From 2002 to 2014, the total global terrestrial NPP was, on average, 212.6 GtCO2 y−1, a value that is in agreement with others (Field et al. 2007). If we assume that 50% of the NPP is allocated aboveground (Frank et al. 2004) and that 2/3 of the aboveground resource can be harvested sustainably to maintain soil carbon, then the sustainably harvestable biomass resource across the terrestrial biosphere is roughly 70.9 GtCO2 y−1 globally (Table 1).
Highly prospective CO2 basins underlie 1786 Mha of the terrestrial surface, but much of this area corresponds to relatively low NPP (Fig. 1). Notably, there are few high confidence storage basins under intensively cultivated regions, such as the United States Corn Belt or the Eurasian Wheat Belt, and there is an almost total lack of highly prospective storage under the equatorial forests of Africa and South America. Total NPP over the highly prospective basins is 22.9 GtCO2 y−1, corresponding to a sustainably harvestable total of 7.6 GtCO2 y−1 or 11% of the global quantity (70.9 GtCO2 y−1).
Future energy crops will need to displace natural and managed ecosystems, and biodiversity protection and food production should be considered. To better understand these constraints, we estimate the maximum potential production of sustainably harvested biomass overlying suitable storage formations for each vegetation type based on the assumption that current NPP is a useful proxy for energy crop NPP (Fig. 2). These estimates represent an upper bound for each land-use type.
Evergreen broadleaf forests are the most productive vegetation class, leading to a total sustainably harvestable resource of 1.3 GtCO2 y−1 grown on 103 Mha, largely in Peru, Colombia, and Indonesia. Other productive vegetation types that substantially overlap highly prospective basins include cropland/natural vegetation mosaic, deciduous broadleaf forest, savannas, and mixed forest. Though croplands are the largest readily available resource and could sustainably contribute up to 1.5 GtCO2 y−1, relatively low sustainable harvests imply the requirement for a large land area.
Because BECCS will be more valuable for CO2 removal rather than as a source of energy (Klein et al. 2014; Sanchez et al. 2015), it will rarely make sense to replace forests, a high carbon value ecosystem, with energy crops. Though the economics might be compelling, unless the initial harvest, prior to energy crop deployment, is utilized for biomass energy, the carbon debt from deforestation is a large setback, difficult to reverse even after decades of management for biomass crops (Fargione et al. 2008). High carbon value ecosystems are often protected in IAMs for this reason, and some models rely on afforestation as a cost-effective carbon removal technology (Popp et al. 2014; Griscom et al. 2017). Forested lands account for 52% of the sustainably harvestable NPP over highly prospective basins (4 GtCO2 y−1; Fig. 3). As a group, forested lands are the most photosynthetically productive. However, the carbon relevance of currently forested lands in producing BECCS feedstocks is likely to depend on (1) whether policies require accounting for the deforestation-related carbon debt and (2) whether the BECCS technology allows utilization of the initial (pre-energy crop) biomass for energy.
More than one third (36%) of the total area overlying basins is either unsuitable for agriculture (e.g., urban or wetlands) or has NPP so low that commercial harvesting is unlikely to be economically viable (e.g., frozen, desert, or shrubland). Though vast (647 Mha), these lands produce a total NPP of only 0.6 GtCO2 y −1 or a sustainably harvestable level of 0.2 GtCO2 y−1 (Fig. 3). Some of these areas, regions in North Africa and the Middle East for instance, with ample solar and basin capacity but negligible NPP could represent important areas for direct air capture deployment.
Croplands and grasslands constitute the remaining 10.2 GtCO2 y−1 of NPP or 3.4 GtCO2 y−1 of sustainably harvestable biomass overlying a storage site, occupying 641 Mha of land (Fig. 3). Although this resource opportunity is approximately equivalent to a stabilization wedge (3.7 GtCO2) (Pacala and Socolow 2008), some of the area will be required for food production (food crops + livestock pasture). The acreage of cropland over basins (262 Mha) represents one fifth of the global cropland area. If these lands continue to be used for food production, the sustainably available material for bioenergy will be some fraction of the agricultural residues left after harvest. If 50% of aboveground NPP is allocated to harvestable material (grain) and 33% must remain to maintain soil fertility, then the remaining 17% of aboveground NPP is available as residue. Under these assumptions, cropland residues could sustainably contribute 0.4 GtCO2 y−1 of BECCS.
Agricultural lands that are not cropped every year may be an attractive target for conversion to energy crops, since they are of limited value for food production. Marginal agricultural lands can be marginal for a number of reasons (Robertson et al. 2017). Building from farmer decisions about cropping, we define land as marginal if it is used for crop production in fewer than 11 years of the 13-year MODIS record. Based on this definition, the total NPP on marginal agricultural land over highly prospective basins is 3.4 GtCO2 y−1, with 1.1 GtCO2 y−1 sustainably harvestable (Fig. 3).
Although 1.1 GtCO2 y−1 is approximately 10% of the supply used in < 2 °C scenarios, this is a meaningful first step towards decarbonization and, with carbon prices exceeding $100 tCO2−1, represents a massive market opportunity. However, this is a conservative estimate because we implicitly target least-cost options with low implementation barriers, without the need to first develop a pipeline or other transportation networks. At low CO2 prices such opportunities could catalyze the development of the infrastructure and finance markets needed to meet < 2 °C scenario targets. Going beyond the BECCS potentials presented here will require long distance travel, and techno-economic analyses at different price points are needed to understand how the potential harvestable resource changes with respect to CO2 price. Yield improvements and energy transformations, such as those included in IAM scenarios, are likely to increase the potential as well.
The intersection of storage basins and agricultural land not cropped continuously is concentrated in a few regions. North America dominates at the continental scale as consequence of widespread storage potential in both the USA and Canada. Australia, Africa, Asia, and South America do not have extensive highly prospective basins (Fig. 4a).
At the national scale, over half (51%) of the sustainably harvestable biomass from marginal agricultural lands is situated in three countries: the USA (0.3 GtCO2 y−1), Russia (0.13 GtCO2 y−1), and Canada (0.1 GtCO2 y−1). More than three quarters of the resource is in 10 countries (the USA, Russia, China, Canada, Myanmar, Mexico, Indonesia, Romania, Germany, and Venezuela) (Fig. 4b). Although these estimates are small relative to global emissions, they could help allow select countries to meet ambitious decarbonization targets. In Myanmar and Hungary for example, as much as 58 and 24% of recent national emissions, respectively, could be offset using the sustainably harvestable biomass from marginal agricultural lands for BECCS.
These estimates are based on one approach to estimating the sustainably harvestable biomass resource. Three main factors could increase or decrease the amount of available biomass co-located with suitable storage basins. First, the area of suitable storage basins may change. Storage in igneous rocks such as basalt may be possible, as demonstrated in Iceland (Matter et al. 2016). In this case, the land area for highly prospective sites would increase dramatically, for example, in India. Storage basins in some regions (e.g., Central Africa) are not well explored, and injection into low to moderate prospectivity basins may be practical in some cases because the low energy density of biomass and limited collection area for an individual project results in relatively small emissions sources, requiring only marginal storage capacity and injectivity. Second, future NPP of biomass crops may be higher or lower than recent NPP. Typically, the conversion of native vegetation to agriculture reduces NPP, largely because annual crops have a shorter growing period (Krausmann et al. 2013). Though perennial energy crops may be impacted less by the length of the growing season, other factors, including the time needed for crop improvement and limited farmer experience, are likely to be constraints (Field et al. 2007). Subsidies in the form of fertilizer or especially irrigation can push the NPP of managed land well above those of unmanaged ecosystems (Krausmann et al. 2013). Such subsidies would increase biomass for BECCS but would also increase costs. Subsidies and improved management would have the greatest NPP impact in regions where the yield gap is largest, in developing countries and marginal lands for instance. Higher CO2 concentrations, climate change, crop selection, or crop improvement could all lead to increased NPP, although climate change and crop selection could also lead to decreased NPP. Third, the sustainably harvestable fraction of NPP could be more than 1/3 in some settings and less in others.
The future success of BECCS depends on strategic plant siting to build up the necessary technology, infrastructure, and markets to support the nascent industry. Although the BECCS opportunity on marginal lands overlying storage basins is substantially less than ambitious climate stabilization scenarios require, it is vastly greater than current CO2 removal and could, in turn, catalyze the development of the transport infrastructure needed to meet long-term climate targets.