Measuring Carbon Footprints of Agri‑Food Products

6. Farm level calculation tools

Abstract

Farm level calculation tools are the third building block for carbon footprints. This chapter discusses the current landscape of farm level calculation tools in terms of tools' scope, process, methods, alignment with international standards and best practice, and user friendliness and accessibility. The chapter also discusses the reliability of existing tools and how this could be improved.

The availability of science-based methods by itself is not sufficient to scale up carbon footprint measurement in food systems. For example, it may be scientifically feasible to use direct measurement methods for farm level emissions, but these methods are costly and difficult to implement and are therefore at the moment not suitable for widespread use. Similarly, some Tier 3 methods may require many input parameters or powerful computing resources, making them harder to use outside of a research context. For this reason, estimating emissions is often done through calculation tools (also referred to as emission accounting tools), which simplify the necessary data inputs and calculation methods to allow more widespread uptake.

The focus here is on farm level tools, in part because a significant share of total emissions in food systems occur on the farm. However, it is worth noting that simplified calculation tools may also be useful in other stages of the supply chain, e.g. to help small- and medium-sized enterprises (SMEs) along the food supply chain estimate their emissions. Even large firms (e.g. firms producing agro-chemical inputs) may use calculation tools to scale up and automate carbon footprint calculations.

Farm level tools play a crucial role in measuring and communicating carbon footprints of food products. As noted earlier, because the farm stage is responsible for a large share of the overall carbon footprint of food products, precise information is important. But carbon footprints can vary considerably from one farm to the next, and various management practices can reduce emissions. Ideally, emissions at the farm stage would therefore be calculated using primary data, including information on management practices. This would make it possible for firms downstream in the supply chain to identify and reward farmers with lower carbon footprints. Farm-level tools can also provide valuable information to farmers on how to reduce their emissions. By contrast, average data from LCA databases make it impossible to shift towards producers with below-average emissions, and remove producers’ incentives to lower their emissions (Richards, 2018[1]) (Deconinck, Jansen and Barisone, 2023[2]).

6.1. The current landscape of farm level calculation tools

Farm level calculation tools are widespread. The Greenhouse Gas Protocol, in developing its Land Sector and Removals Guidance, identified 19 farm level calculation tools, while a review conducted for the UK Department for Environment, Food and Rural Affairs (Defra) mentions the existence of 81 farm level calculation tools worldwide (RSK Adas, 2023[3]).

Not all tools are created equal, however. Tools differ on several dimensions:

• Scope

  • Reporting level: Some tools are designed to calculate product carbon footprints (e.g. the Cool Farm Tool). Most, however, focus on farm level emissions (although tools also often include embedded emissions from feed and fertiliser production). Some tools calculate “whole farm” emissions (i.e. total emissions grouping together all activities on the farm) while others calculate “enterprise” emissions (i.e. emissions of a specific activity on the farm). In some cases emissions are calculated for a user-defined area (i.e. specific fields indicated by the user as belonging to the farm).

  • Commodity/sector coverage: Tools differ with respect to their commodity or sector coverage: some cover more commodities than others, and some are specialised in a single commodity. For example, Eggbase is a dedicated tool for the egg laying and poultry sector.

  • Geographic coverage: Some tools are tailored to specific countries while others have global coverage. For example, the COMET-Farm tool was developed for the United States, while the Cool Farm Tool and FAO’s EX-ACT tool have a global coverage.

  • Inclusion of other sustainability criteria: Some tools include sustainability criteria other than GHG emissions. For example, the Cool Farm Tool also evaluates the water use and biodiversity performance of the farm. The French CAP’2ER tool covers GHG emissions, ammonia emissions, consumption of fossil fuels, water quality, water consumption, erosion, phosphorus consumption and consumption of phytosanitary products.

  • Inclusion of economic criteria: Some tools include economic information for decision support. In fact, some tools were originally created as economic decision support tools for farmers and subsequently added carbon footprint calculation capabilities. Other tools were created in the first instance as carbon footprint tools; these tend not to cover economic aspects.

• Process

  • Transparency: Some tools are more transparent about their assumptions and calculation methods than others. For example, the Holos tool provides an open-source version of its core algorithms on GitHub, an open-source software repository.

  • Governance: Some tools have an independent scientific advisory board or other mechanisms for quality assurance, while for other tools these governance and quality assurance aspects are less clear.

  • Updating: Some tools are regularly updated and clearly indicate their version number and changes made since the previous version, while others are updated irregularly or in less transparent ways.

• Methods

  • System boundaries: Some tools include only on-farm (Scope 1) emissions. Others include emissions from purchased inputs (e.g. electricity, feed, fertiliser), i.e. Scope 2 and upstream Scope 3 emissions. Downstream Scope 3 emissions are usually not included.

  • Emissions categories and accounting metrics: Not all emissions categories are covered in all tools. For example, direct land use change (dLUC) and changes in soil carbon are included in some but not all tools (e.g. included in Farm Carbon Calculator, COMET-Farm, Holos).

  • IPCC Tier methodology used: As noted earlier, IPCC distinguishes between three methodological tiers, in increasing order of complexity and accuracy. Some tools may be limited to Tier 1 or 2, whereas others use Tier 3 or a combination of methodologies (e.g. using Tier 3 for some processes but Tier 2 for others – as also happens in National Greenhouse Gas Inventories).

  • Allocation rules: For tools which calculate product carbon footprints, there is variation in the allocation methods used. Some use economic allocation while others use a hierarchy of methods.

• Alignment with international standards and best practice:

  • Alignment with reporting standards and guidelines: Some tools explicitly state that they are aligned with existing standards such as ISO 14067 or the GHG Protocol, but not all do. Several tools claim compliance with the older PAS 2050 product carbon footprint standard (which was last updated in 2011). Interestingly, this includes tools which do not actually calculate product carbon footprints but which use the other prescriptions of the PAS 2050 standard as guidance for farm level carbon footprint calculations.

  • Alignment with up-to-date scientific best practice: Some tools are aligned with the latest IPCC guidance and metrics, in particular the IPCC 2019 Guidance and the Global Warming Potential (GWP) values of the latest IPCC Assessment Report, while others are not. Similarly, some tools (such as the ECOGAN tool, developed by the Spanish Ministry of Agriculture, Food and Fisheries) use the same methods used in the National Inventory Reports of the countries in their geographic scope, while others do not.

• User friendliness and accessibility

  • Cost: Publicly available tools (e.g. Holos, COMET-Farm) are free, while others may have a free version in addition to one or more paid versions (which may come with additional support). For example, Cool Farm Tool and Farm Carbon Calculator both offer free versions for farmers but paid versions for other commercial users.

  • Languages: Some tools are available in multiple languages (notably the Cool Farm Tool, which is available in 13 languages).

  • Time requirements: Tools with a broader scope, more emission categories and higher-Tier methodologies tend to have higher data requirements and hence take longer to complete. Reported time requirements vary from 30 minutes to 160 minutes.

  • API: Some tools provide an Application Programming Interface (API) to allow users to upload data to the tool directly from other software applications.

  • Exporting results: Some tools allow users to easily export results (e.g. as Excel files).

  • Some other aspects of user friendliness could include whether tools report results in line with reporting requirements of international standards, or whether tools also provide suggestions to farmers on how they can reduce their emissions.

It is important to note that not all of these differences are problematic. Since farm level tools have many potential uses, tools can differentiate themselves by focusing on, for example, a specific scope (such as a specific commodity or geography), or by making a different trade-off between precision and simplicity. For users to make an informed choice requires that tools be clear about these aspects, and about their limitations.

It is out of the scope of this report to present a full review of existing farm level tools on each of the above dimensions. However, a closer look at some tools can help illustrate differences and similarities. For this purpose, six tools were selected from a list of resources provided by the GHG Protocol in the context of its draft Land Sector and Removals Guidance (inclusion on this list does not imply that these tools are necessarily endorsed by the GHG Protocol). Tools were chosen based on their prominence in the literature and to ensure geographic diversity. However, many other tools exist.1

Table 6.1 compares these six tools on a number of the criteria listed above.

Table 6.1. Key characteristics of selected farm level calculation tools

	Agrecalc	COMET-Farm	Cool Farm Tool	Farm Carbon Calculator (Farm Carbon Toolkit)	Holos	OverseerFM
First created	2007	2005	2011	2009	2004	2003
Reporting level	Whole farm; enterprise; product	Enterprise	Product	Whole farm; product	Whole farm	Whole farm; enterprise
Commodity/sector coverage	Multiple types of crops and livestock	Multiple types of crops and livestock	Multiple types of crops and livestock	Multiple types of crops and livestock	Multiple types of crops and livestock	Multiple types of crops and livestock
Geographic coverage	United Kingdom	United States	Global	United Kingdom	Canada	New Zealand
Inclusion of other sustainability criteria	No	No	Water (blue, green); Biodiversity (farm level only)	No	N losses	N and P losses
Full transparency of methods and/or code	No	Yes (based on USDA methods report and DayCent model)	Yes (methods)	Yes (methods)	Yes (complete algorithm available on GitHub)	Yes (methods)
Latest update	2023	2023	2024	2023	2024	2024
IPCC Tier methodology	1, 2	1, 2, 3	1, 2, 3	1, 2	2	2
Allocation rules	Economic	Not applicable	Economic (crops) Biophysical (dairy co-products)	Unclear	Not applicable	Not applicable
Alignment on international carbon footprint reporting standards and guidance	“Broadly aligned” with PAS 2050; ISO 14044; GHG Protocol Ag Guidance draft; FLAG SBTi; moving towards full alignment with ISO14064 and ISO14067 in 2024	No	Broadly aligned with major standards and IDF dairy standard	PAS 2050; considering alignment with other standards	No	No
Alignment on IPCC 2019 and most recent GWP	Yes	Yes (aligned on US National Inventory)	Yes	Yes	Yes	Unclear
Cost	Free option 3 paid options	Free	Free for farmers Paid options for other businesses	Free for farmers Paid options for other commercial users	Free	Paid annual subscription
Languages	English, French, Spanish	English (Spanish available for Comet Planner tool)	More than 13 languages including English	English	English, French	English
Time to complete	160 min (1)	Not available	150 min (1)	30 min – 120 min (2)	10-30 min (2)	Not available
API	Under development	Yes (cropland)	Yes	No	Yes	No
Export of results	No	JSON, XML	Paying members can provide code to farmer to share results	PDF, CSV, JSON	CSV	PDF

Note: (1) from Brake (2021[4]) (2) from tool provider.

Source: Analysis by the authors using publicly available information on each of the tools. Analysis refers to the version of each tool available in February 2024.

Table 6.2 provides additional detail on the system boundaries used by different tools, and the different emission categories covered. While tools again differ, there are some systematic patterns which can be discerned.

In terms of on-farm emissions, tools generally cover the main sources of emissions such as enteric fermentation, manure management, fertiliser application, and fuel combustion, with some exceptions. Most of the tools cover soil organic carbon sequestration. By contrast, residue management, feed loss and the use of organic fertiliser inputs is not included in most of the tools surveyed here.

Table 6.2. System boundaries of selected farm level calculation tools

	Agrecalc	COMET-Farm	Cool Farm Tool	Farm Carbon Calculator	FarmGAS	Holos	OverseerFM
On-farm emissions (Scope 1)
Residue management	Maybe	Yes	Yes	Maybe	Maybe	Yes	Maybe
On-farm feed production	Yes	Yes	Yes	Yes	Maybe	Yes	Yes
Liming	Maybe	Yes	Yes	Yes	Yes	Maybe	Yes
Synthetic fertiliser management	Yes	Yes	Yes	Yes	No	Yes	Yes
Manure management	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Organic fertiliser inputs	Yes	Yes	Yes	Yes	Maybe	Maybe	Maybe
Grazing (manure and urine)	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Enteric fermentation	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Bedding	Yes	Yes	Yes	Yes	Maybe	Yes	Maybe
Soil organic carbon sequestration	Yes	Yes	Yes	Yes	No	Yes	No
Tillage	Yes	Yes	Yes	Maybe	No	Yes	Maybe
Fuel combustion	Yes	Yes	Yes	Yes	No	Yes	Yes
Purchased energy (Scope 2)
Purchased fuels and electricity	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Purchased steam, heating and cooling for farm use	Yes	No	Yes	Yes	Maybe	Yes	Yes
Purchased inputs (upstream Scope 3)
Physical capital	No	No	Maybe	Yes	No	No	Maybe
Embedded livestock emissions	No	No	No	Yes	No	Maybe	Yes
Feed	Yes	No	Yes	Yes	Yes	Yes	Yes
Bedding	Yes	No	Yes	Yes	Maybe	No	Maybe
Fertiliser	Yes	No	Yes	Yes	Yes	Yes	Yes
Lime	Yes	No	Yes	Yes	Maybe	Maybe	Yes
Pesticides	Yes	No	Yes	Yes	Maybe	Yes	Yes
Seed and/or young plant material production	Maybe	No	Maybe	Maybe	Maybe	Maybe	Maybe
Services	Maybe	No	Maybe	Maybe	Maybe	Maybe	Maybe
Transport of farm inputs to farm	Maybe	No	Maybe	Maybe	Maybe	Maybe	Yes

Note: Following the similar approach used by the GHG Protocol Land Sector and Removals Guidance, “Yes” means “High confidence that the resource supports calculation of this metric or accounting category”, “No” means “High confidence that the resource does not support calculation of this metric or accounting category”, and “Maybe” means “Lack of explicit publicly available information on whether or not the resource supports calculation of this metric or accounting category”. The overview here refers to the version of each tool available in February 2024; the year of the latest update of each tool is provided in the previous table.

Source: Analysis by the authors using publicly available information on each of the tools.

Most tools cover emissions from purchased energy (Scope 2), but tools differ in their treatment of embedded emissions related to the production of purchased inputs (upstream Scope 3 emissions from the point of view of the farm). The COMET-Farm tool explicitly excludes all purchased inputs except energy, while other tools generally try to include embedded emissions in purchased feed and fertiliser. Tools also differ in their treatment of embedded emissions in physical capital and in purchased livestock. The Farm Carbon Calculator is most complete in this regard as it includes vehicles, machinery and agricultural buildings, as well as materials used on farm such as metal, wood and plastic. Cool Farm Tool is planning to include capital items in its calculations as well. It is less clear how existing tools account for some other purchased inputs such as seed, services, or transport of farm inputs to the farm.

It is worth noting that emissions from purchased energy (Scope 2) and purchased inputs (upstream Scope 3) are almost by definition not where farm level calculation tools have a comparative advantage. The main promise of farm level calculation tools is to use primary data (e.g. information on management practices) to create a more precise estimate of on-farm emissions than would be possible from secondary LCA databases. However, many tools try to provide a more complete picture by adding an estimate of Scope 2 and upstream Scope 3 emissions. This is typically done by multiplying a relevant measure (e.g. purchased synthetic nitrogen fertiliser) with emission factors from secondary LCA databases (e.g. average production emissions of synthetic nitrogen fertiliser). It would however be preferable if also for these emission sources more precise information were used, such as supplier-specific data on the carbon footprint of the exact fertiliser products purchased by the farmer. Emission factors could then be used whenever such supplier-specific information is not available. Some tools allow for this, but not all.

It is also worth noting that even where tools cover similar emission categories, they may differ in terms of their underlying methodology (e.g. Tier 1, 2, or 3 methods – or different kinds of Tier 3 methods).

Another way of looking at similarities and differences among tools is in terms of alignment with the accounting metrics listed in the draft Land Sector and Removals Guidance of the GHG Protocol. GHG Protocol itself has provided a preliminary assessment of which accounting metrics are covered by which tools. Table 6.3 summarises this information for the tools considered here. (See Chapter 4 on reporting standards and guidelines for an overview of which of the metrics are required and which are optional under the draft Guidance).

Tools are generally able to account for non-land emissions (e.g. combustion) and non-CO₂ emissions from land management (e.g. enteric fermentation). Some tools calculate CO₂ emissions and removals from land management and from land use change, while others do not (Under the draft Land Sector and Removals Guidance, reporting emissions in these categories is required while reporting removals is optional but subject to additional criteria). Tools similarly differ in their ability to report on the (optional) metric of gross biogenic land CO₂ emissions and removals. The largest gaps are found for the land tracking metrics, where it is unclear whether any of the tools covered here can report on at least one of the metrics (as required in the draft Guidance).

Table 6.3. Emission accounting metrics included in selected farm level calculation tools

	Agrecalc	COMET-Farm	Cool Farm Tool	Farm Carbon Calculator	FarmGAS	Holos	OverseerFM
Non-land emissions
On-site energy use GHG emissions	Maybe	No (separate tool available)	Yes	Yes	Maybe	Yes	Yes
Land management: non-CO2 emissions
Enteric fermentation CH4 emissions	Maybe	Yes	Yes	Yes	Yes	Yes	Yes
Manure management CH4 and N2O emissions	Maybe	Yes	Yes	Yes	Yes	Yes	Yes
Managed soils N2O emissions	Maybe	Yes	Yes	Yes	Yes	Yes	Yes
Biomass burning CH4 and N2O emissions	Maybe	Yes	No	No	Yes	Yes	Yes
Rice cultivation or flooded land CH4 emissions	Maybe	Yes	Yes	No	Maybe	No	Maybe
Land management: CO2 emissions and removals
Biomass carbon stocks	Maybe	Yes	Yes	Yes	Maybe	Yes	Yes
Dead organic matter carbon stocks	Maybe	Yes	Yes	Maybe	Maybe	Yes	Maybe
Soil carbon stocks	Maybe	Yes	Yes	Yes	Maybe	Yes	Maybe
Biomass carbon stock changes	Maybe	Yes	Yes	Maybe	Maybe	Yes	No
Dead organic matter carbon stock changes	Maybe	Yes	Yes	Maybe	Maybe	Yes	No
Soil carbon stock changes	Maybe	Yes	Yes	Yes	Maybe	Yes	No
Land use change: emissions and removals
Direct land use change emissions (dLUC)	Maybe	Yes	Maybe	Yes	No	Yes	No
Statistical land use change emissions (sLUC)	Maybe	No	No	No	No	Yes	No
Gross biogenic land CO2 emissions and removals
Gross biogenic land CO2 removals	Maybe	Yes	Maybe	No	Maybe	Maybe	No
Gross biogenic land CO2 emissions	Maybe	Yes	Maybe	No	Maybe	Maybe	No
Land tracking metrics
Indirect land use change emissions (iLUC)	Maybe	No	No	No	No	No	No
Land occupation (LO)	Maybe	No	Maybe	No	No	No	No
Carbon opportunity cost (COC)	Maybe	No	No	No	No	No	No

Note: Table shows indicative assessment by GHG Protocol for emission accounting categories listed in the draft Land Sector and Removals Guidance. Not all of the listed categories are required under the draft Guidance (see discussion in main text). “Yes” means “High confidence that the resource supports calculation of this metric or accounting category”, “No” means “High confidence that the resource does not support calculation of this metric or accounting category”, and “Maybe” means “Lack of explicit publicly available information on whether or not the resource supports calculation of this metric or accounting category”. Not showing three accounting categories with limited relevance for agriculture and food (product carbon stock changes, temporary product carbon storage, and gross biogenic product CO₂ emissions from end of life treatment) and one accounting category relevant only for bioenergy (gross biogenic product CO₂ emissions from combustion).

Source: GHG Protocol.

6.2. How reliable are existing farm level tools?

Given the differences between tools, it is unsurprising that tools can provide different emission estimates for the same farm, as several studies have found (RSK Adas, 2023[3]; Bonasia et al., 2022[5]; Brake, 2021[4]; Grain Growers, 2020[6]; Leinonen et al., 2019[7]; Lewis et al., 2012[8]; Richards, 2018[1]; Sykes et al., 2017[9]; Whittaker, McManus and Smith, 2013[10]).

Two studies for Australia illustrate this. Brake (2021[4]) compared four tools suitable for assessing mixed farming enterprises in Western Australia. The tools varied in their level of detail and generated different results, both at the level of the whole farm and by type of enterprise (crops and sheep). Compared to the average estimate, results varied from 70% below to 62% above the average estimate (or a ratio of 540% between the highest and lowest estimate). Similarly, a Grain Growers (2020[6]) study compared five tools suitable for cereal, pulse and oilseed production in Australia. Using data from two sample farms from opposite sides of Australia (New South Wales and Western Australia), the various tools again produced results ranging from 61% below to 67% above the average estimate (a ratio of 428%). The report concluded that there is a need for a more harmonised approach to emissions accounting.

A recent study commissioned by the UK Department for Environment, Food, and Rural Affairs (Defra) similarly found large variation in results from farm level calculation tools (RSK Adas, 2023[3]). The study used data for a set of 20 “typical” UK farms covering cereals, general cropping, horticulture, mixed farming, pigs, poultry, dairy, grazing livestock in less favoured areas, grazing livestock in lowlands, and two additional farms to test tools’ ability to represent anaerobic digestion and agroforestry (silvopasture) practices in dairy systems. Table 6.4 summarises the findings, showing the ratio between the highest and lowest emission estimate across the different tools tested. These ranges are generally smaller than the ones found by Brake (2021[4]) and Grain Growers (2020[6]) for Australia, but are still considerable.

Table 6.4. Variation in results of farm level calculation tools for emissions in UK agriculture

Model farm	Number of tools	Ratio of highest to lowest estimate
Cereals 1	5	175%
Cereals 1 with carbon stock change	5	220%
Cereals 2	4	128%
General cropping 1	5	171%
General cropping 1 with carbon stock change	5	1093%
Horticulture 1	3	157%
Horticulture 2	3	238%
Pigs 1	4	133%
Pigs 2	4	250%
Poultry 1	6	355%
Poultry 2	5	448%
Dairy 1	5	129%
Dairy 1 with carbon stock change	5	155%
Dairy 2	4	123%
Dairy 3	4	141%
Dairy 4	4	143%
Grazing in less favoured areas 1	4	196%
Grazing in less favoured areas 2	4	109%
Grazing in lowland 1	4	281%
Grazing in lowland 2	4	238%
Mixed farming 1	4	179%
Mixed farming 2	4	217%

Note: For farms where carbon stock changes are included, emissions are net emissions.

Source: RSK Adas (2023[3]).

In general, tools were most aligned for dairy, and least aligned for poultry, grazing in lowland, and cereals when carbon stock changes were included. In some cases, the variation is caused by a single outlier result, but in other cases there was not much agreement between the tools. In some cases, a single factor drove the variation (e.g. embedded emissions of animal feed in the poultry assessments) while in other cases a range of factors was responsible (e.g. for lowland grazing, calculators differed on emissions from enteric fermentation, manure management, nitrous oxide soil emissions, and embedded feed emissions). In some cases, tools agreed on the overall emissions but estimated different contributions of emissions sources. Tools also differed strongly on carbon stock changes. However, it was not the case that one tool systematically delivered higher or lower estimates than others.

Overall, the study identified divergence in the following emission sources:

• Carbon stock changes: Calculators showed a large variation in which aspects are included (e.g. land use change, emissions and removals from above ground biomass, emissions and removals from below ground carbon stocks) and how they were modelled.

• Crop residues: Calculators varied in their assumptions on the quantity produced and quantity remaining on the field.

• Enteric fermentation: Tools differed in how they accounted for livestock numbers and the amount and type of feed, leading to different results.

• Manure: Calculators varied in how they accounted for manure quantity and management practices.

• Embedded emissions from feed and fertilisers: Tools relied on emission factors from secondary data sources for these emissions, but the numbers used varied considerably. For feed, this was particularly true for land use change emissions from soy-based feed. For fertilisers, two calculators were using out-of-date emission factors.

The study also analysed the underlying reasons for the divergence in estimates for these emission sources:

• Calculators applied a range of different system boundaries in their tools. For example, should forested land be included as part of the farm business or not? Different default answers to these questions (or a lack of guidance to users on what should be included) led to different results.2

• Developers of calculation tools must balance precision with user friendliness, which means tools differed in terms of the amount of data they asked users to input (and correspondingly, where tools used assumptions rather than user-provided data).

• For purchased inputs, tools did not always use the best available emission factors from secondary data sources. The study notes that in the UK context, relevant datasets exist for energy (from the UK government), animal feed (from the Global Feed LCA Initiative), and fertilisers (from Fertilizers Europe); but not all tools relied on these sources. It is worth noting that some tools allowed using actual emission factors for feed or fertiliser provided by the manufacturers.

• Not all calculators were aligned on the IPCC 2019 guidelines; some were still relying on the 2006 edition. Even where tools follow the latest IPCC guidelines, tools differ in whether they are using Tier 1, 2, or 3 approaches.

• Tools did not take a consistent approach to carbon removals and emissions from land use change and land management. The study notes that some tools used the IPCC Tier 1 methodology for carbon stock change in mineral soils, even though this approach was not designed to assess effects at farm scale.

• Tools also differed in the extent to which they could account for the use of mitigation options such as nitrification inhibitors for nitrogen fertilisers.

While the assessment thus found major shortcomings to existing tools, the study also notes that several tools have gone through updates in the meantime, which may have reduced divergence. To improve the harmonisation of farm level calculation tools, the study recommends among other things that calculation tools should align with the latest standards and guidelines (e.g. ISO, GHG Protocol), the latest IPCC guidance, and the latest version of emission factor databases. Tools should be regularly reviewed and updated; they should present outputs in accordance with the latest standards; and there should be transparency and third-party verification of the alignment of calculators to minimum standards to build user confidence (RSK Adas, 2023[3]).

6.3. How reliable should tools be?

As noted earlier, a reliable estimate is one with a low systematic error (i.e. a low level of bias) and a low non-systematic error (i.e. a low level of random error).

Farm level tools can be used for various purposes, such as raising awareness, reporting emissions, evaluating projects, and making product claims (Colomb et al., 2012[11]). Not all of these purposes require the same degree of reliability.

For example, where farm level measurement is used by a farmer to track changes over time, potential errors might be less problematic: even if the estimated emissions are systematically above or below the true value, the tool might still provide a reasonable approximation of the changes over time.

Similarly, where farm level tools are used by a downstream firm to estimate the average carbon footprint of a large number of suppliers (e.g. with the goal of reporting its Scope 3 emissions), non-systematic error may be less problematic: with a large enough sample size, non-systematic errors would tend to average out.

However, where the aim is to compare different farms, or where decisions over awarding a contract or providing market access depend on measurement outcomes, a greater level of reliability is needed, both in terms of systematic and non-systematic errors. This is also true where tools are used to estimate the amount of emissions avoided or sequestered through a project or action, and especially where results of such an estimate are the basis for generating carbon credits or offsets.3 As these use cases are becoming more common, the level of reliability expected of farm level tools is increasing as well. It might therefore be useful for tool providers to explicitly adopt a philosophy of continuous improvement, whereby each new iteration of the tool aims to reduce systematic and non-systematic error. Governments could provide an enabling environment, e.g. by regularly engaging in benchmarking exercises such as the one conducted in the United Kingdom (RSK Adas, 2023[3]). These could form the starting point for constructive engagement with tool providers to identify areas for improvement.

The issue of comparability is particularly important where different agro-ecological conditions are concerned. Some methods may have a wide applicability but are relatively coarse. This is the case for IPCC Tier 1 methodologies: the default values are designed to be systematically neither above nor below the true value – but there may be a significant difference between the true value and the estimated value on a particular farm. By contrast, other methodologies may be highly precise for a specific context but less reliable outside of it. This may be the case for certain Tier 3 methodologies: a complex model may have been designed and validated for a specific agro-ecological context, where it may have low systematic and non-systematic error – but outside of this context, the model may not be appropriate. In other words, there is a potential trade-off between reliability and geographic coverage. It might be useful for tools to clearly indicate the geographies (or agro-ecological conditions) for which they work best.

A second trade-off was hinted at earlier: more sophisticated Tier 3 methods may yield more reliable answers than Tier 1 or Tier 2 methods but may require more data from the user. For example, tools to model soil carbon dynamics may require users to provide information on land use for the past ten or twenty years. Not all farmers will have such detailed information readily available, and even if they did, inputting information into the tool can be a time-consuming process (and runs the risk of data entry errors). A tool can always be simplified by substituting assumptions and default values for user-provided data, but this will reduce its reliability. One possible path forward is to explore ways to make data entry as easy as possible for farmers, for example by pre-populating a tool using data from administrative sources or farm management systems, or by providing assistance (e.g. through extension services).

Still, it is likely that estimates for an individual farm will have a considerable random (non-systematic) error. This raises the question of whether tools should report this uncertainty (e.g. by displaying confidence intervals). On the one hand, doing so could create more transparency about the level of precision of estimates. This could be especially relevant in comparing different tools or methods. On the other hand, there may be a pragmatic argument against providing uncertainty ranges to users. For example, the exact emission reduction from using a new feed additive may be highly uncertain on an individual farm because of the intrinsic variability of biological systems. At the level of a region or an entire country, however, it may be clear that the new feed additive is highly likely to reduce emissions (as random errors are “averaged out” across large numbers of farms). If farmers are presented with large uncertainty ranges around the expected emission reduction from the new feed additive, uptake may remain low. It might therefore be more opportune to present users with the average effect only, to provide the right incentives to reduce emissions for the sector as a whole. This would also have the benefit of not penalising farmers for factors which are outside of their control. However, more research is needed on this question.

6.4. A first assessment

Farm level tools could play a crucial role in creating reliable and widespread measurement of carbon footprints, as well as in driving emission reductions. Several tools exist, but as the discussion in this chapter has shown, at present the tools can give widely divergent answers for the same farm. This is due among other factors to differences in system boundaries and emission categories covered; differences in calculation methods; and differences in emission factors used for purchased inputs. Some tools are aligned on relevant standards and are transparent about their methods, but not all. It is currently difficult for users to understand the relative strengths and weaknesses of the available tools and to make an informed choice. However, there appear to be several options to remedy this situation.

To realise the potential of farm level data, more must be done to improve the reliability of existing tools, and to create greater transparency about their performance. A combination of minimum requirements and benchmarking exercises could help. For example, it appears that many differences between the tools would be resolved if tools were all aligned on the latest international standards (notably ISO and GHG Protocol), followed the latest scientific guidance (notably the IPCC 2019 guidelines), and were updated regularly to ensure emission factors for purchased inputs were taken from up-to-date sources. In parallel, benchmarking exercises (where data for the same farm is submitted to different tools to compare estimates) could be useful to help tool developers identify areas for improvement and to inform users.

In addition to the above, two important weaknesses in the current landscape of farm level tools concern geographic coverage and the treatment of soil organic carbon. Most tools have been developed for use in high-income economies. While some tools provide global coverage, in general there are fewer Tier 3 methods available for developing countries which limits the possibilities for developing farm level tools adapted to those contexts.

Many farm level tools calculate soil organic carbon stocks and/or changes. However, as the UK review showed, tools differ widely in how they model this. Some tools use the IPCC Tier 1 method, which was developed for national reporting and may be unreliable for farm level calculations. Other tools are more sophisticated. The Holos and COMET-Farm tools use the precise coordinates of a user’s fields to link to detailed soil maps and weather data, which are then fed into a simulation model. But as noted in Chapter 5 on Science-based methods, properly validated simulation models are not available for all contexts (Garsia et al., 2023[12]).

Finally, most of the existing farm level tools were not originally designed for calculating product carbon footprints. Moving from a farm level assessment to a product-level assessment requires at least two changes. First, since a product carbon footprint takes a life cycle perspective, emissions from other stages need to be added to the farm level emissions. For a cradle-to-farm gate assessment, this means including emissions from purchased energy (Scope 2) and purchased inputs (upstream Scope 3). Second, these emissions must be expressed on a per-product basis. For farms, this will typically require allocating emissions across multiple outputs. Both steps could be done inside a farm level tool, or could be done in a subsequent step. As noted, several tools draw on emission factor databases to include an estimate of emissions from purchased energy and inputs, and some tools (e.g. the Cool Farm Tool) allocate emissions to calculate product carbon footprints. When these steps are embedded in a farm level tool, it is important to ensure that up-to-date emission factor databases are used, and that allocation rules are consistent with reporting standards and guidelines. Moreover, tools should ideally allow using supplier-specific information (from fertiliser firms, feed manufacturers, etc.) to be used in calculations, with emission factors as a fallback option when supplier-specific information is not available.

References

[5] Bonasia, C. et al. (2022), Whole Farm Net Zero: Approaches to quantification of climate regulation ecosystem services at the whole farm scale..

[4] Brake, R. (2021), Evaluation of carbon accounting tools available to mixed farming enterprises in Western Australia (Report for the Department of Primary Industries and Regional Development), Richard Brake Consulting Pty Ltd.

[11] Colomb, V. et al. (2012), Review of GHG Calculators in Agriculture and Forestry Sectors. A Guideline for Appropriate Choice and Use of Landscape Based Tools, ADEME, IRD and FAO.

[2] Deconinck, K., M. Jansen and C. Barisone (2023), “Fast and furious: the rise of environmental impact reporting in food systems”, European Review of Agricultural Economics, Vol. 50/4, pp. 1310-1337, https://doi.org/10.1093/erae/jbad018.

[12] Garsia, A. et al. (2023), “The challenge of selecting an appropriate soil organic carbon simulation model: A comprehensive global review and validation assessment”, Global Change Biology, Vol. 29/20, pp. 5760-5774, https://doi.org/10.1111/gcb.16896.

[13] GHG Protocol (2005), The GHG Protocol for Project Accounting, https://ghgprotocol.org/sites/default/files/standards/ghg_project_accounting.pdf.

[6] Grain Growers (2020), Carbon Calculators: Compared for Australian Grain Growers., Grain Growers and Kondinin Group.

[7] Leinonen, I. et al. (2019), Comparative analysis of farm-based carbon audits.

[8] Lewis, K. et al. (2012), “Carbon accounting tools: are they fit for purpose in the context of arable cropping?”, International Journal of Agricultural Sustainability, Vol. 11/2, pp. 159-175, https://doi.org/10.1080/14735903.2012.719105.

[1] Richards, M. (2018), Measure The Chain: Tools For Assessing GHG Emissions In Agricultural Supply Chains, Ceres and CGIAR Research Program on Climate Change, Agriculture and Food Security, https://hdl.handle.net/10568/98361.

[3] RSK Adas (2023), Harmonisation of Carbon Accounting Tools for Agriculture, UK Department for Environment, Food, and Rural Affairs, https://sciencesearch.defra.gov.uk/ProjectDetails?ProjectId=20967%0A.

[9] Sykes, A. et al. (2017), “A comparison of farm-level greenhouse gas calculators in their application on beef production systems”, Journal of Cleaner Production, Vol. 164, pp. 398-409, https://doi.org/10.1016/j.jclepro.2017.06.197.

[10] Whittaker, C., M. McManus and P. Smith (2013), “A comparison of carbon accounting tools for arable crops in the United Kingdom”, Environmental Modelling & Software, Vol. 46, pp. 228-239, https://doi.org/10.1016/j.envsoft.2013.03.015.

Notes

← 1.  One prominent tool, FAO’s EX-ACT tool, is not included here as it was mainly designed for evaluating impacts of projects rather than calculating attributional carbon footprints.

← 2.  The answer to this particular question is no, from the point of view of product carbon footprint standards, if the forest is not part of the process for producing the agricultural product. For firm level reporting, the answer is more complicated; see the discussion of the Draft Land Sector and Removals Guidance above.

← 3.  Such “project accounting”, which underlies carbon credits and carbon offset schemes, is different from the attributional carbon footprint accounting view which is the focus in this paper. Rather than asking which activities and products account for existing levels of emissions, a project-based approach compares emissions against a counterfactual (e.g. to calculate by how much an initiative has reduced emissions relative to a baseline). See GHG Protocol (2005[13]) for a discussion of project accounting. However, many of the issues regarding measurement and communication are common across the approaches.