Reading carbon accounting parameters: emission factors, GWP, and data quality

Two practitioners applying the same standard can produce very different carbon totals — the difference is in the parameters. Here is how to read the ones that matter.

Emission factors: fit beats freshness

An emission factor converts activity data (kWh consumed, fuel burned) into emissions. Newer isn’t automatically better; fit is the test. Two checks: source match — national factors reflect national averages, but a plant in a hydro-rich southwestern province versus a coal-heavy northern one can differ by over 30% on electricity; use regional-grid or contract-specific factors when the goal is internal decarbonization. And temporal consistency — annually updated factors track the energy mix, but switching factors mid-series distorts trends; hold the factor version constant across a multi-year inventory or document the conversion. A common trap is biomass: some methodologies zero out biogenic CO2, but N2O and CH4 from combustion still count — check whether your methodology separates biogenic from fossil sources and requires non-CO2 equivalents.

GWP: 100 years or 20 — results differ severalfold

Global warming potential converts CH4, N2O, and other gases into CO2-equivalents, and the time horizon is the crux. For external reporting, follow the IPCC or national-inventory prescribed GWP-100 (methane: 28 under GWP-100, but 84 under GWP-20 — a 3× swing on the same tonne). For internal planning focused on near-term impact — say a livestock biogas-recovery project where short-lived methane dominates — a GWP-20 view better shows the value of early action. Watch versioning: the IPCC’s Sixth Assessment updated several GWP values; adopting the new set requires restating historical data for comparability.

Activity data: the boundary matters more than the number

Activity data is the input side — fuel volumes, feedstock, refrigerant charges. Accuracy debates usually miss the boundary and allocation questions where errors actually live. Three failure points: boundary inconsistency — is electricity counted as net purchases, or mixed with a captive plant’s grid feed-in? Methodologies typically require net purchased power, under an organizational boundary defined by operational or financial control. Allocation — multi-product chemical plants can allocate utility energy by output, revenue, or reaction heat, with very different per-product results; understand and disclose which was used. Missing data — default values for un-documented purchased materials come from industry averages that can deviate 20%+; where the gap is material, measure or use analogy rather than defaulting.

Leakage and deductions: the easily-missed negatives

Leakage is the added emissions a project displaces elsewhere — protect one forest and harvesting may rise around it. Methodologies list leakage sources, but applicability is your judgment: a renewables-substitution project displacing biomass power must consider indirect harvesting effects, and some 2026 forestry methodologies now require market leakage (project-driven timber-price effects), usually with default rates of 10–30%. For removals (BECCS and similar), deduction parameters assume capture efficiency and storage permanence — check whether the standard’s data-quality tiers allow adjusting them to measured values. The working rule for any deduction or leakage number: ask whether it was calculated, measured, or defaulted — defaults are for first estimates, not final reports.

Data quality: the soft parameter that decides hard results

Standards such as ISO 14064 and the GHG Protocol grade activity data and factors as high, medium, or low quality: continuous monitoring (CEMS) rates highest and defaults lowest; current-year data beats five-year-old data; regional measured factors beat national averages. Where confidence is low, methodologies allow conservative estimates — choosing the value that avoids understating emissions — but over-conservatism makes reduction performance look worse than it is, so quality assessment exists to explain uncertainty, not just to pick numbers. From 2026, some compliance markets require third-party verifiers to quantitatively grade data quality, with low-scoring items separately justified — internalizing that logic is worth more than accumulating parameter values.

Questions & answers

How do I choose emission factors correctly? Latest national or industry factors first, adjusted for region and technology; for supply-chain accounting, prefer supplier-measured factors.

GWP-100 or GWP-20? External reports follow the official standard (GWP-100); add GWP-20 internally when near-term reductions matter. Stay consistent.

How is the activity-data boundary set? By the methodology’s organizational boundary (operational or financial control); cross-boundary activities require allocation.

Do I calculate leakage myself? Methodologies give default rates; commission model- or survey-based corrections when your project deviates materially.

Does low data quality block reports? It can — verifiers demand sensitivity analysis or supporting evidence; upgrade data sources ahead of time.

Can parameters be mixed across methodologies? No — parameters must match the scope and boundary of one complete methodology or results lose comparability.

What changed for 2026? Updated IPCC GWP values, revised grid factors in several countries, and new leakage-estimation models — track the latest methodology versions.