Overview
The results of your GHG inventory are not completely accurate. Quantifying these uncertainties isn't easy, as not all emission factor databases provide an uncertainty, and there is a lack of consistency between the databases that do. This article explains the different sources of uncertainty, what is currently included in the figures displayed on the platform, how Greenly computes the uncertainty of each emission factor, and how individual uncertainties are aggregated into a single figure for a category, a scope, or your total footprint.
1. Sources of uncertainty
There are several sources of uncertainty in a GHG inventory:
Emission factors: Calculating an emission factor is inherently uncertain and relies on numerous assumptions and calculation choices, so each emission factor may be more or less reliable. Emission factor databases are also not exhaustive — in some cases the temporal, geographical, or technological granularity is not precise enough.
Emission factor selection: For a given quantity or amount, several similar and relevant emission factors with different values may exist, and an incorrect emission factor may be selected.
Activity data: Expenses carry no uncertainty, as these amounts are exact. However, activity data provided by a reporting company can be inaccurate — for example, the distance travelled by truck or the quantity of steel purchased are often estimated values.
Methodology: The expense-based approach has a higher level of uncertainty than the activity-based one. We therefore encourage reporting companies to always favour activity-based studies whenever possible.
2. What is currently included in the uncertainties displayed?
Currently, only the uncertainty linked to the emission factor is included in the computation of uncertainties. Activity data uncertainties (for instance, the uncertainty on the total distance traveled by a vehicle fleet) will be included soon.
3. How are emission factor uncertainties computed?
As not all emission factor databases provide an uncertainty, and there is a lack of consistency between the databases that do, Greenly has implemented an internal process to link every emission factor to an uncertainty.
There are six levels of uncertainty: 5%, 15%, 30%, 50%, 65%, and 80%. The level is selected depending on the type of emission factor and on an emission factor confidence score.
Confidence score (between 0 and 1)
The confidence score depends on the type of emission factor:
Greenly Monetary Emission Factors: These ratios are computed by dividing emission factors from renowned databases (e.g. Ecoinvent) by the average price of the product, service, or activity. The confidence score depends on the bias of the primary data used to compute the ratio and on the variance of the category (e.g. a very specific sector with highly homogeneous products has a low variance).
Input-output Monetary Emission Factors: These ratios are computed using input-output tables for a given sector and country. The confidence score depends on the variance of the sector and on the relevance of the sector for the Greenly category.
Company-Specific Monetary Emission Factors: These ratios are computed by dividing the GHG inventory of a company by its revenue (sectoral averages can also be computed). The confidence score depends on the reliability of the revenue and on the reliability, transparency, and completeness of the GHG inventory.
Activity Emission Factors: These ratios are provided by external sources. The confidence score depends on the reliability of the source — for example, a renowned and trustworthy database has the highest score, while a non peer-reviewed article has a low score.From confidence score to uncertainty
Once the confidence score is computed, the uncertainty level is selected depending on the type of emission factor and the confidence score. For example, an activity emission factor with a confidence score of 1 has an uncertainty of 5%, while a Greenly monetary emission factor with a confidence score of 0 has an uncertainty of 80%.
4. How are aggregated uncertainties computed?
Each activity in your GHG inventory is associated with an uncertainty, derived from the uncertainty of its emission factor. To compute the aggregated uncertainty at a higher level — whether for a visualisation category, a regulatory category, a scope, or the total footprint — a weighted average is used, where each activity's uncertainty is weighted by its share of total emissions:
Aggregated uncertainty = (Emissions₁ × Uncertainty₁ + Emissions₂ × Uncertainty₂ + …) / (Emissions₁ + Emissions₂ + …)
This means that activities with larger emissions have more influence on the aggregated uncertainty. An activity representing a small share of emissions has little impact on the group's uncertainty, even if its individual uncertainty is high.
Example — a Scope 3 with two activities:
Activity A: 900 tCO₂e, uncertainty 5%
Activity B: 100 tCO₂e, uncertainty 50%
Aggregated uncertainty = (900 × 5% + 100 × 50%) / (900 + 100) = (45 + 50) / 1000 = 9.5%
Why not the Root-Sum-of-Squares (RSS) method?
An alternative approach used in some standards (such as the IPCC Guidelines for national GHG inventories) is the root-sum-of-squares (RSS) method, which combines uncertainties as follows:
Aggregated uncertainty = √((Emissions₁ × Uncertainty₁)² + (Emissions₂ × Uncertainty₂)² + …) / (Emissions₁ + Emissions₂ + …)
Using the same example, RSS would give √(45² + 50²) / 1000 = √4525 / 1000 ≈ 6.7%, a lower result than the weighted average (9.5%).
RSS assumes that individual uncertainties are statistically independent and will partially cancel each other out, producing a more optimistic combined uncertainty. The weighted average makes no such assumption; it treats uncertainties as fully additive, which is more conservative and more appropriate given that:
Emission factor uncertainties at Greenly are discrete levels (5%, 15%, 30%, 50%, 65%, 80%), not statistical distributions derived from rigorous measurements
The independence assumption behind RSS is difficult to verify in practice, as many activities share similar emission factors or data sources
A more conservative figure is more honest when communicating uncertainty ranges
To understand what your overall uncertainty score means and how to improve it, see the article “What does my uncertainty score mean?”