Gains:
- Ability to place an activity in the correct scope (1/2/3) and calculate it in tonnes CO₂e with the logic of 'activity data × emission factor'
- Ability to confirm emission factor from official source by country and year and convert to CO₂e with correct GWP set for methane/N₂O
- Ability to produce results with deterministic accuracy by having artificial intelligence perform multi-line inventory totals with executable code rather than manually
Establishing an organization's carbon footprint is the cornerstone of environmental management. Without a footprint, you cannot set a target, make a reduction, or report. Greenhouse gas inventory is an account in which all greenhouse gases emitted by an institution in a certain period are converted into a common unit (ton CO₂e). In this unit, you will learn how inventory is set up, what scopes mean, and how to safely use AI to accelerate this accounting.
First, the world's common language: GHG Protocol (Greenhouse Gas Protocol) is the most widely used standard in the world that defines how inventory is calculated. ISO 14064 is the international standard equivalent of the same work. They both divide emissions into three scopes — and this distinction is the basis of everything.
What are three scopes?
Scope 1 — Direct emissions. Emissions from sources owned or controlled by the organization. Example: natural gas burned in the factory's boiler, diesel in company vehicles, chemical reactions in the production process.
Scope 2 — Indirect emissions of purchased energy. Emissions released elsewhere during the production of electricity, steam, heating and cooling that the organization purchases. You do not burn electricity, but your consumption causes emissions in the power plant.
Scope 3 — All other indirect emissions (value chain). The broadest and often greatest scope. Purchased goods and services, business travel, employee transportation, use of products sold, waste, supply chain. GHG Protocol Scope divides 3 into 15 categories.
Tip: In most utility and retail companies, 70-90% of emissions are in Scope 3. Reporting only Scope 1 and 2 is just showing the tip of the iceberg, and more and more legislation is mandating Scope 3.
How are emissions calculated? Basic formula
At the heart of inventory is one simple multiplication:
Emission = Activity data × Emission factor
- Activity data: how much you did — how many m³ of gas, how many kWh of electricity, how many km of roads, how many tons of steel.
- Emission factor: the amount of greenhouse gases released per unit of activity — for example, kg CO₂e per kWh of electricity.
For example, 480,000 m³ of natural gas × 2.02 kg CO₂e/m³ = 969,600 kg = approximately 970 tonnes of CO₂e.
Where does the emission factor come from? IPCC (Intergovernmental Panel on Climate Change), countries' official national inventory tables, DEFRA (UK) and EPA (US) factor sets, or the country's grid emission factor for electricity (the average carbon content of that country's electricity). This factor is different in each country and each year.
Step by step: setting up an inventory
1. Draw the organizational boundary. Which facilities, which companies are involved? Determine control approach (operational/financial control).
2. List the activities. Write the emission sources for each scope: fuels, electricity, travel, supply.
3. Collect activity data. Invoice, meter, voyage record, purchase data. This is the most time-consuming part of the job and it is essential that the data is traceable.
4. Match the correct factor. Assign the official factor for the correct country, year and fuel to each activity.
5. Multiply, turn, add. Convert to CO₂e (for methane and N₂O, GWP — Global Warming Potential, coefficient of global warming potential is used), sum by scope.
6. Document and verify. Record the source of each number; Prepare for independent verification (assurance).
Here AI; Powerful in extracting activity list, writing account code, scanning for inconsistencies in data and producing draft reports. But factor selection, coverage classification, and final approval are yours.
three mini cases
Case 1 — Grid factor error. An analyst asked the AI to calculate the electricity emissions of a factory (Scope 2). The factory is in Türkiye, but YZ defaulted to the US average grid factor (about 0.37 kg CO₂e/kWh). When the analyst applied the correct local factor (around 0.44 kg CO₂e/kWh, varies by year), the result increased by approximately 560 tonnes of CO₂e for a consumption of 8 million kWh. It was essential to clearly state the country factor.
Case 2 — The power of methane. A farm consultant had an AI count methane from livestock “like CO₂.” Whereas the 100-year GWP of methane is approximately 28; So 1 ton of methane means 28 tons of CO₂e. When the consultant implemented the GWP, this item of the inventory grew 28 times. If GWP is forgotten, the calculation will be completely wrong.
Case 3 — Code gained validation. A sustainability team had the AI manually collect emissions from a 40-row fuel table; The AI got a subtotal wrong and the result was off by 6%. When the team had the AI collect the same data with Python code (using Pandas), the code returned the deterministic correct total. Lesson: Don't make LLM students do arithmetic; Write the code and run it.
Weak prompt / Strong prompt
Weak prompt:
Find your carbon footprint from this fuel data.
Why it is weak: No scope, country, factor source and unit. AI collects by hand, fits factors, makes mistakes.
Powerful prompt:
Your role: GHG inventory specialist. Calculate the Scope 1 emissions for the fuel table below. Country: Türkiye, year: 2024. I will give the emission factors (there is a factor column in the table); FITTING. Do the calculation by writing Python code with pandasi; Give the emissions and total for each row in tonnes CO₂e. If Methane/N₂O is present, apply GWP and indicate which GWPset you are using. Data: [table]
Four copyable templates
1) Multi-line inventory account (with code):
Your role: carbon inventory expert. Process the following activity table with pandas. Columns: activity, amount, unit, factor, scope. For each row calculate emission = amount × factor, convert to tonnes CO₂e, by scope and give the grand total. If the factor is empty, mark [INCOMPLETE]. Give the code and the result together. Table: [here]
2) Scope classification help:
Divide the following emission sources into Scope 1, 2 or 3 according to the GHG Protocol, and if it is Scope 3, write down which of the 15 categories it falls into. Mark the source you are not sure of as "must be verified", making it fake. Sources: [list]
3) GWP conversion check:
Convert the following gas quantities to CO₂e. State clearly which GWP set you used (e.g. IPCC AR6, 100 years) and show the GWP value you used for each gas. DO NOT MAKE UP THE VALUES; Use standardIPCC values and specify the name of the set. Gases: [list]
4) Data quality screening:
Look for inconsistencies in the following activity data: missing values, unit mix-ups (kg/tonne, kWh/MWh), extreme outliers, duplicate records. List what you find in a tabular form, stating why it is suspicious. Changing data; just mark it. Data: [here]
Common mistakes
- Calculating local consumption with the default (e.g. US) factor. Always specify the country and year.
- Bypassing GWP. Methane and N₂O must be converted to CO₂e; Write which GWP set you are using.
- Skipping Scope 3 altogether. In most institutions, the largest share is in Scope 3.
- Getting LLM to do arithmetic. Have multi-row totals written into code; Do not collect by hand.
- Not documenting the source. Auditing requires a traceable source of each issue.
Attention: Emission factors are updated. The factor you used last year may have changed this year; Always record which factor version you are using so that inventories are comparable from year to year.
In summary
The greenhouse gas inventory is the collection of "activity data × emission factor" multiplied by the right scope, in the right unit, and with the right GWP. AI; It is an accelerator in listing activities, writing code, scanning data, and generating drafts. But the source of the factor, coverage class, GWP selection and final verification rest with the expert. Have the code do the arithmetic, get the factor from the table.
Application task
Prepare a five-line activity chart (e.g. natural gas, diesel, gasoline, purchased electricity, business flight). Find a factor in each line and write its source. Using the 1st template, have the AI calculate the emissions with the pandas code and collect them on a scope basis. Then compare the scope classification of the AI with the 2nd template to the one you made manually. If different, investigate why.
checklist
- [ ] I determined the organizational boundary and scopes.
- [ ] For each activity I used the official factor appropriate to the country and year.
- [ ] I converted to CO₂e with the correct GWP set for methane/N₂O.
- [ ] I had the multi-line summation done by the executing code, not by LLM.
- [ ] I didn't ignore Scope 3, at least I scanned the categories.
- [ ] I documented the source and factor version of each number.