Why Your Emission Factors May Be Wrong: Bottom-Up vs Top-Down GHG Accounting

Open any corporate carbon inventory and you will find a spreadsheet of activity data — litres, kilowatt-hours, tonnes, kilometres — each multiplied by a coefficient. The activity data usually gets the scrutiny: is the meter reading right, did we capture every invoice, is the boundary complete? The coefficients rarely do. They arrive from a published table, carry a citation, and are treated as fact.

They are not facts. They are assumptions with error bars, and the error bars vary by two orders of magnitude across the sources in a typical inventory.

What an emission factor actually is

An emission factor is a claim of the form: an activity of this type, on average, releases this much greenhouse gas per unit. Grid electricity in India, roughly 0.7 kg CO₂e per kWh. Diesel combustion, roughly 2.68 kg CO₂ per litre. Landfilled municipal waste, some quantity of methane per tonne — and here the numbers start getting interesting.

Every factor is the product of a research process: someone measured a sample of real sources, aggregated the results, and published a central value. The critical questions — almost never asked by the people using the factor — are how many sources were measured, how similar were they to yours, and how much did they vary?

The reliability gradient

Not all factors are equal. This ordering is more useful than any single accuracy figure:

Source type	Typical uncertainty	Why
Stationary fuel combustion	Low (a few %)	Simple stoichiometry; carbon content of fuel is well characterised and measurable
Purchased electricity	Low–moderate	Grid factor is well documented, though it varies by region, time of day and year
Transport fuel	Low	Same combustion chemistry; complications are in activity data, not the factor
Industrial process emissions	Moderate	Process-specific but usually measurable and well studied at the plant
Refrigerant leakage	Moderate–high	Depends on equipment condition and maintenance quality, not just charge size
Landfill and wastewater methane	High	Depends on waste composition, moisture, temperature, cover, capture efficiency and microbial activity
Agricultural soil N₂O	High	Depends on soil type, moisture, temperature, timing and form of application
Rice cultivation methane	High	Depends on water regime, organic amendments, soil, cultivar and season
Fugitive and venting emissions	High	Often intermittent and unmeasured; a few "super-emitters" can dominate the total
Biomass and open burning	Very high	Highly variable combustion conditions; factors often rest on very few studies

The pattern is not random. Where emissions come from combustion chemistry, factors are reliable. Where they come from biology or from things going wrong, they are not. Microbes respond to temperature and moisture in ways a global average erases; leaks and equipment failures are, by nature, not the average case.

How large can the gap get?

Large enough to change conclusions. The clearest recent demonstration comes from atmospheric science rather than corporate reporting, which is what makes it useful — nobody had an incentive in the outcome.

In Zhu et al., Science (2026), researchers estimated methane emissions over Siberia from satellite and surface atmospheric measurements, working backwards from what is actually in the air rather than forwards from activity data. For eastern Siberia, where the emissions are wildfire-driven, their estimate came out 6.7 to 41 times higher than the standard global biomass-burning inventories — GFED, FINN, GFAS and QFED.

Their diagnosis of why is the transferable part, because all three causes are ordinary:

1. The factor rested on very few studies. The methane emission factor for boreal and peat fires derives from a small number of field and laboratory measurements. That was the available evidence, so it was applied globally.

2. The activity data was itself uncertain. Burned area and fire radiative power are satellite-derived estimates with their own error, and they multiply straight through.

3. A whole source was missing. The inventories count methane from combustion. They do not count methane released from permafrost thawed by the fire — a real emission outside the accounting boundary.

The three failure modes, translated

Strip away the Arctic context and each cause maps onto something in a normal corporate inventory.

Thin evidence base

You are using a default factor from IPCC, DEFRA, EPA or a national database. It was derived from measurements in a different climate, at different scale, under different management. For combustion this hardly matters. For a biological or fugitive source it can matter enormously — an N₂O factor from temperate European agriculture is a weak guide to a monsoon-fed Indian field.

Uncertain activity data

The factor may be fine while the quantity it multiplies is estimated. Waste tonnage inferred from truck counts, refrigerant charge assumed from equipment specifications rather than service records, supplier volumes taken from purchase orders rather than delivery data. The output inherits every input error.

Sources outside the boundary

The emission is real, physically caused by your operations, and appears nowhere in your inventory because no line item exists for it. Fugitive releases nobody measures, methane from a wastewater lagoon classified as "water treatment", emissions during startup, shutdown and upset conditions that the steady-state factor was never meant to cover.

What actually changes because of this

For most companies the honest answer is: less than you might fear, in a narrow accounting sense. Your BRSR filing is not invalid because a landfill factor is uncertain, and defaults remain accepted practice. What changes is the defensibility of your number when it comes under pressure:

• Assurance is tightening. Providers increasingly test whether a default factor is appropriate for a material source, rather than only checking that the arithmetic used it correctly.
• Buyers want primary data. If you sit in someone's Scope 3, your generic factor becomes their reported number, and large buyers with science-based targets are pushing back on that.
• Independent measurement now exists. Satellite methane monitoring means a third party can produce a number for a facility without your cooperation. When theirs and yours disagree, "we used the published default" is a thin defence.
• Targets built on wrong baselines mislead you. This is the one that costs money regardless of anyone else's opinion. If a material factor is significantly off, you may be directing abatement capital at the wrong source entirely.

A proportionate response

Not every factor deserves attention. Most deserve none. Rank them:

Step 1 — Score each source on materiality × uncertainty. Take each line's share of your total footprint and multiply it by a rough uncertainty band for its factor (use the table above). A source at 40% of your footprint with an order-of-magnitude uncertainty is the priority. A source at 0.5% with a combustion factor is not, however easy it would be to measure.

Step 2 — Trace the provenance of the top few. For each high-scoring factor, establish where it came from, what was measured, how many sites, in what conditions, and how much they varied. Most published factors document this if you look. If you cannot find it, that is itself the finding.

Step 3 — Measure the ones that matter. For a genuinely material source with a weak factor, site-specific measurement is the fix — flux chambers for a landfill, continuous monitoring for a process vent, field trials for an agricultural practice. This is real money, which is exactly why steps 1 and 2 come first.

Step 4 — Widen the boundary check. Walk the site with an engineer, not an accountant, and ask where gas physically escapes. Compare that list against your inventory line items. The gaps are your missing sources.

Step 5 — Document uncertainty rather than hiding it. State the basis of each material factor and its uncertainty range in your methodology note. A number with an honest error bar is more credible than a suspiciously precise one, and it protects you when the figure is challenged.

Step 6 — Re-check when something changes. New factor versions, a changed process, a new site or a shifted grid factor all invalidate prior work. Build a review cycle rather than treating the inventory as settled.

The realistic standard

The goal is not a perfect inventory — there is no such thing, and pursuing one wastes money that should be spent on abatement. The goal is an inventory whose weak points you know and can explain. A company that can say "this is 45% of our footprint, the factor has high uncertainty, here is why we use it and here is our measurement plan" is in a far stronger position than one presenting six significant figures with no idea which of them are load-bearing.

Carbon Credit Consulting builds greenhouse gas inventories on evidence rather than defaults, and helps companies decide where measurement is worth the cost. Explore our GHG accounting and ESG & BRSR reporting services, or talk to us.