1. Carbon Footprint of Food Ingredients
Our carbon footprint calculations use well-established environmental assessment methodologies to provide accurate greenhouse gas emission estimates for recipes. We combine multiple datasets to ensure comprehensive coverage of different food items while maintaining consistency in measurement approach.
1.1 Data Sources and Scope
Primary Dataset: Agribalyse
We use the Agribalyse dataset [1] as our primary source for emission factors for each food item in a recipe. The advantage of Agribalyse [1] lies in its comprehensive cradle-to-grave Life Cycle Assessment (LCA) approach, which includes:
- Production of raw materials (agricultural processes)
- Processing of food items
- Packaging requirements
- Distribution networks
- Retail operations
- Storage conditions
- Preparation at the consumer end
- Transport between each stage of the value chain
Agribalyse [1] excludes transport between retail outlets (shops, supermarkets) and the consumer’s home. Food waste and losses are accounted for at various stages of the life cycle, except at the consumer’s home. Importantly, inedible losses are still accounted for even at the consumer’s home, while only edible losses (household food waste) are excluded. The dataset is representative of food consumption and preparation patterns in France.
Supplementary Dataset: CarbonCloud ClimateHub
In addition to Agribalyse [1], we utilize the CarbonCloud ClimateHub [2] to:
- Cross-check emission factors for verification
- Source emission factors for food items not found in Agribalyse [1]
The CarbonCloud [2] dataset primarily focuses on Sweden and the United States, but also includes LCA data for food items from other countries. Its scope includes the lifecycle from farm to retail store. Consumer preparation is never included in their assessments. For our purposes, we use emission factors measured at store level, which excludes:
- Transport between retail outlet and home
- Home preparation and storage
- Food waste occurring at home
1.2 Methodology Considerations
Standardizing Emission Factors
To maintain uniformity in our values, we consistently use farm-to-shop emission factors and exclude the use phase from our footprint calculation scope. This decision is based on two key factors:
- The significant local variation in energy emission factors (e.g., the carbon footprint of cooking with electric appliances would be very small in Sweden or France, but significant in Germany)
- The substantial individual variations in how food is transported from retail outlets to homes
Calculation Process
The emission factors are provided in the unit ‘kgCO₂e/kg’ and include all greenhouse gases emitted in the lifecycle of 1 kg of the product within the defined scope. Our calculation process follows these steps:
- Convert all ingredient quantities to kilograms
- For volume-to-weight conversions, we utilize the Aqua-Calc food volume to weight calculator [3]
- Calculate the total GHG emissions using the following formula:
1.3 Limitations and Future Work
Our current calculations focus exclusively on climate impact through greenhouse gas emissions. For a truly comprehensive assessment of environmental impact, additional factors such as pollution or biodiversity impact should be included. We plan to incorporate these aspects in future iterations of our methodology.
A more thorough analysis of the lifecycle steps that produce the largest emissions might also be done in future iterations. For now we only highlight emission hotspots within the recipe notes together with suggestions how to mitigate those emissions.
Remember that while these calculations provide valuable insights into the environmental impact of recipes, they represent estimates based on standardized data and may not capture all nuances of individual food sourcing and processing methods.
2. Accounting for the Digital Carbon Footprint
In addition to the food-related emissions, we also include the digital carbon footprint generated during the creation and sharing of each recipe. As the use of generative AI and digital tools becomes increasingly common, their environmental impact can no longer be ignored.
2.1 The Impact of Generative AI
Training large AI models is highly resource-intensive, as highlighted by numerous studies since 2019 [4]. With the rapid development and deployment of ever more sophisticated models, the energy and hardware demands of AI continue to soar. Generative AI is now a commodity, integrated not only into dedicated apps like ChatGPT but also into everyday tools such as search engines and office software. While the carbon footprint of a single AI query is small, the sheer volume of global queries means that the cumulative impact is significant—potentially rivaling the emissions of entire industries.
Efforts are underway to mitigate these impacts, including the development of more efficient models (such as DeepSeek), improvements in hardware and data center efficiency, and increased use of renewable energy. However, these advances are often outpaced by the exponential growth in AI usage.
On the positive side, generative AI can also help reduce emissions in other sectors by optimizing processes, improving logistics, and supporting sustainable innovation. For example, using AI to develop recipes, write stories, and generate images for this blog reduces the need for repeated cooking, photography sessions, and manual content creation, which would otherwise consume more time and resources. However, it is important to note that, since this blog would not exist without generative AI, these avoided emissions are theoretical.
2.2 Estimating the Digital Footprint
We account for the digital emissions of each recipe as follows:
- Generative AI queries:
- Each query (text or image) is estimated to produce 2.2–5 grams of CO₂e, depending on the model and complexity [5,6].
- For each recipe, we typically use:
- 1 query for the recipe itself
- 5–10 queries for the story
- 2–3 queries for the introduction
- 1 query for the recipe description
- 2–3 queries for the review
- A few queries for notes and image alt texts
- Up to 10 queries for food photography
- This results in a total of 17–33 queries per recipe, or 0.037–0.165 kgCO₂e. We use an average value of 0.1 kgCO₂e per recipe.
- Content creation and research:
- Writing, researching, and creating illustrations for a blog post typically takes about 1 hour on an iMac M1, consuming around 60 Watts of electricity.
- With the 2024 German energy grid emission intensity of 0.321 kgCO₂e/kWh [8], this adds 0.019 kgCO₂e per recipe.
- Total production footprint:
- The sum of generative AI and content creation emissions is 0.119 kgCO₂e per recipe.
- Allocation per user:
- Assuming each recipe is cooked by 10 people over its lifetime, the per-user production footprint is 0.012 kgCO₂e.
- Website usage:
- Loading a recipe page produces approximately 0.27 grams of CO₂e [13].
- The additional energy used by mobile devices to display the page is considered negligible.
2.3 Summary
The total digital footprint per meal is calculated as
\texttt{Digital Carbon Footprint} = \frac{\texttt{Production Carbon Footprint} + \texttt{Use Phase Carbon Footprint}}{\texttt{Number of portions}} \approx \frac{0.0123\texttt{kgCO2e}}{\texttt{Number of portions}}This value is included in the meal carbon footprint displayed with each recipe, providing a transparent and comprehensive view of the environmental impact of both the food and the digital processes behind it.
3. Estimating the Transportation Carbon Footprint
Transportation from the shop to your home is another important factor in the overall carbon footprint of a meal. The greenhouse gas (GHG) emissions associated with this step depend on the mode of transport, the distance traveled, and the amount of groceries purchased.
3.1 Emission Factors by Transportation Type
| Transportation Type | Emission factor (kgCO₂e/km) |
|---|---|
| Walking/Bike | 0.0 |
| Motorbike | 0.1137 |
| Bus | 0.1085 |
| Train | 0.0045 |
| Battery Electric Vehicle (BEV) | 0.0435 |
| Petrol Car | 0.1645 |
| Diesel Car | 0.1698 |
These emission factors are sourced from the DEFRA/BEIS 2024 database [7], which is based on UK data. For electric vehicles (BEVs), the actual value depends on the local energy mix and may vary by country or region.
3.2 Calculating Your Transportation Footprint
Since most shopping trips involve buying more than just the ingredients for a single meal, it’s important to allocate emissions proportionally. Bulk shopping is generally more environmentally friendly than frequent trips for small quantities. The formula to estimate your transportation carbon footprint for a meal is:
\texttt{Transportation Carbon Footprint} = \texttt{Emission factor in kgCO2e/km}\times \texttt{Distance in km}\times\frac{\texttt{Weight of one portion in kg}}{\texttt{Weight of purchased goods in kg}}The weight of one portion can be found on the nutrition labels displayed below each recipe. If tap water is added (e.g., for soups), subtract this from the total.
3.3 Example Calculation
Suppose you do your weekly grocery shopping with an electric vehicle (BEV) at a supermarket 10 km away (20 km round trip), and you purchase 15 kg of goods. The meal you prepare has a portion weight of 0.5 kg. In Germany (2024), the average grid emission intensity is 0.321 kgCO₂e/kWh [8], and your EV consumes 16 kWh/100 km. Assuming a 5% charging loss, the calculation is:
\texttt{Transportation Carbon Footprint (BEV)} = \texttt{Emission factor in kgCO2e/kWh}\times \frac{\texttt{Vehicle power consumption in kWh/km}}{1-\texttt{Loss rate}}\times\texttt{Distance travelled in km}\times\frac{\texttt{Weight of one portion in kg}}{\texttt{Weight of purchased goods in kg}} = 0.321\texttt{kgCO2e/kWh}\times\frac{0.16\texttt{kWh/km}}{0.95}\times 20\texttt{km}\times\frac{0.5\texttt{kg}}{15\texttt{kg}} = 0.036\texttt{kgCO2e}This is about 5–10% of the footprint of a typical vegan meal. If you lived in Sweden, where the energy mix is much greener, the same trip would result in only 0.002 kgCO₂e—less than 1% of a vegan meal.
Location- vs. Market-Based Emissions
- Location-based approach: Assumes the electricity used to charge your EV comes from the local grid mix (renewables, fossil fuels, nuclear, etc.).
- Market-based approach: If you, for example, have a contract for 100% renewable energy, the emission factor is 0 kgCO₂e/kWh, resulting in zero transportation emissions. However, the electricity would still come from the grid.
- Solar charging: If you charge exclusively with your own solar panels, both approaches yield zero emissions.
Comparing Vehicle Types
If you use a diesel car instead, the calculation is:
\texttt{Transportation Carbon Footprint (Diesel)} = \texttt{Emission factor in kgCO2e/km}\times\texttt{Distance travelled in km}\times\frac{\texttt{Weight of one portion in kg}}{\texttt{Weight of purchased goods in kg}} = 0.1698\texttt{kgCO2e/km}\times 20\texttt{km}\times\frac{0.5\texttt{kg}}{15˜\texttt{kg}} = 0.113\texttt{kgCO2e}This is substantially higher than the footprint for an electric vehicle.
3.4 Considering Vehicle Production (Supply Chain Emissions)
The DEFRA/BEIS emission factors [7] only account for fuel or energy use, not the production or end of life of the vehicle itself. Supply chain emissions (scope 3) from vehicle manufacturing of an owned or leased vehicle can be significant [9]:
- Internal combustion engine vehicle (ICEV): ~6.6 tCO₂e over 200,000 km lifetime
- Electric vehicle (including battery): ~10 tCO₂e over 300,000 km lifetime
This adds roughly 0.03 kgCO₂e/km for either vehicle type. For the above example:
\texttt{Car production emissions} = 0.03\texttt{kgCO2e/km}\times 20\texttt{km}\times\frac{0.5\texttt{kg}}{15\texttt{kg}} = 0.02\texttt{kgCO2e}3.5 Summary
Transportation emissions are highly individual and depend on your location, shopping habits, and vehicle type. For this reason, transportation is not included in the recipe’s default carbon footprint calculation. However, you can use the guidance and formulas above to estimate your own transportation footprint and make more informed, sustainable choices.
4. Estimating the Use Phase Carbon Footprint
The use phase carbon footprint covers all emissions generated during the preparation, storage, and cleanup of your meal at home. This includes the energy consumed by kitchen appliances, storage devices, and cleaning equipment. The actual footprint depends on many factors: your location, the age and efficiency of your appliances, your energy source, how much food is wasted, and how long and how you store your food.
4.1 Average Appliance Electricity Use
| Equipment | Energy Use (kW) |
|---|---|
| Cooking stove top | 1.5 |
| Oven | 2.4 |
| Water heater | 4.0 |
| Microwave | 1.2 |
| Toaster | 1.2 |
| Rice cooker | 0.86 |
| Air fryer | 1.5 |
| Refrigerator/Freezer combi | 0.02 |
| Dishwasher | 1.8 |
These values are sourced from the Energy Use Calculator [10] and the Reishunger website [11] for the digital rice cooker. They provide a basis for estimating the energy consumption of typical kitchen activities.
4.2 Example Calculation
The use phase may include several activities: storing ingredients, washing, cooking, cleaning, and storing leftovers. Here’s a detailed example:
| Activity | Energy Calculation | Allocation | Emissions Calculation |
|---|---|---|---|
| Storage in fridge after purchase (2 days) | 2 × 24 h × 0.02 kW = 0.96 kWh | Fridge occupation: 5% | 0.96 kWh × emission factor × 0.05 |
| Washing rice and vegetables with warm water (natural gas heating) | 3 l water × 25°C increase × 4.18 kJ/kg°C = 313.5 kJ | Efficiency: 85% | 313.5 kJ × 0.000032 m3/kJ × 1.94 kgCO2/m3 × 1.17 = 0.023 kgCO2 |
| Cooking rice in rice cooker (30 min) | 0.5 h × 0.86 kW = 0.43 kWh | 0.43 kWh × emission factor | |
| Cooking in wok (20 min) | 0.33 h × 1.5 kW = 0.5 kWh | 0.5 kWh × emission factor | |
| Freezing leftovers for a week | 7 × 24 h × 0.02 kW = 3.36 kWh | Fridge occupation: 1% | 3.36 kWh × emission factor × 0.01 |
| Heating leftovers in microwave (10 min) | 0.167 h × 1.2 kW = 0.2 kWh | 0.2 kWh × emission factor | |
| Cleaning dishes in dishwasher (2 h) | 2 h × 1.8 kW = 3.6 kWh | Dishwasher occupation: 50% | 3.6 kWh × emission factor × 0.5 |
Example Results
- Location-based (Germany, 2024):
Total use phase emissions = 0.99 kgCO2e / number of portions (emission factor = 0.321 kgCO2e/kWh) - Location-based (Sweden, 2024):
Total use phase emissions = 0.056 kgCO2e / number of portions (emission factor = 0.018 kgCO2e/kWh, electricity-heated water) - Market-based (100% renewables, Germany):
Total use phase emissions = 0.023 kgCO2e / number of portions (emission factor = 0)
4.3 Key Insights
- The use phase carbon footprint can be significant—sometimes comparable to the food emissions themselves—depending on your location and energy source.
- Ovens and dishwashers are typically the largest contributors due to their high power consumption and long operating times.
- Real-time carbon intensity of electricity can be checked on platforms like Nowtricity [8] or Electricity Maps [12]. Running appliances when the grid is greener (e.g., sunny afternoons) can further reduce your footprint.
4.4 Summary
The use phase carbon footprint is highly individual, influenced by your appliances, habits, and local energy mix. For this reason, it is not included in the default recipe carbon footprint. Additionally, food waste at home is also not included in the default calculation. We encourage you to avoid food waste whenever possible—check out the sustainability tips included with each recipe for practical ideas on how to make the most of your ingredients and minimize your environmental impact. The guidance and formulas above allow you to estimate your own use phase emissions and make more sustainable choices in your kitchen.
References
[1] Agribalyse Dataset: Cornelus, Mélissa; Auberger, Julie; Rimbaud, Audrey; Ceccaldi, Mathilde, 2024, “AGRIBALYSE® version 3.2”, https://doi.org/10.57745/XTENSJ, Recherche Data Gouv, V7
[2] CarbonCloud ClimateHub, https://apps.carboncloud.com/climatehub/
[3] Aqua-Calc, https://www.aqua-calc.com/calculate/food-volume-to-weight
[5] https://smartly.ai/blog/the-carbon-footprint-of-chatgpt-how-much-co2-does-a-query-generate
[6] Tomlinson, B., Black, R.W., Patterson, D.J. et al. The carbon emissions of writing and illustrating are lower for AI than for humans. Sci Rep 14, 3732 (2024), https://www.nature.com/articles/s41598-024-54271-x
[7] DEFRA emission factors: https://www.gov.uk/government/publications/greenhouse-gas-reporting-conversion-factors-2024
[8] Nowtricity: https://www.nowtricity.com
[9] Regett, A., Mauch, W., Wagner, U., Carbon footprint of electric vehicles – a plea for more objectivity. Forschungsstelle für Energiewirtschaft e.V. https://www.ffe.de/wp-content/uploads/2020/10/Carbon_footprint_EV_FfE.pdf
[10] Energy use calculator: https://energyusecalculator.com/
[11] https://www.reishunger.de/wissen/article/676/stromverbrauch-von-reiskochern
[12] Electricity Maps: https://app.electricitymaps.com/map/72h/hourly
[13] Website Carbon Calculator https://www.websitecarbon.com
