Synthesis and Review: Advancing agricultural greenhouse gas quantification

Researchers have developed a range of methods for quantifying—measuring and modeling—GHGs from agricultural systems, however, field sampling can be costly especially for larger scales and models may not have sufficient calibration and validation data to be applied in new areas. The articles in this issue identify innovations that can advance the field.

Vägen and Winowiecki (2013) demonstrate methods for using cumulative soil mass (calculated as the product of bulk density and soil depth per unit ground area) to estimate soil organic carbon stocks, which is simpler and more robust statistically than typical measures of bulk density to a fixed depth. Field sampling determines the relationship between carbon and soil type and can be linked to satellite data for landscape-scale assessment of soil carbon and drivers of soil loss to target management options. Mapping soil carbon patterns using satellite data and calibrating this with landscape level field sampling of soils is a leap forward.

Quantifying nitrous oxide fluxes ⁶ is challenging because of high costs of equipment that can be easily damaged in the field, and high spatial and temporal variability of fluxes. Fluxes are highest around brief wetting or freeze-thaw events, with most of the annual flux occurring in only a few days. Current methods involve chambers placed on fields, which are good for detecting relative differences across treatments and hence explaining processes. However, they also have low spatial and temporal resolution, which makes accurate quantification of field-level emissions difficult or requires expensive sampling. Micrometeorological methods (Eddy flux towers) can be used for integrated measurements over large areas, but are limited to landscapes with uniform surfaces and are expensive. Where research infrastructure is minimal, Hensen et al (2013) suggest low-cost mass balance and plume methods to measure localized significant sources. Rapid advancements in analytical technologies are expected to allow improved field measurements in the next few years (Hensen et al 2013).

Wetlands—including flooded lands, irrigated rice and mangrove fisheries—are challenging for their complexity and variability due to changing hydrological conditions and large gaps in knowledge. Lloyd et al (2013) provide guidance on combining field measurement, modeling and remote sensing. They use automatic weather station data in simple energy and GHG exchange (SVAT) models, with Eddy covariance data over two to three days to calibrate the model during pivotal periods. Remote sensing is used to validate the results. Used together, the models and remote sensing can extend estimates to landscape and regional scales.

Paustian (2013) discusses the potential for biogeochemical modeling to rapidly advance quantification. The information to support these models in low income countries is just beginning to emerge with resources like the World Digital Soils Map (Sanchez et al 2009) and climate and remote sensing data. At the same time, improving the quality and the quantity of underlying activity data—the type and amount of different crops, animals and management practices—is also critical for modeling. While available (but incomplete) in well-resourced countries like the United States, and available in the form of long-term national statistical information at FAO (FAOSTAT 2014), some critical information may be virtually absent in many low-income countries. Paustian (2013) proposes collecting activity data through existing mobile telephones to crowd-source geo-referenced land management data (Paustian 2012). A crowd sourcing iPhone app successful in collecting soils data across the UK, has now been expanded to the EU and is planned for introduction into Africa (Shelley et al 2013, Robinson 2013). Similar activities are being tested at FAO, as an extension of already successful crowdsourcing technologies used to collect price data (www.amis-outlook.org/). Combining farmer-provided activity data with new data sources on soils and climate may greatly improve models.

Agricultural emissions are tracked and monitored at different scales—national, landscape, production system and farm –for many purposes, including national GHG reporting to the UNFCCC, corporate sustainability, or income generation (e.g., voluntary market or international investment). Most approaches track single gases and some do not address system-level effects such as leakage or interactions among management practices.

National statistics are not consistent or compiled regularly and the last global emissions assessment was almost ten years ago (Tubiello et al 2013). FAO's new global GHG database provides a time series of country-level statistics from 1961 to present for agriculture, and 1990–2010 for land use change, based on FAOSTAT activity data and IPCC Tier 1 methodology. The database also provides emission projections for 2030 and 2050. Available to anyone with Internet access, the database can be used to update information on agricultural emissions, compare sources of emissions or look at trends– critical information for national, regional and global strategies to address climate change. Given that the FAOSTAT database can only be as accurate as its underlying activity data, countries have an added incentive to promote national coordination processes among relevant agencies in order to collect and report improved country-level activity data.

National reporting of emissions, including from agriculture, will be required for all countries starting in 2014, with biennial update reports. As agriculture is the major source of emissions in many developing countries, better quantification and reporting capacity is needed for tracking national emissions and managing viable mitigation responses. The IPCC guidelines are applicable to all countries, but inventories require knowing which data to collect, where to find it in censuses and surveys, and how to use modeling to extrapolate and target priorities for field data collection. Many low-income countries lack these capacities and suffer from high turnover in their inventory teams, making it difficult to maintain a consistent, repeatable system. One approach is to begin building robust Tier 1 inventories, using nationally available agriculture statistics reported to FAO. This process provides a valid mechanism to begin identifying data gaps, and can still be useful when developing qa/qc data analyses with higher Tier inventories (FAO 2014). At the same time, countries should be encouraged to develop higher tier inventories and move beyond national level scales. Dedicated software products can guide the inventory process and make them more consistent and manageable (Ogle et al 2013).

National inventories would be significantly improved with targeted investments from the international community to improve and consolidate information in two areas: (1) enhance national agency collaboration, for example between ministries of agriculture and environment to bridge fundamental institutional gaps towards the development of coherent national data systems; and (2) augment the IPCC Emissions Factor Database with more accurate country-specific or IPCC 'Tier 2' factors, especially for high emission activities (IPCC EFDB). Sharing innovations for producing Tier 2 factors will support these aims. For example, Canadian researchers track how emissions change with management on the farm by collecting at least one year of data from representative farms that reflect the diversity of farm management practices and regional climates nationally to develop country-specific emission factors. They also collect information directly from farmers on manure management practices that mitigate emissions, like improved storage of manure (VanderZaag et al 2013).

When moving to smaller scales, lack of local activity data and relevant emissions factors can reduce accuracy. Quantification at the landscape scale--smaller than a region, but larger than a few farms—can help aggregate changes in emissions or removals across diverse land uses and enhance flexibility in mitigation options. Little attention has been given to landscape scale accounting. While direct measurements using Eddy flux towers or chamber-based measurements seem appealing, they are costly and have limited feasibility in complex landscapes. Milne et al (2013) thus suggest using remote sensing to stratify the landscape and develop a nested sampling scheme, followed by targeted measurement to calibrate a model that can be used to estimate emissions.

Accounting at farm scales often serves multiple objectives that reflect farmers' needs for food security and livelihoods. Rosenstock et al (2013) are building a protocol for low-cost standard measurement of GHG emissions of smallholder agriculture that integrates emissions with production and cost information. Berry and Ryan (2013) use a simplified biogeochemical model–the Small-Holder Agriculture Monitoring and Baseline Assessment (SHAMBA) methodology—to reduce cost and effort, while accepting higher levels of uncertainty. The methods were developed for the Plan Vivo Standard, which values emissions generated as a co-benefit to development and hence does not require precise estimates (Plan Vivo 2012). Where higher uncertainty is tolerable, such lower cost models and methods could remove barriers to accounting and support broader participation.

Many calculators and tools have been developed for agricultural emissions. Colomb et al (2013) reviewed 18 calculators that were developed for raising farmers' awareness, informing management, reporting, determining carbon credits for markets, or assessing products. They operate at the farm, regional or supply chain scales. Many calculators did not fully account for land use change emissions (e.g., soil C loss with conversion to a new land use)—and those that did, only included direct land use change (no indirect calculated due to displaced activities). Some only had a partial accounting of nitrous oxide sources. Only five calculators included a measure of uncertainty. The authors suggest that given the variability in calculators, certifying calculators or assessors who use them would be needed to enhance quality assurance. More inter-tool and model comparison can also assist in guiding their use. Seebauer (2014) compared the 'Sustainable land management practices' (SALM) methodology from the Verified Carbon Standard (VCS) to the Cool Farm Tool, a simple spreadsheet, to account for emissions due to changes in residue burning, fertilizer use, mulching, composting, and tree planting in farms in Western Kenya. Despite differences, both tools predicted significant net GHG benefits of the applied practices across the farms and in most cases captured this across the variability of farms and practices.

Common units are needed for comparing outcomes in accounting. Global warming potentials (GWP), allow aggregating the impacts of gases in units of CO₂ equivalents, by using weighting factors (UNFCCC 2009a, b). However, outside of current UNFCCC agreements, alternative weighting approaches are being explored that might better represent the potential for temperature change. An alternative that weights methane as four times the value of CO₂ versus the conventional twenty five (Manning and Reisinger 2011) would significantly affect agricultural emissions calculations. Reisinger and Ledgard (2013) explored the implications of this weighting for New Zealand dairy farms and found that the significance of individual gases would change as would the total emissions, but relative effectiveness of different management strategies would remain unchanged.

If farmers' objectives are to achieve food security and/or optimal production of food under both current and changing climate conditions, can this be achieved in a way that also reduces emissions? Quantification approaches and models to assess the multiple objectives of food security, resilient agriculture, and GHG mitigation are still emerging.

Lobell et al (2013) explore how much GHG mitigation would result from investment in adapting agriculture to climate change. The enhanced crop productivity and thus reducing clearing of land for agriculture are assumed to reduce emissions from forest clearing. The authors estimate that a $225 billion USD investment in adaptation would result in 61 Mha less conversion globally, resulting in 15 Gt CO₂e less emissions by 2050, at costs comparable or less than direct investment in mitigation activities. The authors' results highlight the high uncertainty of current model estimates, with a need for improved estimates of emissions factors for converting land to agriculture, price elasticity of land supply, and elasticity of substitution between land and non-land inputs (e.g., fertilizer, labor).

Valin et al (2013) explore how much GHG mitigation would result if we focus on increasing crop yield and livestock feed efficiencies. They show that closing yield gaps by 50% for crops and 25% for livestock by 2050 would decrease agriculture and land use emissions by 8% overall or 12% per calorie produced. However, the pathway matters: a fertilizer intensive pathway results in lower GHG reductions; a focus on crop yield rather than livestock productivity would bring larger food benefits, whereas a focus on livestock would bring higher GHG reductions.

Understanding such inevitable trade-offs can guide policy. Hussein et al (2013) suggests that policies focused on forest carbon alone, when implemented in low-income countries can increase poverty and reduce food security of the local populations due to reduced agricultural expansion and increased local food prices.