What is uncertainty in carbon emissions measurement?
The carbon footprint: an inherently inaccurate estimate
The various existing methodologies for calculating carbon footprints were developed to enable organizations to estimate their greenhouse gas emissions on the basis of their activity data: energy consumption, quantity of materials or products purchased, kilometers traveled, etc. These data are then used to calculate their carbon footprint.
While there are devices and techniques for measuring greenhouse gas emissions directly in the air, this approach is clearly not viable for measuring the emissions induced by the activities of even the smallest companies, quickly and efficiently. In order to encourage organizations to reduce their emissions, it was therefore necessary to develop calculation standards that would give them "simple" access to an estimated value of their gas emissions. This is how the various greenhouse gas assessment methodologies came into being.
What is commonly referred to as a company's "carbon footprint" is therefore the result of various data collections, themselves measured, and operations (multiplications, sums, extrapolations, etc.) performed on these data. By its very nature, the calculation of a carbon footprint is an estimate, and therefore involves a degree of uncertainty.
It's important to remember that the aim of drawing up a company's carbon footprint is not to compare the results obtained with those of other players, which would give rise to endless debates on calculation methods, but rather to help identify decarbonization levers and repeat the exercise over time, in order to observe the relevance and effectiveness of the decarbonization efforts made by the organization.
Reminder: how is a carbon footprint calculated?
In simple terms, calculating a company's carbon footprint means adding up the emissions induced by its activity. The emissions of each induced activity are obtained by multiplying a measured physical activity datum ( KWh of energy consumed, euros spent, kilos of materials purchased, quantities of products bought and sold, m2 of building space or land occupied, etc.) by an emission factor enabling conversion into a single reference unit, the "CO2 equivalent" (CO2e).
Carbon footprint calculation methods are therefore based on equivalences established between a known quantity of resources consumed by the company and an equivalent quantity of carbon dioxide released into the atmosphere for the production of these resources or activities. For example, we generally consider that the consumption of 1 kg of tomatoes corresponds to 0.624 kgCO2e emitted (during agricultural processes, transport, distribution, etc.).
This conversion is made possible by emission factors. These are calculated using laboratory research, case studies, averages and extrapolations, etc.
What are the sources of uncertainty when calculating a carbon footprint?
When carrying out a greenhouse gas (GHG) balance sheet, uncertainty is found in two main areas: in the activity data collected, and in the emissions factors (EF) chosen . The result is an uncertainty in the final result of the total GHG emissions calculated.
Since GHG accounting is based on the multiplication and summation of data, each of which carries a level of uncertainty, the final result, in terms of the total quantity of GHGs emitted, also carries a level of uncertainty. In this case, the uncertainty is aggregated. This is why it is also important to always consider a Bilan Carbone as an order of magnitude and not as an exact value.
There are many reasons for the uncertainty associated with activity data and emission factors :
- Lack of completeness : the amount of source data influences accuracy.
- Lack of reliability: measurement quality or estimation level.
- Lack of temporal representativeness: the "freshness" of data whose last update may be dated.
- Lack of geographical representativeness: the correspondence of a data used to estimate the reality on the ground.
Uncertainty in activity data
The level of uncertainty of activity data is set empirically according to its origin and quality.
- If the data is measured or "specific", we estimate the uncertainty at between 0 and 5%. This is the case, for example, for electricity consumption data taken from a meter.
- If the measurement is extrapolated or "semi-specific", the uncertainty is estimated at 30%. If, for example, an electricity meter reading is only available for 3 of a company's 5 sites, the data for the remaining 2 sites can be extrapolated, and the result will therefore carry a greater uncertainty.
- If the measurement is statistical or "generic", the estimated uncertainty is 50%. This corresponds, for example, to statistics on the average French commute to work.
This categorization of uncertainty is established in France by the Association pour la Transition Bas Carbone (ABC), which also supports and disseminates the Bilan Carbone® reference methodology.
Uncertainty in emissions factors
For emission factors, uncertainty stems from the way in which the emission factor has been determined, as well as the number and precision of the parameters involved in its calculation. ADEME, which offers one of the world's largest emission factor databases, provides an estimate of the uncertainty of most of its emission factors, as well as documentation on the origin of the factor in order to explain how each emission factor is calculated.
A "low" uncertainty of the order of 5% corresponds, for example, to the emission factor for fuel combustion. The quantity of CO2 emitted by the combustion of a liter of fuel has been measured fairly accurately. Depending on combustion conditions, not all the fuel consumed will be transformed into CO2 (unburned matter, VOCs, etc.), which explains the low level of uncertainty.
A high uncertainty of the order of 50% corresponds, for example, to the CO2e emission factors per kilometer of a road freight vehicle. In this case, actual emissions may vary according to driving style, weather, topography and the truck's actual load factor.
The uncertainty associated with monetary emission factors is much greater than that associated with physical emission factors. Indeed, monetary emissions factors generally carry an uncertainty of around 80%. This is notably the case for most of the monetary emissions factors in the ADEME database.
That's why it's preferable to use data specific to your own case rather than general data. Taking the example of road freight again, using data on your own fuel consumption is better than general data on average emissions per km.
To reduce the uncertainty of a balance, reference organizations recommend the use of physical and specific activity data, and the choice of the most specific emission factors possible. It is therefore strongly recommended to use monetary data only as a last resort, in order to ensure that the Bilan Carbone is usable over time.
Why is it important to take uncertainty into account when carrying out a Bilan Carbone?
In order to offer excellent reporting, the Bilan Carbone® methodology recommends respecting, among other things, the principles of accuracy and exhaustiveness in measured emissions. This means that biases and uncertainties must be reduced as much as possible, while seeking to cover as many emissions as possible in the course of measurement.
Measuring uncertainty also gives you the ability to improve the quality of your Bilan Carbone over time. It should be noted that precise calculation of uncertainty is not mandatory for all methodological standards, although it is generally encouraged, at least in qualitative terms. For example, it is compulsory to provide information on the uncertainties of emission items when carrying out a Bilan Carbone®, but this is only optional for the BEGES and the GHG Protocol.
Tracking and reducing uncertainty about your carbon footprint allows you to :
- identify sources of uncertainty andimprove the quality of its greenhouse gasbalance from year to year.
- obtain a carbon footprint that is more reliable over time , and therefore more useful for making the transition and reducing its impact.
- more homogeneous and therefore more easily comparable balance sheets between different years or different entities.
- not to be biased in identifying its decarbonization levers and prioritizing its decarbonization action plan.
How are overall uncertainty and uncertainty associated with emission items calculated?
The uncertainty associated with activity data and emission factors is generally associated with a 95% confidence interval, i.e. the probability that the actual value lies within this interval (e.g. + or - 30%) is 95%.
Once all the data have an associated uncertainty, two formulas are used to aggregate the uncertainties. The first formula (A) gives the uncertainty of a sum, expressed as a percentage:
The second formula (B) gives the uncertainty of a product, also expressed as a percentage:
To obtain the uncertainty associated with a company's carbon footprint, we will combine the uncertainties in the following steps:
- Use the first formula (A) to obtain the uncertainty associated with the measurement of an activity data item. The formula is applied to the uncertainties of the various points of measurement of the same activity data (e.g. the quantity of electricity consumed by all the sites of a company).
- Use the second formula (B) to obtain the uncertainty associated with the CO2e value of an emission item. In this case, the formula is applied to the uncertainty of the activity data calculated in the previous point and to the uncertainty of the emission factor used to obtain a value in CO2e.
- Finally, the first formula (A) is used again to obtain the uncertainty associated with the company's total emissions value. In this case, the formula is applied to the uncertainties of all the emissions item data calculated by repeating the previous two points.
The limits of calculating uncertainty in carbon footprint analysis
We can see from the previous formula that the greater the number of values we have, the more the total uncertainty tends to decrease. This is both interesting, as it obviously encourages exhaustiveness in data collection, but it can also potentially lead to biased analyses of the result by reducing the weight of uncertainty associated with certain data or certain emission items.
At the level of calculating an emission item, for example. If a company were to collect numerous expenditure data from different suppliers of the same product, and use a single, highly uncertain, monetary emission factor to obtain equivalent emissions, the final uncertainty on its emissions would be average or even low. The problem here is the resulting capacity for analysis and decision-making, making it impossible to answer questions like "Which supplier should I work with first to reduce my scope 3 emissions?".
The formulas for calculating aggregate uncertainty presented above are based on the assumption of low correlation between data points, which is not always the case in practice, depending on the carbon methodology applied to the source data. In some cases, if no attention is paid to the correlation between data points, the use of a large number of source data points can artificially lower the aggregate uncertainty and thus bias the analysis of the Bilan Carbone and the action levers to be considered. In the case of a high correlation between source data, each of which carries a high level of uncertainty, it is preferable not to use the above formulas in order to obtain a realistic and "exploitable" resulting uncertainty.
- a carbon balance with 10 emission items, each with a 30% uncertainty, would give an overall uncertainty on the carbon balance of 20%.
- a carbon balance with 1,000 emission items, each with an uncertainty of 30%, would give an overall uncertainty on the carbon balance of 1%.
In the following example, provided by the ABC in its Bilan Carbone methodological guide, the overall uncertainty of the company's carbon footprint would only be 12%, even though the uncertainty associated with the main emission items (travel and inputs) is over 20%.
It is therefore very important to always keep a step back from the uncertainty data calculated during analyses, and to always try to obtain a low uncertainty by reducing the uncertainty on the source data rather than solely on the basis of the volume of data. In other words, more accurate activity data and more accurate emission factors.
How does Traace integrate uncertainty calculation?
The Traace platform natively integrates the consideration of uncertainty in emissions factors, activity data and, of course, in the final carbon emissions results. Our customers can freely and easily enter the levels of uncertainty they wish on the original data, as well as on their own emissions factors. What's more, we maintain an extensive database of emissions factors whose uncertainties are filled in automatically.
Emissions analysis results on our dashboards systematically display total uncertainty levels, and granularly by category and emission item.
Our carbon accounting philosophy to limit uncertainty
As we have seen, there are two levels of uncertainty in calculating the carbon footprint, both of which can be influenced: The activity data collected and the choice of emission factors.
At Traace, we understand the importance of relying not only on clear, accurate activity data, but also on reliable emissions factors that vary as little as possible according to conditions, even if this means regularly updating activity data to make them ever more accurate according to your company's situations and contexts.
This is why we at Traace always strive to use physical emission factors on the one hand, and to support our customers in collecting accurate and exhaustive activity data on the other. This is the price of obtaining a quality carbon footprint that can be used as a starting point for a decarbonization strategy. The Traace teams are also responsible for updating the emissions factors in the databases available on the platform, so that our customers' carbon footprints are always as up-to-date as possible.
A data collection module designed to facilitate this stage and limit uncertainties
Data collection (gathering activity data and the corresponding emissions factors) is the most time-consuming phase in the process of drawing up a GHG balance sheet. While the methodologies recommend balancing data collection efforts against objectives, they also specify that data with a high degree of uncertainty (statistical averages, for example) may be easier to find, but their use is not recommended.
At Traace, we've developed a high-performance data collection module to help our customers avoid compromises between quality and quantity, and collect as much accurate data as possible . This module enables us to schedule data collection campaigns based on 100% customizable questionnaires .
Our module makes it easy to solicit the relevant stakeholders, collect accurate data, limit the sources of error that are very common in Excel, and manage the progress of data collection in a collaborative way, so as to meet your carbon footprint measurement and analysis objectives.
To find out more about how Traace can help you collect climate impact data, contact us!
- IPCC - Good Practice Guidance and Uncertainty Management in National Greenhouse Gas Inventories - https://www.ipcc-nggip.iges.or.jp/public/gp/english/
- GHG - Quantitative Inventory Uncertainty Guidance
- Bilan Carbone® V8: Methodological Guide (Appendices) - https://abc-transitionbascarbone.fr/ressource/bilan-carbone-v8-guide-methodologique-annexes