Over several decades in North America and Europe, short rotation woody crops (SRWC), such as willow and poplar, have been increasingly recognized as important renewable energy sources because of their multiple environmental and rural development benefits. As a low-carbon energy source, willow biomass is helpful in decreasing greenhouse gas (GHG) emissions when it is used to replace fossil fuels, benefitting local rural communities by providing employment and improving energy independence, and providing multiple ecosystem services by decreasing soil erosion, improving water quality, and increasing biodiversity [1
]. For example, willow can support biodiversity by functioning as natural habitats or as ecological corridors connecting patches in increasingly fragmented landscapes in the US [4
]. In some situations, there are concerns that large scale deployment of willow could impact landscape aesthetics and diversity if a large proportion of one particular land type, such as natural grassland, in a given areas is converted to willow. The most recent US Billion-Ton report projected that 411 million to 736 million dry tons of energy crops, including willow, could be produced annually in 2040 with a price of $
60 per dry Mg or less under baseline and high-yield case assumptions, respectively [5
]. New York State recently adopted aggressive goals to reduce greenhouse gas emissions by 85% by 2050 [6
Environmental indicators of renewable energy systems include GHG emissions, productivity of bioenergy crops [8
], and energy return on fossil fuel inputs. Life cycle assessment (LCA) is a widely accepted method used to quantify the environmental impacts for processes or products across their life cycle. LCA is a tool used to assess the GHG emissions and other environmental impacts of bioenergy crop production including cultivation, harvesting, and transporting biomass to end users [9
]. In addition, in the US Energy Investment and Security Act (P.L. 110–140, subtitle A), LCA has been identified as the standard methodology to determine if biofuels meet requirements for GHG reductions required for the National Renewable Fuel Standard.
In North America and Europe, a large number of studies have used LCA to evaluate environmental impacts and energy return on energy investment (EROI) of willow biomass crop production. Djomo et al. (2011) reviewed 26 papers and reported that the range of EROI for willow was between 3:1 and 16:1 and GHG emissions ranged from 0.7 to 10.6 g CO2
at the farm gate [9
]. Since 2011, additional LCAs on willow crop production systems have been conducted for the US [1
], Asia [12
], and Europe [13
]. The EROI in these studies ranges from 11.0:1 to 43.4:1 for willow biomass delivered to facility that converts the biomass to bioenergy, biofuels, or bioproducts.
The life cycle GHG emissions range from 33.7 to 100 kg CO2eq
per oven dry Mg of willow chip production including delivery to power plants without including belowground carbon sequestration, or changes in soil organic carbon (SOC) associated with land use change (LUC) [10
]. However, it has been demonstrated that belowground carbon, (i.e., carbon sequestered in the coarse roots and stool) is substantial [18
] and can reduce GHG emissions by as much as 138.4 kg CO2eq
per Mg biomass [1
]. Results of SOC changes under willow cultivation are inconclusive to date in New York State because changes have only been measured at a limited number of sites [19
]. Recently, detailed studies have been conducted by Argonne National Laboratory to predict SOC change associated with LUC using the CENTURY model, thus providing estimates of this critical component of GHG emissions of bioenergy crop feedstock production [20
]. However, LCAs that incorporate SOC changes in willow feedstock production have been lacking in the US.
In previous LCAs that did not include changes in SOC, the three largest contributors to the life cycle GHG emissions are fertilization, transport, and harvesting [11
]. In two LCAs that include SOC change in Sweden, yield was found to be the most important factor for climate change mitigation [13
]. These four factors are subject to a large magnitude of uncertainty in LCA of willow production in different geographical locations [1
]; therefore, incorporating spatial variability in LCA modeling is one approach to address this uncertainty. An LCA that considered spatial variability and changes in SOC content had an average EROI of 30:1 and determined that 84.3 Mg carbon ha−1
would be sequestered in the soil during a 100-year time frame in Sweden [14
]. However, studies with SOC change from land use changes with finer spatial scales that can incorporate more variability on landscape levels are lacking in the US and other parts of the world.
The objectives of this study were to conduct spatial LCA of willow biomass production and delivery of chips to an end user including SOC change and location-specific modeling of GHG emissions and fossil fuel consumption to investigate variations across five counties in central and northern New York State. In this study, biomass growth data from 480 ha willow biomass crops and other recently collected primary and GIS data from the region were used as data sources to create an LCA model of willow biomass production that yields cradle-to-gate GHG emissions and EROI.
This study illustrates that willow crops are a biomass feedstock that is carbon-negative across the landscape when it is grown on land that was formerly in cropland/pasture, which makes up 88.6% of the suitable parcels identified (Table S3
), and is a low-carbon feedstock when grown on grassland. When cropland/pasture is converted to willow, the baseline GHG emissions are −126.8 kg CO2eq
biomass while the spatial analysis shows that some parcels have GHG emissions below −200 kg CO2eq
biomass. The negative life cycle emissions for willow grown on cropland occur across almost all of the parcels in the five county regions, except for a couple of percent that are above 0 kg CO2eq
when one way transportation distances exceed 126 km. When willow is grown on former grassland, which makes up 11.4% of the identified suitable parcels, the baseline life cycle GHG emissions are 27.7 kg CO2eq
biomass and over 99% of all parcels are positive. The feedstock from these parcels can be considered a low carbon source of biomass. A previous LCA of willow grown in the region reported negative GHG emissions for all the scenarios, which included different transportation distances, high and low yields, and the application of N fertilizer or no fertilizer [1
]. A review of LCAs in Europe and North America reported slightly positive GHG emissions (14–207 kg CO2eq
), but most of these studies excluded an assessment of belowground carbon and SOC changes associated with direct land use change [9
]. Both these factors were included in this study using the best available, but limited, data and their influence on the results highlights a need for additional studies to refine this information. For willow grown on former cropland/pasture in this study there is the potential to sequester carbon while producing feedstock that can be used to provide a range of bioenergy and bioproducts and simultaneously create jobs and provide a range of other ecosystem services like improved water and soil quality and enhanced biodiversity [40
The variations in life cycle GHG emissions between and within counties are substantial and driven by multiple parameters that have different resolutions ranging from the sub parcel (e.g., SSURGO at 10 m resolution) to county level (e.g., SOC change). For the cropland 30-cm soil scenarios, the GHG emissions differed by county ranging from −176.9 kg CO2eq
in Lewis to −53.2 kg CO2eq
in St. Lawrence. For the grassland 30-cm soil scenarios, Jefferson has the lowest mean GHG emissions: 49.4 kg CO2eq
compared to 134.2 kg CO2eq
in Oswego (Figure 7
). Although Jefferson and Lewis counties are both close to the two end users, the life cycle GHG emissions for grassland in Lewis County had larger values due to higher rates of SOC change (Table 3
) and lower yield (Table S3
) than Jefferson County (49.4 kg CO2eq
). The transportation distance for Lewis County is generally much smaller than Oneida County, however, the life cycle GHG emissions of Oneida County in grassland are lower than Lewis County due to higher yields and a similar SOC change rate. The impact of transportation distances, yield estimates, which also impacted values for belowground biomass and harvesting operations, and changes in SOC on GHG emissions varied parcel to parcel in this study, creating complex patterns across the landscape (Figure 7
). Using LCA and other tools that use a systems approach to analysis and create a framework for examining the interactions among different components is important to understand the overall GHG emissions from these systems and the factors that need to be addressed to further improve and reduce uncertainty for willow systems.
The net-positive life cycle GHG emissions from grassland compared to the negative net emissions for cropland are largely due to the different modeled changes in SOC associated with land use change. The baseline results use an average SOC change rate across the five counties of 0.28 Mg carbon ha−1
(ranging from 0.19 to 0.34 Mg carbon ha−1
) sequestration rate for 30-cm soil in cropland/pasture and −0.29 Mg carbon ha−1
change rate (ranging from −0.19 to −0.43 Mg carbon ha−1
) (Table 3
) in hay/grasslands after conversion to willow crops [19
]. The SOC change rate in 100-cm soil is higher than that in 30-cm soil, because the SOC modeling of 100-cm simulates both topsoil (0–30cm) and subsoil (30–100cm) including additional soil organic matter pool for the subsoil [19
]. However, the majority of willow in northern NY is growing on marginal land and the soil depth is often limited due to perched water tables, hardpans, or bedrock. Previous assessments of rooting depth in willow in the region limited sampling to 45 cm because soil conditions limited root development in almost all cases [17
]. Thus, while SOC changes are reported here to 100 cm depth, the number of sites where willow will be grown in this region and changes will occur to this depth are probably limited. Incorporating the best available data on soil carbon change due to land use change for a willow energy crop into a spatial LCA is an important step forward in the understanding of these systems, but the data is only available on the county level and does not capture the range of variation across the landscape. The results of this study emphasize the importance of this change in the overall GHG balance of these systems and that better data on these changes are needed. For example, SSURGO soil data is at spatial scale of 10 m, and potential future improvement of current county-level SOC data from CENTURY would support more detailed parcel-level soil SOC modeling from LUC including soil variations at the parcel level.
We incorporated SOC change associated with direct land use change in this study because of the impact it has on the results for willow biomass as well as biofuels that are produced from this material [41
]. The baseline GHG emissions in this study (30 cm depth) without SOC change was −52.3 kg CO2eq
for cropland and −49.9 kg CO2eq
for grassland, which were similar to the −52.7 kg CO2eq
value for the high yield-fertilized-long transportation distance scenario reported in an earlier study [1
]. When SOC change is included, the baseline values in this study become −126.8 kg CO2eq
for cropland and 27.7 CO2eq
for grassland. For both grassland and cropland, SOC change is the second largest factor, after belowground carbon, contributing to the overall life cycle GHG emissions. Caputo et al. (2013) included belowground biomass but not SOC changes and had negative GHG emissions in all their scenarios [1
]. Few of the studies reviewed by Djomo et al. (2011) included changes in soil carbon or belowground carbon stored in coarse roots and stools, and the GHG impact of these studies were all positive, ranging from 0.7 to 10.6 g CO2eq
(13.7–208.8 kg CO2eq
]. The inclusion of the SOC and belowground carbon included in this study results in negative GHG emissions for cropland and slightly positive values for grassland that are at the low end of the range reported by Djomo et al. (2011) [9
]. In a more recent LCA that included soil carbon sequestration [42
], the life cycle GHG emissions of willow biomass production were estimated as 0.8 g CO2eq
for arable land, −10.4 g CO2eq
for marginal pastureland, and −31.8 g CO2eq
for marginal abandoned land in 100-cm soil depth. For the three scenarios there were changes in SOC (28.5 kg C ha−1
for arable land, SOC gain of 153.4 kg C ha−1
for marginal pastureland, and SOC loss of 31.5 kg C ha−1
for marginal abandoned land) in contrast to the SOC increases in cropland but decreases in grassland used in this study. However, the differences of life cycle GHG emissions by SOC change were relatively minor compared to other processes such as transportation [42
]. The inclusion of changes in SOC carbon associated with land use change is important to understand the impacts and benefits of these systems, but better data is needed to understand these patterns.
There is a limited amount of data available on changes in SOC under willow, especially in North America, and the results are inconclusive. A meta-analysis showed that a transition from cropland to short rotation coppice willow or poplar increased SOC by 5.0% ± 7.8% and by 3.7% ± 14.6% when these crops were planted on grassland [43
]. These analyses only had 18 data points for cropland and seven for grassland, so there is large amount of uncertainty. An earlier analysis suggested an increase in SOC over time when poplar or willow are planted on cropland and no change or a slight decrease when planted on grassland, but model results vary based on the time period of the assessment [20
]. For example, losses of 7–8% of soil C were suggested for data collected in the first 5 years after willow is planted on grassland, but for studies that had time frames longer than 5 years there is no significant change from the baseline [20
]. A side by side study of willow and grassland over a 2-year period showed that willow was a net carbon sink even when biomass was harvested while the grassland was a net source of carbon [44
]. This study is limited to one site and, as is the case with all studies, is influenced by site and management practices both before and during the study. A soil C chronosequence study, ranging from 5 to 19 years old, for willow in the region of the US where this study occurs suggests little change over that time period [19
]. The data on SOC changes associated with willow biomass crops is limited and uncertain. We used the best available data based on SOC change associated with direct land use change based on modeling done by the Argonne National Lab using the CENTURY model, but this data was only available at the county level. More information on the dynamics of direct land use change for woody crops as well as a better understanding of long-term change in soil C are needed to reduce uncertainty and the overall GHG balance of willow systems across the landscape.
The inclusion of belowground biomass in this and other studies is important because of the magnitude of its impact. In this study, it is the largest single carbon sink, but as is the case with SOC data, information on this important component of the system is limited. In the 30-cm soil grassland scenario, belowground carbon accounts for −144.71 kg CO2eq
(45.6% of the total impacts in absolute numbers) and the SOC change accounts for 77.55 kg CO2eq
(24.5% of the total impacts). One study in the US of the carbon stored in belowground biomass suggests it changes over several rotations and is substantial, 45.3 ± 4.4 Mg CO2eq
(25.4 Mg C ha−1
]. This is 1.6 to almost 3 times greater than the increase in soil carbon in willow relative to cropland and exceeded the losses in SOC replacing grassland. In another recent study, root:shoot ratios varied among cultivars (0.46–0.72) at one site but not at a second site (0.63–0.65) [31
] but some of these differences were reduced when yields were used as a covariate in the analysis. The belowground biomass values used in this study are proportional to the aboveground biomass [31
]; therefore, increasing the yield of stem biomass in turn increases belowground carbon sequestration. This allowed us to apply these belowground values on a parcel by parcel basis across the landscape, but there is a need for more detailed information on how the belowground portion of these system changes across sites and with different cultivars and management systems.
Aboveground biomass yield is the most widely studied aspect of SRWC because it is the main product generated from these systems. In this study, changes in yield across the landscape had a direct impact on belowground biomass values, throughput of the harvester, and influenced the impact of changes in SOC, which was a fixed value per ha for grassland or cropland across an entire county. By using the relationship between NCCPI and yield from a previous study, we were able to assign yield at the sub parcel level because NCCPI data is available at a 10 m resolution and the land cover data we used was available at 30 m. This data effectively represented yields that have been measured in research and commercial scale trials across the region. Recently developed models used to predict yield across the continental US used 800 m grid (64 ha) for climate data and a 1024 ha parcel size for soils data [38
]. Using this model to predict yields in this study would have further limited our ability to assess changes across the landscape.
While transportation distance from the field to the gate of the end use facility has been noted as an important factor in the GHG emissions of willow and other energy crops in the past, this current spatial LCA analysis illustrates the importance of understanding how transportation distance interacts with other factors. In Caputo et al. (2014), there are two haul distance scenarios to biorefineries, namely long haul distance and short haul distance [1
]. The baseline case in this study is similar to the short haul distance scenario in Caputo et al. (2014) with an average 35.5 km to biorefineries. The life cycle GHG emissions result of −126.8 kg CO2eq
biomass in this study is at the lower end of the GHG emissions range from −83.1 to −138.3 CO2eq
of short haul distances scenarios in Caputo et al. (2014) [1
]. Sensitivity analyses in recent LCA studies of willow biomass crops in Sweden [14
] and in Spain [45
] showed that transportation distance had the largest influence on life cycle GHG emissions. In this study, transportation was consistently one of the top three factors influencing GHG emissions, but its impact was modified by other factors. For example, sensitivity analysis showed that transportation distances changed GHG emissions less than 5% in Jefferson county but by more than 10% for Oneida county. However, the resulting distribution of GHG emissions for Jefferson and Oneida county in the uncertainty analysis are similar because of other factors such as modeled SOC increases on cropland, which were 30% higher in Oneida than in Jefferson. Integrating multiple factors in a spatial LCA provides valuable information to support decision making on where to establish willow biomass crops if one of the goals is to minimize GHG emissions.
Harvesting accounts for 9.48 kg CO2eq
in all scenarios and is the second largest impact among crop management activities after fertilizer, so improvements in harvester efficiency can contribute to GHG emission reductions. This number is about 27% lower than the 13 kg CO2eq
value in a previous study [1
] due to improvements in harvester efficiency of single pass cut and chip operations modeled in this study and improved data for harvester operation and fuel consumption that have been made over the past few years [29
]. Site management is an integrated category including all site preparation, planting, and site maintenance processes, specifically including processes #1 to # 15 and process #20 in Table S1
. Site management activities are a small component of the total life cycle GHG emissions because they only occur once in the 25-year time span of this system, accounting for 4.40 kg CO2eq
in 30 cm grassland scenarios and are slightly lower for cropland (1.97 kg CO2eq
) scenarios because fewer site preparation operations are required.
Willow biomass has a high energy ratio in central and northern New York State; the energy consumption was estimated as 969.6 MJ for 1 Mg biomass, and the net EROI is 19.2. For cropland, this number dropped to 937.9 MJ for 1 Mg biomass and increased to 19.8 as EROI. The overall GHG impact and energy balance of an energy system will depend on the conversion technology and final end products generated from the willow biomass.
The energy consumption values are mainly affected by transportation distance to the end users (Black River and Lyonsdale). The parcels near the power plants show less than 500 MJ Mg−1
energy consumption compared with over 2750 MJ Mg−1
in energy consumption in parcels further away (Figure 9
). The geographical distributions of energy consumption demonstrate that parcels further away from end users tend to have higher energy input largely due to increasing the transportation distance. However, the pattern of increasing energy consumption is not strictly circular centering around the end users because it is also impacted by the spatial distribution of yield and land cover types and the tortuosity of the roads in portions of the study region.
Harvesting and fertilization are the two largest energy inputs to willow biomass production, which agrees with other studies [1
]. These management activities contribute 14% (harvesting) and 17% (fertilization) of the total energy consumption of 969.6 MJ Mg−1
for the baseline grassland scenario. The energy demand of willow biomass production in Caputo et al. (2014) ranges from 445.0 to 1052.4 MJ Mg−1
biomass for eight scenarios with variations in transportation distance, yield, and fertilizer application [1
]. Energy inputs in this study are larger than six of the eight scenarios in Caputo et al. (2014), with the four lowest energy inputs being scenarios that exclude fertilizer inputs and the associated input of 164.6 MJ Mg−1
. Another difference is the inclusion in this study of a process to load chips from piles on the ground into trucks for delivery to an end user, which is the most common practice in the region. In this study, the energy input associated with loading chips was about 35% higher than site management energy inputs for grassland and 2.8 times greater for the cropland scenario because it is an activity that occurs multiple times over the life of the crop. As a result, the range of EROI values in this study (8.95 in the St. Lawrence grassland scenario to 18.6 in the Lewis county cropland scenario) is on the lower end compared to the range of EROIs of 18.3 to 43.4 in Caputo et al. (2014) [1
]. However, this study’s EROI values are at the high end of the 3–16 values for cradle-to-plant LCA studies reviewed in Djomo et al. (2011) [9
]. Harvesting throughput values in our study are based on more recent data that showed a substantial improvement in harvester throughput [29
] that results in lower energy consumption per Mg of biomass and a higher EROI. Further improvements in harvesting and chip loading operations may be possible to further lower energy inputs into the system.
One approach to increase EROI in this system, and reduce GHG emissions, is to replace synthetic fertilizer with organic amendments or eliminate fertilizer applications, as long as the yield can be maintained. However, the yield response of willow to different rates of fertilizer is not well defined with most previous studies in the region (except for Adegbidi et al. (2001) [47
]) showing no statistical improvement in yield with fertilizer additions [48
]. However, a recent meta data analysis of willow fertilization trials across North America and Europe indicated that there was a lot of variation in yield response, but an overall positive trend was present [50
]. Another option to shift fossil fuels based fertilizer energy inputs is to use organic waste materials like manure or biosolids, which Heller et al. (2003) suggested could increase EROI by 33–45% [51
]. Regional studies showed that these materials produce yields comparable to commercial fertilizer [49
]. While we included a range of fertilizer additions in the sensitivity analysis of this study, we did not include an associated change in yield because of the uncertainty of this relationship. Being able to further define the yield response to fertilizer additions across a range of soil types and site conditions would allow a better assessment of the impact of fertilizer on EROI and GHG emissions in willow systems.