Limitations
An MRV platform is judged on this page more than on any other. Three bands, four verification gates, and for each gate what it does not show — because “internally consistent” and “verified against measurement” is the distinction most easily lost, and losing it is how a screening estimate ends up quoted as a measurement.
The platform stands behind these, within the stated uncertainty.
- Source-resolved methane and combustion CO2 for the reported period
- Carbon intensity per unit of saleable energy, with propagated uncertainty
- Relative comparison between facilities on a common accounting basis
- Pathway analysis under defined policy levers
Useful for prioritisation and discussion. Not a determination of anything.
- Fleet operability and marine-access assessment
- Regulatory readiness scoring — a capability assessment, not a legal compliance determination
- All four pathway targets, which test against policy benchmarks and a market comparator rather than statutory maxima
Outside the boundary. Absent from every figure on the platform.
- Emissions verified against direct field measurement
- Non-routine and abnormal events, which can dominate a single facility-month
- Drilling and well-intervention campaigns
- Marine logistics, shuttle-tanker propulsion and aviation
- N2O from combustion, and CO2 already present in the produced gas stream
Known gaps
Stated with the consequence and the remedy, because a gap listed without either is an admission rather than a plan.
Flare combustion efficiency is not measured at any facility
It governs roughly 70% of national methane and is the dominant term in the uncertainty; it is carried as a bounded distribution rather than a measured value.
Remedy: A flare-performance measurement campaign
Component counts and leak-detection surveys do not exist
Eight equipment sources remain at IPCC Tier 1.
Remedy: A component inventory and an LDAR programme
Verification gates
Four automated checks the engine must pass. Each is reported with what it establishes and what it does not, in that order.
Independent facility cross-check (SGN)
passesobserved 15 against a threshold of 15 % divergence on the field total
Establishes: Two independent implementations of the method agree on one field over 99 months.
Does not establish: That either figure matches measured emissions at the facility.
Different factors, software and analyst; no shared code.
Independent re-implementation
passesobserved 0.2 against a threshold of 0.2 % maximum relative error
Establishes: Two implementations of the same equations produce the same numbers.
Does not establish: That the equations describe the facility correctly.
Production hold-out (2024-2025)
passesobserved 17.2 against a threshold of 25 % mean absolute error
Establishes: The driver model predicts withheld years within its stated error, which is what the projection band carries.
Does not establish: That the emission factors applied to those drivers are correct.
oil 16.7%, gas 17.2%
Scenario emulator agreement
passesobserved 0.979 against a threshold of 0.95 R^2, out-of-sample
Establishes: The emulator can stand in for the engine inside sensitivity and VoI analysis.
Does not establish: That an emulated value may be published as an emissions result.
CI R^2 0.979, methane R^2 0.9998
Every figure on this platform is a bottom-up, activity-based estimate with propagated uncertainty. None of them has been checked against a measurement of the thing it estimates, because no such measurement exists for these facilities. That is the gap the platform is built to close, and until it closes, saying so is part of the product.