Supplementary Material

Willow Hallgren, C.Adam Schlosser, Erwan Monier, David Kicklighter, Andrei Sokolov, Jerry
Melillo

Climate Impacts of a Large-Scale Biofuels Expansion

Methods  

Data
The data used in the study consists of 1-year snapshots at 1990 and 2050 of transient land cover change as simulated by the Terrestrial Ecosystem Model (TEM, Melillo et al., 1993; Felzer et al., 2004; Sokolov et al., 2008), coupled to the MIT Emissions Predictions and Policy Analysis model, (EPPA, Paltsev et al., 2005; Gurgel et al., 2007), as described in Melillo et al. (2009a). In order to utilize the EPPA-TEM-IGSM land use/cover data we translated the IGSM land cover classification scheme (Schlosser et al., 2007) consisting of 34 land cover types, to the CLM3.0 classification scheme, consisting of 17 plant functional types (PFTs) required by the CLM3.0 input land surface dataset (Hallgren et al., 2012). In addition to the land-use information, radiative forcing variables (trace gas, and aerosol concentrations, solar constant, etc.) from a 550 ppm stabilization scenario climate simulation by the IGSM (Melillo et al., 2009a) were used in CAM to simulate the climate in 2050. 

Since the aim of our study was to look at the specific impact of biofuels on climate in 2050, and not at how transient land-use change impacts climate over time, we only used snapshots of the land use change over time from the coupled economics-ecosystem model (EPPA-TEM).  Thus we performed equilibrium simulations for 2050, not transient simulations leading up to 2050. Although in reality, the climate is never in equilibrium, our study can serve as an upper bound in terms of the climate response to land cover change due to biofuels, and as a prelude to an analysis of the impact of transient vegetation change. Nevertheless, it should be noted that in the Melillo et al. (2009) study that served as the basis for this equilibrium analyses, land-use changes and their associated impacts were not necessarily permanent. Managed lands could be abandoned and revert to natural vegetation as dictated by the economic policy. While the global trends indicate an increase biofuel production, certain regions show a decrease in production for both cases considered, such as: the United States, India, and Indonesia (e.g. Kicklighter et al., 2012). Such occurrences then support climate mitigation through terrestrial carbon sequestration through the latter half of the 21st century.

Modification of the Land surface in CLM
In order to utilize the EPPA-IGSM-TEM derived land cover data we had to translate the IGSM-TEM land cover classification scheme, consisting of 34 land cover types, to the CLM3.0 classification scheme, consisting of 17 plant functional types (PFTs) required by the CLM3.0 input land surface dataset.  Table S1 shows how this was done.

In order to translate the IGSM-TEM land cover classification scheme to that required by CLM3.0, some decisions about how to reclassify the IGSM-TEM land cover types 18-35 had to be made. Since 100% of the IGSM-TEM land cover had to be accounted for in the reclassification process, some of the reclassifications were not ideal; for example, the IGSM-TEM classification scheme had several types of wetland, and with no equivalent CLM vegetation type, these wetland land cover types were reclassified as bare ground, even though this is clearly not a suitable category. There were other land cover types in IGSM-TEM which had no equivalent: land cover types 30, 31 and 34 are registered by the IGSM -TEM output dataset but they are not assigned any land area, so these have safely been recategorized as bare ground. Where an IGSM-TEM land cover type could be justifiably reclassified as more than one of the CLM PFTs, it has been split into two or more percentages of CLM PFTs (e.g. IGSM-TEM vegetation type #23 (mangroves) has been split evenly into constituting 50% of CLM PFT# 4, and 50% of PFT# 5. Table S1 details exactly how the 35 IGSM-TEM land cover types were translated into 17 CLM PFTs.

We changed the currently unparameterized crop type in CLM3.0 (PFT #17) to reflect the second generation biofuel land cover output from TEM. In the context of the present study, the term Òsecond-generationÓ refers primarily to the economics of the technologyÕs penetration into the global energy portfolio.  Therefore, the area of the land cover regarded as Ôsecond-generation biofuelsÕ that is input into the climate model, reflects the economic response Ð and the biogeophysical (i.e. albedo) and biogeochemical (i.e. carbon exchange) aspects reflect a generic biofuel crop. We are currently working on a project that looks at the differential climate impact of first and second generation biofuel crops and of different second-generation biofuel crops."

Further details about the models and experiments 

All component models of the integrated modeling framework used in this study have been thoroughly evaluated. CAM has been thoroughly evaluated in the literature  (Collins et al., 2004, 2006, Gettelman et al., 2006; Bonan, G. B., and S. Levis, 2006; Boville et al., 2006) and is as ÔvalidatedÕ as any other climate model currently in widespread use in the scientific 
community.  TEM is a calibrated ecosystem model (i.e. it is calibrated to observations), and EPPA is grounded on GTAP data, which are observed econometric covariations, correlations and elasticities based on observation. The GTAP Data Base is a fully documented, publicly available global data base which contains complete bilateral trade information, transport and protection linkages among 113 regions for all 57 GTAP commodities. The EPPA-TEM coupling has been examined in other literature (Gurgel et al 2007), and the IGSM has been evaluated  (Prinn et al., 1999; Sokolov et al., 2005, 2008).

A total of six simulations were done using CAM3.1, at a spatial resolution of 2¡ x 2.5¡, and with a climate sensitivity of 2.25. These consisted of simulations with both the climate forcing and the land surface dataset set to 2050 conditions, for both of the two land use scenarios (case 1 and case 2). We performed two simulations for each land use scenario; one with a cellulosic biofuel-based energy policy implemented in 2026 in the original IGSM-TEM data, and one where this energy policy is not implemented.  

In order to compare the biogeophysical surface temperature response to this global biofuel energy policy with the biogeochemical temperature response, using the carbon balance results of Melillo et al (2009b), and N2O emissions from subsequent IGSM-TEM simulations (Kicklighter, 2012), another two simulations were conducted which added onto the default CO2 used by CAM3.1, the total (i.e. direct and indirect) carbon gain (or loss) to the earth system (in ppmv CO2) from both case 1 and case 2, respectively, as calculated by Melillo et al. (2009b), as well as this N2O emission data from (Kicklighter, 2012) which was added on to the default N2O input into CAM.  Although the direct and indirect biogeochemical impacts of biofuels on climate could also have been analyzed separately (Melillo et al. (2009) did distinguish between direct and indirect carbon debt associated with biofuels and the carbon emitted to the atmosphere due to growing biofuels), we believe this important issue is best explored in a separate paper. 

The modelling framework employed in this study builds on that of Melillo et al., 2009b, and can be illustrated by the following figure (S1):

Author Contributions  
The primary research was carried out by W. H. with ongoing intellectual, technical and editorial input from C.A. S.  E. M., supplied advice on the design of the modeling experiments, as well as technical, scientific and editorial advice. A. S. provided the GHG data for use in CAM and technical and scientific advice.  D. K. and J. M. supplied the input land-use data and provided the theoretical underpinning of the land- use change context of this work, as well as advice experimental design, and editorial input.

Additional References for Supplementary Material
Gettelman, A., Collins, W.D., Fetzer, E.J., et al. (2006), Climatology of upper-tropospheric relative humidity from the Atmospheric Infrared Sounder and implications for climate. Journal of Climate, 19(23): 6104-6121. ÊÊ