GCAM Modeling of Bioenergy and Carbon Emissions from Land Use Change MARSHALL WISE, LEON CLARKE, KATE CALVIN, PAGE KYLE, HAEWON MCJEON Joint GCAM Community Modeling Meeting and GTSP Technical Workshop Joint Global Change Research Institute College Park, Maryland, USA September 20, 2012
Bioenergy and Land Use Emissions: GCAM Background! GCAM has always included terrestrial carbon emissions from land use change for biomass energy, as well as from all land use activities.! Terrestrial carbon is tracked for each use of land in each of the model s regions (currently 151), and changes to those uses result in land use change emissions or uptake.! In all scenarios, net terrestrial carbon emissions or uptake is part of the computation of atmospheric CO2 concentrations.! In scenarios where we value terrestrial carbon (e.g., our UCT as in Wise et al 2009 in Science), terrestrial carbon impacts are factored into the economic land use decisions.
Bioenergy and Land Use Emissions: Objective/Motivation! To isolate and quantify the land use carbon emissions impact from specific biomass sources/crops, using GCAM.! Provide better understanding, diagnostics, and explanation for what goes on in myriad calculations behind the scenes in GCAM.! Compare our results with the literature.! Contribute to a better understanding of the numbers, drivers, and factors related to biomass energy and land use emissions.! Perhaps we can help simplify the understanding,! Or if that doesn t work, we can try to simplify the explanation of how complex it is.
IPCC-Cited Estimates of Biofuels and Land Use Change Emissions Figure taken from IPCC Special Report on Renewables, Chapter 9. LUC-Related GHG Emissions [g CO 2 eq/mj] 200 150 100 50 0 Total Uncertainty Range Central Tendency Range -50 European Wheat Ethanol US Maize Ethanol Brazilian Sugarcane Ethanol Soybean Biodiesel Rapeseed Biodiesel References: 3 5 2 3 2 Figure 9.11 Illustra1ve es1mates of direct and indirect LUC- related GHG emissions induced by several first- genera1on biofuel pathways, reported here as ranges in central tendency and total reported uncertainty. Es1mates reported here combine several different uncertainty calcula1on methods and central tendency measures and assume a 30- year )me frame. Reported under the x- axis is the number of references with results falling within these ranges (Sources: Searchinger et al., 2008; Al- Riffai et al., 2010; EPA, 2010b; Fritsche et al., 2010; Hertel et al., 2010; Tyner et al., 2010).
Bioenergy and Emissions in the Literature! Studies range from illustrative case studies (e.g., Fargionne, Searchinger), very detailed Life Cycle Analyses (e.g., Argonne GREET, Oak Ridge), near-term detailed CGE modeling (e.g., GTAP), both (Wang et al 2011: Argonne, GTAP, and DOE) and some longterm modeling (e.g., MIT EPPA, others).! Much US focus on nearer-term ethanol and impacts of RFS and Low Carbon Fuel Standards! E.g., California Air Review Board! US EPA Office of Transportation and Air Quality.! Almost all focus has been on biofuels.! Though analyses that include lignocellulosic crops would also be applicable to electric power generation.! On the one hand, there is a wide range of estimates of carbon emissions change from biomass energy, as expected from different models with different methods, assumptions, and approaches.! However, most of the economic modeling results are at least in the same universe.
Other Estimates of Biofuels and Land Use Change Emissions! Figure 2 taken from Prins, Stehfest, Overmars and Ros. 2010. Are Models Suitable for Determining ILUC Factors? Netherlands Environmental Assessment Agency, Bilthoven.! Note that these studies were not done under uniform assumptions.
Land Use Change Emission: Definitions! Here, we use GCAM to measure the carbon emissions from land use change per unit of biomass energy : i.e., terrestrial carbon emissions that result from any changes in land use in all categories of land across the globe.! These are total emissions including direct and indirect emissions.! Direct emissions: Land use change (LUC) emissions that occurs in place where the biomass is grown, displacing other land that stores carbon (could potentially be positive or negative)! Indirect emissions: LUC emissions that occur away from where the biomass is grown when a biomass energy crop displaces food crops or wood production in one region that results in clearing of land to replace this production in other regions (requires global economic modeling)! Within the region of biomass growth, the distinction between direct and indirect emissions may be ambiguous, and the distinction is not as important as the total (also expressed by Wang et al 2011).! What is important is to get the total right, not the categorization.
Typical Pattern of Carbon Emissions from Land Use Change due to Biomass Expansion! Initial pulse from net change in vegetative carbon followed by long tail of lagged changes in soil carbon.
GCAM Approach to Computing LUC Emissions Attributable to Biomass! Modeling approach is to run GCAM to measure the global land use change emissions that results from an incremental (or marginal or additional) amount of a specific type of biomass in a specific region. (i.e., a numerical derivative approach, requires 2 runs).! This approach isolates the impact of a specific amount of biomass energy on land use change (i.e., a numerical partial derivative approach).! Avoids confounding emissions with biomass expansion in following years and their corresponding emissions pulses.! Avoids averaging of emissions impacts over different types of resources, where resources with zero emissions like residues would be penalized.! Results will vary depending on several factors including whether the biomass is a residual product or a dedicated crop, the type of biomass crop, where the crop is grown, and land use policy.
GCAM Experiment: Switchgrass Crops in the US! For this study, we model an incremental amount (0.1 EJ) of switchgrass grown in US AEZs 7 through 12.! We assume model year 2020.! We assume no climate policies or land use policies globally.
Emissions Results for US AEZ 7-12 (for just this one scenario with no policy)! Right hand chart simply lops off the first year to show that there are differences in net soil carbon in subsequent years.! They are smaller but not at all negligible.! Next slides look at AEZ 7 vs 10. AEZ 7 has a lower switchgrass yield than 10, yet AEZ10 has higher land use change emissions
GCAM US Land Use Regions
AEZ 7 and 10: 2005 Land Use! Differences in extent of cropland, and amounts of such as forests, shrubland, and pasture in non-commercial uses
GCAM s Response to Biomass Increase and Corresponding Land Use Change Emissions! One key factor that is independent of the model is the assumed yield of the biomass crop.! Future and current values are uncertain.! Assumption about future productivity in all crops also important.! GCAM responds based on its economic algorithms and on the regional physical land and agricultural technology characteristics.! Paradigm is equal profitability of all land uses at the margin within a land region, with market equilibration in all products across land regions.! Calibration to historical base year provides inferences on economics of agriculture and land use in each region.! All else equal, expansion of any one use or crop faces diminishing returns to profit at the margin.
AEZ 7 and 10: Land Use Response to Switchgrass Supply! Much less indirect effect from AEZ 7! Less cropland displaced, and some that is displaced is lower-yielding! (Note: 0.1 EJ is the marginal increment used for calculation)
AEZ 7 and 10: net Cumulative Change in Terrestrial Carbon! Higher carbon density of forest becomes evident.! Negative net change in stock means positive terrestrial carbon emissions
AEZ 7 and 10: Net Cumulative Emissions inside and Outside of the Region! To compare to studies cited in IPCC, assuming 30-year horizon, these are equivalent to about 18 to 32 grams CO2/MJ of primary biomass! To compare to studies for liquid fuels, would have to make assumptions about conversion efficiencies as well as how much biomass vs fossil energy is used in conversion.
Repeat: Other Estimates of Biofuels and Land Use Change Emissions! Figure 2 taken from Prins, Stehfest, Overmars and Ros. 2010. Are Models Suitable for Determining ILUC Factors? Netherlands Environmental Assessment Agency, Bilthoven.! Note that these studies were not done under uniform assumptions.
Dealing with the Varying Time Profile of land Use Change Emissions! Unlike fossil fuel emissions, land use change emissions have a time profile that makes it difficult to assess physically and economically.! Upfront carbon emissions may be much higher than fossil fuels, but eventually they asymptotically approach zero even as biomass energy continues to be produced on that land each year.! EPA, CARB, and some other studies have adopted a straightforward approach (for now) of assuming a fixed 30-year planning horizon (based on a typical life of a project).! This is not unreasonable, but it is admittedly completely arbitrary with respect to the impact of emissions on the atmosphere.! What happens after 30 years?! Discounting is also considered, and to a first order an assumed time horizon for payback implies a discount rate.! Again, 30 years may not be unreasonable wrt discount rate.
AEZ 10: Switchgrass Emissions Averaged over Fixed Time Horizons! 30 year horizon is common convention but just that.! Go out far enough and the terrestrial carbon emissions rate would approach zero (though there would likely be some nitrogen and other gases from agricultural operations).
AEZ 10: Switchgrass Emissions Discounted Annual Equivalent, Considering CCS! For this crop in this scenario, can find a discount rate where the biomass is equivalent to fossil fuels.
AEZ 10: Switchgrass Emissions Discounted Annual Equivalent Values! With BioCCS, must also consider the carbon content of the biomass itself, not just the land use change emissions.! GCAM, under UCT policies (i.e., where terrestrial carbon is valued), makes thousands of such calculations in land allocation, and it does include a discount rate as an input factor.
Further Considerations! We will also run emissions computations on other bioenergy crops to see any insights or differences beyond simple yield differences.! Recall that these land use emissions measures were made under the assumption of no land use or terrestrial carbon policy.! Emissions are taken as is, with no incentive to lower them.! Under a UCT policy in GCAM, there are economic incentives to reduce direct and indirect emissions by! choosing less carbon intensive lands for growing biomass,! and reducing indirect effects by reducing indirect land use change and targeting it to less carbon intensive lands.! But there may be also carbon trade-offs when bumping up against an afforestation margin at some level of biomass.! These scenarios, as well as other land use policies should be run to illustrate the biomass decisions in these cases.