Identifying the location and amount of grain crop residues (stubble) in Australia is necessary for determining the viability of potential biofuel plant locations. We combined 22 years of crop statistics with harvest indices and land use to arrive at spatially explicit stubble productivity figures. Stubble quantities using different focal radii and from different seasons provide an insight into the feasibility of its use for bioenergy. We focus on areas where the stubble concentrations within a 50 km radius were at least 500 kt per year; the amount suggested for a viable lignocellolosic bioethanol facility. The outcome of this study has been to show, for the first time, where there are large amounts of stubble in Australia. Whether the supply of stubble is sufficiently constant over time and indeed available at a price that is economic for a biofuel plant must be subject to future work.
With the increasing world demand for energy, and the declining supply of low cost fossil energy resources, there is increasing interest in biomass for energy conversion (Richard, 2010). Worldwide estimates of biomass potential range from 10% to 60% of the current world energy demand (Richard, 2010). The IEA estimated that, a halving of the world's GHG emissions by 2050 requires an almost fourfold increase in biomass use to contribute 23% to the world's energy production (Taylor, 2008). Estimates of the ethanol production potential in the Asia-Pacific Economic Cooperation countries on marginal lands alone is 540 GL, although this would require a significant effort in establishing dedicated production systems outside current agricultural production areas (Milbrandt & Overend, 2009). Estimates for the worldwide technical bioenergy potential of biomass exist, but the feasibility of realising these depends on a range of issues including the availability of significant quantities of feedstock to make biofuels at a price competitive with fossil fuels (Sims et al., 2008; Richard, 2010).
Lignocellulosic (non-edible) biomass has a relatively low energy content compared to fossil fuel and its availability is spatially and temporally variable. Energy content, spatial density and temporal variation are the primary influences on security of biomass supply, harvest and transport costs and are thus key factors determining the viability of potential biofuel plant locations. Using crop residues for biofuels reduces the competition between biomass production and high value cropping for arable land because it does not directly impact on crop or fibre production (de Gorter & Just, 2010; Kerr, 2010; Richard, 2010). Thus, quantifying location and amount of crop residue resources is critical in assessing the potential of a region or nation to use biomass for energy.
Worldwide estimates of feedstock availability generally provide insufficient detail for determining potential biofuel plant locations. This is because of the diametrically opposing scale issues associated with biomass for biofuel conversion. Economies of scale for lignocellulosic biofuel facilities dictate a minimum size of a generally large fuel output in the order of at least 200 ML, which has an associated high biomass demand. Biomass production on the other hand is locally variable and has limited capacity, constrained by land use, soil, climate and nutrient supply (Sims et al., 2008; Richard, 2010). Thus, there is a need for estimates of biomass amounts by resource location at fine spatial and temporal resolution. This is an initial step to underpin development of biofuel industries to improve national energy security or GHG mitigation strategies.
Currently, there is limited biofuel production from the lignocellulosic part of biomass, but this conversion technology is predicted to play an important role in the bioenergy future because of its limited competition for food production when compared to first generation biofuels (Sommerville et al., 2010). Bionergy facilities of various sizes for producing ethanol from a range of biomass sources including food wastes and industrial waste streams exist. However, for pure lignocellulosic biofuel plants to be economically competitive they require a minimum size of around 200–1000 ML of yearly fuel production and respective large biomass supply (Richard, 2010). Conversion factors are around 310–380 L of ethanol per metric tonne of biomass (Sommerville et al., 2010), which translates into approximately 0.5–3.2 Mt of yearly biomass requirement. Thus, scaling up biofuel production requires knowledge of feedstock resource location and density to identify areas with sufficiently large biomass production within an economic transport distance to sustain a biofuel facility.
Australian farmers contribute significant amounts of the world's grain production using conservation farming practices. While stubble retention is widely adopted, farmers also utilise and remove grain residues to overcome planting difficulties under no-till systems (Llewellyn et al., 2009; Scott et al., 2010). In Australia, the in-field residue components of grain crops are commonly referred to as stubble, which is the terminology we use subsequently in this paper. Farine et al. estimate the national stubble potential to be approximately 25 Mt per year based on a 10 year average (Farine et al., 2011).
Approaches assessing biomass resource potential differ markedly (Smeets et al., 2009). In this manuscript we use the staged approach outlined in Herr et al., which starts with a top-down assessment of theoretical biomass production potential, then applies technical and environmental constraints, followed by social and economic constraints. Each appraisal reduces the available amount of biomass that is available for a bioenergy processing facility (Herr et al., accepted). During each stage there is an assessment to determine if a feedstock is worthwhile investigating.
The purpose of this paper is to establish if there is sufficient stubble resource available in Australia to warrant further research investment. While Farine et al. have provided national estimates for stubble in Australia, they fall short of providing the spatial distribution of the resource (Farine et al., 2011). The work we present here is at the environmental–technical assessment stage and includes a spatial analysis of stubble, spatial densities as well as temporal variations for the Australian continent.
Materials and methods
Figure 1 shows the location of Australia's major grain production zones (GRDC, 2003; BRS, 2006). This area covers approximately 1 million km2 with major crops being wheat, barley, and oats which account for 90% of national grain production. The other 10% consists mainly of sorghum, canola, lupins, peas and beans (Unkovich et al., 2009; see e.g. Herr et al., 2010).
In this analysis, 22 years of national crop production statistics provided the basis for calculating stubble production. These data come at the statistical division level based on surveys from the Australian Bureau of Statistics, which the Bureau of Rural Sciences concorded to a common year 2001 geography (J. Walcott, BRS, personal communication). Stubble production is not routinely measured in Australia, so this study estimates it from grain yield data and knowledge about the ratio of grain to total above-ground biomass (harvest index (HI)). While yield data are widely available, HI data are less available at this scale and little is known about how HI varies spatially, between years and with management. Hence we embarked on determining if HI is the main source of variation in estimating stubble.
Assessing HI variation for stubble estimation
Before calculating stubble from the harvest index for the whole of Australia, it is necessary to establish the main sources of variation that influence production. Analysis of a large crop harvest index database indicates that of the main grain crops, reported mean HI values varied between 27% for canola and 46% for sorghum. Average HI for wheat was 36% with a coefficient of variation of 19% across 194 sites by year combinations (Unkovich et al., 2010). We further investigated the potential influence of soil, climate and fertiliser inputs using the crop simulation model APSIM (Keating et al., 2003) to simulate wheat production for one site in each of the five main wheat growing States for the period 1940–2007. Details of location and parameters are outlined in Dunlop et al. (Dunlop et al., 2008). At each site we simulated two soils typical of the area with contrasting water holding capacity. Crop growth was for a no-till system with four nitrogen fertiliser application rates: 0, 50, 100, and 150 kg-N ha-1. This range is more extreme than would typically be applied, but it is useful for investigating the potential impact of nutrient stress and nitrogen application on HI. Each simulation assumed there was minimum carry-over of soil water and nitrogen after harvest. When estimating crop residues (stubble) available for off-take for biofuel, 15% of the total residues was considered unharvestable trash, and a further 1 t ha-1 (southern cropping regions) or 1.5 t ha-1 (northern cropping regions) retained for soil erosion protection, with the balance being available for biofuels.
Estimating the national stubble production spatially
Agricultural statistics, and socio-economic data in general, come with different boundaries and resolutions compared to biophysical data and land use, so there is need for spatial integration methods (Herr, 2007). Details of the method employed here are outlined elsewhere (Herr & Dunlop, 2011).
We combined stubble estimates with land use information to identify the spatial distribution of stubble amounts and limited these by technical and environmental constraints related to soil, water and soil organic carbon conservation. In this context erosion protection requirements are to retain 30% of the stubble yield in the field to prevent soil erosion. This equates broadly to a minimum stubble retention of 1.5 t ha-1 in the northern grain production zone (to account for increased decay in areas with summer rainfall) and 1 t ha-1 in the southern and western region (see Fig. 1). This enabled us to derive an average stubble estimate that is potentially available for harvest (subsequently referred to as stubble PAfH) in a spatial form (see Herr et al., 2010 for details).
As a surrogate for dealing with transport distance directly, we used the radius of a focal analysis. We used radii ranging from 40 km to 160 km to sum the stubble amounts around each focal point. While the concentrations of these pixels are not additive, these radii can be seen as an index of transport distance as they provide the concentration of stubble around each focal point for areas of 7900 km2 and 80 425 km2 respectively.
Differences in HI are the main variant for estimating stubble from different crops. There were no clear differences in modelled HI between sites with winter-dominant, equi-seasonal or summer-dominant rainfall. Higher nitrogen application rates tended to increase simulated HI, and there was no evidence of an interaction between soil water holding capacity and nitrogen application rates (Fig. 2). Simulated HI varied considerably from year to year. The coefficient of variation for HI ranged from 13% to 23% for 150 kg-N ha-1, to 20–42% for 0 kg-N ha-1 (for high water holding capacity soils). This was largely due to HI being reduced in high rainfall years when nitrogen was not sufficient, and to a lesser extent due to increased variability across the whole rainfall spectrum at very low nitrogen rates. In general, rainfall in the different years affected crop yield much more than HI: the simulated yield coefficient of variation was 35–47% for 150 kg-N ha-1 and 44–84% for 0 kg-N ha-1. Therefore it is highly likely variation in crop growth (yield) over the years will have a much greater impact on calculated stubble production than variation in HI.
We used the average harvest index from the crop harvest index database to estimate stubble PAfH. The stubble PAfH is highly variable over the years, which is a reflection of the weather variation in Australia. For example in the low rainfall cropping season of 2004 and the drought season 2003, stubble PAfH was highest (approximately 35 Mt) and lowest (approximately 7.5 Mt) respectively (Fig. 3).
There are four areas where stubble amounts are largest, ranging between 10 and 70 kt per 400 km2. These are in the western and southern cropping zones (indicated as clusters of orange-red in Fig. 4). These larger amounts are a result of higher density of cropping land, high grain yields or a combination of both (Fig. 4).
The minimum resource needs of a 200 ML enzymatic ethanol plant is approximately 500 kt (Richard, 2010; Sommerville et al., 2010). Figure 5 shows the total amount of harvestable stubble in areas where the focal concentration is above 500 kt for the given radius. The amounts are for a year with low grain production (2003), a high grain production year (1999) and for the yearly average from 1993 to 2005. At 50 km radius the total amount of harvestable stubble in a high production year is around 5 Mt. To achieve this amount in an average year the radius must be over 60 km and in year of low rainfall/low production, stubble PAfH of 5 Mt is not achieved within any radius of up to 160 km (Fig. 5).
The spatial distribution of the stubble resource varies with the radius of the focal analysis. This is shown in Fig. 6, where the stubble amounts (yearly averages) for a 70 km radius and a minimum of 500 kt PAfH stubble concentration are located in four main areas of the southern and western grain production zones (Fig. 6a). Using a 140 km radius, many more areas achieve 500 kt stubble concentrations; here spatial distribution becomes connected in the southern zone with some additions from the northern zone (Fig. 6b). The total average stubble PAfH in these areas is approximately 17 000 kt (Fig. 5, red squares).
The removal of stubble requires thorough consideration across the full range of benefits and impacts–which include potentially higher soil erosion, reduced water infiltration due to reduced surface cover (Bristow et al., 1996), and depletion of soil carbon and nutrients. The application of a simple set of constraints to calculate stubble PAfH is an attempt to ensure that these impacts are not so large as to outweigh the benefits, but does not constitute a full analysis of the potential broader impacts (Herr et al., 2010). Our spatial estimates of stubble PAfH incorporate these simple limitations and provide values at the statistical division level. Further details on the issues and calculations are located in Herr et al. (Herr et al., 2010). This radius range depends on the stubble production, which is variable from year to year (Fig. 3). For an average year, there are four areas that have ≥500 kt stubble in a 70 km radius and these areas are located in the main grain production zones (Fig. 6) and have approximately 7.5 Mt of total available stubble (Fig. 5, red squares). No stubble concentration exceeded 5 Mt in 2003 for the 70 km radius, which indicates that supply may become an issue in many areas during a drought year. Even with an extended radius, total available stubble will not reach 5 Mt during a low production year (Fig. 5). Thus, a lignocellulosic biofuel plant is unlikely to obtain a continuous supply of feedstock unless it increases its catchment area to more than 80 km radius or there are options of storage or feedstock substitution. For example, the co-location with non-crop residues from forest biomass would be advantageous. Australia has also large areas of (irrigated) cotton production. Regionally, this crop could also provide an additional residue source in large quantities (see eg. Fraser & New South Wales Office of Energy, 1995; Rodriguez et al., 2011).
In Australia, grain production is not evenly distributed in terms of yield and intensity of cropping. For example in the Western region with its mostly sandy soils, production is generally lower than in the southern region, but the density of cropping is higher in the Western zone (Herr & Dunlop, 2011). This has implications for transport costs, which is an important factor in the economics of biofuel production. Transport costs in Australia are in the vicinity of A$0.18–A$0.27 per t and km travel distance in the 50–200 km range (Rodriguez et al., 2011; 2009 oilprice A$79 per barrel), so this would add up to approximately A$27 t-1 to the feedstock price at the ethanol plant gate. Currently there is a mismatch in cost of stubble production and the amount a biofuel producer is able to pay for the feedstock: Full cost recovery of stubble extraction from the paddock is approximately A$100 t-1, which includes harvesting and minimal nutrient (N, P, K) replacement costs (Herr et al., 2010), while the feedstock price an ethanol production facility could afford under current conditions (i.e. oilprice and capital costs in Australia) is in the range A$70–A$100 (Rodriguez et al., 2011). Future oil price and carbon price increases may improve the ability of energy producers to pay for feedstock. Still, much of the stubble resource is unlikely to be harvested for environmental and agronomic reasons. On average only 21 Mt of the total stubble production are calculated to be potentially harvestable. When requiring a minimum of 500 kt of stubble concentrations at a local plant, only 7500 kt (36%, 70 km radius) and 17 500 kt (83%, 140 km radius) are available on average. If the latter were converted into ethanol it could replace between 8 and 20% of Australia's 2010 petrol demand (18.7 GL), using 310–380 L t-1 of biomass to petrol conversion factor and accounting for the lower energy content of ethanol. This could save between 3.4–8.6 Mt CO2 emissions (DRET, 2010; DCCEE, 2011).
Our work identified that there are significant amounts of stubble PAfH at several locations in Australia, which would be suitable for commercial ethanol plants at 200 ML capacity. However, this is only a first step towards the feasibility of an Australian biofuel industry that incorporates crop stubble as feedstock. We have assessed the environmental and technical constraints for stubble production potential as a first step. Farmers will limit their engagement in biomass for bioenergy production if the trade-offs from stubble removal are unclear and hence perceived as higher risk. There is also need to address cultural and social reasons, as farmers’ attitudes and peers influence their management (see e.g. Herr et al., 2004).
This study has highlighted the approximate range and upper limit of stubble availability for Australia and put this into the context of current commercially viable lignocellulosic ethanol plant requirements. Our work (1) investigates the suitability of using HI to estimate stubble; (2) uses this to provide a spatially and temporally explicit account of stubble that is potentially available for harvest; (3) applies some simple constraints to stubble removal and assess transport distances to highlight areas where there may be sufficient stubble; and (4) highlights the supply concerns with temporal variability in production due to climate variability. Such supply concerns could be alleviated by using storage and/or other feedstock such as forestry residues at locations where both biomass types are within reasonable transport distance. For example, stubble can be baled tightly (to reduce infiltration of moisture) and stored to maintain year around supply. In addition, many overseas facilities (e.g. Abengoa in Salamanca, Spain) are looking at combined grain/stubble facilities, where the ethanol production inputs can be flexibly shifted between stubble and grain depending on availability and price (O'Connell, 2008).
The study demonstrated that there are four areas in Australia which could be considered for a lignocelulosic ethanol plant of an annual 200 ML capacity. This assumes that it is economically feasible to obtain the biomass in a 70 km radius or that there are options for storage and biomass substitution and that it is environmentally acceptable to harvest stubble. Our analysis does rely on average production figures, so there may be issues with the consistency of supply within individual years. Future work would need to focus on improved field verification of the stubble estimates at specific locations and also quantify the impacts of stubble removal on soil erosion and soil carbon stocks at these locations. Economics of harvest, transport and substitution with other feedstocks are also of future interest when investigating the feasibility of a location for a biofuel plant.
The CSIRO Sustainable Agriculture and the Energy Transformed Flagships provided funding for this research in collaboration with the Grain Research and Development Corporation. We thank Rick Llewellyn and Roger Lawes for their critical reviews, and Peter Hairsine and Carlos Gonzales-Orozco for early inputs into the development of this paper. We heartily thank Johannes Bauer for title suggestions and Monika Van Wensveen for editing.