Supplementary Information Air Pollutant Emissions Inventory of Large Scale Production of Selected Biofuel Feedstocks in 2022 Authors: Yimin Zhang, Garvin Heath, Alberta Carpenter, Noah Fisher Table of Contents Table of Contents .......................................................................................................................................... 1 Table of Tables ............................................................................................................................................. 3 Table of Figures ............................................................................................................................................ 4 1 Biomass Production .............................................................................................................................. 5 1.1 Background on the Billion Ton Study Update ................................................................................. 5 1.2 Justification of the Selection of Cellulosic Feedstock Price Target Examined................................ 8 1.3 Comparison of Estimated Total Cellulosic Feedstock Availability in the BTS2 to That Considered in Our Study at the Selected Price Target .............................................................................. 9 1.4 Allocation of Cellulosic Feedstocks Considered to Meet the Cellulosic Biofuel Mandate in the RFS2 in 2022 .......................................................................................................................................... 12 1.5 2 Key Assumptions Used to Estimate County-Level Feedstock Availability .................................. 13 1.5.1 Agricultural residue............................................................................................................. 13 1.5.2 Switchgrass ......................................................................................................................... 14 1.5.3 County-Level Yield Comparison Among CG, WS and SG ................................................ 15 1.5.4 Forest residue ...................................................................................................................... 16 1.6 Number of Counties that Have Projected Production for Each Feedstock in 2022 ....................... 16 1.7 Corn Grain Data Limitations ......................................................................................................... 17 Emissions From Equipment Operation ............................................................................................... 17 2.1 NONROAD Model ........................................................................................................................ 17 2.1.1 2.1.1.1 NH3 Emissions ............................................................................................................ 19 2.1.1.2 PM2.5 Emissions........................................................................................................... 19 2.1.1.3 VOC Emissions ........................................................................................................... 19 2.1.2 2.2 Combustion Emissions Calculation .................................................................................... 18 Age distributions for equipment populations ...................................................................... 20 Equipment Lists ............................................................................................................................. 20 2.2.1 Corn Grain Non-Harvest Activity ....................................................................................... 20 2.2.1.1 3 Corn Grain Irrigation .................................................................................................. 22 2.2.2 Corn Stover and Wheat Straw Harvest ............................................................................... 24 2.2.3 Switchgrass ......................................................................................................................... 26 2.2.4 Forest Residue ..................................................................................................................... 27 Fertilizer and Pesticide Application .................................................................................................... 28 3.1 Nitrogen Fertilizer.......................................................................................................................... 28 1 3.2 Pesticide ......................................................................................................................................... 30 4 Fugitive Dust Emissions ..................................................................................................................... 32 5 Metrics Evaluated ............................................................................................................................... 33 5.1 Emission per gallon of ethanol (Figure 1 in the main text)............................................................ 33 5.2 Contribution Analysis (Figure 2 in the main text) ......................................................................... 34 5.3 Comparison to National Emissions Inventory (NEI) For Cellulosic Feedstocks (Figure 3 in the main text) ................................................................................................................................................ 35 5.4 NEI Ratio Thresholds Compared to Attainment Status for National Ambient Air Quality Standards (NAAQS) (Figure 4 in the main text) .................................................................................... 36 6 Additional Maps and Figures .............................................................................................................. 37 6.1 Biomass Production Density .......................................................................................................... 37 6.2 Emissions Density.......................................................................................................................... 38 6.2.1 6.3 Cellulosic ............................................................................................................................ 39 Individual Feedstock Comparison to NEI ...................................................................................... 47 6.4 Counties with R exceeding thresholds compared to NAAQS Nonattainment Areas ..................... 50 7 Comparison To Other Studies ............................................................................................................. 51 8 References ........................................................................................................................................... 53 9 List of Acronyms...................................................................................................................................61 2 Table of Tables Table S1. Comparison of estimated 2022 production volumes for all cellulosic feedstocks in the BTS2 to those considered in our scenario..................................................................................... 11 Table S2, Switchgrass fertilizer application rates. ........................................................................ 15 Table S3. Number of counties projected to produce each cellulosic feedstock............................ 17 Table S4. Conversion factors for default NONROAD output ...................................................... 20 Table S5. Corn grain conventional till, non-harvest equipment operations ................................. 21 Table S6. Corn grain reduced till, non-harvest equipment operations ......................................... 21 Table S7. Corn grain no-till non-harvest equipment operations ................................................... 21 Table S8. Irrigation equipment as modeled by the NONROAD program .................................... 24 Table S9. Corn stover and wheat straw harvest and transport equipment .................................... 26 Table S10. Switchgrass non-harvest equipment ........................................................................... 26 Table S11. Switchgrass harvest equipment .................................................................................. 27 Table S12. Forest residue equipment ............................................................................................ 27 Table S13. Emission factors for five commonly used nitrogen fertilizers ................................... 30 Table S14. Switchgrass pesticide application schedule (P, lb active ingredient/ac) .................... 32 Table S15, Fugitive dust emission factors (lb of PM10/ac)1.......................................................... 32 Table S16. Constants used to estimate fugitive PM emissions from transportation ..................... 33 Table S17. Ethanol Yields ............................................................................................................ 34 Table S18. NAAQS criteria air pollutant and paired air pollutants or precursors ........................ 36 Table S19. Switchgrass comparison between our estimates and GREET Version 1_2012 ......... 52 Table S20. Corn stover comparison between our estimates and GREET Version 1_2012 .......... 52 Table S21. Forest residue comparison between our estimates and GREET Version 1_2012 ...... 53 Table S22. Forest residue comparison between our estimates and GREET Version 1_2012…...53 3 Table of Figures Figure S1. POLYSYS core modules (UT 2012) ............................................................................. 7 Figure S2. Biofuel quantities produced from total cellulosic feedstocks at selected farm-gate price targets in the BTS2 baseline scenario . ................................................................................ 10 Figure S3. Percentage of respective feedstock among the cellulosic feedstocks considered in this study at $55/dt price target. ........................................................................................................... 12 Figure S4. Cellulosic biofuel (represented by ethanol) production (16 bgy) using 52% of cellulosic feedstocks considered in this study .............................................................................. 13 Figure S5. Distribution of county-level yields (dry tons/acre) for corn stover, switchgrass, and wheat straw. .................................................................................................................................. 16 Figure S6. Single pass combine collection rates (Jacobson 2012) ............................................... 24 Figure S7. County-level distribution of nitrogen application rate ranges for conventional till and no-till corn grain production (no-till and reduced till acres are assumed to have the same nitrogen application rate). ........................................................................................................................... 29 Figure S8. Share of nitrogen fertilizer applied to CG, CS and WS .............................................. 30 Figure S9. Production density for all feedstocks (dry metric ton/square kilometer) .................... 38 Figure S10. Cellulosic emissions density maps (kg of pollutant per km2 county area)................ 41 Figure S11. Corn grain emissions density maps (kg of pollutant per km2 county area)............... 42 Figure S12. Corn stover emissions density maps (kg of pollutant per km2 county area) ............. 43 Figure S13. Wheat straw emissions density maps (kg of pollutant per km2 county area) ............ 44 Figure S14. Switchgrass emissions density maps (kg of pollutant per km2 county area) ............ 45 Figure S15. Forest residue emissions density maps (kg of pollutant per km2 county area) ......... 46 Figure S16. Ratio of switchgrass emissions to NEI non-point and non-road emissions .............. 47 Figure S17. Ratio of corn stover emissions to NEI non-point and non-road emissions. ............. 48 Figure S18. Ratio of wheat straw emissions to NEI non-point and non-road emissions.............. 49 Figure S19. Ratio of forest residue production emissions to NEI non-point and non-road emissions. ...................................................................................................................................... 50 4 1 Biomass Production The selected feedstocks considered for producing cellulosic biofuels in this study are corn stover (CS), wheat straw (WS), forest residue (FR) (logging residues from non-federal forests, in particular), and switchgrass (SG) (a perennial dedicated energy crop). All assumptions in this section are derived from the Billion Ton Study Update (BTS2), which provides a strategic assessment of potential biomass supply for many major feedstocks within the contiguous United States (DOE 2011). SG is assumed to be grown on rain-fed land east of the 100th Meridian. This meridian matches the western boundary of Oklahoma (excluding the panhandle) and bisects North Dakota, South Dakota, Nebraska, Kansas, and Texas. CS and WS are only harvested from acreage where reduced tillage or no till is practiced; collection of crop residues is prohibited from conventionally-tilled acres. All crops and emissions are modeled on an annual basis. Because SG is a perennial crop, details are presented in Section S.3.2.3 to describe how annual emissions are calculated. This section provides additional information on feedstock production in the Methods (Section 2) of the main text. 1.1 Background on the Billion Ton Study Update The first Billion Ton Study (BTS) report (Perlack et al 2005) estimated the “potential” biomass within the contiguous US considering a number of assumptions about current and future inventory and production capacity, availability, and technologies. The first BTS was a nationallevel study based on 2005 US Department of Agriculture (USDA) agricultural projections, 2000 forestry Resources Planning Act (RPA) and Timber Products Output (TPO). The BTS2 (DOE 2011) updated the estimates in the first BTS and addressed a number of shortcomings. One particular update, which is critically important to our air quality analysis, is that the BTS2 estimates provide county-by-county inventory of each individual feedstock examined in this study. The BTS2 evaluated two scenarios—baseline and high-yield. The baseline scenario assumes a continuation of the USDA 10-year forecast for the major food and forage crops and a continuation in trends toward no-till and reduced cultivation. It also extends the study an additional 10 years to 2030. Energy crop yield reflects learning or experience in 5 planting energy crops and limited gains that can be made through breeding and selection of better varieties. The high-yield scenario is more closely aligned to the assumptions in the 2005 BTS; the projected increase in corn yield averages almost 2% annually over the 20-year simulation period while the energy crop productivity increases are modeled at three levels—2%, 3%, and 4% annually. These gains are due not only to experience in planting energy crops, but also to more aggressive implementation of breeding and selection programs. We assume the baseline scenario for analyses of this study. The BTS2 used the Policy Analysis System (POLYSYS) model (UT 2012) to estimate production of agricultural biomass. POLYSYS is a modeling framework developed to simulate changes in policy, economic, or resource conditions and to estimate the resulting impacts on the US agricultural sector (UT 2012). It is structured as a system of interdependent modules simulating crop supply for 305 production regions, national crop demand and prices, national livestock supply and demand, and agricultural income (UT 2012). The modules also include crop rotation and management practices, environmental impacts, and production of energy crops on multi-year production cycles. POLYSYS uses as anchors for its analyses national baseline projections and related assumptions from the USDA, Food and Agriculture Policy Research Institute (FAPRI) and the US Congressional Budget Office (CBO). The land base for agricultural resources in the BTS2 covered 311 million acres of cropland and 140 million acres as cropland pasture and permanent pasture (DOE 2011, Section 5.2.1). Sustainability criteria were used in determining the availability of crop residues. The limiting factors for residue removal include soil organic carbon, wind and water erosion, plant nutrient balance, soil water and temperature dynamics, soil compaction, and offsite environmental factors. 6 Figure S1. POLYSYS core modules (UT 2012) The non-federal forest biomass and wood waste resources considered in this BTS2 (DOE 2011) include: forest residues from integrated forest operations of timberland, other removal residue, thinnings from other forestland, unused primary and secondary mill processing residues, urban wood wastes, and conventionally sourced wood. Estimation of forest biomass in the BTS2 was derived from various data sources, including an annual survey of industrial users of roundwood by USDA Forest Service, Resource Planning Act (RPA) projections and forest inventory and analysis data of the USDA Forest Service on logging residues, thinnings, and other removals. The BTS2 considered 504 million acres of non-federal timberland and 91 million acres of other forestland in assessing the forestland resources. Forestry resources considered in BTS2 are only those that are not currently utilized. In all cases, practices for harvesting forest biomass were expected to have the goals of protecting ecological functions and minimizing negative 7 environmental impacts (DOE 2011). The forest estimates in BTS2 consider residue access, recovery and merchantability, and other requirements for resource sustainability. BTS2 (DOE 2011) is used in this study for estimating county-level feedstock quantities required to meet the cellulosic biofuel mandate in the updated Renewable Fuel Standard (RFS2) for 2022. While other data sources (such as the USDA National Agricultural Statistics Service (NASS)) are available for estimating biomass feedstock production, the BTS2 data are selected because they provided consolidated county-level estimates with an extensive feedstock coverage with projections through 2030, and allowed for selecting feedstock price targets to model potential feedstock availability by different feedstock uses. The price selected here is used solely for estimating air emissions from supplying this amount of biomass to meet anticipated demand from biopower and cellulosic biofuel production in 2022, and likely will change in the future due to factors such as inflations. The price and quantities utilized in this assessment reflect estimates when the BTS2 data was officially released to the public. 1.2 Selection of Cellulosic Feedstock Production Scenario The quantity of biomass feedstock which could be made available for biofuel production varies by the price the biofuel producers are willing to pay. Biofuel producers compete with other biomass users such as biopower generators in the market. Hence, the market price of biomass will be dictated by the combined demand of the major users. To understand the minimum price required to meet demand from two potentially largest biomass uses in the near future, i.e., advanced biofuels and biopower, Langholtz et al (2012) employed POLYSYS to estimate the minimum prices (in 2011 dollars) under price-run and demand-run scenarios, respectively. The price-run scenario mimicked long-term contracting conditions, where the biomass users lock a fixed price for an extended period while the demand-run scenario assumed a gradual price change due to varying demand over time. The combined demand modeled by Langholtz et al (2012) took into account biomass quantity required to meet additional 96.6 billion kWh of biopower and 21 billion gallons of advanced biofuels required by the RFS2 by 2022. The estimated minimum farm-gate price is $53/dry short ton ($58.43/ dry metric ton) and $62/dry short ton ($68.36/dry metric ton), respectively, under the price-run and demand-run scenarios by 2022. County-level cellulosic feedstock availability used in this study is based on a $55/dry short 8 ton ($60.64/dry metric ton) farm-gate price, which is slightly lower than the estimated price under the demand-run scenario by Langholtz et al (2012) because we only consider the cellulosic biofuel mandate of the RFS2, which is 16 billion gallons by 2022. We assume some advanced biofuels will be produced from non-lignocellulosic feedstocks such as algae feedstock. 1.3 Comparison of Estimated Total Cellulosic Feedstock Availability in the BTS2 to That Considered in Our Study at the Selected Price Target Although the BTS2 assessed a large number of cellulosic feedstocks (Table S1), our study focuses on three agricultural feedstocks (i.e., CS, WS, and SG) and one forest feedstock (i.e., FR). CS and WS are the two largest agricultural residues among all agricultural residue resources. SG is selected because it is expected to be the major contributor to biomass from dedicated energy crops. Logging residues (from non-federal lands) we examine account for 46% of forest biomass. Other feedstocks (not considered here) can be evaluated in the future when data are available. In addition, some feedstocks (such as manure) might be more appropriate for biopower or biomethane production (vs. liquid biofuels assessed here). As noted in the previous section, these quantities in Table S1 are subject to change in the future. Figure S2 shows the percentages of feedstocks we examine among total cellulosic feedstock availability estimated for the baseline scenario by BTS2. The figure also shows the amount of biofuels, which can be produced from the biomass estimated for the baseline scenario of the BTS2. 9 Billion gallons produced 70 60 50 40 50% 30 20 53% 68% 54% 10 0 $45 $50 $55 Farmgate Price Target ($/dry ton) Scenario feedstocks $60 All BTS2 feestocks Figure S2. Biofuel (represented by ethanol) quantities produced from total cellulosic feedstocks at selected farm-gate price target in the BTS2 baseline scenario (BLY+EC1_BLT). The percentages located above ‘Scenario feedstocks’ bars represent the percentages of feedstocks in our scenario compared to all BTS2 cellulosic feedstocks. Table S1 shows the production volume for all cellulosic feedstocks in the baseline scenario of BTS2, as well as cellulosic feedstocks considered in our study at $55/dry short ton ($60.64/dry metric ton). As shown in Table S1, cellulosic feedstocks considered in this study account for 53% of total cellulosic annual production estimated by BTS2 for the baseline scenario (DOE 2011). Figure S3 shows the distribution of mass of each cellulosic feedstock considered at $55/dry short ton ($60.64/dry metric ton) by 2022. SG is estimated to account for 43% of the selected cellulosic feedstocks, followed by corn stover (34%). 10 Table S1. Comparison of estimated 2022 production volumes for all cellulosic feedstocks in the BTS2 to those considered in our scenario. BTS2 CELLULOSIC FEEDSTOCK DESCRIPTION Agricultural Biomass/Residuesa Barley straw Corn stover Cotton residue Hay Oat straw Orchard and vineyard prunings Rice hulls Rice straw Sorghum stubble Sugarcane trash Wheat dust Wheat straw Subtotal Dedicated Energy Cropsb Perennial grassesc Switchgrassd High-yield sorghume Subtotal Forest Biomass Conventional wood Coppice and non-coppice woody crops Integrated operations (LOGRLOGT)f Dry short tons/year Scenario % of total BTS2 1,932,300 114,697,500 5,875,600 133,766,600 19,100 5,541,300 1,668,700 7,361,500 622,100 1,121,200 580,100 28,875,800 302,061,800 0.3% 18.2% 0.9% 21.3% 0.0% 0.9% 0.3% 1.2% 0.1% 0.2% 0.1% 4.6% 48.0% 144,495,800 23.0% 8,915,000 153,410,800 1.4% 24.4% 933,950 61,105,600 34,626,800 0.1% 9.7% 5.5% Dry short tons/year % of total BTS2 114,694,378 18.2% 28,872,601 143,566,979 4.6% 22.8% 144,498,509 23.0% 144,498,509 23.0% 44,992,000 7.2% 44,992,000 7.2% Logging residues (LOGR)g Treatment thinnings, other forest lands Subtotal 906,250 97,572,600 0.1% 15.5% Waste Biomass Manure Urban wood waste, C&D Urban wood waste, municipal solid waste Subtotal 41,913,350 23,388,100 10,754,900 76,056,350 6.7% 3.7% 1.7% 12.1% - 629,101,550 100% 333,057,488 TOTAL 52.9% a Agriculture biomass and residues includes feedstocks currently being produced with production quantities available as bioenergy feedstocks. b Dedicated energy crops are feedstocks that are not currently in production but are projected to be available as bioenergy feedstocks. c Perrenial grasses include switchgrass (Panicum virgatum), big bluestem (Andropogon gerardii), and indian grass (Sorghastrum nutans)(DOE 2011). d The BTS2 total for perennial grass production is the sum of county production estimates each independently rounded to the 100’s prior to summation. The scenario total for switchgrass is from raw data (Eaton, 2013), which explains the difference between the two estimates. 11 Called “annual energy crop” in BTS2 (DOE 2011). The integrated operations production numbers reflect 50% of available logging residues (tops and limbs, LOGR) and 50% of available forest thinnings (LOGT). g Logging residues (LOGR) for the scenario reflects 100% of available logging residues from non-federal lands only. e f WHEAT STRAW 9% SWITCHGRASS 43% CORN STOVER 34% FOREST RESIDUES 14% Figure S3. Percentage of respective feedstock among the cellulosic feedstocks considered in this study at $55/dt price ($60.64/dry metric ton) target. 1.4 Allocation of Cellulosic Feedstocks Considered to Meet the Cellulosic Biofuel Mandate in the RFS2 in 2022 One of the goals of this study is to understand the impact of additional air emissions due to the production of 16 billion gallons of cellulosic biofuels for meeting the mandate in RFS2 by 2022. To estimate the percentage of selected cellulosic biomass needed for producing 16 billion gallons of cellulosic biofuels, we use a generic biofuel (represented by ethanol) yield of 89.9 gallons/dry short ton (99.1 gallons/dry metric ton) for agricultural cellulosic feedstocks (CS, WS, and SG) and a yield of 75.7 gallons/dry short ton (83.5 gallons/dry metric ton) for FR as per Hsu et al (2010). Based on Equation S1, feedstock quantity in Table S1 and assumed ethanol yields, we estimate that about 52% of the BTS2 projected biomass production for the four feedstocks we examine will be required to meet RFS2’s cellulosic biofuel mandate in 2022. % ππππ’ππππ = π πΉπ2 π£πππ’πππππ ππππππ‘π πππ πππππ’πππ ππ πππππ’ππ ππ 2022 πΈπ‘βππππ ππππ π΅ππ2 πππππ’ππ‘πππ π£πππ’πππ πππ ππππ. πππππ π‘ππππ ππ₯ππππππ Equation S1 12 Allocation of feedstocks to different end uses (e.g., biofuel, biopower) is not established in the BTS2 (DOE 2011). Consequently, we apply this estimated national average (i.e., 52%) equally to all counties and all cellulosic feedstocks considered (as discussed in Section 2.3.1 of the main text). Figure S4 shows the amount of biofuels (represented by ethanol), which can be produced from each feedstock examined (after accounting for the 52% share) based on the BTS2 results. 8 7.1 7 5.7 Billion Gallons 6 5 4 3 1.9 2 1.4 1 0 SG CS FR WS Figure S4. Cellulosic biofuel (represented by ethanol) production using 52% of the BTS2 production of cellulosic feedstocks considered in this study (total equals 16 bgy). 1.5 Key Assumptions Used to Estimate County-Level Feedstock Availability 1.5.1 Agricultural residue The amount of agricultural residues generated as a by-product of grain production can be estimated from harvest index (HI), which is defined as the mass ratio of grain to total biomass (DOE 2011, Text Box 4.3). For example, if a HI is 0.5, the stover to grain ratio is 1:1. The BTS2 assumes a static harvest index of 0.5 for CS and WS for all counties in the baseline scenario modeled in this study. To protect soil quality, BTS2 prohibits residue collection from acres using conventional tillage. The amount of agricultural residue that can sustainably be removed from agricultural cropland is subject to two constraints; 1) removals cannot exceed the tolerable soil loss limit as recommended by the USDA’s Natural Resource Conservation Service (NRCS) and 2) removals 13 cannot result in long-term loss of soil organic matter as estimated by the Revised Universal Soil Loss Equation (RUSLE2) and the Wind Erosion Prediction System (DOE 2011). These two constraints were taken into consideration for estimating county-level sustainable residue removal rate in the BTS2. Based on data from existing literature, the BTS2 (DOE 2011) estimated an average nutrient composition of removed CS. Nutrient values used were 14.8 pounds (lb) of nitrogen, 5.1 lb P2O5 (phosphate), and 27.2 lb K2O (potassium) per dry short ton of corn stover (16.3, 5.6, and 30.0 lb/dry metric ton of CS for N, P2O5 and K2O respectively). Similarly, average nutrient values for wheat straw were 15.4 lb nitrogen, 2.7 lb P2O5 and 25.4 lb K2O for each dry short ton removed (17.0, 3.0 and 28.0 lb/dry metric ton of WS for N, P2O5 and K2O respectively). We assume the equivalent nutrients will need to be compensated during corn and wheat production by applying additional fertilizers (i.e., in addition to what is needed without residue removal). Because agricultural residues are a by-product from crop production, we chose the productpurpose allocation approach (Wang et al. 2011) for estimating emissions associated with residue harvesting. In other words, only additional inputs exclusively attributable to residue removal (e.g., additional fertilizer application) are allocated to residues. Because we model a scenario for 2022, a single pass residue collection system (i.e., grain and residue collected simultaneously) designed by INL (Jacobson 2012) is assumed in our study. 1.5.2 Switchgrass The BTS2 assumes SG is grown only on rain-fed lands, which constrains growth to counties east of the 100th meridian except for one county in the Pacific Northwest (DOE 2011). The BTS2 also assumes that to grow SG, the returns must be greater than the pasture rent plus additional establishment and maintenance costs (DOE 2011). Best management practices for establishment, cultivation, maintenance, and harvesting are assumed when determining price-point availability (DOE 2011). The total amount of cropland in any given county that can be converted to SG is limited to 25% (DOE 2011). SG stands are expected to last at least 10 years, after which time the stands could be renovated and replaced with new, higher-yielding cultivars (DOE 2011). The assumed stand life in BTS2 is 14 10 years. No-till planting is assumed for SG establishment (1st year) as it has significant cost and environmental benefits (DOE 2011). After the establishment year, well-established SG stands require limited herbicides. Nitrogen fertilizer is not recommended during the planting year since nitrogen encourages weed growth. Data on fertilizer and herbicide usage for both establishment and maintenance years were provided by Turhollow (2011). A summary of this information is shown in Table S2. Table S2, Switchgrass fertilizer application rates. Year Applied N P2O5 Establishment Year (Year 1) None 40 lb/ac1 Maintenance Years (Years 2-10) 13 lb N/dry short ton harvested (14.3 40 lb/ac1 lb N/dry metric ton) 1 Apply in the Northeast, Lake States, Corn Belt, Appalachia, Southeast, Delta States K2O 80 lb/ac1 80 lb/ac1 Because activities associated with establishment year and maintenance years differ, we assume 1/10th of SG acres in any given county are in each year of the production cycle (i.e., 1/10th in 1st year, 1/10th in 2nd year, so on so forth). For a given year (e.g., 2022), county-level emissions are summed up over acres in each year of the production cycle using Equation S2. 10 πΈπππ π ππππ = ∑ π=1 π΄ × πΈπ 10 Equation S2 Where: ο· Emissions = Emissions from SG production in a county in a given year ο· A = Total SG acres in a county in a given year ο· Ei = Sum of all activity emissions from 1 acre of SG production in a given year (i) of the 10-year cycle (Ei varies by year due to different activities and chemical requirements) 1.5.3 County-Level Yield Comparison Among CG, WS and SG Figure S5 displays a histogram of yields for WS, CS, and SG. SG has the highest yields, with CS having comparatively higher yields than WS. 15 180 160 Number of counties 140 120 100 CS 80 WS SG 60 40 20 0.20 0.60 1.00 1.40 1.80 2.20 2.60 3.00 3.40 3.80 4.20 4.60 5.00 5.40 5.80 6.20 6.60 7.00 7.40 7.80 8.20 8.60 9.00 9.40 9.80 0 Yield (dry tons/acre) Figure S5. Distribution of county-level yields (dry short tons/acre) for corn stover (CS), switchgrass (SG), and wheat straw (WS). 1.5.4 Forest residue The FR scenarios were based on the integrated operations (LOGR) feedstock category of the BTS2 harvested from non-federal lands. In the BTS2, integrated operations (LOGRLOGT) combine both residues and thinnings with quantities reflecting 50% of each (50% LOGR, 50% LOGT). ORNL performed a custom run of POLYSYS to provide results for 100% residues for use in this study. The data provided by ORNL included the Federal Information Processing Standard (FIPS) county code and the production quantity per county in dry tons (Eaton 2012). 1.6 Number of Counties that Have Projected Production for Each Feedstock in 2022 Table S3 shows the number of counties that have projected production of a certain feedstock and the number of counties which have production of at least one cellulosic feedstock for 2022 based on the BTS2 data (DOE 2011). The total number of counties in the contiguous US is also shown 16 alongside for comparison. These numbers can help interpret the results of Figure 3 in the main text. Table S3. Number of counties projected to produce each cellulosic feedstock Corn Stover 1104 1.7 Forest Residue 2169 Switch grass 1183 Wheat Straw 1169 Cellulosic feedstock (at least one) 2841 Number of Counties in Contiguous US 3109 Corn Grain Data Limitations Corn grain (CG) crop budgets are based on data between 1996 and 2007 and are on a state level (not a county level). Irrigation data is also based on state averages. We assume the crop budgets and irrigation requirements are identical among all counties within a state. The existing crop budget database shows significant discontinuity between regions in their inputs and practices. To correct for this, University of Tennessee (UT) and ORNL have developed a new methodology that uses fewer, but more reliable, budgets to smooth the transition between regions (Hellwinckel 2012). 2 Emissions From Equipment Operation This section provides additional information on NONROAD2008, which is used for modeling emissions from operating agricultural machinery for non-harvest, harvest and transport activities. 2.1 NONROAD Model The EPA’s NONROAD2008a (available at http://www.epa.gov/oms/nonrdmdl.htm#model) model is the primary tool used in this study to compute air pollutant emissions from machinery operation for field activities because the model generates emission inventories from individual counties to the entire nation, covers all the major air pollutants of interest (CO, NOX, SOX, PM10, total hydrocarbon [THC]) except for NH3 (which is calculated separately based on fuel consumption and emission factors), and takes into account emission controls required by regulations over time (from 1970 to 2050) (EPA 2005b). In particular, the NONROAD model is designed to account for the effect of the federal emissions regulations. However, it does not cover any California emissions standards or any proposed federal emissions standards. 17 The model includes more than 80 basic and 260 specific types of non-road equipment, and it further stratifies equipment types by horsepower rating. Fuel types include gasoline, diesel, liquefied petroleum gas (LPG), and compressed natural gas (CNG). NONROAD2008a supersedes all previous versions of the model. Compared to the original NONROAD2005 model, the main revision of NONROAD2008a consists of new exhaust and evaporative emission controls; the new version predicts lower THC, CO, NOX, and PM emissions than the previous version (EPA2005a) with comparable scenario inputs owing to new emission controls based on regulations promulgated since the original version. Further model changes include (but are not limited to) the following: ο· Added ability to model effects of ethanol blends ο· Revised emission factors for small spark ignition engines ο· Revised fuel consumption for some engines. The NONROAD2008a model will hereafter be referred to simply as the NONROAD model. 2.1.1 Combustion Emissions Calculation The NONROAD model uses the Equation S3 (EPA 2010b) to calculate combustion exhaust emissions. Emissions = (POP) ∗ (Power) ∗ (LF) ∗ (A) ∗ (EF) Equation S3 where: ο· Emissions = the total emissions (g); ο· POP = the equipment population in a given county (#); ο· Power = the average horsepower of the machinery (hp); ο· LF = the load factor, the fraction of available power (%); ο· A = the activity of the equipment (hours/year); ο· EF = the emission factor (g/(hp*hr)). 18 The NONROAD program uses source classification codes (SCC) to distinguish the different engine types and horsepower (hp) ranges to further classify an engine. It is important to understand that the program does not model specific pieces of equipment, but engines of varying power ranges (EPA 2005a). For example, a 135-hp tractor is modeled in a 100–175 hp range. More information on how the NONROAD model calculates emissions and the default values used in the program may be found in the model’s technical documentation (EPA 2010b). 2.1.1.1 NH3 Emissions Ammonia emissions are calculated based on fuel consumption. First, the fuel consumption, reported in gallons by NONROAD, is converted to Btu using the lower heating value. After the conversion, a fuel-specific emission factor per Btu of fuel is applied, as shown in Table S4. 2.1.1.2 PM2.5 Emissions The size distribution of the particulate matter is given in the NONROAD model’s technical documentation (EPA 2010d, EPA 2010e). As shown in Table S4 below, PM2.5 emissions are derived from PM10 and are distinguished by fuel type (EPA 2010c). 2.1.1.3 VOC Emissions Hydrocarbon emissions can be reported using several metrics depending on the desired use of the emission estimates and the measurement technique used in the underlying data. Most hydrocarbon emissions data from mobile sources are measured as total hydrocarbon (THC, compounds existing entirely of hydrogen and carbon) (EPA 2005a). The NONROAD program adds THC to oxygenated compounds (alcohols and aldehydes commonly found in engine exhaust) then subtracts the methane and ethane components to get VOC (EPA 2010a). The definition of VOC excludes methane, ethane, acetone, and compounds not commonly found in large quantities in engine exhaust, like chlorohydrocarbons. Although acetone is not subtracted, it is present in smaller quantities compared to methane and ethane, and will have a negligible effect on the results (EPA 2010a, EPA 2010g). The THC to VOC conversion factors are shown below in Table S4. 19 Table S4. Conversion factors for default NONROAD output Diesel Gasoline LPG CNG PM2.5 LHV (Btu/gallon) 0.97*PM101 0.92*PM102 1.0*PM102 1.0*PM102 128,4503 116,0903 84,9503 20,2683 NH3 (g/MMBtu) 0.684 1.014 (not reported) (not reported) VOC 1.053*THC5 0.933*THC5 0.995*THC5 0.004*THC5 1 EPA 2010d EPA 2010e 3 DOE 2012 4 EPA 2011h 5 EPA 2010a 2 2.1.2 Age distributions for equipment populations The NONROAD model calculates age distributions for equipment populations for given equipment types and scenario years. This calculation is necessary for the model to account for several factors which affect emissions over time, including emissions deterioration, new emissions standards, technology changes, changes in equipment sales and total equipment population, and scrappage programs. Further information on the details may be found in the NONROAD model’s technical documentation (EPA 2005b, EPA 2005c, EPA 2004, EPA 2010f). 2.2 Equipment Lists 2.2.1 Corn Grain Non-Harvest Activity We only model CG under current practices (for year 2011) due to data limitations (e.g., no reliable projected crop budget for 2022). Three tillage types were considered for CG: conventional till, reduced till, and no-till (based on conventional tillage with moldboard plow, limited till and no-till crop budget data (APAC 2007)). Conventional till requires the most frequent use of farm machinery among all tillage types. Tables S5-S7 show the CG equipment usage for each tillage type. 20 Table S5. Corn grain conventional till, non-harvest equipment operations Field Activity Dry Fertilizer Spreader (trailer mounted)1 Chemical Applicator GE30ft (trailer mounted)1 Chemical Applicator GE30ft (trailer mounted)1 Fertilizer Applicator (attached to implement)1 8 Row Planter (regular)1 Field Cultivator GE15ft1 Tandem Disk (regular) 14-18ft1 Moldboard Plow (regular) 4-6b1 1 Tractor Used (hp) Hours/Acre 205 205 205 205 205 205 205 205 0.0873 0.0391 0.0391 0.2619 0.2619 0.063 0.1611 0.342 APAC 2007 Table S6. Corn grain reduced till, non-harvest equipment operations Field Activity Dry Fertilizer Spreader (trailer mounted)1 Dry Fertilizer Spreader (trailer mounted)1 Row Cultivator GE15ft1 8 Row Planter (regular)1 Chemical Applicator GE30ft (trailer mounted)1 Tandem Disk (regular) 14-18 ft1 Offset Disk/Light duty 14-18 ft1 1 Tractor Used (hp) Hours/Acre 100 100 135 100 100 135 135 0.0873 0.0873 0.1281 0.1 0.0391 0.1611 0.1384 APAC 2007 Table S7. Corn grain no-till non-harvest equipment operations Field Activity Dry Fertilizer Spreader (trailer mounted)1 Dry Fertilizer Spreader (trailer mounted)1 Chemical Applicator GE30ft (trailer mounted)1 Dry Fertilizer Spreader (trailer mounted)1 Chemical Applicator GE30ft (trailer mounted)1 7 Row No-till Planter1 1 Tractor Used (hp) Hours/Acre 160 160 160 160 160 160 0.0873 0.0873 0.0391 0.0873 0.0873 0.2001 APAC 2007 21 2.2.1.1 Corn Grain Irrigation The USDA Farm and Ranch Irrigation Survey is conducted every five years to collect US onfarm irrigation data (USDA 2007). The survey covers all farms that produce $1,000 or more of agricultural products. US farms and ranches utilize gasoline, diesel, liquefied petroleum gas (LPG), compressed natural gas (CNG), and electricity for their irrigation systems. This applies to both well and surface water sources as well as pressure and gravity irrigation systems. Although the dominant energy sources for the irrigation systems are electricity (60%) and diesel (27%) (USDA 2007), the energy mix can vary by state. The following tables from the 2008 survey were utilized to understand irrigation pumping requirements for CG (USDA 2007). The survey provides data on a state level. County-level data are derived from the state level based on CG county-level farm acreage. ο· Table 15 - Irrigation Wells Used on Farms: 2008 and 2003 ο· Table 16 - Characteristics for Irrigation Wells Used on Farms: 2008 and 2003 ο· Table 18 - Irrigation Pumps on Farms for Wells: 2008 and 2003 ο· Table 19 - Irrigation Pumps on Farms Other Than for Wells: 2008 and 2003 ο· Table 20 - Energy Expenses for On-Farm Pumping of Irrigation Water by Water Source and Type of Energy: 2008 and 2003 ο· Table 28 - Estimated Quantity of Water Applied and Primary Method of Distribution by Selected Crops Harvested: 2008 and 2003. For each state the following data were extracted: ο· Crop (corn only) ο· State ο· Irrigation method (well, non-well, discharge, reservoir, and boost) ο· Irrigated acres ο· Amount of water used for irrigation per acre (uH2O, acre-ft/acre) ο· Fuel type (gasoline, diesel, LPG, natural gas, electricity) ο· Percentage of acres by fuel type and irrigation method 22 ο· Average flow (q, gpm) ο· Static water depth (d, ft) ο· Load factor of engine (lf, %) ο· Pump efficiency (pe, %) ο· Gear drive efficiency (gde, %) ο· System pressure (p, lb/in2) ο· Friction head (FH, ft) ο· Velocity head (VH, ft) ο· Pressure head (PH, ft) Once the data were extracted from the USDA irrigation survey, time required for irrigation (hours/acre) and the required horsepower (hp) are calculated using Equations S4 and S5 (CARB 2006) and fed into the NONROAD model to determine irrigation equipment emissions. πππ π»2 π βππ’ππ π’π»2π 32581 ππππ − ππ‘ = ∗ ππππ π 60 πππ/βππ’π Equation S4 βπ = π ∗ (π + ππ» + πΉπ» + ππ») 1 1 ∗ ∗ 3961.80 πππ ∗ ππ ππ Equation S5 where: ο· 3,961.8 ft/min-gal H2O= (55 ft-lb/sec) * (60 sec/min) / (8.34 lb/gal H2O) ο· FH = 2.54 ft ο· VH = usually negligible ο· PH = p * 2.31 in2/lb Irrigation equipment is categorized into four NONROAD runs according to the fuel type: gasoline, diesel, LPG, and CNG (see Table S8). 23 Table S8. Irrigation equipment as modeled by the NONROAD program Field Activity Size of Equipment (hp) Hours/Acre Gasoline Irrigation Equipment Diesel Irrigation Equipment LPG Irrigation Equipment CNG Irrigation Equipment 3-100 6-300 100-175 25-300 Varies By State Varies By State Varies By State Varies By State 2.2.2 Corn Stover and Wheat Straw Harvest Currently, there is no large-scale collection of agricultural residues for industrial applications. A single-pass methodology is used to model the harvest of CS and WS, which are harvested from reduced and no-till acres only. In this system, a combine collects the grain (corn or wheat) while towing a baler, which collects the residue. The residue feeds directly into the baler, making its collection highly efficient (Jacobson 2012). The maximum operating speed of the combine is found to occur when the yield is less than 1 dry short ton/acre (Jacobson 2012). As the yield increases, the harvest time per acre also increases (see Figure S6). (Hr/acre) Hours Per Acre vs. Yield 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 Maximum operating speed 0 1 2 3 4 5 Yield (dry short ton/acre) Figure S6. Single pass combine collection rates (Jacobson 2012) Based on Idaho National Laboratory’s recent research, it was determined that a 540-hp combine would be required for the single pass methodology to collect both grain and residue without 24 significantly slowing down grain harvest as in the normal harvesting system whereas only grain is harvested (Jacobson 2012). Compared to harvesting only the grain (corn or wheat), harvesting both residues and grain significantly reduces the speed of the combine. Based on estimates (Jacobson 2012), we allocate about 380-hp out of the 540-hp to residue harvest while 160-hp is allocated to grain harvest. This calculation is shown in the Equations S6 and S7: πΈππππ π π‘ππ£ππ = πΈππππ πππ ∗ ( 380 ) 540 Equation S6 Or for WS πΈπβπππ‘ ππ‘πππ€ = πΈππππ πππ ∗ ( 380 ) 540 Equation S7 Table S9 shows the equipment for the harvest and on-farm transport for CS and WS. Because harvest of crop residues is not currently practiced at a large-scale for biofuel production, alternative harvest and processing systems may be developed and implemented using different equipment compared to that modeled in our analysis. Readers are encouraged to refer to Hess et al. (2009) for more information about likely future crop residue harvest and processing systems. 25 Table S9. Corn stover and wheat straw harvest and transport equipment Field Activity Machinery Used (hp) Hours/Acre Corn stover, wheat straw harvest Combine Corn stover and wheat straw on- farm transport 540 100-175 varies by yield 13 dry short tons/hr max throughput 2.2.3 Switchgrass SG is produced in a 10-year rotation, and different equipment is used for establishment and maintenance years. As stated previously, because of the perennial nature of SG, our analysis assumes that 1/10 of the total SG acres in 2022 is in the first, second, third, fourth, fifth, sixth, seventh, eighth, ninth, and tenth year of the rotation, respectively. The equipment list for all nonharvest operations for each year of the cycle is shown in Table S10. Table S10. Switchgrass non-harvest equipment Machinery Complement Times Over Field Tractor Used (hp) Hours/ Acre Establishment - Year 1 Offset disk, 12 ft . 2 1 Fertilizer and lime spreader, 4T, 40ft 2 Boom sprayer, 50ft1 3 1 No-till drill, 15 ft 1 Maintenance - Years 2-10 130 60 60 130 0.286 0.084 0.117 0.183 Reseeding (year 2 only)1 1 130 0.046 Fertilizer and lime spreader, 4T, 40ft1 Boom sprayer, 50ft (year 5 only)1 Tractor 2,4-D application (year 5 only)1 1 1 1 60 60 60 0.042 0.039 0.089 1 1 Turhollow 2011 For SG harvest, the year 1 yields are 1/3 and year 2 yields are 2/3 of mature yields (years 3–10); only one harvest per year is assumed. The harvest equipment list is shown in Table S11. Similar to crop residues, switchgrass is not currently produced and harvested at a commercial scale for biofuel production, alternative harvest and processing systems may be developed and implemented using different equipment compared to that modeled in our analysis. Readers are encouraged to refer to Hess et al. (2009) for more information about likely future harvest and processing systems, advanced for switchgrass harvest and processing. 26 Table S11. Switchgrass harvest equipment Machinery Complement Tractor Used (hp) 1 Mower - conditioner Rake1 Baler1 Bale mover1 1 130 60 130 130 Hours/Acre Max Throughput (dry short tons/hour) Years 1 - 10 0.138 0.286 0.169 0.117 20 13 Turhollow 2011 2.2.4 Forest Residue FR collection occurs only where other logging operations are the driving economic force, or in other words, where there are pre-existing logging operations (DOE 2011). For this reason, the equipment list for FR consists only of a chipper and a loader as the FR is already transported to the forest landing. A product-purpose allocation is applied to FR thus only activities at the landing are considered in our inventory; all activities to get the FR to the landing is attributed to the harvested logs (not considered in this study). The equipment specifics of chipper and loader are based on Leinonen (2004) (See Table S12). Because NONROAD estimates emissions from forestry equipment operations based on volumes processed, we converted mass to volumes by assuming a conversion factor of 30 pound (lb) per cubic feet (ft3) for forest residue. Logging residues are sometimes piled and open burned on location. This practice varies depending on a number of factors including ownership, location, type, regeneration and forest productivity. Without access to spatial data for how to represent current residue management, this analysis did not include offsetting of open burning, which result in conservative estimates of emissions. Table S12. Forest residue equipment Field Activity Loader1 Chipper1 Tractor Used (hp) 124.5 414 ft3/hr 18,878 1,905 1 Leinonen 2004. 27 3 Fertilizer and Pesticide Application Production of biomass often requires use of fertilizers and pesticides. Nitrogen (N), phosphorous (P), and potassium (K) fertilizers are commonly applied to CG and SG during their production (DOE 2011). Harvesting CS and WS remotes nutrients, which could be available for next crops if residues are left in the field. As such, additional N, P, K fertilizers are assumed to be applied to compensate for the loss of nutrients due to residue removal. Of fertilizers, only NOx and NH3 emissions from N fertilizer application are estimated; we were unable to locate EFs for P and K fertilizers for the air pollutants examined. A pesticide is used to control weeds (herbicides), insects (insecticides), fungi (fungicides), and rodents (rodenticides). Pesticides are made of active ingredients (i.e., the pest-killing material) and inert ingredients (i.e., solvents used as carriers for the pest-killing materials). Both active and inert ingredients may potentially release VOC emission either during application or as a result of evaporation (ERG 2012). 3.1 Nitrogen Fertilizer N-application rate for corn grain varies significantly by locations. A distribution of county-level N-application rate (used in our modeling) for conventional till and no till is shown in Figure S7. Reduced till N-application rate is assumed to be the same as conventional till N-application rate. 28 Figure S7. County-level distribution of nitrogen application rate ranges for conventional till and no-till corn grain production (no-till and reduced till acres are assumed to have the same nitrogen application rate). There are many N fertilizers, with prices varying considerably among these sources. Data from the USDA Economic Research Service indicates the nitrogen fertilizers with highest consumption in 2010 are anhydrous ammonia, ammonium nitrate, urea, ammonium sulfate, and nitrogen solutions (USDA 2012). We assume these five N fertilizers are used for CG as well as for CS and WS (to compensate for the loss of nutrients due to residue removal). Because each N fertilizer type emits different levels of NOx and NH3, we assume the share of each N fertilizer among total N usage is identical to that in 2010 (Figure S8). For SG, nitrogen solutions will likely be the primary fertilizer used in the model year (Turhollow 2011) and are assumed as the only N fertilizer applied to SG in this analysis. 29 Urea 25% Anhydrous Ammonia 34% Nitrogen solutions 35% Ammonium Nitrate 3% Ammonium Sulfate 3% Figure S8. Share of nitrogen fertilizer applied to CG, CS and WS (USDA 2010a) Table S13 shows the emission factors for NH3 and NOx for the N fertilizers based on existing literature. Table S13. Emission factors for five commonly used nitrogen fertilizers N fertilizer type Anhydrous ammonia Ammonium nitrate (33.5% N) Urea (44%-46% N) Ammonium sulfate Nitrogen Solutions NO a, b (% of N in Nitrogen Fertilizer] 0.79 3.8 0.9 3.5 0.79 NH3 (% N volatized as NH3)c,d 1.0 2.0 15.0 8.0 8.0 a FOA (2001) GREET (ANL 2010) c Goebes et al 2003 d Davidson et al 2004 b 3.2 Pesticide The estimation of emissions from pesticides is challenging due to the wide range of formulations (e.g., emulsifiable concentrate, aerosol, solution, flowable, granule), application equipment, and application type (e.g., band, broadcast, serial, spot). Presently, the most widely accepted guidance for the estimation of emissions from pesticide application is Emission Inventory Improvement Program (EIIP) guidance (EPA 2001). The preferred methodology uses the vapor pressure of the active ingredient to determine the 30 appropriate emission factor, the amount of pesticide applied to an area, and the percent of the active ingredient in the pesticide applied. Although this methodology is preferred, it typically requires a significant amount of data regarding the fraction of active ingredient in the pesticide applied and the type of formulation by which the active ingredient is applied. For a given active ingredient, there may be dozens of pesticide labels with various formulation types and active ingredient fractions. In addition, the list of pesticides in the EIIP guidance is far from comprehensive (ERG 2012, EPA 2012a). Given the lack of information necessary to estimate emission factor using the preferred method, we used an alternative approach from the EIIP guidance. This alternative methodology uses Equation S8, shown below. πΈ = π ∗ πΌ ∗ πΈπ ∗ πΆπππΆ Equation S8 Where: ο· E= Total emissions (lb VOC/acre) ο· R= Lb of pesticide applied per year per harvested acre ο· I= Lb of active ingredient per lb of pesticide ο· ER = Evaporation rate (ratio) (default value = 0.9) ο· CVOC = VOC content (default value = 0.835) (lb VOC/lb active ingredient). This method was also used to estimate VOC emissions from pesticide in the 2008 National Emission Inventory (NEI) (Huntley 2012). In the NEI, the default evaporation rate is 0.9 and the default VOC content is 0.835 lb VOC/lb of active ingredient. These default values are used in our analysis. Data on pesticide use for CG vary by state and are derived from USDA (2010b). The same methodology is applied to SG to estimate VOC emissions from herbicide applications. During the establishment year (year 1) of SG production, herbicides are applied to eliminate the competition of weeds. During the maintenance years (years 2-10), only one treatment is assumed to be needed (Turhollow 2011). Herbicide application rate and schedule are summarized in Table S14. The pesticide application data is from scenario modeling since switchgrass does not have an established crop budget and therefore is continually being refined and updated. The data used in this manuscript reflects the model as of 2011. 31 Table S14. Switchgrass pesticide application schedule (P, lb active ingredient/ac) Application Time Establishment (year 1) Year 5 2,4-D-amine 1.0 lb/ac 1.5 lb/ac Atazine 1.0 lb/ac Quinclorac 0.5 lb/ac 4 Fugitive Dust Emissions Fugitive dust sources may be separated into two broad categories: process sources and open dust sources (WRAP 2006). Agricultural tilling and harvesting fall under open dust sources and apply to all feedstocks considered in this study except FR, for which product-purpose allocation obviates the need to estimate fugitive dust. The agricultural tilling source includes airborne soil particulate matter (PM) emissions produced during the preparation of agricultural lands for planting, harvest and other activities. Dust emissions are produced by the mechanical disturbance of the soil by the implement used and the tractor pulling it (WRAP 2006). PM emissions vary by activities. Depending on the type of activity performed and the number of passes required for performing the activity (e.g., discing, weeding), the PM emission factors are computed for nonharvest and harvest activity categories for each feedstock on a basis of per acre (Table S15). Emission factors of PM10 specific to each activity are taken from WRAP (2006) and Gaffney and Yu (2003). The PM2.5/PM10 ratio is assumed to be 0.20 for fugitive dust emissions based on MRI (2006). FR chipping is likely to generate fugitive emissions, however, no EF was identified for the operation. Table S15, Fugitive dust emission factors (lb of PM10/ac)1 Non-Harvest Harvest 1All Corn Grain Conventional Reduced Till Till 8 7.2 2.4 2.4 CS WS SG 1.8 1.8 1.6 2.4 No-Till 5.2 2.4 values calculated based on WRAP (2006) and Gaffney and Yu (2003), and the crop budget for each feedstock. Fugitive dust emissions from non-paved roads are estimated using Equation S9. These emissions arise from vehicles traveling on unpaved roads that pulverize the surface material. These particles can get lifted into the air and become PM emissions (EPA 1998). The constants are listed in table S16 for specific sizes (EPA 1998, table 13.2.2-2). 32 πΈ= π π π π π( ) ( ) π£ 12 3 π π ( ) 0.2 Equation S9 Where: E= size specific emissions (lb PM2.5 or PM10) s= Surface material silt content (%) W= Mean vehicle weight (short tons) M= Surface material moisture content (%) v= Vehicle miles traveled (miles) k= Constant (lb/VMT) a= Constant b= Constant c= Constant Table S16. Constants used to estimate fugitive PM emissions from transportation k (lb/VMT) a b c PM2.5 0.38 0.8 0.4 0.3 PM10 2.6 0.8 0.4 0.3 5 Metrics Evaluated This section provides additional information on construction of Figures 1-4 in the main text. 5.1 Emission per gallon of ethanol (Figure 1 in the main text) County-level air emissions are calculated for each feedstock, and are divided by the total amount of ethanol, which can be potentially produced from a given feedstock to estimate emissions on the basis of 1 gallon of ethanol produced for comparison across feedstocks (Equation S10). Ethanol yields for the feedstocks examined are shown in Table S17. 33 πΈπππππππππ = ∑πΈπππ π ππππ ππππ πππ πππ‘ππ£ππ‘πππ πππ π πππ£ππ πππππ π‘πππ πΆππ’ππ‘π¦ πππππ π‘πππ πππππ’ππ‘πππ ∗ πΈπ‘ππ»π¦ππππ Equation S10 Table S17. Ethanol Yields Feedstock Corn Grain Corn Stover, Wheat Straw, Switchgrass Forest Residue 1 2 Ethanol yield 2.76 gallons/bushel1 89.9 gallons/dry ton2 75.7 gallons/dry ton2 Average corn ethanol yield in 2011 (EPA 2011) Hsu et al 2010 5.2 Contribution Analysis (Figure 2 in the main text) The contribution analysis estimates the contribution of emissions from one of the following five activity categories: non-harvest, N fertilizer, pesticide, harvest, and transport. The contributions from each feedstock are summarized below. Non-Harvest Category: - Emissions from irrigation, machinery operations associated with fertilizer and chemical application, and machinery operation for field preparation (e.g., cultivating, discing, plowing). N Fertilizer Category: - NO and NH3 emissions from N fertilizer application (e.g., through nitrification and volatilization). Chemical Category: - VOC emissions from pesticide application Harvest Category: 34 - Emissions from machinery operations associated with feedstock harvesting. For example, SG harvesting includes machinery operations such as mower, rake, and baler. Forest residue harvesting includes chipper and loader. Transport Category: - Combustion emissions from tractor used to transport feedstock to temporary on-farm storage facility. To calculate the contribution from a specific activity category in a county, the sum of emissions from that category is calculated and then divided by the total emission from all activities from the production of a feedstock in the county (Equation S11). Figure 2 in the manuscript displays the distribution of contribution calculated per county. The ranges in Figure 2 are attributed to the spatial variation of the data inputs. Note that for fertilizer and chemical applications, the combustion emissions associated with applying the fertilizers/chemicals are accounted for in the non-harvest activity category. πΆπππ‘ππππ’π‘ππππππ‘ππ£ππ‘π¦ = 5.3 ∑ πππππ’π‘πππ‘ ππ¦ πππ‘ππ£ππ‘π¦ ∑ πππππ π πππ πππ‘ππ£ππ‘πππ Equation S11 Comparison to National Emissions Inventory (NEI) For Cellulosic Feedstocks (Figure 3 in the main text) Equation S12 and S13 describe the methods used to compare emissions from cellulosic feedstock to the baseline 2008 NEI. π ππ+ππ = πΈπππ π ππππ ππππ πππππ’ππ‘πππ ππ πππππ’πππ ππ πππππ’ππ πππππ π‘πππ πππππππ + ππππππππ‘ ππππ π ππππ ππππ 2008 ππΈπΌ Equation S12 π ππΈπΌ π‘ππ‘ππ = πΈπππ π ππππ ππππ πππππ’ππ‘πππ ππ πππππ’πππ ππ πππππ’ππ πππππ π‘πππ πππ‘ππ ππππ π ππππ ππππ 2008 ππΈπΌ 35 Equation S13 The Rs (Rnr+np and RNEI total) are the calculated ratio of the modeled emissions to 2008 NEI emissions. Emissions from production of cellulosic biofuel feedstock are the sum of emissions in a given county for all relevant cellulosic feedstocks examined. Nonroad (NR) + Nonpoint (NP) emissions from NEI are the sum of the nonroad and nonpoint category emissions from the NEI (EPA 2008). Total emissions from 2008 NEI are a sum of the nonroad, nonpoint, onroad and source point category emissions from the 2008 NEI. R is calculated for every county where there are emissions from cellulosic feedstock production. The R analysis described in the manuscript refers to Rnr+np. 5.4 NEI Ratio Thresholds Compared to Attainment Status for National Ambient Air Quality Standards (NAAQS) (Figure 4 in the main text) To inform consideration of the potential effect cellulosic biofuel feedstock production might have on a county’s NAAQS non-attainment status (EPA 2014), the NEI comparison results are visualized along with locations of counties in non-attainment for ozone or PM2.5 NAAQS. The ratio thresholds are set at 5%, 10% and 20%, which were thought to be potentially significant. Overlap in the counties exceeding the ratio thresholds and non-attainment might indicate potential challenges to meet the NAAQS. Counties exceeding these thresholds for the precursor pollutants are compared to counties that were in non-attainment for the NAAQS for the relevant pollutants (ozone, PM2.5 and, SO2). For pollutants with multiple precursors (see Table S18), the county was included in the comparison if any one of the precursors exceeded a threshold level. No counties are in non-attainment for NO2 and CO NAAQS. The maps are generated by overlaying layers exhibiting counties in non-attainment and counties exceeding the different thresholds. Table S18. NAAQS criteria air pollutant and paired air pollutants or precursors NAAQS Criteria Pollutant Air pollutant or precursor Ozone PM2.5 and PM10 SO2 NO2 CO NOX, VOC NOX, VOC, SO2, PM2.5 or PM10, NH3 SO2 NO2 CO 36 6 Additional Maps and Figures 6.1 Biomass Production Density Maps displaying the density of biomass production are created based on Equation S14 (note: county area is different than harvested area, or planted area). πππππ’ππ‘πππ π·πππ ππ‘π¦ = ππππππ π πππππ’ππ‘πππ (πππ π ) π‘ππ‘ππ πππ’ππ‘π¦ ππππ Equation S14 The production density maps (Figure S9) are useful for gaining perspective on where emissions might concentrate due to the large production of several cellulosic feedstocks. The map labeled cellulosic takes into account the sum of production from CS, FR, SG, and WS. There is a high level of overlap in production of SG, WS, and CS in some counties in Kansas, North Carolina, Virginia, and Maryland, making these areas more likely to see higher increase in emissions than counties producing only one feedstock. As previously stated, SG production is limited to counties west of the 100th meridian because SG is assumed to be grown on rain-fed lands only (DOE 2011). The 100th meridian (roughly) runs along the eastern border of Montana, Wyoming, Colorado and New Mexico; there is no production west of this border, except for one county in the Pacific Northwest. FR is assumed to be collected from non-federal lands and has high production in the Midwestern, Northwestern, and Eastern states. Numeric data on county-level biomass production density related to the projected production of cellulosic feedstock (individual feedstock and aggregated for all cellulosic feedstock) examined in this analysis can be obtained at http://en.openei.org/datasets/dataset/air-pollutant-emissionsinventory-supporting-data. 37 CELLULOSIC CG CS FR SG WS Figure S9. Production density for all feedstocks (dry metric ton/square kilometer) 6.2 Emissions Density The following maps (Figures S10-S15) display the emissions of each pollutant per county area. Generally speaking, the high production density areas (see Figure S9) are also high emissions density areas (Figures S10-S15). A few common patterns are briefly described here to facilitate 38 understanding of the results. PM10 and PM2.5 always follow the same spatial trend because all PM2.5 emissions are modeled as a fraction of PM10 emissions. In general, CO and SOX, which occur only from fuel combustion emissions, share the same spatial patterns. NOX and NH3 follow the same spatial trends because the N fertilizer application is the dominating source of these emissions. Note that even when spatial trends are similar, magnitude of emissions can differ (as seen in the scales). 6.2.1 Cellulosic Figure S10 is the emission density maps for all cellulosic feedstocks combined. The NOX and NH3 have different spatial patterns for the aggregated cellulosic figure, even though these two pollutants have identical spatial trends for each of the individual cellulosic feedstocks (See Figures S11-S15). For NOX, the highest emissions density areas are estimated to be in Kansas, Oklahoma, and Iowa. SG dominates the NOX emissions in Oklahoma and Kansas because the N fertilizer applied to SG is assumed to be nitrogen solutions, which have higher NOX emissions than the weighted average N fertilizers used for corn grain, CS and WS. However, nitrogen solutions have slightly lower NH3 emissions compared to the weighted average emission factors for the N fertilizers used for corn grain, CS and WS. In addition, both CS and WS are produced in some parts of Kansas and Oklahoma, contributing to the total emissions. The high emissions density area in Iowa for NOX and NH3 is entirely attributable to higher CS production density. CO and SOX have high emissions density in Iowa, Kansas, Oklahoma, and Illinois because of the production of multiple feedstocks in these states. The high emissions density in Iowa is primarily attributable to high production density of CS. In Illinois, the high emission density is mainly because of the overlapping production of WS, CS, and FR; and in Kansas, it is attributable to overlap in production from SG, WS, and CS. The highest PM emissions density areas are in Kansas, Oklahoma, and Texas. At all of these locations, the emissions can be mainly attributed to SG because SG has both non-harvest and harvest activities with high fugitive dust emissions from both categories, compared to other cellulosic feedstocks. 39 The high VOC emissions density in Kansas, Oklahoma, and Texas are due to the application of multiple pesticides for SG production, which emits VOC. Numeric data on county-level air emission densities from the projected production of cellulosic feedstock can be obtained at http://en.openei.org/datasets/dataset/air-pollutant-emissionsinventory-supporting-data. 40 NH3 NOx CO SOx PM10 PM2.5 VOC Figure S10. Cellulosic emissions density maps (kg of pollutant per km2 county area) 41 NH3 NOX CO SOX PM2.5 PM10 VOC Figure S11. Corn grain emissions density maps (kg of pollutant per km2 county area) 42 NOX NH3 CO SOX PM10 PM2.5 VOC 3 Figure S12. Corn stover emissions density maps (kg of pollutant per km2 county area) 43 NOX NH3 CO SOX PM10 PM2.5 VOC Figure S13. Wheat straw emissions density maps (kg of pollutant per km2 county area) 44 NOX NH3 CO SOX PM10 PM2.5 VOC Figure S14. Switchgrass emissions density maps (kg of pollutant per km2 county area) 45 NOX NH3 CO SOX PM10 PM2.5 VOC Figure S15. Forest residue emissions density maps (kg of pollutant per km2 county area) 46 6.3 Individual Feedstock Comparison to NEI Figures S16-S20 are individual feedstock emissions compared to the 2008 national emissions inventory (NEI) non-point and non-road categories (Rnp+nr). NH3 and NOX emissions from SG, CS, and WS production are the main factor to cause R to exceed the three thresholds. In addition, PM emissions from SG production are responsible for a few exceedances. Figure S16. Ratio (Rnpr+nr) of switchgrass emissions to 2008 NEI nonpoint and nonroad emissions 47 Figure S17. Ratio (Rnpr+nr) of corn stover emissions to 2008 NEI nonpoint and nonroad emissions. 48 Figure S18. Ratio (Rnpr+nr) of wheat straw emissions to 2008 NEI nonpoint and nonroad emissions. 49 Figure S19. Ratio (Rnpr+nr) of forest residue production emissions to 2008 NEI nonpoint and nonroad emissions. 6.4 Counties with R exceeding thresholds compared to NAAQS Nonattainment Areas Counties with R exceeding selected thresholds for any pollutant potentially contributing to ambient ozone and PM2.5 concentrations are displayed in Figure 4. Emissions of ozone and PM2.5 precursors are relevant to secondary pollutant (ozone and PM2.5) formation at regional scales (order of 100-1000 kilometer),37 thus all counties in nonattainment areas (NAAs) for the ozone and PM2.5 NAAQS are also displayed in Figure 4. 50 The list of counties exceeding the three thresholds for ozone precursors along with reference to counties in nonattainment for 1997 or 2008 8-hour ozone NAAQS and the list of counties exceeding the three thresholds for PM2.5 and its precursors along with reference to counties in nonattainment for 1997 or 2006 PM2.5 NAAQS are available at http://en.openei.org/datasets/dataset/air-pollutant-emissions-inventory-supporting-data. 7 Comparison To Other Studies Tables S19-S22 show comparisons of the results of this study to estimates from the GREET Version 1_2012 model (ANL 2012) and Tessum et al (2012) when data can be derived based on one gallon of ethanol produced. To facilitate the comparisons, the boundary used for GREET modeling is before feedstock is shipped to biorefinery (i.e., transportation of feedstocks is not included). Note that our study does not include emissions from upstream activities (e.g., electricity generation); therefore only emissions emitted directly from feedstock production activities are included. Yields reported in Table S17 are used to convert GREET emissions to mass per gallon of ethanol for comparison to ours. Our results are reported as maximum, mean and minimum while results from GREET are only reported as means. Tessum et al (2012) used an earlier version of the GREET model (GREET 1.8d1) (ANL 2010) and the results are expressed in g/mile. The default fuel economy for year 2010 in GREET 1.8d1 is used to convert results in Tessum et al (2012) to a comparable unit used in our study (g/gallon of ethanol) (Tessum 2013). The system boundary used for comparison is up to the farm-gate before feedstock is shipped to biorefinery. While Tessum et al (2012) considered co-product credits in estimating emissions for corn production used for ethanol; the credits are excluded in our comparisons. Table S19 shows the comparison of mean emissions per gallon of corn ethanol from Tessum et al (2012) (S1b-corn.xls). Since GREET does not track ammonia emissions, Tessum et al (2012) estimated NH3 emissions based on EFs and the types of N fertilizer used. The types and shares of N fertilizer used by Tessum et al (2012) (Hill et al 2009) are 70.7% ammonia, 21.2% urea, and 8.2% ammonium nitrate. Once fertilizer is applied, an “industry-weighted” average of 0.038 kg of NH3 per kg N 51 applied (i.e., the average of 0.027 and 0.049) was used by Tessum et al (2012) to estimate NH3 emissions. Observations of differences between our results and those from GREET/Tessum et al (2012) in air pollutant emissions examined are discussed in section 3.5 of the main text, along with reasons to potentially explain these differences. Table S19. Corn grain comparison among our estimates, GREET Version 1_2012 and Tessum et al (2012) Corn Grain g/gal EtOH VOC CO NOx PM10 PM2.5 SOx NH3 GREET (year 2011) Mean 0.17 0.96 4.29 0.11 0.10 0.01 N.A. Max 1.72E+01 7.22E+01 2.22E+01 4.80E+01 1.00E+01 8.33E-02 5.99E+01 Our estimates (year 2011) Mean 3.60E+00 5.82E+00 5.81E+00 1.17E+01 2.52E+00 8.89E-03 1.43E+01 Min 1.00E+00 4.84E-01 2.25E+00 6.02E+00 1.33E+00 8.27E-04 5.03E+00 Tessum et al (year 2010) Mean 0.152 0.841 3.61 9.70E-02 8.80E-02 1.10E-02 5.48 Table S20. Switchgrass comparison between our estimates and GREET Version 1_2012 Switchgrass g/gal EtOH VOC CO NOX GREET (year2020) Mean 0.04 0.17 1.46 PM10 PM2.5 SOX 2.9E-02 2.6E-02 6.9E-04 Max 9.04E-01 2.59E-01 2.89E+00 Our estimates (year 2022) Mean 2.86E-01 1.00E-01 2.53E+00 Min 1.67E-01 6.97E-02 2.46E+00 1.13E+01 2.29E+00 9.08E-04 3.50E+00 7.14E-01 3.70E-04 2.00E+00 4.10E-01 2.66E-04 52 Table S21. Corn stover comparison between our estimates and GREET Version 1_2012 g/gal EtOH Corn Stover VOC CO NOx PM10 PM2.5 SOx GREET (year 2020) Mean 0.07 0.29 1.83 0.05 0.04 1.2E-03 Max 4.26E-02 1.92E-01 1.99E+00 6.48E+00 1.32E+00 6.32E-04 Our estimates (year 2022) Mean 3.44E-02 1.53E-01 1.89E+00 3.33E+00 6.85E-01 5.11E-04 Min 3.13E-02 1.38E-01 1.85E+00 1.43E+00 3.05E-01 4.65E-04 Table S22. Forest residue comparison between our estimates and GREET Version 1_2012 Forest Residue g/gal EtOH VOC CO NOx PM10 GREET (year 2020) Mean 0.10 0.41 0.9 0.07 PM2.5 SOx 0.06 1.65E-03 Max 7.47E-04 1.17E-03 2.95E-03 1.37E-04 Our estimates (year 2022) Mean 7.47E-04 1.17E-03 2.95E-03 1.37E-04 Min 7.47E-04 1.17E-03 2.95E-03 1.37E-04 1.33E-04 1.39E-05 1.33E-04 1.39E-05 1.33E-04 1.39E-05 8 References Agriculture Policy Analysis Center (APAC). 2007. Economic and environmental impacts of movement toward a more sustainable agriculture in the United States, Appendix 3: Data sources used in developing the ABS budgets. Project No. 43-3AEK-3-80080. Knoxville, TN: University of Tennessee. ANL (Argonne National Laboratory). 2010. The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Model version 1.8d1. Lemont, IL: Argonne National Laboratory. [online] No longer available for download [Accessed August 2010]. 53 ANL. 2012. Green House Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Model version 1 2012 revision 2. Lemont, IL: Argonne National Laboratory. [online] Available at: http://greet.es.anl.gov/ [Accessed July 2012]. California Air Resources Board (CARB). 2006. Emission Inventory Methodology, Agricultural Irrigation Pumps – Diesel, Appendix D. [online] Available at: http://www.arb.ca.gov/regact/agen06/attach2.pdf [Accessed October 2013]. Davidson, C., Adams, P., Strader, R., Rinder, R., Anderson, N., Geobes, M. and Ayers, J. 2004. CMU Ammonia Model, Version 3.6. Pittsburgh, PA: The Environmental Institute, Carnegie Mellon University. DOE (US Department of Energy). 2011. US Billion-Ton Update: Biomass Supply for a Bioenergy and Bioproducts Industry. R.D. Perlack and B.J. Stokes (Leads), ORNL/TM2011/224. Oak Ridge, TN: Oak Ridge National Laboratory. DOE. 2012. Alternative Fuels Data Center. DOE, Energy Efficiency and Renewable Energy (EERE). [online] Available at: www.afdc.energy.gov/afdc/fuels/properties.html [Accessed December 2012]. Eaton, L. 2012. “BTS2 projections for forest residue (tops and limbs only/LOGR) production for 2022 at $55/dry ton”. Oak Ridge National Laboratory. [Email] (Personal communication with A. Carpenter on August 23, 2012). Eaton, L. 2013. “Perennial grass and switchgrass production”. Oak Ridge National Laboratory. [Email] (Personal communication with A. Carpenter on January 4, 2013). EPA (US Environmental Protection Agency). 1995. Air Pollutant Emissions Factors, Volume 1: Stationary Point and Area Sources (AP42), chapter 13. 5th Edition. Research Triangle Park, NC: Office of Air Quality Planning and Standards. [online] Available at: http://www.epa.gov/ttn/chief/ap42/ch13/index.html [Accessed January 2012]. EPA. 2001. Emission Inventory Improvement Program (EIIP). PESTICIDES - AGRICULTURAL AND NONAGRICULTURAL. Research Triangle Park, NC: Office of Air Quality Planning and 54 Standards. [online] Available at: http://www.epa.gov/ttnchie1/eiip/techreport/volume03/iii09_jun2001.pdf [Accessed January 2012]. EPA. 2004. Nonroad Engine Growth Estimates, NR-008c. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2005a. User’s Guide for the Final NONROAD2005 Model. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2005b. Exhaust Emission Effects of Fuel Sulfur and Oxygen on Gasoline Nonroad Engines, NR-003c. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2005c. Calculation of Age Distributions in the Nonroad Model – Growth and Scrappage, NR-007c. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2005d. Geographic Allocation of Nonroad Engine Population Data to the State and County Level, NR-014d. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2006. AP 42, Fifth Edition, Volume I Chapter 13: Miscellaneous Sources. 13.2.2 Unpaved Roads. [online] Available at: http://www.epa.gov/ttn/chief/ap42/ch13/final/c13s0202.pdf [Accessed August 2013]. EPA. 2008. National Emissions Inventory 2008 Data. Research Triangle Park, NC: Office of Air Quality Planning and Standards. [online] Available at http://www.epa.gov/ttn/chief/net/2008inventory.html [Accessed January 2012]. 55 EPA. 2010a. Conversion Factors for Hydrocarbon Emissions Components NR-002d. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010b. Median Life, Annual Activity, and Load Factor Values for Nonroad engine Emissions Modeling, NR-005d. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010c. Nonroad Engine Population Estimates, NR-006e. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010d. Exhaust and Crankcase Emission Factors for Nonroad Engine Modeling – Compression Ignition, NR-009d. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010e. Exhaust Emission Factors for Nonroad Engine Modeling – spark-Ignition, NR010f. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010f. Nonroad Spark-Ignition Engine Emissions Deterioration Factors, NR-011d. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010g. Nonroad Evaporative Emission Rates, NR-012d. Washington, DC: Office of Transportation and Air Quality. [online] Available at: http://www.epa.gov/oms/nonrdmdl.htm [Accessed January 2012]. EPA. 2010h. Renewable Fuel Standard Program (RFS2) Regulatory Impact Analysis. Assessment and Standards Division Office of Transportation and Air Quality. [online] Available at: http://www3.epa.gov/otaq/renewablefuels/420r10006.pdf [Accessed January 2015]. 56 EPA. 2012a. 2008 National Emissions Inventory, v2, Technical Support Document, DRAFT. Research Triangle Park, NC: Office of Air Quality Planning and Standards. [online] Available at http://www.epa.gov/ttn/chief/net/2008neiv2/2008_neiv2_tsd_draft.pdf [Accessed January 2012]. EPA. 2012b. The Green Book Nonattainment Areas for Criteria Pollutants. Research Triangle Park, NC: Office of Air Quality Planning and Standards. [online] Available at http://www.epa.gov/oar/oaqps/greenbk/index.html [Accessed December 2012]. EPA. 2012c. “Regulatory Actions.” EPA Revises the National Ambient Air Quality Standards for Particle Pollution. Research Triangle Park, NC: Office of Air Quality Planning and Standards. [online] Available at http://www.epa.gov/pm/actions.html [Accessed December 2012]. EPA. 2013. Index of/EmisInventory/2011nei/doc/. Agricultural_pesticides_2011.zip. Data modified 11/25/2013. Available at ftp://ftp.epa.gov/EmisInventory/2011nei/doc/. Accessed January 2015. EPA. 2014. The Green Book Nonattainment Areas for Criteria Pollutants. Available at http://www.epa.gov/airquality/greenbook/. As of December 5, 2013. Accessed January 2014. FOA. 2001. Global estimates of gaseous emissions of NH3, NO and N2O from agricultural land. International Fertilizer Industry Association. Food and Agriculture Organization of the United Nations. Available at http://www.fao.org/docrep/004/y2780e/y2780e00.HTM. Accessed January, 2014. Gaffney, P. and Yu, H. 2003. Computing Agricultural PM10 Fugitive Dust Emissions Using Process Specific Emission Rates and GIS. US EPA Annual Emission Inventory Conference. [online] Available at: www.epa.gov/ttnchie1/conference/ei12/fugdust/yu.pdf [Accessed January 2012]. Goebes, M., Strader, R. and Davidson. C. 2003. An Ammonia Emission Inventory for Fertilizer Application in the United States. Atmospheric Environment, 37(18):2539–50. Hall, S. and Matson, P. 1996. NOX Emissions from soil: Implications for Air Quality Modeling in Agricultural Regions. Annual Review of Energy and the Environment, 21:311–46. 57 Hellwinckel, C. 2012. “Methodology to update the corn crop budgets.” University of Tennessee, Agriculture Policy Analysis Center. [Email] (Personal communication with A. Carpenter on July 30, 2012). Hess, J. R., Kenney, K. L., Ovard, L. P., Searcy, E. M., & Wright, C. T. (2009). CommodityScale Production of an Infrastructure-Compatible Bulk Solid from Herbaceous Lignocellulosic Biomass. (Idaho National Lab, Ed.). Uniform-Format Bioenergy Feedstock Supply System Design Report Series. Hill, J., Polasky, S., Nelson, E., Tilman, D., Huo, H., Ludwig, L., Neumann, J., Zheng, H. and Bonta, D. 2009. Climate change and health costs of air emissions from biofuels and gasoline. Proceedings of the National Academy of Sciences USA, 106(6):2077−82. Huntley, R. (2012). 2008 National Emissions Inventory (NEI) methodology documentation. Research Triangle Park, NC: US Environmental Protection Agency, Emission Inventory and Analysis Group. Hsu, D. D., Inman, D., Heath, G. A., Wolfrum, E. J., Mann, M. K. and Aden, A. 2010. Life Cycle Environmental Impacts of Selected US Ethanol Production and Use Pathways in 2022. Environmental Science & Technology, 44(13):5289-97. NREL Report No. JA-510-45915. doi:10.1021/es100186h. Jacobson, J. 2012. “Biomass Logistics Model Output”. Idaho National Laboratory [Email] (Personal Communication with Yimin Zhang, January 2012). Langholtz, M., Graham, R., Eaton, L., Perlack, R., Hellwinkel, C. and De LaTorre Ugarte, D. 2012. Price projections of feedstocks for biofuels and biopower in the US. Energy Policy, 41:484-93. Leinonen, A. 2004. Harvesting Technology of Forest residues for fuel in the USA and Finland. VTT Research Notes 2229. Helsinki, Finland: VTT Technical Research Center of Finland. MRI (Midwest Research Institute). 2006. Background Document for Revisions to Fine Fraction Ratios Used for AP-42 Fugitive Dust Emission Factors. Western Governors’ Association, 58 Western Regional Air Partnership (WRAP) Report. MRI Project No. 110397. [online] Available at www.epa.gov/ttnchie1/ap42/ch13/bgdocs/b13s02.pdf [Accessed January 2012]. NRC. 2011. Renewable Fuel Standard: Potential Economic and Environmental Effect of US Biofuels Policy. Washington, DC: The National Academies Press. ISBN-10: 0-309-18751-6. Perlack R.D., Wright L.L., Turhollow A.F., Graham R.L., Stokes B.J. and Erbach D.C. 2005. Biomass as feedstock for a bioenergy and bioproducts industry: the technical feasibility of a billion-ton annual supply. DOE/GO-102995-2135 or ORNL/TM-2005/66. Oak Ridge, TN: Oak Ridge National Laboratory. Tessum, C.W., Marshall, J.D. and Hill, J.D. 2012. A Spatially and Temporally Explicit Life Cycle Inventory of Air Pollutants from Gasoline and Ethanol in the United States. Environmental Science & Technology, 46(20):11408−17. Tessum, C.W. 2013. “Conversion of emissions to per gallon of ethanol produced”. University of Minnesota. [Email] (Personal communication with Y. Zhang on January 16, 2013). Turhollow, A., 2011. “Switchgrass crop management and harvesting equipment requirements”. Oak Ridge National Laboratory. [Email] (Personal Communication with A. Carpenter on July 14, 2011). USDA (US Department of Agriculture). 2007. Farm and ranch irrigation survey. USDA, Editor. [online] Available at: http://www.agcensus.usda.gov/Publications/2007/Online_Highlights/Farm_and_Ranch_Irrigatio n_Survey/index.php [Accessed August 2012]. USDA, 2010a. Economic Research Service, Fertilizer Use and Price Survey, US consumption of selected nitrogen materials (Table 4). [online] Available at http://www.ers.usda.gov/dataproducts/fertilizer-use-and-price.aspx#26720 [Accessed May 2012]. USDA. 2010b. Corn, Upland Cotton and Fall Potatoes, Pesticide Use. [online] Available at: http://www.nass.usda.gov/Data_and_Statistics/PreDefined_Queries/2010_Corn_Upland_Cotton_Fall_Potatoes/index.asp [Accessed January 2013]. 59 USDA. 2012. US Consumption of nitrogen, phosphate, and potash. [online] Available at http://www.ers.usda.gov/data-products/fertilizer-use-and-price.aspx#26720 [Accessed January 2013]. University of Tennessee (UT). (2012). The POLYSYS Modeling Framework: A Documentation. Knoxville, TN: Institute of Agriculture, Agricultural Policy Analysis Center. [online] Available at http://www.agpolicy.org/polysys.html [Accessed January 2012]. Veldkamp, E., and Keller, M., 1997. Fertilizer-induced Nitric Oxide Emissions from Agricultural Soils. Nutrient Cycling in Agroecosystems, 48(1-2):69-77. Wang, M., Huo, H., and Arora, S. 2011. Methods of dealing with co-products of biofuels in lifecycle analysis and consequent results within the US context. Energy Policy, 39: 5726-36. Western Regional Air Partnership (WRAP). 2006. Fugitive Dust Handbook. Prepared for Western Governors’ Association by Countess Environmental. [online] Available at: http://www.wrapair.org/forums/dejf/fdh/content/fdhandbook_rev_06.pdf [Accessed December 2012]. 60 9 List of Acronyms AQ BTS BTS2 CBO CG CNG CO CS DOE EF EIIP EPA FAPRI FIPS FR GHG GREET HI HP INL K K2O LB LCA LPG N NAAQS NAAs NASS NEI NH3 NONROAD NOX NP NR NRCS ORNL Air Quality First Billion Ton Study Billion Ton Study Update US Congressional Budget Office Corn Grain Compressed Natural Gas Carbon Monoxide Corn Stover Department of Energy Emission Factor Emission Inventory Improvement Program US Environmental Protection Agency Food and Agriculture Policy Research Institute Federal Information Processing Standard Forest Residue Greenhouse Gas Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation Harvest Index Horsepower Idaho National Laboratory Potassium Potassium oxide Pounds Life Cycle Assessment Liquefied Petroleum Gas Nitrogen National Ambient Air Quality Standards Non-Attainment Areas National Agricultural Statistics Service National Emission Inventory Ammonia EPA NONROAD2008a Model Oxides of Nitrogen Nonpoint Nonroad Natural Resource Conservation Service Oak Ridge National Laboratory 61 P P2O5 PM PM10 PM2.5 POLYSYS R RFS2 RPA RUSLE2 SCC SG SI SOX THC TPO USDA UT VOC WS Phosphorous Phosphate Particulate Matter Particulate Matter Size fractions less than 10 µm in diameter Particulate Matter Size fractions less than 2.5 µm in diameter Policy Analysis System Ratio Updated Renewable Fuel Standard Resource Planning Act Revised Universal Soil Loss Equation Source Classification Codes Switchgrass Supporting Information Document Oxides of Sulfur Total Hydrocarbons Timber Products Output US Department of Agriculture University of Tennessee Volatile Organic Compounds Wheat Straw 62