year coppice
0
40
80
b
t • h
a-1
T. Camphoratus - 15 year coppice
0
40
80
t • h
a-1
E. Grandis (moderate) - 10 year coppice
0
40
80
120
160
Years
t • h
a-1
0 10 20 30
0 10 20 30
0 10 20 30
AG* = above-ground; BG* = below-ground
b
d
f
f T. camphoratus and E. grandis*.
Table 3 – Carbon emissions from pyrolysis after 30 yearsof charcoal production on each hectare of land (t haL1 inCO2 equivalent units using 100-year GWPs).
Baseline
Tarch5
Tarch10
Tarch15 Euc(slow)
Euc(moderate)
Traditional
kilns
12 17 18 15 53 74
Improved
kilns
8 12 13 11 38 53
Tarch10 Euc (slow)a
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 21498
the study area. This initial clearance is defined as Year-1.
Subsequently, management practices are defined as in Table 1.
Each simulation was run for 30 years (the maximum lifetime of
an A/R project in the CDM). The evolution of carbon stocks is
shown in Fig. 6a–f.
The two E. grandis scenarios, which have peak growth
rates four and seven times larger than the peak growth rate
of T. camphoratus respectively, have vertical scales that are
double the stands of T. camphoratus. Fig. 6a shows the
baseline scenario, in which initial AG and BG biomass are
harvested. After harvest, stocks of living biomass drop to
near zero and the soil receives a pulse of carbon in the form
of fine roots and leaf litter. Soil carbon decays rapidly at first,
then more slowly as carbon is shifted to longer-lived pools in
the soil. Wheat is cultivated during the entire period,
creating small stocks of biomass carbon during each growing
season. This is estimated at roughly 2 t ha�1 based on
average wheat yields in Narok district (w3.3 t ha�1 [62]) and
published estimates of stalk to grain and root-to-shoot ratios
in wheat [57, Ch. 4].
Fig. 6b–d show the simulations for each stand of T. cam-
phoratus. Carbon is lost relative to Year-0 because the trees
never reach the size of the mature stand. The models also
show gradual declines in BG and soil carbon during the
modeling period, because natural turnover of the fully
formed roots outpaces the ability of the tree, with smaller
quantities of AG biomass, to allocate BG carbon. Inputs of leaf
litter are also lower than in the mature stand of trees. These
declines are most pronounced in the 5-year coppice cycle.
Fig. 6e and f show slow and moderate rates of growth for E.
grandis under a 10-year coppice cycle. Stocks of carbon in AG
and BG biomass are far larger than in T. camphoratus
scenarios. In addition, soil carbon increases over time
because leaf litter and root turnover are larger in stands of
E. grandis.
-50
0
50
100
150
to
ns o
f carb
on
Euc (moderate)
0 10 20 30
5.2. Emissions from charcoal production
Total quantities of charcoal produced in each scenario are
shown in Fig. 7. Emission factors given in Fig. 3 are multiplied
by charcoal production in each scenario to define emissions
from pyrolysis (shown in Table 3).
0
50
100
150
Baseli
ne
Baseli
ne-IK
Tarch5
Tarch1
0
Tarch1
5
Euc (m
odera
te)
Euc (s
low)
to
ns o
f ch
arco
al p
er h
a Traditional kilnsImproved kilns
Fig. 7 – Total charcoal production after 30 years of
management.
5.3. Assessing changes in carbon stocks relative to thebaseline
Each emissions profile can be assessed at the stand level
assuming a fixed land area. Carbon stocks at each point in time
are subtracted from the carbon stocks in the baseline. This
yields time-dependent emission reductions for each scenario,
shown in Fig. 8a and b. The plots extend to year-31 to show the
effect of a final harvest on carbon dynamics in each scenario.
The net change in carbon closely mirrors the coppice cycle.
However, peak carbon storage declines over time as a result of
successive cycles of charcoal production. Each time charcoal
is produced, CH4 and N2O are released, which are not fully
offset by the next cycle of biomass growth. Thus, with each
successive cycle of charcoal production, the net sequestration
relative to the baseline is diminished. This effect is more
pronounced in Fig. 8a, which shows the results for EM kilns,
because more CH4 and N2O are released per unit of charcoal
produced.
Fig. 9 shows changes in carbon stocks and emissions from
charcoal production relative to the baseline scenario after 30
-50
0
50
100
150
0 10 20 30to
ns o
f carb
on
Baseline w/IK Tarch10Euc (slow) Euc (moderate)
Note: Tarch5 and Tarch15 have been omitted for clarity.
b
Fig. 8 – a. Changes in carbon stocks relative to the baseline
scenario, including atmospheric emissions from charcoal
production using traditional EM kilns. b. Changes in carbon
stocks relative to the baseline scenario, including
atmospheric emissions from charcoal production using
improved kilns.
Euc (slow)
AG BG
Soil Charc(IK)
Charc(EM)
-150
-100
-50
0
50
Baseline-IK
Carb
on
em
issio
n red
uctio
ns ( t • h
a-1)
Tarch5
Tarch10
Tarch15
Euc (moderate)
AG BG
Soil Charc(IK)
Charc(EM)AG BG
Soil Charc (IK)Charc (EM)
Fig. 9 – Net carbon emissions per hectare relative to the
baseline scenario in aboveground (AG) pools, belowground
(BG) pools, and soil, as well as emissions from charcoal
production (Charc EM or IK) in year-30 of each scenario.
Positive changes represent carbon emissions; negative
changes represent carbon sequestration. Emissions are
shown for charcoal production from EM kilns (dashed
lines) and from improved kilns (solid lines).
0
200
400
600
Baseline-IK
Tarch5
Tarch10
Tarch15
Euc (slow)
Euc (moderate)
To
ns o
f carb
on
em
issio
n red
uctio
ns
an
d seq
uestratio
n in
year-30
AG CBG CSoil CCharc C
AG CBG CSoil C
EM kiln scenario Improved kiln scenario
Fig. 10 – Net carbon emission reductions and sequestration
relative to the BAU scenario, demonstrating the impact of
increasing the land area affected by BAU land clearance so
that the same quantity of charcoal is produced during the
30-year project activity. Each scenario includes charcoal
production from EM kilns (shaded entries) and from
improved kilns (solid entries).
Table 4 – Average annual carbon emissions reductionsusing Baseline-1, a stand-level assessment with fixedland area, and Basline-2, an assessment of NRBsubstitution (t haL1).
Scenario EM kiln Improved kiln
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 2 1499
years of management. Stock changes are shown for each pool
(AG, BG, and soil). Emissions are shown for both EM and IK
scenarios. Simply shifting to an IK with no change in post-
harvest management results in carbon emission reductions of
3.4 t ha�1. Shifting to coppice management and using an EM
kiln reduces carbon emissions by 20–105 t ha�1. Shifting to IK
production yields reduces carbon emissions by an additional
4.4–21.4 t ha�1. Averaged over the 30-year project lifetime,
these scenarios generate annual carbon offsets ranging from
0.1 to 3.5 t ha�1. For comparison, the median annual carbon
offsets among the 30 afforestation/reforestation projects
currently in the CDM pipeline is w3 t ha�1 [23].
As discussed above, quantifying emission reductions
based on the amount of energy from NRB that is displaced
gives a different picture. Coppice management yields 1.6–14
times the amount of deliverable energy per hectare relative to
the baseline scenario. Thus, to deliver the same quantity of
energy from the BAU scenario as in the coppice management
scenarios, additional land must be cleared. Fig. 10 shows the
results of adding additional land. Here, net emission reduc-
tions and carbon sequestered increase by 150–500% relative to
the fixed land scenario. Further, shifting from EM to improved
kilns increases the net emission reductions and sequestered
carbon by an additional 60–120%. Table 4 summarizes the
average annual carbon savings achieved by each scenario
under the two baselines.
Baseline-1 Baseline-2 Baseline-1 Baseline-2
Baseline-IK – – 0.1 1.2
Tarch5 0.5 1.4 0.7 3.2
Tarch10 0.8 1.8 1.0 3.6
Tarch15 1.0 1.5 1.2 3.1
Euc (slow) 1.7 8.3 2.2 13.6
Euc (moderate) 2.8 12.9 3.5 20.4
6. Discussion
This analysis demonstrates that GHG emissions reductions
can be realized by shifting from one-time charcoal
production as a by-product of land clearance to a dedicated
coppice-management system in which either native vegeta-
tion or exotic species are periodically harvested and carbon-
ized. The magnitude of carbon emissions reductions that
may be achieved varies depending on the coppice cycle as well
as the choice of tree species and carbonization technology.
However, these variations are smaller than the variations that
arise from the choice of baseline methodology.
The large differences in ERs associated with different
baseline scenarios raise questions about how to best assess
changes in GHG emissions associated with activities involving
fuel substitution and land cover change. Calculating emissions
reductions relative to the baseline scenario based on a single
unit of land leads to systematically lower emissions reduc-
tions than calculating emissions reductions based on the
amount of NRB displaced.
Recent analyses have shown that similar complications
arise in the modern energy sector. When biofuels replace
fossil fuels, losses in biomass and soil carbon can result in
a ‘‘carbon debt’’ that takes decades or centuries to repay
[63]. However, as this analysis shows, when the conversion
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 21500
of natural ecosystems defines the BAU scenario, alternate
post-clearance land management options may have the
opposite effect: leading to fewer emissions than would have
otherwise occurred. For land managers, deviations from
BAU behavior must bring tangible benefits. This is not
possible under Kenya’s current market conditions; land
managers see much greater returns from one-time charcoal
production followed by intensive grain cultivation, than they
would see from any of the charcoal production scenarios
explored here [20].
6.1. Charcoal and food production
The scenarios explored here shift land out of wheat produc-
tion. This is not a threat to local food security as wheat is
a negligible part of the local diet. However, wheat from Narok
is processed for consumption in urban and peri-urban
markets. Thus, shifting land that was destined for wheat
production into less carbon-intensive uses may create some
unmet food demand. If this were to occur on a large scale,
a segment of the population could be adversely affected.
Alternatively, it may induce other land clearance to meet
demand, which could emit additional GHGs. These indirect
effects have been estimated to have substantial impacts when
biofuel production occurs at very large scales [64], but such
effects are difficult to characterize in small markets.
6.2. Assessing the choice of baseline
The disparity between emission reductions arising from
different choices of baseline methodology reflect the difficulty
in combining emissions reductions derived from LUC and
energy-based activities. One might simply ask ‘‘which base-
line most accurately reflects reality?’’ The problem is that both
can be justified with reasonable arguments. For example, land
planted with E. grandis managed on a 10-year coppice cycle
over 30 years would produce 10 times more charcoal than is
produced from one-time clearance of native vegetation. In
theory, providing that charcoal to the market would lessen
demand for additional virgin land to be cleared for charcoal
production. The problem with this logic is that land clearance
is only partially driven by demand for charcoal. It is also
driven by a desire to expand crop cultivation. Thus, the
clearance of additional land also depends on the extent to
which cultivation presents an attractive land use option.
This argument may make the assessment based on fixed
land area seem more justifiable. However, wheat is not
consumed locally. Like charcoal, wheat is simply a commodity
sold to non-local markets. As was mentioned above, current
policies in Kenya are not amenable to sustainable land
management for charcoal production [20]. However, this may
change as a result of recent legislation [2,3]. If sustainable land
management for charcoal production were a viable option,
then it is possible that fewer land owners would opt to plant
wheat, making the assessment of emission reductions based
on energy the more realistic option.
It should also be noted that the methodologies that have
been developed to calculate emission reductions in existing
carbon offset projects are not necessarily realistic. In some
cases, realism is traded for simplicity and replicability. For
example, the methodologies that have been approved for
reducing or substituting the use of NRB estimate emission
reductions by defining a baseline in which users switch to
fossil fuels like LPG or kerosene, an unlikely scenario in most
settings where NRB is common [65]. Thus, while assessment
of net carbon savings based on fixed land area is certainly
a more conservative assessment method, in the absence of
genuine experience with land management for sustainable
charcoal production, we can not determine which method
better reflects reality. In any case, careful monitoring would be
required to elucidate how Kenya’s charcoal market would
respond to more sustainable production practices.
Acknowledgements
The author acknowledges the support of the Link Foundation
Energy Fellowship which supported field work that provided
vital insights into this analysis. In addition, Professor Omar
Masera of the National Autonomous University of Mexico
provided useful advice on how to model NRB and Murefu
Barasa provided research and editorial assistance, but any
errors or omissions are the sole responsibility of the author.
r e f e r e n c e s
[1] Bailis R, Ezzati M, Kammen DM. Mortality and greenhousegas impacts of biomass and petroleum energy futures inAfrica. Science 2005;308:98–103.
[2] Government of Kenya. Forests Act, 2005; 2005.[3] Government of Kenya. The Energy Act, 2006. Available at:
http://www.erb.go.ke/energy.pdf; 2006.[4] Ministry of Energy (MoE). Study on Kenya’s energy demand,
supply and policy strategy for households, small scaleindustries and service establishments: final report. KAMFORCompany Limited; 2002. p. 158.
[5] Mugo F, Poulstrup E. Assessment of potential approaches tocharcoal as a sustainable source of income in the arid andsemi-arid lands of Kenya, DANIDA and RELMA: final reportno. 72; 2003.
[6] FAO Statistics Division. FAOSTAT forestry. Food andAgriculture Organization of the United Nations; 2007.
[7] IEA. Energy statistics of non-OECD countries 2003–2004.Paris: The Organization for Economic Cooperation andDevelopment (OECD); 2006.
[8] Mugo FW. Charcoal trade in Kenya, RELMA/SIDA: workingpaper No. 5; 1999. p. 35.
[9] Mutimba S, Barasa M. National charcoal survey of Kenya.Final draft project report, No. 56. Africa: Energy forSustainable Development; 2005.
[10] House JI, Hall DO. Savannas: human influence. Available at:.King’s College London http://www.savannas.net/savhi.htm;2003 [accessed 12.07.05].
[11] Breman H, Kessler J- J. Woody plants in agro-ecosystems ofsemi-arid regions. Berlin: Springer-Verlag; 1995.
[12] Chidumayo EN. Estimating fuelwood production and yield inregrowth dry miombo woodland in Zambia. Forest Ecologyand Management 1988;24:59–66.
[13] Chidumayo EN. Zambian charcoal production: Miombowoodland recovery. Energy Policy 1993;21:586–97.
[14] Hosier RH. Charcoal production and environmentaldegradation: environmental history, selective harvesting,
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 2 1501
and post-harvest management. Energy Policy 1993;21:491–509.
[15] Bailis R, Ezzati M, Kammen D. Greenhouse gas implicationsof household energy technologies in Kenya. EnvironmentalScience and Technology 2003;37:2051–9.
[16] Bond T, Venkataraman C, Masera O. Global atmosphericimpacts of residential fuels. Energy for SustainableDevelopment 2004;8:20–32.
[17] Pennise D. Greenhouse gas, indoor air pollution, and wooduse implications of the charcoal fuel cycle, Environmentalhealth sciences. Berkeley, CA: University of California; 2003.p. 194.
[18] Smith K, Uma R, Kishore VVN, Lata K, Joshi V, Zhang J, et al.Greenhouse gases from small-scale combustion devices indeveloping countries phase IIa: household stoves in India.No. EPA-600/R-00-052. US Environmental Protection Agency;2000. 98.
[19] IPCC. Contribution of working group I to the fourthassessment report of the intergovernmental panel onclimate change. In: Solomon S, Qin D, Manning M, Chen Z,Marquis M, Averyt KB, Tignor M, Miller HL, editors. Climatechange 2007: The physical science basis. New York, NY:Cambridge University Press; 2007.
[20] Bailis R. Fuel from the savannah: social and environmentalimplications of commercial charcoal production. Energy andResources Group. Berkeley, CA: University of California; 2005.p. 387.
[21] CDM-EB. Report of the twenty-seventh meeting of theexecutive board of the clean development mechanism,UNFCCC, No. CDM-EB-27. 15. Available at: http://cdm.unfccc.int/EB/027/eb27rep.pdf; 2006.
[22] CDM-EB. Report of the twenty-eighth meeting of theexecutive board of the clean development mechanism,UNFCCC, No. CDM-EB-28. 23. Available at: http://cdm.unfccc.int/EB/028/eb28rep.pdf; 2006.
[23] Fenhann J. CDM pipeline overview [February 2009 edition]Available at:. CD4CDM - UNEP Risø Centre http://cd4cdm.org/Publications/CDMpipeline.xls; 2009 [accessed 24.02.09].
[24] CDM-EB. EB 23 report annex 18: definition of renewablebiomass, UNFCCC. Available at: http://cdm.unfccc.int/EB/023/eb23_repan18.pdf; 2006.
[25] Nabuurs GJ, Garza-Caligaris JF, Kanninen M, Karjalainen T,Lapvetelainen T, Liski J, et al. CO2FIX V2.0-manual of a modelfor quantifying carbon sequestration in forest ecosystemsand wood products, ALTERRA, No. 1068. 48. Available at:http://www2.efi.fi/projects/casfor/>; 2001.
[26] Masera OR, Garza-Caligaris JF, Kanninen M, Karjalainen T,Liski J, Nabuurs GJ, et al. Modeling carbon sequestration inafforestation, agroforestry and forest management projects:the CO2FIX V.2 approach. Ecological Modeling 2003;164:177–99.
[27] Schelhaas MJ, Esch PWv, Groen TA, Kanninen M, Liski J,Masera O, et al. CO2FIX V 3.1-A modelling framework forquantifying carbon sequestration in forest ecosystems,ALTERRA, No. 1068. 51. Available at: http://www2.efi.fi/projects/casfor/>; 2004.
[28] Bertschi IT, Yokelson RJ, Ward DE, Christian TJ, Hao WM.Trace gas emissions from the production and use ofdomestic biofuels in Zambia measured by open-path Fouriertransform infrared spectroscopy. Journal of GeophysicalResearch-Atmosphere 2003;108:5–13. 5–1.
[29] Brocard D, Lacaux C, Lacaux JP, Kouadio G, Yoboue V.Emissions from the combustion of biofuels in WesternAfrica. In: Levine JS, editor. Biomass burning and globalchange. Cambridge, MA: MIT Press; 1996. p. 350–60.
[30] Pennise D, Smith KR, Kithinji JP, Rezende ME, Raad TJ,Zhang J, et al. Emissions of greenhouse gases and otherairborne pollutants from charcoal-making in Kenya and
Brazil. Journal of Geophysical Research-Atmosphere 2001;106:24143–55.
[31] Smith KR, Pennise DP, Khummongkol P, Chaiwong V,Ritgeen K, Zhang J, et al. Greenhouse gases from small-scalecombustion in developing countries. US EPA, No. EPA-600/R-99–109. Charcoal Making Kilns in Thailand; 1999.
[32] Masera O, Ghilardi A, Drigo R, Trossero MA. WISDOM: A GIS-based supply demand mapping tool for woodfuelmanagement. Biomass and Bioenergy 2006;30:618–37.
[33] Masera OR, Drigo R, Trossero MA. No. Y4719/E, 52. Available at:.Woodfuels integrated supply/demand overview mapping(wisdom): methodological approach for assessing woodfuelsustainability and supporting wood energy planning. UNFAOhttp://www.fao.org/DOCREP/005/Y4719E/Y4719E00.HTM; 2003.
[34] Okello BD, O’Connor TG, Young TP. Growth, biomass estimatesand charcoal production of Acacia drepanolobiom in Laikipia,Kenya. Forest Ecology and Management 2001;142:143–53.
[35] Young TP, Francombe C. Growth and yield estimates innatural stands of Leleshwa (Tarchonanthus camphoratus).Forest Ecology and Management 1991;41:309–21.
[36] de Jong BH, Masera O, Olguin M, Martinez R. Greenhouse gasmitigation potential of combining forest management andbioenergy substitution: a case study from central highlandsof Michoacan, Mexico. Forest Ecology and Management 2007;242:398–411.
[37] Lettens S, Muys B, Ceulemans R, Moons E, Garcia J, Coppin P.Energy budget and greenhouse gas balance evaluation ofsustainable coppice systems for electricity production.Biomass and Bioenergy 2003;24:179–97.
[38] Cooper JS. Specifying functional units and reference flows forcomparable alternatives. The International Journal of LifeCycle Assessment 2003;8:337–49.
[39] CDM-EB. Indicative simplified baseline and monitoringmethodologies for selected small-scale CDM project activitycategories: attachment C to Appendix B of Annex II ofDecision 4/CMP.1 (Guidance relating to the cleandevelopment mechanism), UNFCCC, No. I.C./Version 11,Sectoral Scope: 01, EB 32. Available at: http://cdm.unfccc.int/UserManagement/FileStorage/CDMWF_AM_NZ0ZPDY2NRQS6A8ZLT9JSXTBZNFKX2; 2006.
[40] Bhattacharya SC, Sett S, Shrestha RM. State of the art forbiomass densification. Energy Sources, Part A: Recovery,Utilization, and Environmental Effects 1989;11:161–82.
[41] Maundu P, Tengnas B. Useful trees and shrubs for Kenya.Nairobi: World Agroforestry Center; 2005.
[42] Odour N, Wekesa L. Marketing aspects of charcoal fromCommiphora species and Terminalia orbicularis. Kenya ForestResearch Institute (KEFRI) with support from the Kingdo ofBelgium (Belgian Technical Cooperation) and the KenyanForest Department; 2003. No. 83–94 [Ch. 6 in Feasibility studyon Commiphora spp].
[43] Kituyi E, Marufu L, Wandiga S, Jumba I, Andreae M, Helas G.Biofuel availability and domestic fuel use patterns in Kenya.Biomass and Bioenergy 2001;20:71–82.
[44] FAO. Industrial charcoal making. Available at:. UN Food andAgriculture Organization http://www.fao.org/docrep/X5555E/X5555E00.htm; 1985. FAO Forestry Paper, No. 63.
[45] Foley G. Charcoal making in developing countries. Technicalreport, no. 5. International Institute for Environment andDevelopment; 1986. p. 214.
[46] Bailis R, Pennise D, Ezzati M, Kammen DM, Kituyi E. Impactsof greenhouse gas and particulate emissions from woodfuelproduction and end-use in sub-Saharan Africa. In: 2nd worldconference and technology exhibition on biomass for energy,industry and climate protection, Rome, Italy; 2004.
[47] Hibajane SH. Assessment of earth kiln charcoalproduction technology. In: Energy, environment, and
b i o m a s s a n d b i o e n e r g y 3 3 ( 2 0 0 9 ) 1 4 9 1 – 1 5 0 21502
development series, vol. 39. Stockholm EnvironmentalInstitute; 1994. p. 39.
[48] Schenkel Y, Bertaux P, Vanwijnbserghe S, Carre J. Anevaluation of the mound kiln carbonization technique.Biomass and Bioenergy 1998;14:505–16.
[49] ESMAP. Rwanda: commercialization of improved charcoalstoves and carbonization techniques - mid-term progressreport, The Energy Sector Management AssistanceProgramme of the World Bank, Report no. 141/91; 1991. p. 96.
[50] Antal MJ, Allen SG, Dai X, Shimizu B, Tam MS, Grønli M.Attainment of the theoretical yield of carbon from biomass.Industrial Engineering and Chemical Research 2003;39:4024–31.
[51] Scholes RJ, Hall DO. The carbon budget of tropical savannas,woodlands, and grasslands. In: Breymeyer AI, Hall DO,Melillo JM, Agren GI, editors. Global change: effects onconiferous forests and grasslands. Chichester, New York:Wiley; 1996. p. 69–100.
[52] World Agroforestry Center. Agroforestree database. WorldAgroforestry Center; 2007.
[53] Beentje H. Kenya trees shrubs and lianas. Nairobi: NationalMuseums of Kenya; 1994.
[54] IPCC. Available at:. Revised 1996 IPCC guidelines for nationalgreenhouse gas inventories: reference manual, vol. 3. BlackwellPress www.ipcc-nggip.iges.jp/public/gl/invs6.htm; 1997.
[55] FAO. Eucalypts for planting. Available at:. In: FAO forestryseries, vol. 11. UNFAO http://www.fao.org/docrep/004/AC459S/AC459S00.HTM (web version is only available inSpanish); 1981. p. 677.
[56] Liski J, Nissinen A, Erhard M, Taskinen O. Climatic effects onlitter decomposition from arctic tundra to tropical rainforest.Global Change Biology 2003;9:575–84.
[57] IPCC. Good practice guidance for land use, land-use changeand forestry. Hayama, Japan: Institute for GlobalEnvironmental Strategies (IGES) for the IPCC; 2003.
[58] Ringius L. Soil carbon sequestration and the CDM:opportunities and challenges for Africa. Climatic Change2002;54:471–95.
[59] Woomer PL, Palm CA, Qureshi JN, Kotto-Same J. Carbonsequestration and organic resources management in Africansmallholder agriculture. In: Lal R, Kimble JM, Follett RF,Stewart BA, editors. Management of carbon sequestration insoil. Boca Raton, FL: CRC Press; 1997. p. 153–73.
[60] Smith DM, Larson BC, Kelty MJ, Ashton PMS. The practice ofsiviculture: Applied forest ecology. 9th ed. New York: Wiley;1997.
[61] Evans J. Plantation forestry in the tropics. 2nd ed. Oxford:Clarendon Press; 1992.
[62] Government of Kenya. Narok District development plan.Ministry of Finance and Planning; 2002.
[63] Fargione J, Hill J, Tilman D, Polasky S, Hawthorne P. Landclearing and the biofuel carbon debt. Science 2008;319:1235–8.
[64] Searchinger T, Heimlich R, Houghton RA, Dong F, Elobeid A,Fabiosa J, et al. Use of U.S. croplands for biofuels increasesgreenhouse gases through emissions from land-use change.Science 2008;319:1238–40.
[65] UNFCCC. Report of the executive board of the cleandevelopment mechanism: thirty-seventh meeting, UNFCCC,No. CDM-EB-37. 21. Available at: http://cdm.unfccc.int/EB/037/eb37rep.pdf; 2008.
[66] IEA. Energy statistics of non-OECD countries 2000–2001.Paris: The Organization for Economic Cooperation andDevelopment (OECD); 2003.
[67] Nyang F. Household energy demand and environmentalmanagement in Kenya. Amsterdam: Thela ThesisPublications; 1999.
[68] IPCC. Climate change 2001: the scientific basis. Cambridge,UK and New York, NY, USA: Cambridge University Press;2001.
Abbreviations used in the text
A/R: afforestation and reforestationAG: abovegroundBG: belowgroundBAU: business-as-usualCDM: Clean Development MechanismCDM-EB: CDM Executive BoardEM: earthmound (kiln)ER: emissions reductionEUC: eucalyptus (modeling scenario)GIS: Geographic Information SystemGHG: greenhouse gasGWP: global warming potentialIK: improved kilnIPCC: Intergovernmental Panel on Climate ChangeLUC: land-use changeLPG: liquid petroleum gasNRB: non-renewable biomassTARCH: Tarchonanthus camphoratus (modeling scenario)