Farm carbon calculators – why do they give such different results?

CLA Land Use Policy Adviser Matthew Doran provides an in-depth analysis of carbon accounting for agriculture and the variations to consider between carbon calculators
Farmer ploughing the field with sunset

Carbon calculators can help farm businesses estimate their greenhouse gas footprint and identify ways to reduce it. They often highlight inefficiencies in production and possible financial savings. Increasingly, supply chains are asking their producers to complete a carbon assessment as a mandatory condition of purchase.

However, a big frustration for users of farm carbon calculators has been the variation in outputs from different calculators for the same farm. This variation is not such an issue if you use the same calculator to identify actions track your progress, but it can make it hard to initially choose a calculator or trust its output. Users who make net zero claims from farm carbon calculators need to be careful of potential liability for greenwashing, given this variation.

The carbon accounting for agriculture report

In 2022, Defra contracted environmental consultancy firm ADAS to research why calculators give such different results, and recently, the report ‘Harmonisation of Carbon Accounting Tools for Agriculture’ was published. ADAS ran data from 20 model farms through five different calculators at the end of May 2023 (Agrecalc Ltd’s Agrecalc; The Cool Farm Alliance’s Cool Farm Tool v2.0; Farm Carbon Toolkit’s Farm Carbon Calculator; Trinity Agtech’s Natural Capital Navigator, Sandy v4.0; and Solagro’s The Farm Carbon Calculator v3.1) – plus a sixth calculator for poultry enterprises, Eggbase. The 20 model farms covered 9 farm types with each farm type having an intensive and less-intensive scenario for comparison. ADAS also tested two extra dairy scenarios to examine functionality for agroforestry, anaerobic digestion, and methane-suppressing feed additives.

The report confirms what farmers have long known: there are staggering differences in outputs between calculators. Inconsistencies come from use of different emission factors, different levels of detail in user-entered data, treatment of soil carbon, boundaries of what is included and what is not, apportionment of emissions that are not annual, and omission of some processes.

From the perspective of users, the key takeaway from the report is that farmers and advisors should choose the calculator most appropriate to their needs, rather than worrying about its accuracy, which cannot actually be known. This means choosing the calculator that (1) matches the level of detail on data which the user has access to, (2) matches the task to which the end output will be put, and (3) includes the ability to test ‘what if’ scenarios to help prioritise investments to reduce emissions. Some calculators also come with information and advice about emissions reductions, another reason to choose a calculator over another. The CLA guidance note ‘Carbon Accounting for Landowners’ provides information to assist in choosing the most appropriate calculator for your needs.

In other words, carbon calculators are best viewed as management tools to help farmers reduce emissions rather than definitive assessments of emissions and carbon sequestration. The ADAS report emphasises that simpler, more generic calculators exist for a reason: they are a starting point which allow farmers to take action without the barriers of completing a more complex carbon calculator. Reducing emissions is invariably much more important than the quality of the figures.

Main results from the report

The biggest variation between the calculators was in free-range poultry, lowland grazing and crop production on peat soils. In the free-range poultry scenario, there was a 4.5-fold difference between the highest and lowest calculator outputs despite beginning with the same dataset. This was linked to whether emissions from land-use change to produce soya in poultry feed were included in the calculator. For intensive and extensive lowland grazing, there was a 2.8-fold and 2.4-fold difference respectively between the highest and lowest calculator outputs. In the cereal crop scenario on cultivated peat, there was a nearly 11-fold difference between the highest and lowest calculator outputs, because two calculators ignored emissions from cultivated peat.

There was more similarity on other farm types, namely all the dairy farms, intensive pigs, organic and regenerative cereal cropping, and extensive upland grazing. For the latter scenario, all models were within 9% of each other. Nevertheless, when the researchers broke down the emission profiles for these sectors into separate sources, they found divergent values. This suggests that calculators arrived at similar values out of chance rather than because they are harmonised.

No calculator consistently output the highest or lowest values. Overall, it seems that different treatments of a single factor in each scenario drove most of the variation. This provides hope for more harmonised calculators if government or accredited standards can set clear protocols for what processes and level of detail must be included.

How do farm carbon calculators work?

Before examining the reasons the researchers found for why calculators vary, it is worth taking some time to understand how they work and what their purpose is.

With the rare exceptions of scientific research sites, no farms have equipment to measure the true concentrations of greenhouse gases being emitted from soils and livestock. Nobody knows what these values are on a given farm. Therefore, farmers need a model – a carbon calculator – which can estimate greenhouse gas emissions from a relatively small amount of user data. Carbon calculators simplify the farm’s complex biological system into a set of linked equations that capture its most important processes, which are then tailored to a given farm through user-specified variables.

The basic equation underlying carbon calculators is:

activity x emission factor = emissions from that activity

An activity is a process on the farm, such as the amount of fertiliser applied per hectare. An emission factor is an average value for the amount of greenhouse gases emitted or removed from the atmosphere per unit of that activity. Emission factors are derived from measurements and estimates made elsewhere. They are subject to frequent revision as scientists and industry publish new research and/or change their products.

To model more complex processes and increase correspondence to real farms, calculators contain equations to link several pieces of input data before multiplying the product by its relevant emission factor. Any processes which are not included in the calculator’s design, or which users leave blank, are either ignored or plugged with major, generic assumptions.

Farm carbon calculators involve a delicate balancing act. They must ask users for sufficient data to increase their correspondence to reality, but not so much that users cannot complete them – whether for lack of data, lack of time, or lack of technical expertise. The simpler the requests on the user, the more assumptions the calculator must make, and the more generic that calculator’s output.

That’s not to say that simple calculators are ‘bad’ per se. They allow farmers to estimate their emissions even if they have limited data. Furthermore, given that nobody knows the true emissions from a farm without extremely expensive equipment, there is no practical way to tell whether a more complex calculator actually gives a more accurate output. It is likely that the more detailed the calculator, the more tailored its output will be to the farm in question, but this is not assured.

For more information about how carbon calculators work, and advice on how to choose and complete one, please see CLA guidance note ‘Carbon Accounting for Landowners’.

Why was there so much divergence between calculators?

With these basics covered, we can explore the reasons that the researchers identified for divergence.

Firstly, calculators use different emission factors – the average amount of greenhouse gases emitted during a given process. This naturally leads to different outputs. For instance, two calculators (at the time of research) were using outdated emission factors for fertilisers. Similarly, even if the emission factors are consistent, different assumptions used to fill in missing or unavailable data, such as the volume of crop residues left in fields, leads to divergence.

Secondly, some calculators specify their emission factors with greater precision. The Intergovernmental Panel on Climate Change (IPCC) categorises the specificity of emission factors according to their ‘Tier’. Tier I are stock emission factors with the same value everywhere in the world (acceptable when no country-specific data is available). Tier II emission factors are country-specific or better. Tier III methods improve on Tier II by calculating bespoke emission factors for the situation in question through equations which use a process-based model to link various data inputs.

The granularity of data entry, i.e., how much detail users can specify, determines the Tier which a calculator can use to estimate emissions for a given process. Differences here lead to major differences in final output. A good example is nitrous oxide emissions from soils following fertiliser applications: calculators range from Tier I (using emission factors derived from soils in dry climates) to Tier III (bespoke according to climate, soil type and fertiliser applications). Enteric fermentation is an important area where different levels of detail on livestock age, sex, weight, and feed characteristics cause divergence between calculators.

Thirdly, calculators handle sequestration in soils beneath crops differently. On mineral soils, some calculators simply apply the IPCC Tier I emission factor for carbon flux, which “does not take into account details of soil type, existing soil organic carbon or have any sensitivity to different ways soil management practices are implemented”, according to the report. Other calculators employ Tier II or better methods, using user-measured values for soil organic carbon content and other contextual factors such as climate. Oxidation of peat soils was ignored by two calculators at the time of the research.

A fourth source of inconsistency is that calculators set different boundaries for their emissions: i.e., they exclude different things. No clear protocols exist as to what carbon sequestration on a landholding should be included in a farm-level assessment. One farm carbon calculator is solely product-level and does not have functionality to include sequestration in farm hedges, trees, and land entered into agri-environmental schemes. Other calculators include this functionality – and the complexity with which they account for woodland etc. differs from Tier I to Tier III.

In addition, calculators do not share a consistent way of apportioning emissions which do not fit into neat annual cycles, including liming, rotational manuring, perennial crop processes, and non-annual livestock lifecycles.

Finally, some calculators omit whole processes which others include. This isn’t a separate category, but an extension of the factors above because omission is driven by the level of detail in user inputs, different boundaries, and different emission factors. For example, some calculators omit:

  • Nitrous oxide emission from the breakdown of crop residues.
  • Emissions embedded in capital items, materials, and medicines which the farm purchases.
  • Land-use change in embedded emissions of bought-in livestock feed (a particular problem for soya-based feeds).
  • Enteric emissions from monogastric livestock.
  • Technologies like methane-supressing feed additives, nitrification inhibitors, and urease inhibitors.
  • Agroforestry.

Indeed, the researchers found that all the calculators miss some important processes, particularly the specifics of manure management (e.g., amount exported from the farm; separation into different lagoons; acidifying and cooling slurry; ammonia scrubbing of ventilation gases etc).


The preceding discussion might make it sound like farm carbon calculators are not worth completing. However, this is not the conclusion that the report draws. There are, it argues, some relatively quick fixes to the divergence that calculator providers, accreditation bodies, and government could implement:

  • Emission factors for emissions embedded in bought-in energy, fuel, feed, and fertilisers could be standardised in databases.
  • Standard guidance for setting boundaries and assessing carbon removals could be issued to improve consistency between calculators.
  • Certain functionality (e.g., emissions from peat soils) could be made mandatory.
  • Calculators could be required to add more granularity in areas with high emissions, such as enteric methane emissions.

Fundamentally, all the calculators are models; they are partial representations of reality, each with a different intended user. Even with greater efforts to harmonise calculators, differences will remain due to complementary ways of modelling complex systems, driven by different end user requirements.

The CLA’s advice is to choose the calculator which best meets your (and/or your supply chain’s) needs given the data you have available, then stick to this one in subsequent audits. This way, you will be able to make a reasonable comparison over time to track improvements.