Emissions Forecasting for Australian NGER and AASB...

Emissions forecasting gives you a likely year-end number in Q2 instead of October. It projects the rest of the year from your monthly emissions records, with a lower and upper bound for each month. It then compares the projected total with your targets, which leaves time to act on it.

A common pattern in Australian reporting cycles goes like this. August: "we don't have full Scope 3 yet." September: a scramble. October: the working file gets its hardest review of the year and the annual number moves while the sustainability manager tries to explain why. For the 2025-26 year the NGER report is due by midnight AEDT on Monday 2 November 2026, because 31 October falls on a Saturday, and the CER has said it will not grant extensions. The AASB S2 disclosure follows with the financial report. By January the CFO is asking a question that could have been answered in April: "how did we end up here?"

That question is why Carbonly includes emissions forecasting. It is not scenario analysis under AASB S2 paragraph 22, which tests climate resilience and is a different exercise. It is a projection of the reported number from the ledger itself, shown as a range rather than a single figure.

If you can see a likely year-end number in Q2, you can act on it. If you first see it in October, you mostly just file it.

What you see: a year-end range, month by month

The forecast reads the emissions ledger. Every invoice, docket and meter read that gets extracted becomes a record with a date, a scope and a project. The forecast totals those records by month over the trailing twelve months, for the whole organisation or filtered to one project or one scope, and projects the months ahead.

The projection is made by an AI model that is asked to look for trend, seasonality and recent step changes in the monthly series and to return a range for each future month. If that call fails or returns anything malformed, the forecast falls back to a straight-line regression over the same twelve months, with a band of one standard deviation of the residuals either side. The result records which method produced it, and the AI path includes a short written explanation of what it saw in the data. With fewer than three months of data, or a series that is all zeros, only the regression runs.

Seasonality is the reason a simple annualisation can mislead. A remote project site might burn far more diesel during an earthworks phase than in the months either side of it. An office might use more electricity in the months when HVAC runs hardest. Refrigerant top-ups follow a service cycle. Multiplying year-to-date by 12/n ignores all of that.

The output is a range, not a single number. The CFO sees a projected figure with a lower and an upper bound for each month, and the range should be wider the further out the month is. A range is the honest thing to put in a board pack. A single point estimate implies more precision than a part-year ledger can give.

Why this matters for Safeguard Mechanism reporters

If you operate a facility covered by the Safeguard Mechanism, its baseline declines each year, by 4.9% a year to 2030 for most facilities. Emissions above the baseline have to be managed, usually by surrendering ACCUs or Safeguard Mechanism Credits, and an unmanaged excess exposes the responsible emitter to civil penalties under the NGER Act. The Safeguard Mechanism 2026 changes post covers the options and current penalty figures.

Finding out in October that a facility is well over its baseline leaves few options. Finding out in March leaves time to plan.

With Carbonly you can see whether a facility's projection approaches or crosses its declining baseline. The Safeguard Trajectory Check runs on demand for a plan and compares projected emissions against each covered facility's baseline.

The check does not tell you what to do about it. That is a carbon reduction planning question. What it does is start the conversation while options are still open: re-routing a fleet to cut diesel, bringing forward a planned gas-to-electric change, or buying units earlier in the year rather than in an October rush. Those decisions need visibility in Q2.

Why this matters for AASB S2 disclosure

After the first reporting period, the climate disclosure includes comparative information for the prior period. That sounds routine until the current-year figure is still moving in September with assurance work scheduled soon after.

A forecast helps here. With a projection next to the prior-year figure, the CFO can see the likely shape of the disclosure months before assurance starts. If the current-year projection is well above the prior year, there are two reasonable responses: look into the drivers now, or start preparing the explanation for why it moved.

The disclosure also has to make sense next to any transition plan the entity has described. If the plan says emissions are falling toward a 2050 net zero target and the Q2 projection shows a year-on-year rise, the board will want to know before the report is published, not after. Our climate transition plans post covers what a transition plan disclosure involves.

What makes the forecast worth trusting

The main input is the monthly ledger: every extracted document and activity record, tagged to a project and a month. The forecast is only as good as that ledger. A month where fuel dockets have not been loaded yet will read as a low month, and the projection will reflect it, so check the Data Health Agent's view of scope and month coverage gaps before reading too much into a forecast.

The second thing to understand is factor versions. The NGA Factors workbook is updated each year, and the NGA explainer covers how editions apply. Each Carbonly record keeps the emission factor version it was calculated with, so the historical months in a forecast reflect the factors that were applied when each record was calculated.

The same goes for GWPs. NGER uses IPCC AR5, the ledger calculates in AR5, and so the forecast is on an NGER basis. If part of your AASB S2 disclosure needs AR6 values, treat that conversion as a separate, documented step outside the forecast.

Whether you will hit your target, and what closing the gap costs

If you have set carbon targets in Carbonly, the forecast compares the projected total for each target year (actual months plus forecast months) with the target and shows whether the projection is above or below it, and by what percentage.

Carbon planning also includes a marginal abatement cost curve: the actions in a plan, ranked by estimated cost per tonne abated, with a cumulative reduction. When a projection sits above a target, the MACC is where you look for the cheapest ways to close the gap. The MACC uses the cost and reduction estimates you enter for each action; it does not invent them.

A hypothetical example: rooftop solar at a distribution centre might be estimated at 800 tonnes a year at $12 a tonne, a small fleet electrification trial at 400 tonnes at $180 a tonne, and an HVAC upgrade at 200 tonnes at negative $30 a tonne because the energy saving outweighs the cost. The MACC ranks them, and the forecast tells you roughly how many tonnes the gap is.

Put together, the forecast shows the gap and the MACC shows what closing it might cost, which gives the board pack something more specific than a trend arrow.

Reading the range

The range is the part most worth understanding.

A narrow range over a complete, stable ledger suggests the projection is a reasonable planning number.

A wide range, or a range built over months with missing data, means the mid-point is not reliable. Use the range, not the mid-point, and do not commit to a single figure in the board pack.

A forecast built on only a few months of history is a directional signal at best. Expect it to move as more data arrives.

Be open about this with whoever reviews your numbers. A tight projection six months out invites questions about its assumptions. A wider range, with a clear note on what data would narrow it, usually leads to a better conversation.

What forecasting is not

Emissions forecasting is not scenario analysis under AASB S2 paragraph 22. Scenario analysis is a climate resilience exercise that runs the business through different warming and transition pathways to test its exposure to transition and physical risk. It is strategic, usually done annually, and has qualitative and quantitative outputs.

Forecasting projects the reported number from the ledger. It answers an operational question ("what is our year-end likely to look like?"), not a strategic one ("how would we fare in a disorderly transition?"). An AASB S2 report needs the scenario analysis. The forecast is a management tool that helps you get there without surprises.

People mix them up because both look forward. Keeping them separate matters for how you use each output and how you present it.

Asking ChatGPT or Claude how you are tracking

A CFO between meetings can ask "how are we tracking on the 2030 target?" and get the answer from the ledger without opening a dashboard. That works through Carbonly's MCP server. Once ChatGPT or Claude is connected to your workspace, it can call target progress (with trajectory status and the remaining gap to target) and the MACC for a plan.

A practical starting point

If you have never forecast against your emissions ledger, start with your highest-emitting project, not the whole portfolio.

Load at least six months of documents (invoices, dockets, meter reads) in any of the eight file formats the document engine accepts, review the extractions and the low-confidence material matches, and then run the forecast for that project. A full twelve months of history gives the forecast more to work with.

Look at the range. If it is narrow and the ledger is complete, roll the approach out to other projects. If it is wide, check which months are thin or missing before treating the projection as a planning number.

For the 2025-26 NGER year the forecast is largely academic now: the year closed on 30 June and the report is due by midnight AEDT on 2 November 2026. Our NGER October 2026 deadline preparation guide covers the final weeks. Forecasting earns its keep in the 2026-27 year that is already under way.

Carbonly puts a year-end emissions range in front of the CFO in Q2, built from your own documents. Consultants running an ASRS or NGER engagement can set up a client workspace, load the ledger and show a forecast without building a model from scratch. Pricing is per project (Small, Medium, Large or Enterprise) with a $100 per month workspace minimum and no per-seat fees. To see it on your own data, email hello@carbonly.ai.

Emissions Forecasting for Australian NGER and AASB S2 Reporters: Knowing Your Q4 Number in Q2