Estimates and Data Quality Under the UAE Climate Law: What Is Defensible?

Prepared in British English as a practitioner Knowledge Card Package: answer, explanation, application, evidence, connections and publishing layer.

A missing invoice or failed meter is not only a technical inconvenience. It affects the reliability of the inventory, the completeness of a filing, the interpretation of year-on-year changes and potentially whether a threshold or source classification is crossed. Weak estimates can also undermine reduction claims: a fall in reported emissions may be caused by a new proxy rather than actual operational improvement.

The practical objective is not to eliminate all estimation. It is to control where estimation enters, how much of the inventory it affects and how quickly the underlying evidence will improve.

Figure 1. Estimate defensibility decision flow. Branded educational visual by London Reporting Academy.

It is tempting to rank data automatically as measured, then calculated, then estimated. That is too crude.

A poorly maintained meter with an unknown boundary can be less reliable than a controlled calculation based on complete supplier invoices. A supplier estimate may be better than an internal proxy if the supplier provides product-specific evidence. A mass balance can be the prescribed method for a process source rather than a second-best substitute.

The method should therefore be assessed against six questions:

1. Authority status. Is the method prescribed, approved or permitted?

2. Source fit. Does it quantify the actual source, gas and boundary?

3. Completeness. Does it cover the full period and all relevant source streams?

4. Representativeness. Are the proxy, factor, geography, technology and time period appropriate?

5. Reproducibility. Can another person rebuild the result from retained evidence?

6. Control. Has the method been reviewed, approved and tracked for improvement?

Article 6 establishes the MRV duty for determined Sources but leaves detailed standards, forms and electronic mechanisms to MOCCAE and competent authorities. The law does not itself set:

• a universal percentage of estimated data allowed in an inventory;

• one mandatory missing-month formula;

• a fixed company-wide materiality threshold for data errors;

• a national four-tier uncertainty scale for every Source;

• a blanket prohibition on spend-based or production-based proxies; or

• a rule that an estimate must always be conservative.

Local systems can be more specific. The Environment Agency - Abu Dhabi's March 2026 workshop described a facility-level tier approach for activity-data uncertainty: Tier 1 below +/-7.5 per cent, Tier 2 below +/-5.0 per cent, Tier 3 below +/-2.5 per cent and Tier 4 below +/-1.5 per cent. It also described higher expectations for major sources, Tier 1 as potentially suitable for minor sources and possible exclusion of de minimis sources. Those values are an Abu Dhabi local implementation example and must not be presented as a universal federal scale.

A strong estimate file answers the following questions.

Why was an estimate necessary?

Record the exact gap: failed meter, unreceived invoice, unavailable landlord data, supplier non-response, closed site, new acquisition, data corruption or technically infeasible direct measurement. "Data unavailable" is not enough.

What hierarchy was followed?

A common hierarchy is:

1. recover the actual source record;

2. obtain a supplier or authority replacement;

3. calculate from an independent operational record;

4. interpolate from comparable periods;

5. apply a physical proxy;

6. apply a spend or broad industry proxy as a last resort.

The hierarchy is a practitioner model. The applicable authority may prescribe a different fallback or require approval in a monitoring plan.

Why is the proxy representative?

A proxy should preserve the physical relationship with the missing activity. Useful examples include:

• production tonnes for a process that has stable energy intensity;

• operating hours and rated load for equipment with controlled utilisation;

• tonne-kilometres for freight;

• occupied floor area and days for a stable building load; or

• adjacent-month consumption adjusted for production and weather.

A proxy is weak where the relationship has changed because of maintenance, fuel switching, shutdown, expansion, abnormal weather or a change in product mix.

How was uncertainty assessed?

Uncertainty does not need to be reduced to false precision. The organisation should identify the main sources: meter accuracy, sampling, factor uncertainty, allocation, missing period, proxy relationship and manual transformation.

For a material estimate, document either:

• a quantified range or sensitivity;

• a comparison to an independent benchmark;

• a scenario test using alternative reasonable methods; or

• a qualitative rating explaining why numerical uncertainty cannot be estimated reliably.

Who reviewed and approved it?

Separate preparation from review. The reviewer should check the source gap, formula, proxy fit, units, factor, period, duplication and effect on totals or thresholds. Material estimates should be approved by the methodology owner or reporting controller, not only the data preparer.

What will improve next cycle?

Every recurring estimate should have an owner, action and target date. Examples include installing a sub-meter, amending a supplier contract, collecting monthly landlord data, integrating fuel cards or obtaining laboratory analysis.

Missing-month estimates are common and easy to mishandle. A controlled process is:

1. verify that the month is truly missing rather than posted under a different account or facility;

2. identify whether consumption follows production, days, weather, hours or another driver;

3. select comparable months before and after the gap;

4. adjust for shutdowns, maintenance, acquisition, product mix or abnormal operations;

5. calculate the estimate and an alternative reasonableness range;

6. record the share of annual emissions affected;

7. obtain review and approval; and

8. replace the estimate if actual data arrive before the filing or correction window closes.

Avoid annualising an average blindly. One month of natural gas during a summer shutdown is not comparable with a winter production peak. Likewise, electricity in a hotel, shopping centre or district-cooling system can be highly seasonal.

"Materiality" in data-quality control is not automatically the same as a statutory reporting threshold or a verifier's materiality threshold. The company should define internal escalation criteria that consider:

• absolute and percentage effect on total emissions;

• effect on a facility, source or gas subtotal;

• whether the estimate changes threshold or designation conclusions;

• whether it changes a reduction trend, target result or public claim;

• whether the method is recurring or isolated;

• whether the gap indicates a control failure; and

• whether the authority specifically requested the data.

Do not use an internal materiality threshold to leave known errors uncorrected where the reporting instruction requires correction. The EAD March 2026 workshop, for example, described correction of errors within 30 days of discovery for the Abu Dhabi facility route.

Figure 2. Illustrative LRA data-quality score for climate data. Branded educational visual by London Reporting Academy.

The following score is an internal management tool, not a legal UAE rating. Score each dimension from 1 to 5 and apply the weights.

Illustrative score calculation: multiply each 1-5 score by its weight and add the results. A score of 4.5-5.0 is strong; 3.5-4.49 is generally usable with minor improvement; 2.5-3.49 requires explicit limitation, approval and remediation; below 2.5 requires escalation and an alternative method or authority discussion. These bands are LRA practice only.

Do not average away a critical weakness

A high total score should not override a critical issue. A source may score well overall but still be unusable because:

• the boundary is wrong;

• the factor is not accepted;

• the estimate crosses a reporting threshold;

• the source record appears manipulated;

• the method double counts another source; or

• the authority requires direct measurement or approval.

Add a mandatory red-flag field alongside the numeric score.

Illustrative scenario - not company data. A facility has monthly gas invoices and production data, but the August invoice is missing. The plant operated at reduced output for a planned shutdown.

A simple average of July and September would overstate August. The team first confirms that the volume was not recorded under another account. It then estimates consumption using August operating hours and the plant's stable gas-per-operating-hour relationship from comparable shutdown months, cross-checked against production and a supplier statement.

The file records:

• the missing invoice and recovery attempts;

• the selected proxy and two rejected alternatives;

• the formula, units and emission factor;

• a sensitivity range;

• the percentage of annual facility emissions affected;

• reviewer approval; and

• a control action requiring direct monthly meter extraction next cycle.

When the supplier invoice arrives after filing, the team compares actual and estimated values, applies the relevant correction procedure and preserves both versions.

Cycle 1 - make gaps visible

Create source, estimate and evidence registers. Separate actual and estimated values. Implement basic reconciliations and management sign-off. Disclose material limitations.

Cycle 2 - replace high-emission proxies

Prioritise major sources, recurring missing months and estimates affecting thresholds or public claims. Add meters, supplier clauses, data interfaces and stronger factors.

Cycle 3 - automate controls and quantify uncertainty

Automate completeness, duplicate and variance checks. Quantify uncertainty for priority sources. Test change controls and prepare verifier-ready evidence packs.

The roadmap should be risk-based. A low-value travel estimate may remain proportionate while an estimated furnace fuel stream needs urgent remediation.