Supporting Online Information (SOM)
A - Detailed description of the land use modules
Global Change Assessment Model (GCAM)
GCAM includes a spatially-disaggregated land use model that models land cover, land use, and
production of agricultural and forest products, as well as ecosystem types. Energy, agriculture, forestry,
and land markets are integrated in GCAM, along with unmanaged ecosystems and the terrestrial carbon
cycle. GCAM determines the demands for and production of products originating on the land and the
carbon stocks and flows associated with land use. GCAM 3.0 divides the globe into 151 land use
subregions based on a mapping of up to 18 AEZs (Monfreda et al, 2009) within each of GCAM’s 14 global
geo-political regions, as shown in Figure 1. These AEZs are defined as zones with similar temperature
and precipitation levels, and as such are a useful division of land for modeling agriculture and other land
use. A complete description of GCAM 3.0 modeling of agriculture and land use is provided in Wise and
Calvin (2011) and Kyle et al (2011).
Figure 1. GCAM 3.0 Geopolitical and Land Use Region Map
Within each of these 151 subregions, GCAM categorizes land into approximately a dozen types based on
cover and use. Some of the land types in the subregions, such as tundra and desert, are not considered
arable. Among arable land types, further divisions are made for lands historically in non-commercial
uses such as forests and grasslands as well as commercial forestlands and croplands. Within each
subregion, arable land is allocated across a variety of uses based on expected profitability, which
depends on the productivity of the land, the non-land costs of production (labor, capital, fertilizer, etc.),
and the price of the product. Production of approximately fifteen crops and commercial forest product
is currently modeled, with yields of each specific to each of the 151 subregions. Technical change in
GCAM is exogenously specified, with yield increase rates differentiated by region and time period.
Increases in technical change reduce pressures for land expansion. While this version of GCAM does not
track nitrogen fertilizer inputs explicitly, N2O emissions from agriculture are included and respond to
technical change. We assume that N2O emissions are constant per unit of crop produced, and thus, they
will increase per unit of land as yields increase. The model is designed to allow specification of different
options for future crop management for each crop in each subregion, and the model structure itself
allows for other regional breakdowns besides the AEZs. In the version used for the EMF study, GCAM
does not distinguish between different agricultural inputs (such as fertilizers, energy or machinery).
Therefore, the impacts of increasing energy prices on agricultural production costs are not taken into
account. Furthermore, the energy to dry cellulosic biomass for energy production is not considered.
Because it is an IA model, land use modeling in GCAM is not limited to agriculture but is instead
comprehensive in scope of modeling land use and land cover. Land in each of each subregions is divided
into one of several land use and land cover categories, with the soil and vegetative terrestrial carbon in
each category in each subregion modeled (Kyle et al, 2011). These land categories include lands for
commercial uses such as cropland, commercial pasture, forest products, and bioenergy crops, as well as
non-commercial but arable lands such as non-commercial forestlands, grasslands, and shrublands. For
both commercial and unmanaged forestlands, the GCAM models the temporal accumulation of
terrestrial carbon based on growth profiles specific to each subregion. Non-arable lands such as tundra,
desert, and urban land are also tracked but considered fixed for this study. The amounts of land in each
of the arable land categories, including the distinction between commercial and non-commercial land
coverage, are not rigid in GCAM. As GCAM models future periods, the amount of land devoted to each
of these categories and uses changes in response to socioeconomic, policy, and technology drivers, and
the net terrestrial carbon emission or net carbon uptake from these land use changes are computed.
As with the GCAM energy system, the economic modeling approach for GCAM agriculture, forest, and
land is that of an integrated economic equilibrium in the products, sectors, and factors that are
modeled. Markets, for products such as corn, wheat, wood, or bioenergy crops must be cleared so that
supplies are equal to demands in each model period. Depending on the product or on user
specifications, markets can be cleared globally, regionally, or across groups of regions.
GCAM models the production of several types of bioenergy: traditional bioenergy (straw, dung, fuel
wood, etc.), bioenergy from waste products (including crop residues, municipal solid waste, and black
liquor from the pulp and paper industry), and purpose-grown bioenergy crops. Purpose grown
bioenergy crops, including perennial grasses like switchgrass, and woody crops such as willow, are
modeled as economically competing for land with all other agriculture, forestry, and other uses of land.
Food crops, such as corn, soybeans, and sugar, can also be used as energy feedstocks to be supplied to
GCAM’s energy transformation and use sectors.
Finally, GCAM incorporates non-CO2 emissions from agricultural production and CO2 emissions from
land use change. Land-use change CO2 emissions are calculated by first computing the change in
equilibrium carbon stock resulting from a change in land cover, and then allocating that change in stock
across time. We assume that reductions in vegetation carbon stock happen instantaneously, but
accumulations of vegetation carbon and changes in soil carbon occur over long time horizons. The
specific time depends on the land type and region, with colder regions taking longer to equilibrate. For
non-CO2 emissions, we track emissions from agricultural production, agricultural waste burning, forest
fires, deforestation, and savannah burning. Emissions coefficients are computed in the base year using
the inventory data compiled for the RCPs (Thomson et al., 2011). When a carbon price is applied, the
price is directly applied to CO2 emissions from land-use change. The carbon price will also affect the
emissions factors for non-CO2 emissions from agricultural production. However, the carbon price only
affects the cost of agricultural production through the pricing on land-use change CO2 emissions. The
cost of non-CO2 mitigation is not included in the cost of agricultural production.
MAgPIE (Model of Agricultural Production and its Impact on the Environment)
The global land-use model MAgPIE (Lotze-Campen et al., 2008, 2009, Popp et al. 2010) is a recursive
dynamic optimization model with a cost minimization objective function, which has been coupled to the
grid-based dynamic vegetation model LPJmL (Bondeau et al., 2007). It takes regional economic
conditions such as demand for agricultural commodities, level of agricultural technology, and production
costs as well as spatially explicit data (0.5 degree data aggregated to 200 clusters ) on potential crop
yields, land and water constraints (from LPJmL) into account and derives specific land use patterns,
yields and total costs of agricultural production for each grid cell.
LPJmL computes potential irrigated and non-irrigated yields for each crop within each grid cell as an
input for MAgPIE. In case of pure rain-fed production, no additional water is required, but yields are
generally lower than under irrigation. In addition, LPJmL has been applied a priori to simulate cell
specific available water discharge under potential natural vegetation and its downstream movement
according to the river routing scheme implemented in LPJmL. Then, if a certain area share is assumed to
be irrigated in MAgPIE, additional water for agriculture is taken from available discharge in the grid cell.
MAgPIE endogenously decides on the basis of minimizing the costs of agricultural production where to
irrigate which crops.
The objective function of the land use optimization model is to minimize total cost of production for a
given amount of regional food and bioenergy demand. Regional food energy demand is defined for an
exogenously given population in 10 food energy categories (Bodirsky et al. 2012). Bioenergy is supplied
from specialized grassy and woody bioenergy crops, i.e. Miscanthus, poplar and eucalyptus. All demand
categories are estimated separately for 10 world regions and have to be met by the world crop
production. Additionally, the regions have to produce a certain share of their demand domestically to
account for trade barriers (Schmitz et al., 2012).
Four categories of costs arise in the model: production costs for livestock and crop production, yield
increasing technological change costs, land conversion costs and intraregional transport costs.
Production costs, containing factor costs for labour, capital, and intermediate inputs, are taken from
GTAP (Narayanan and Walmsley, 2008). In order to increase total agricultural production, MAgPIE can
either invest in yield-increasing technological change or in land expansion. The endogenous
implementation of technological change (TC) is based on a surrogate measure for agricultural land use
intensity (Dietrich et al., 2012). Investing in TC leads not only to yield increases but also to increases in
agricultural land-use intensity, which in turn raises input costs (proportional to agricultural land use
intensity) and costs for further yield increases (Dietrich et al. 2013). As MAgPIE does not distinguish
between different agricultural inputs (such as fertilizers, energy or machinery) impacts of increasing
energy prices on agricultural production costs are not taken into account. The other alternative for
MAgPIE to increase production is to expand cropland into non-agricultural land (Krause et al., 2009;
Popp et al., 2011). Cropland expansion involves land conversion costs for every unit of cropland, which
account for the preparation of new land and basic infrastructure investments. Land conversion costs are
based on country-level marginal access costs generated by the Global Timber Model (GTM) (Sohngen et
al., 2009). Moreover, land expansion in MAgPIE is restricted by intraregional transport costs which
accrue for every commodity unit as a function of the distance to intraregional markets and the quality of
the infrastructure. The data set is based on GTAP transport costs (Narayanan and Walmsley, 2008) and a
30 arc-second resolution data set on travel time to the nearest city (Nelson, 2008). Finally, MAgPIE
incorporates non-CO2 emissions from agricultural production (Popp et al. 2010, Bodirsky et al. 2012)
and CO2 emissions from land use change (Popp et al. 2012). CO2 emissions from land use change occur
if the carbon content of the previous land use activity exceeds the carbon content of the new land-use
activity. Like other crops, bioenergy crops require nitrogen inputs for their production as nitrogen
fertilizers restock the nutrients removed from the soil by harvest and increase biomass in nutrient poor
soils. But nitrogen input inevitably leads to emission of N2O, which is a byproduct of soil nitrification and
denitrification. Therefore, to assess N2O emissions from the soil due to large-scale bioenergy production
the model first calculates the application of nitrogen fertilizers based on the assumption that nitrogen
withdrawn from the agricultural system due to the harvest of biomass has to be replaced by synthetic
fertilizers to allow for permanent production. Emission factors are connected directly to the
corresponding nitrogen flows of inorganic fertilizer application and emission calculations are in line with
the 2006 IPCC Guidelines of National Greenhouse Gas Emissions (Eggleston et al. , 2006 ), accounting for
NOx, NHy as well as direct and indirect N2O emissions. In MAgPIE, the energy and associated emissions
to dry cellulosic biomass for energy production is not considered. All considered GHG emissions increase
agricultural production costs if GHG emissions are getting priced in the land system. The model has the
option to reduce agricultural non-CO2 emissions due to improvements in management based on
marginal abatement costs of Lucas et al (2007) or avoiding land expansion into high carbon ecosystems
such as tropical forests.
IMAGE (Integrated Model to Assess the Global Environment)
The IMAGE framework (Bouwman et al., 2006) describes various global environmental change issues
using a set of linked submodels describing the energy system, the agricultural economy and land use,
natural vegetation and the climate system. The socio-economic models distinguish 24-26 world regions,
while the natural ecosystems mostly work at a 0.5 x 0.5 degree grid. The model is set-up as a simulation
model.
The use of bio-energy plays a role at several components of the IMAGE system. First of all, the potential
for bio-energy is determined using the land-use model. The land-use model calculates the potential
production of different crops (food crops or bio-energy crops) at a 0.5x0.5 spatial grid based on climate
and soil factors and assumed changes in technology (using a modified version of the GAEZ model). The
land-use model is normally run in interaction with the MAGNET computable general equilibrium model,
where IMAGE provides information on potential productions and MAGNET determines the demand for
food crops and animal products. Next, the IMAGE land use model allocates the required land for food
production to the 0.5 x 0.5 grid, using a set of rules that determine the suitability include potential
yields, distance to infrastructure and distance to water. In IMAGE, technological change, affecting
agricultural yields, is exogenously defined based on FAO projections (Bruinsma, 2003), with yield
increase rates differentiated by region and time period. Additional to this autonomous technological
change, the MAGNET model calculates price-driven substitution between the production factors land,
labour and capital, and endogenously modifies the exogenous yield increase. Increases in agricultural
yields reduce pressures for land expansion.
The interaction with the energy system is implemented by providing information on potential bioenergy production. In this step, several sustainability criteria can be taken into account, i.e. the exclusion
of forests areas, agricultural areas and nature reserves (see van Vuuren et al., 2009). In fact, in default
model runs bio-energy is only allowed on abandoned agricultural land and natural grasslands. Based on
these yield maps, regional supply curves for bio-energy are derived, in 5 year time steps.
In the energy submodel, the demand for bio-energy is assessed by describing the cost-based
competition of bio-energy versus other energy carriers (mostly in the transport, electricity production,
industry and the residential sectors). In the production of bio-energy two “generic” products are
considered: bio-liquid fuels (so use in the transport and residential sector) and bio-solid fuels (as feed
stock for industry and power generation). Bio-liquid fuels can be produced from both first generation
and second generation crops – but the latter dominate the results. In the production of bio-energy costs
for agricultural production, conversion and transport are accounted for. Moreover, the model also
accounts for potential emissions in the production chain. This includes both the N2O emissions
associated with fertilizer use and the expected impact on land-use change emissions. Both emission
sources are estimated on the basis of the expected emission consequences in the land-use system and
are added as a function of total bio-energy production. As most of the bio-energy is used in the form of
woody bio-energy, the N2O emissions were assessed to be low (given assumed low use of fertilizer
compared to food crops) and respond to yield increases. We assume that N2O emissions are constant
per unit of crop produced, and thus, they will increase per unit of land as yields increase. As IMAGE does
not differentiate between different agricultural inputs (such as fertilizers, energy or machinery)
increasing energy prices do not affect agricultural production costs in the model. As bio-energy crops
are only allowed on abandoned agricultural land and non-forest lands, the land-use related emissions
CO2 emissions as calculated in the land system are rather low, and amount to about 20% of the
emissions associated with coal use based.. In addition to bio-energy crops, the model also allows for the
use of residues as feedstock into second generation biofuel production chains and power production.
However, the energy to dry cellulosic biomass for energy production and associated GHG emissions are
not considered in IMAGE.
Next, the resulting demand for bio-energy crops is combined with the demand for other agricultural
products within a region to determine future land use (in the land-use component of IMAGE). For this,
again, the modified version of the GAEZ model and the IMAGE allocation model are used to determine
land use on the 0.5 x 0.5 degree grid. It should be noted that in IMAGE grid cells not used for crop
production or pasture or urban land contain natural vegetation. Here the biome model is used to
determine the vegetation type on the basis of climate and soils. The carbon cycle model in a subsequent
step determines the carbon pools and fluxes associated with each grid cell, based on accounting for
different carbon stocks in each grid cell as a function of land-use and vegetation type. The production of
bio-energy in a grid cell will therefore always imply that this grid cell cannot return to its natural
vegetation state (independent of whether in the baseline this grid cell is abandoned agricultural land or
natural grassland). This implies some net loss of CO3 from the carbon pools (compared to baseline) of
this grid cell to the atmosphere. Production of biomass for bio-energy purposes is also associated with
direct and indirect emissions from N2O due to fertilizer use (as indicated these emissions are assumed to
be relatively low for woody bio-energy). Finally, the emissions associated with land use and land-use
change and the energy system are used in the climate model (MAGICC-6) to determine climate change,
which affects in the next time step all biophysical submodels, including the growth of bio-energy crops.
References
Bodirsky B., Popp A., Weindl I., Dietrich J., Rolinski S., Scheiffele L., Schmitz C., Lotze-Campen H. (2012)
N2O emissions of the global agricultural nitrogen cycle - current state and future scenarios.
Biogeosciences 9 (10): 4169–4197
Bondeau, A., Smith, P.C., Zaehle, S., Schaphoff, S., Lucht, W., Cramer, W., Gerten, D., Lotze-Campen, H.,
Müller, C., Reichstein, M., et al. (2007). Modelling the role of agriculture for the 20th century global
terrestrial carbon balance. Global Change Biology 13, 679–706.
Dietrich JP, Schmitz C, Müller C, Fader M, Lotze-Campen H, Popp A (2012): Measuring agricultural landuse intensity - A global analysis using a model-assisted approach. Ecological Modelling 232: 109–118
Dietrich J.P., Schmitz S, Lotze-Campen, H., Popp, A. and Müller, C. (2013): Forecasting technological
change in agriculture - An endogenous implementation in a global land use model accepted by
Technological Forecasting and Social Change.
Eggleston, H. S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K., and Hayama, K. (Eds.): 2006 Guidelines for
National Greenhouse Gas Inventories, Prepared by the National Greenhouse Gas Inventories
Programme, Institute for Global Environmental Strategies, Kanagawa, Japan, 2006.
Kyle, G. Page et al., 2011. GCAM 3.0 Agriculture and Land Use: Data Sources and Methods. Pacific
Northwest National Laboratory. PNNL-21025.
http://wiki.umd.edu/gcam/images/2/25/GCAM_AgLU_Data_Documentation.pdf
Lotze-Campen, H., Müller, C., Bondeau, A., Jachner, A., Popp, A., Lucht, W., 2008. Food demand,
productivity growth and the spatial distribution of land and water use: a global modeling approach.
Agric. Econ. 39, 325–338.
Lotze-Campen, H., Popp, A., Beringer, T., Mu¨ ller, C., Bondeau, A., Rost, S., Lucht, W., 2010. Scenarios of
global bioenergy production: the trade-offs between agricultural expansion, intensification and trade.
Ecol. Model. 221, 2188–2196.
Monfreda,C.,N. Ramankutty, and T.W. Hertel, “Global Agricultural Land Use Data for Climate Change
Analysis,” in T.W. Hertel, S. Rose, and R. Tol (eds), Economic Analysis of Land Use in Global Climate
Change Policy, Abingdon, Oxon: Routledge (2009).
Nelson, A., 2008. Estimated travel time to the nearest city of 50,000 or more people in year 2000. Global
Environment Monitoring Unit – Joint Research Centre of the European Commission, Ispra Italy.
http://bioval.jrc.ec.europa.eu/products/ gam/ (accessed 30.07.11).
Popp, A., Lotze-Campen, H., Bodirsky, B., 2010. Food consumption, diet shifts and associated non-CO2
greenhouse gases from agricultural production. Global Environ. Change 20, 451–462.
Popp A., Krause M., Dietrich J., Lotze-Campen, H., Leimbach M., Beringer, T., Bauer N. (2012): Additional
CO2 emissions from land use change – forest conservation as a precondition for sustainable production
of second generation bioenergy. Ecological Economics, 74: 64-70
Schmitz, C., Biewald, A., Lotze-Campen, H., Popp, A., Dietrich, J.P., Bodirsky, B., Krause, M., Weindl, I.
(2012): "Trading more Food - Implications for Land Use, Greenhouse Gas Emissions, and the Food
System", Global Environmental Change, 22(1): 189-209.
Sohngen, B., Tennity, C., Hnytka, M., 2009. Global forestry data for the economic modeling of land use.
In: Hertel, T.W., Rose, S., Tol, R.S.J. (Eds.), Economic Analysis and Use in Global Climate Change Policy.
Routledge, New York, pp. 49–71.
Narayanan, B., Walmsley, T.L., 2008. Global Trade, Assistance, and Production: The GTAP 7 Data Base.
Center for Global Trade Analysis, Purdue University, West Lafayett.
Wise, Marshall and Kate Calvin. 2011. GCAM 3.0 Agriculture and Land Use: Technical Description of
Modeling Approach. Pacific Northwest National Laboratory. PNNL-20971. Available at
http://wiki.umd.edu/gcam/images/2/25/GCAM_AgLU_Documentation.pdf
B – Tables SOM
GCAM
Cropland
IMAGE
ReMIND/MAgPIE
1043
1539
1609
0
12
0
Forest
4486
3987
3919
Other Land
4336
4181
2662
Pasture
3306
3280
4721
Bioenergy Cropland
Total land
12824
12999
12911
Table SI1: Global land cover (million ha) in GCAM, IMAGE and ReMIND/MAgPIE for the year
2005.
C – Figures SOM
Figure SI1: Total global food energy demand (measured in terms of energy). Please note that in
GCAM food demand reacts to food prices and lower food demand is observed in 450 FullTech
scenario with higher pressure on the land use system due to land use based mitigation. In
IMAGE and ReMIND/MAgPIE, food demand is prescribed exogenously and therefore does not
differ between scenarios.
Figure SI2: Total global demand for livestock products (measured in terms of energy). Please
note that in GCAM food demand reacts to food prices and lower demand for livestock products
is observed in 450 FullTech scenario with higher pressure on the land use system due to land
use based mitigation. In IMAGE and ReMIND/MAgPIE, food demand is prescribed exogenously
and therefore does not differ between scenarios.
Figure SI3: Regional land pools [million ha] in GCAM, IMAGE and ReMIND/MAgPIE for the year
2005.
Figure SI4: Change in regional land pools [million ha] between 2005 and 2030.
Figure SI5: Change in regional land pools [million ha] between 2005 and 2050.
Figure SI6: Change in regional land pools [million ha] between 2005 and 2100.
Figure SI7: Total cumulative greenhouse gas fluxes from agricultural production (N2O – yellow
bars) and land-use change (CO2 – brown bars) in the reference scenario without bioenergy
production for the Base FullTech scenario aggregated for the time periods 2005-2050 and 20502100 in GCAM, IMAGE and ReMIND/MAgPIE.
Figure SI8: Total cumulative greenhouse gas fluxes from agricultural production (N2O – yellow
bars) and land-use change (CO2 – brown bars) in the reference scenario without bioenergy
production for the 450 FullTech scenario aggregated for the time periods 2005-2050 and 20502100 in GCAM, IMAGE and ReMIND/MAgPIE.