These practices, which largely overlap with “conservation agriculture” practices, are alternatively referred to as “regenerative agricultural”, “climate-smart” or “carbon farming” practices. They include but are not limited to: no-till, cover cropping, precision nitrogen fertilizer management, biochar and compost application, enhanced mineral weathering, new crop rotations, agroforestry, controlled drainage and some edge-of-field practices . The urgency in combating climate change and achieving sustainable development has spurred climate-pledges by individual companies to cut their carbon footprints and stimulate the growth of agricultural carbon markets to incentivize farmers to adopt these practices . Accurate quantification of carbon emissions and removal resulting from adopting various practices is the basis for carbon insetting and offsetting programs related to agriculture. However, the existing scientific literature is not yet conclusive as to where, when, if, and by how much these practices might lead to genuine GHG reduction or carbon removal . While some may debate the effectiveness of these practices for GHG reduction and carbon removal, various public and private sector initiatives are driving substantial investment in policy and incentivization programs to motivate agricultural carbon outcomes, driven by strong political, investor, corporate, and consumer pushes in the European Union, the U.S., China, and other nations . It is thus more urgent than ever for the scientific community to develop robust and scalable strategies for the credible quantification of agricultural carbon outcomes.

These estimates will form the basis for assessment of the climate mitigation potential of these practices, and guide investment in incentivization tools, and perhaps more importantly,plastic pots 30 liters to ensure the market rewards mitigation actions fairly and accurately. Here, we propose that field-level quantification of agricultural carbon outcomes is not only fundamental to a trustworthy, transparent, and cost-effective agricultural carbon market, but also critical to any other sustainability-oriented program for ecosystem services. The existing literature has illuminated the scientific and technical issues and challenges related to the rigor of these assessments of carbon outcomes in agricultural land , but actionable road maps and pathways to quantify field-level carbon outcomes are scarce. From the scientific perspective, existing approaches, such as direct measurement , emission factors, and process based modeling, face fundamental challenges that prohibit them from achieving the accuracy, scalability, and cost-effectiveness demanded by both public and private sectors of the society . Given the growing demand for solutions to the climate crisis, the market is eager to rely upon existing and/or outdated quantification methods for rapid deployment without sufficiently considering their accuracy or scalability. This poses a major risk for large-scale public and private investment in market based emission reduction and carbon sequestration strategies in the agricultural sector such as food/beverage supply-chain intervention, carbon intensity of bio-energy feed stock, climate-smart commodity certification, carbon crediting, and carbon markets – the credibility of these market-based emission reduction instruments, and the quantification of their outcomes is foundational to their success. Thus there is an urgent need to develop the right scientific tools for quantifying carbon outcomes in working lands in order to minimize the risks of large-scale public and private investment in initiatives that do not provide actual climate benefits.

In this regard, we provide a framework for scalably quantifying field level agricultural carbon outcomes that addresses many of the issues and uncertainty associated with the status quo approaches. Specifically, we first discuss the criteria for a successful quantification solution , then propose a new framework to scalably quantify field-level agroecosystem outcomes , and lay out the underlying disciplinary foundation , followed by identifying the scientific challenges in existing approaches . We then present a “System-of-Systems” solution for achieving the field-level quantification of agricultural carbon outcome in an accurate, cost-effective and truly scalable way . Finally, in Section 4 we propose an R&D agenda that can substantiate not only agricultural carbon markets but also sustainable indicators for agroecosystem management. Effective carbon quantification technology applied at the field level must be accurate, scalable, and cost-effective. “Field-level accuracy” is needed if individual farmland’s carbon outcomes may be monetized in the carbon market; it is also required for traceability of any aggregated carbon outcomes in supply-chain quantification . “Scalable” here means that the quantification solution must have an independently verified uncertainty measure across all possible locations; in other words, showing that a solution works well at a few demonstration sites, as many existing Measurement-Reporting-andVerification efforts do, is not enough. Instead, true “scalability” means one method must demonstrate an acceptable accuracy of the solution at randomly selected real-world sites. Another benefit of “scalability” is the potential to map the benefits of different possible practice interventions across the landscape, so investments can be prioritized in places they are most likely to succeed. Though some practitioners argue that aggregated-level accuracy is sufficient because most market-based emission reduction mechanisms nowadays only require carbon outcomes quantified at the aggregated level, we argue that aggregated-level accuracy, which is almost impossible to validate, must come from field-level accuracy.

Finally, for any technology, there is a trade off between cost and accuracy, and the desired solution should be sufficiently cost-effective to achieve the needed accuracy . Based on the above framework and disciplinary foundations, we can identify shortcomings of existing carbon outcome quantification methods, including: direct measurements, such as soil sampling for SOC change , and eddy-covariance technique to measure GHG emissions ; emission factor estimation, in which a fixed linear factor is used to approximate “carbon outcomes” based on different management practices ; and process-based modeling . Direct measurements have long been the primary tool for quantifying carbon outcomes and have significantly advanced our understanding of carbon cycling in the agroecosystems, although they are in general cost-prohibitive and thus not scalable. Specifically, direct measurements, such as using soil sampling to measure changes in SOC storage or using eddy-covariance flux towers to measure carbon fluxes , have been widely used to quantify agroecosystem GHG flux and soil carbon dynamics at site levels . However, it is impractical to collect direct measurements for every field due to the high financial and labor costs. While direct measurements of SOC through soil sampling and labbased soil tests have been widely perceived as the most trustable measurements to verify soil carbon outcomes, soil sampling has inherent limitations. Conventional measurement of SOC stocks requires quantification of SOC concentrations and bulk density. Soil organic C concentration is measured by dry combustion, in which C is converted to CO2 for quantification by gas chromatography. Small sample sizes can challenge the accuracy of measurements if soils are not properly homogenized. Measuring bulk density straddles field and lab requires foresight to sample soils of a defined volume in the field with subsequent lab-based measurement of water-free soil mass to calculate density. Soil bulk density measurement can be complicated by needing to account for non-soil components in samples, adjusting for >2 mm size particles , avoiding compaction of soil during sampling,round plastic pots difficulty in comparing across bulk density measurement methods, and acquiring samples at subsurface depths. An ex-situ technology that has matured in the past decade is laboratory optical and mid-infrared spectroscopy, which has significantly reduced the cost of quantifying SOC concentrations as well as labor-intensive SOC fractions and has been promoted by Global Soil Partnership of the UN Food & Agriculture Organisation . Emerging technologies such as in-situ spectroscopy or geophysical measurements to estimate soil C can reduce sampling cost, but their accuracy is significantly lower than classical laboratory tests. The fact that spatial variation within any given field can be larger than year-to-year changes in SOC contributes substantial uncertainty inherent in direct measurement of SOC stock . As a result, soil sampling is infeasible as a short-term quantification method but is well positioned to set the baseline or periodic verification after 5+ years of practice changes . While not a direct measurement, satellite or other remote sensing techniques have shown potential to monitor SOC , but deploying these techniques for real-field SOC monitoring remain challenging.

This is because: remote sensing only detects soil carbon and associated soil properties at the soil surface, not the the soil profile to full depth ; and crop residues, green vegetation cover, and soil moisture have a large confounding impact on spectral signals, thus making the estimation of bare soil surface carbon concentration difficult in practice . Moreover, direct measurement can not simultaneously measure changes in SOC or GHG fluxes under a practice change versus a counterfactual business-as-usual scenario, but both are needed for estimating their induced “carbon credits” by definition . Direct measurements may be useful when paired experiments are properly implemented in the same field – an approach which has not historically been adopted by market systems. Using the cover crop adoption as an example , the “additionality” criterion for carbon credits requires us to know the SOC stock in the two scenarios, one with newly adopted cover cropping in which SOC stock can be directly measured, and the counterfactual scenario for “business-as-usual” in which SOC stock can no longer be measured directly, but can only be estimated through modeling. Because soil sampling cannot measure ΔSOC that involves a hypothetical “business-as-usual” scenario, the standard soil sampling methods for assessing carbon credits are actually not able to directly quantify the realized carbon benefits . This issue also applies to other direct measurements , as the “carbon credit” quantification always requires counterfactual scenarios for calculating the difference, and agricultural practice inevitably has such a challenge unless farmers are willing to carve out part of their field for two different practices to create the counterfactual scenarios. Emission factor methods are the most widely used approaches in past IPCC reports and also the easiest method to use. While useful for large-scale carbon emission accounting, they suffer from the inability to capture spatial and temporal heterogeneity of E and C and cannot comprehensively track the dynamics embedded in the interactions between E, M and C. The assumption of the same emission or sequestration outcomes based on a particular “action” across different fields is not only inaccurate, but may also disincentivize farmers from participating in a carbon market. Emission factor methods also assume constant crop conditions , while interannual/decadal variability in crop and carbon budget could be significant and not captured. Emission factor methods thus can be hardly used for field-level carbon outcome quantification. For some recent efforts of applying process-based modeling to generate emission factors for more granular spatial and temporal scales , we treat that approach as “process-based modeling” in the next section. Process-based modeling has been regarded as the most mechanistic method to quantify carbon outcomes. Since process-based models can simulate “business-as-usual” scenarios and other counterfactual scenarios, this approach arguably addresses the counterfactual issues of the direct measurement approach laid out above and can allow direct calculation of the actual carbon benefit. Although there has been an increase in the use of process-based modeling as the main approaches to quantify agricultural carbon outcomes , existing modeling approaches have various critical gaps to address, especially related to the absence of necessary processes and the lack of constraints to reduce uncertainties in model parameters. As to the latter point, few existing process-based models include observational constraints, especially when applied to locations beyond calibration/validation sites. The performance of process-based models is ultimately determined by two groups of parameters, i.e. process-specific and location-specific parameters. Process-specific parameters usually do not vary over space and time , therefore can be obtained through calibration and validation based on extensive lab or field experiment data. In contrast, location-specific parameters vary at different locations. For example, photosynthetic capacity is a variable that is spatially and temporally variant with a key control on the photosynthesis process, unfortunately such a major carbon-related process is missing in most process-based models currently used for agricultural carbon quantification. For the limited number of models that include the photosynthetic process explicitly, they are still using crop-specific or even plant-functional-type-specific values of photosynthetic capacity without considering the variabity of photosynthetic capacity in space and time , which can lead to 21% error in estimating photosynthesis . More broadly speaking, location-specific parameters also include local information of model inputs , boundary conditions, and management practices at the field level, without which the field-level accuracy is impossible to achieve.