Case Study: A Corporate GHG Inventory in the Food Sector

We partnered with a US-based producer of organic, plant-based frozen desserts to test drive our hybrid methodology for compiling a corporate greenhouse gas (GHG) inventory. This is the partner company’s first GHG inventory, intended to establish a baseline for annual emissions and a basis for potentially offsetting those emissions.

The hybrid methodology combines two kinds of life-cycle inventory data to quickly and efficiently produce a GHG inventory using CarbonScope:

  • All of the emissions in scope 1 (direct fuel combustion) and scope 2 (purchased electricity) are modeled using our process LCI (PLCI) database, which converts physical quantities of energy use into GHG emissions.
  • The wide range of emissions in scope 3 are modeled largely using our environmentally extended input-output LCI (EEIOLCI) database, which converts dollar amounts from purchase records into equivalent GHG emissions based on the industry sector and distribution channel.

Corporate structure

The company is classified as a small company (less than 100 employees and/or less than $50M in annual revenue). The company outsources its product manufacturing to two co-packers both of whom are outside of the company’s organizational boundary on an operational control basis, so the entirety of the co-packer emissions will be categorized as scope 3.

Activity data

The company provided activity data for this baseline inventory based on their operations in 2019, using our standard data template for hybrid GHG inventories. The actual data is confidential, but here is an example of what the input data might look like for a hypothetical company. The activity data typically includes:

  • Fuel and electricity consumed in company operations – in physical units such as kWh, gallons, etc.
  • Goods and services purchased (inflows) – in 2019 US dollars excluding taxes
  • Waste management services purchased– in 2019 US dollars excluding taxes
  • Freight transport services purchased– in 2019 US dollars excluding taxes
  • Warehousing and storage services purchased– in 2019 US dollars excluding taxes
  • Energy used in processing and use of sold products– in physical units such as kWh, gallons, etc.
  • Employee travel and commuting—in passenger miles or km

The larger of the two co-packers was able to provide their detailed activity data separately. The smaller co-packer was modeled as a supplier providing frozen dessert manufacturing services within the company’s purchased goods and services.

The company sources 100% of its electricity from renewable sources supplied by the local utility. The larger co-packer produces about 3.2% of its electricity using on-site solar, with the remainder sourced from the local grid.

Some of the plant-based ingredients used in the dessert products are imported, which are modeled using domestic production as proxy by assuming that imported commodities have the same input structure and the same production characteristics as comparable products of equal value produced domestically (see this methodology note). While all of the purchased ingredients are organically produced, the inventory uses emission factors for industry sectors as a whole without distinguishing between conventional and organic production. This can be justified by the fact that organic farming does not necessarily have lower carbon footprints than conventional farming when systems are in steady state.

Inventory results

The results of the GHG inventory analysis show that the company’s operations from all three emission scopes amount to 8387 metric tonnes of CO2e for the year 2019. The table and chart below show a detailed breakdown (units: Carbon in Kg CO2e; Energy in MJ; Water in L).

A sensitivity analysis shows that the natural gas used by the larger co-packer is by far the largest contributor to the corporate GHG emissions, accounting for 49% of the total emissions and reported under scope 3.3. The next four contributors are the ingredients sourced by the larger co-packer (16.6% of total emissions; scope 3.1), manufacturing services provided by the second co-packer (9% of total emissions; scope 3.1), truck transport services used to ship finished products (7.6% of total emissions; scope 3.9), and cardboard containers used to ship finished products (3.8% of total emissions; scope 3.1). Overall, scope 3 emissions account for over 99% of the total emissions and 72% of the total emissions are attributable to the larger co-packer.

For a hypothetical company with revenue at the midpoint of the range for the company size, the GHG emissions intensity is 0.36 Kg CO2e per dollar of revenue. This intensity is less than 50% of industry average for the ice cream and frozen dessert sector in the US.


We said at the outset that our hybrid methodology provides for a quick and efficient way to compile a corporate GHG inventory.  Here is a quick accounting of the time and effort that went into generating the inventory:

  • About 2.5 weeks of work by the company’s sustainability team to collect activity data from within the company and from the co-packers.
  • A few hours to automatically import the data into CarbonScope, generate results, and review/interpret the results.

With a baseline GHG emissions inventory established using a simple and well-understood methodology, annual updates to the inventory are expected to take less time/effort and turn into a routine accounting task.

Case Study: Hydration Options

Think before you drink: The carbon footprints of four different hydration options

Susan Cholette and Hoa Nguyen

Project summary

Hydration is a necessity, and the growing consumer shift away from soft drinks towards water should please dentists and physicians alike.  However, as worldwide demand has surpassed half a trillion bottles per year, single use plastic bottles are not the healthiest choice for the planet.  We compare several scenarios for quenching thirst on the go, and show how our purchasing habits can make a substantive impact.  

Systems modeled in this study

We evaluate four different ways to provide a thirsty Bay Area consumer a half-liter of water:  imported bottled water, more locally sourced water bottled in both virgin and 100% recycled PET bottles, and a reusable container that can be refilled as needed.

The following figures illustrate the supply chains for a 500ml bottle of Evian, imported from Switzerland, and Arrowhead™, which sources from Californian springs. According to their website all of Arrowhead™’s individually sized bottles sold in California are currently comprised of a 50/50 mix of virgin and recycled PET, but they indicate their bottle design can support use of 100% recycled content and will eventually.  Some other drink companies, such as Snapple™ have recently re-designed their bottles to use 100% recycled content.  We consider both extremes for recycled content- 0% and 100%- to illustrate the relative impact that recycling has, and we assume that the recycled PET is sourced from the same location as virgin PET. 

Figure 1:  System diagram for a 500ml bottle of Arrowhead™ water

Figure 2:  System Diagram for a 500ml bottle of Evian™ water

While consumers have many options for reusable containers, we select a Nalgene™ bottle to keep within the same family of materials, as it is made of Tritan™, a popular form of PETG plastic.  Unlike the prior three scenarios, where the functional unit is a half liter bottle of water, the functional unit is just the Nalgene™ bottle itself, as it is purchased empty and then filled at home or at a drinking fountain, as shown in Figure 3. We assume that no additional filtration or treatment is used.

Figure 3:  System diagram for a Nalgene™ reusable container

All four scenarios share the same system boundary, cradle-to-grave, where we assume that the bottles are trucked to landfill once discarded, as only about 30% of plastic containers are recycled in the US.

LCA tool and LCI database

We used our carbon modeling tool, CarbonScope, to conduct the LCAs in this project. The life-cycle inventory (LCI) database underlying the analysis is CarbonScopeData.


The three life-cycle impact categories that can be quantified are embodied carbon (kg CO2e), embodied energy (MJ) and embodied water (L), but to keep the comparisons simple, we report only embodied carbon.  Other studies published elsewhere discuss the additional problem of landfill usage and pollution from bottles that escape proper disposal.

Figure 4 shows the relative impacts of each of the bottles and the contribution for each stage towards embodied carbon.  The domestically sourced water bottled in 100% recycled PET has the lowest footprint, a 30% reduction over the virgin PET bottle.  Imported water has almost three times the footprint, thanks to the international transportation required.  Figure 4 also shows that distance trumps recycled content: even if we were to buy an imported brand bottled in 100% recycled materials, it is clear that it would have more embodied carbon than the domestic water bottled in virgin PET. The reusable container has the most embodied carbon due to its greater weight, more energy-intensive material, and the need to transport it across the country.

Figure 4:  Embodied carbon associated with a single use of each of the four bottles

Of course, it would be silly to purchase a reusable container for one time use.  Figure 5 illustrates the how cumulative carbon footprint increases with the number of uses. We include the .001 kg CO2e associated with pumping and treating tap water, a miniscule impact that effectively results in a flat line for the Nalgene™ reusable bottle no matter how many times it is refilled.  

Figure 5: Cumulative GHG Impact: Consumption of 500ml of Water

The initial investment in a reusable bottle pays off quickly: we need only use it three times over the purchase of imported water to accumulate less embodied carbon. We would need to reuse a Nalgene™ bottle just more than six or nine times to have a lower footprint than domestically sourced water bottled in virgin or recycled PET.  Given that such containers sell for $10 or more, it is likely that we would break even environmentally before we do financially.  

While metal or glass containers will have different footprints, the environmental benefit of using reusable drinking containers will be even more advantageous than it is for shopping bags, with one study showing it may take more than 170 uses to offset the investment in a cotton bag over the typical HDPE bags provided at checkout. This is understandable since the transportation of water is inherently emissions intensive. For example, even though the domestic water is sourced from relatively nearby springs, the transport of the water comprises over 20% of the total footprint, while the pumping and treatment of the water is less than 1%.  Other domestic brands that use out of state water sources will have a higher transportation footprint.


In summary, the best choice for hydration is to develop a habit of bringing along a reusable container.  If that is not an option, then buy a brand of bottled water that is more locally sourced, as distance has a larger impact than the percent of recycled content in the bottle.  Imported water should be consumed sparingly, regardless of how it is packaged. Thankfully, imported bottled water has a small and shrinking share of the overall market for still water: only one brand (Fiji) is represented in the top brands that comprise 75% of the US market, and it has a small (3%) share.

 Table 1: What-if analysis for alternate purchasing behavior

Americans currently purchase an average of 36.5 gallons of bottled water annually, about 275 500ml bottles.  While some purchases of single-use water bottles may be necessary, if 200 million Americans obtained a reusable container and replaced half of their yearly bottled water purchases by refilling these containers at taps or drinking fountains, Table 1 shows we could avoid over 1.3 million metric tons of CO2e emissions annually.  While this represents only 0.025% of the national annual emissions, this would be a relatively painless behavior to adopt.  Small drops do add up.

Case Study: Sourdough Crackers

Project summary

Regenerative agriculture utilizes techniques like conservation tillage, cover crops and crop rotation to build up and sequester carbon in the soil. If yields can be maintained similar to conventional agriculture, then this could be the future of food production. So when a startup snack food manufacturer asked us to quantify the carbon benefits of their crackers made with wheat and other ingredients produced using organic/regenerative methods, we quickly leveraged our tools and database to put together a  life cycle assessment (LCA) study comparing crackers made with conventional and organic inputs.

System modeled in this study

The diagram below illustrates the supply chain for a packaged box of sourdough crackers. 

The functional unit is a packaged box of crackers (net weight: 4 oz). The system boundary is cradle-to-gate from the perspective of the cracker manufacturer. We used the same supply chain to compare a box of crackers made with conventionally produced wheat and sunflower seed (these are the two primary agricultural inputs to the cracker production) vs. a box of crackers made with organically produced wheat and sunflower seeds. We had actual secondary data for the conventional production systems in our life cycle inventory (LCI) database. We modified these conventional systems to create hypothetical LCA models for organic production systems in order to quantify the benefit of switching to organics. 

Organic farming systems are generally characterized by the types of inputs used, such as fertilizers and pesticides that are not synthetically produced and are non-toxic. Based on information from the actual farms that produce the organic ingredients considered in this study, we know that these organic systems also use regenerative farming methods to build and preserve soil carbon. Therefore, we refer to these systems as “organic/regenerative” in this report (or just “organic” for short).

General assumptions

An initial LCA study often starts with a number of assumptions and placeholders that are later refined using higher-quality data. For the two cracker production systems modeled in this study, the data available initially included the list of ingredients (including the weights of ingredients), the production and processing locations of the ingredients, and the co-packer location where the final production and packaging take place. We used secondary data from our LCI database to model the production and processing of the conventionally produced ingredients.

In addition, we made several assumptions and used a number of placeholders as documented below in order to fill in the necessary preliminary data for organic production, waste disposal, processing, packaging, and various minor ingredients:

  • For the organic/regenerative farming systems, the on-farm energy use and yield are assumed to be the same as conventional systems for which we have data in our LCI database. Past LCA studies (including one of our own) have shown that organic farming systems often suffer from lower yields relative to a conventional system; however, farming techniques are improving and we expect our assumptions to be replaced with actual data from the specific farms in a future iteration of this LCA study.
  • For the organic/regenerative farming systems, the primary fertilizer is assumed to be manure-based compost supplying the same total nitrogen per acre annually as the synthetic fertilizers used in a comparable conventional system. We elaborate on this further in the “Farm-level LCAs” section below.
  • For the baking process at the co-packer facility, crackers are assumed to use comparable energy as bread rolls for which we have data in our LCI database.
  • Sunflower oil production is assumed to be located near the seed farm, and the waste is composted close to the production facility.
  • Although extensive, the LCI database has a finite number of entries, so substitution of products with similar profiles follow: we swapped malt for the yeast extract, acetic acid for lactic acid (since they have similar fermentation processes), and soda powder for baking powder.  Each of these substitutions are for an ingredient that comprises 1% or less of the formulation.  
  • Packaging material is assumed to consist of a 29.08g paperboard carton and a 5g polyethylene (HDPE) plastic liner.

LCA tools and LCI database

We used our new carbon modeling tool, CarbonScope, to conduct the two product LCAs in this project. FoodCarbonScope was used for the farm-level LCAs of the hypothetical organic/regenerative wheat and sunflower seed crop systems. The LCI database underlying the analysis is CarbonScopeData.

Farm-level LCAs

Since we did not have detailed production data for the two primary agricultural inputs used in the manufacture of the crackers — organic wheat and organic sunflower seeds — we created hypothetical organic/regenerative systems based on the conventional production systems for which we had data in CarbonScopeData. We assumed that the energy use and yield per acre would remain the same between conventional and organic systems, and the primary differences would be in the fertilizer application and soil carbon sequestration. The organic farms are assumed to use manure-based compost, based in part on information received from one of the farms, as the primary fertilizer (NPK percentages =  1.5:1:1.5). 

We made a few additional and reasonable assumptions about compost:

  • Compost mineralizes and releases 10% of the total compost nitrogen per year.
  • Compost has been applied long enough that sufficient amounts of total mineralized nitrogen,  phosphorus and potassium are available to the crops each year.
  • Compost added each year must supply the equivalent of 20% of the synthetic nitrogen through mineralization; the other 80% would come from previous years’ compost applications.

The table below summarizes the key parameters used in the farm-level LCAs of the two commodities. The wheat farming system is considered “transitional” because it is still within a 20-year period following a switch from conventional to organic production. The sunflower seed farming system is considered to be in “steady-state” because the organic production was established more than 20 years ago. The climate and moisture regimes were set based on the locations of the farms. The soil type was set based on information supplied by the farms. The last column in the table sets the carbon inputs to soil as one of four discrete levels (low, medium, high, high-organic), and we chose the “high” level based on our understanding of the farming systems. These parameters are based on IPCC tier 1 methodology, and are used in the farm-level LCAs by FoodCarbonScope to estimate changes in soil carbon during the transition from conventional to organic/regenerative farming.

Of the two crop systems, only the wheat gets credit for soil carbon sequestration as it transitions from a conventional system to an organic/regenerative system. Soil carbon generally increases during such a transition as illustrated below and is calculated using the parameters summarized above. The numbers in the diagram below are for illustration only and do not represent the actual systems modeled in this study. 

The table below summarizes the cradle-to-farmgate LCA results for the two agricultural commodities considered in this study:

Product LCA results

The table below compares the life-cycle impacts of a box of crackers made with organic ingredients produced using regenerative methods vs. crackers made with conventionally produced ingredients. Organic production results in 30% lower greenhouse gas emissions (quantified as Kg CO2e), and all of this is attributable to the difference in agricultural methods. This difference arises from the “Inflows” category which includes the purchased agricultural commodities. The LCA results clearly show the benefits of switching to organic/regenerative production, but with the caveat that credit for soil carbon accumulation can only be taken during the transition period.

The other interesting insight from the results is that packaging dominates the life-cycle impacts of the products. When the ingredients are plant-based with fairly low carbon emissions, transportation and packaging can sometimes take on an outsized role. The table below shows the top four contributors to carbon emissions in the life cycle of the organic cracker product. Two of these are the packaging materials. The paperboard used to make the carton contributes about half of the total carbon footprint of the finished product. 


This LCA study has demonstrated the significant potential for reducing the cradle-to-gate greenhouse gas emissions of the client’s cracker product by sourcing agricultural commodities from farms that are using organic/regenerative methods. It should be noted that the actual supply-chain data for the production was supplemented with several assumptions and placeholders, so these results should be treated as the first step in the process of quantifying and optimizing the climate impacts of the cracker products.

Case Study: Coffee Drinks

Project summary

Americans lead the world in coffee consumption with 400 million cups of coffee consumed per day. Per-capita coffee consumption is even higher in European countries. With coffee consumption comes carbon emissions, and we thought it would be interesting to do a life-cycle comparison of the environmental impacts of three popular coffee drinks that you could order at a neighborhood cafe: a latte with 2% milk, a latte with soy milk, and a cappuccino with 2% milk.

Systems modeled in the study

The diagrams below illustrate the supply chains for both lattes modeled in this study. Milk is sourced relatively locally in half-gallon containers, whereas soy milk is produced more distantly and in smaller containers.  The cappuccino’s ingredients and sourcing are inherently similar to that for the latte with 2% milk, but just with less milk added and more power used for frothing the milk. The functional unit for the LCA is a 12oz latte and a 6oz cappuccino. The cappuccino has 6oz less milk in it, but otherwise has the same coffee/water ratio. Although we could have modeled the cappuccino with a shorter cup, we decided to use the same cup size for all drinks because coffee cups of different sizes are surprisingly similar in weight. The system boundary is cradle-to-grave in all cases, with the to-go cup disposed of at the cafe.

Figure 1: System diagram for a to-go latte with 2% milk
Figure 2: System diagram for a to-go latte with soymilk

LCA tool and LCI database

We used our new carbon modeling tool, CarbonScope, to conduct the LCAs in this project. The life-cycle inventory (LCI) database underlying the analysis is CarbonScopeData.


The three life-cycle impact categories considered in this study are embodied carbon (Kg CO2e), embodied energy (MJ) and embodied water (L). We first consider the embodied carbon associated with a Latte made with 2% milk, grouping all processes as associated with the primary ingredient.   Figure 3 shows that the largest component is from the milk, with the second largest impact from the packaging.   Despite being sourced from afar, coffee’s impact is  relatively small.    

Figure 3: Embodied carbon by process

Replacing to-go materials with one’s own trusted travel mug would allow one to enjoy a free latte every 6th time from a carbon equivalence perspective, if we ignore the impact of producing and then repeatedly washing the mug.  But for now we will for now focus on the highest impact ingredient: the milk.  Our first alternative replaces the 2% milk with soymilk, and the second is the cappuccino, which uses only 3 ounces of milk instead of the 9 used by a latte.

The table below summarizes the LCA results.  While a latte with soy milk has less embodied water than milk, there is no effective difference in the embodied carbon.  Part of this is due to the longer supply chain for soymilk, the greater share of the smaller package, and the inherent energy intensity of soymilk production.  Other alternative milks would have different impacts.  

We then consider a different drink for the third alternative.  The cappuccino shows the significantly lower environmental impacts from consuming less milk, only having 55% of the embodied carbon and, not-surprisingly, one-third the water.  The increased energy usage from frothing the milk is more than offset by reducing the milk used.

Overall, this study shows that milk (or soymilk) dominates the carbon footprint of a latte. The coffee itself is a minor contributor. So habitual latte drinkers should not fret about their coffee addiction (at least from an environmental perspective), but they might consider switching to a less milky drink, and everyone can consider bringing back reusable mugs when it is again safe to do so.

Case Study: Organic vs Conventional Farming

organic farming

Project summary

Given the growing importance of organic food production, there is a pressing need to understand the relative environmental impacts of organic and conventional farming methods. This study applied standards-based life cycle assessment (LCA) to compare the cradle-to-farm gate greenhouse gas emissions of 12 crop products grown in California using both organic and conventional methods.

Systems modeled in the study

We modeled 12 different crops produced using both organic and conventional methods in California. The system boundary was defined as cradle-to-farmgate. In addition to analyzing steady-state scenarios in which the soil organic carbon stocks are at equilibrium, this study modeled a hypothetical scenario of converting each conventional farming system to a corresponding organic system and examined the impact of soil carbon sequestration during the transition.

LCA tool and LCI database

We used our comprehensive food/agriculture LCA tool, FoodCarbonScope, to conduct the LCAs in this project. The life-cycle inventory (LCI) database underlying the analysis is CarbonScopeData.


The results showed that steady-state organic production has higher emissions per kilogram than conventional production in seven out of the 12 cases (10.6% higher overall, excluding one outlier). Transitional organic production performed better, generating lower emissions than conventional production in seven cases (17.7% lower overall) and 22.3% lower emissions than steady-state organic. The results demonstrated that converting additional cropland to organic production may offer significant GHG reduction opportunities over the next few decades by way of increasing the soil organic carbon stocks during the transition. Non-organic systems could also improve their environmental performance by adopting management practices to increase soil organic carbon stocks.

Cradle-to-farm gate GHG emissions for conventional (steady-state), organic (steady-state) and organic (transitional) production

Detailed report

Full text:

Detailed LCA reports available here as “supplemental”:

Case Study: US Food Waste

WASHINGTON, DC - AUGUST 2: Food waste material processed by Compost Cab workers to create compost at Howard University Community Compost Cooperative on Wednesday, August 2 , 2017, in Washington, D.C. (Photo by Salwan Georges/The Washington Post via Getty Images)

Project summary

This pioneering study analyzed the climate change and economic impacts of food waste in the United States. Using loss-adjusted national food availability data for 134 food commodities, it calculated the greenhouse gas emissions due to wasted food using life cycle assessment (LCA) and the economic cost of the waste using retail prices.

Systems modeled in the study

We modeled a total of 134 food commodities using average US production data for each of them. The system boundary was defined as cradle-to-grave. The life-cycle model of material flow through the food system is depicted in Figure 1.

LCA tool and LCI database

We used our comprehensive food/agriculture LCA tool, FoodCarbonScope, to conduct the LCAs in this project. The life-cycle inventory (LCI) database underlying the analysis is CarbonScopeData.


The analysis showed that avoidable food waste in the US exceeds 55 million metric tonnes per year, nearly 29% of annual production. This waste produces life-cycle greenhouse gas emissions of at least 113 million metric tonnes of CO2e annually, equivalent to 2% of national emissions, and costs $198 billion.

Detailed report

Case Study: Meat Eater’s Guide

Project summary

Environmental Working Group (EWG) is an influential non-profit organization that has published groundbreaking research on environmental health. Several years ago, they asked us to use our LCA tools, database and expertise to generate definitive results on the relative carbon footprints of a wide range of food commodities.

Systems modeled in the study

We modeled the full cradle-to-grave life cycles of 25 major food commodities, including typical food waste and cooking. Overall we conducted life-cycle assessments (LCAs) on 53 actual product systems in order to calculate average life-cycle impacts for the 25 commodities. This includes meats such as beef, lamb, poultry and seafood. The list also includes dairy products such as milk, yogurt and cheese, and plant-based foods such as beans, rice, vegetables and tofu.

LCA tool and LCI database

We used our comprehensive food/agriculture LCA tool, FoodCarbonScope, to conduct the LCAs in this project. The life-cycle inventory (LCI) database underlying the analysis is CarbonScopeData.


This graphic from EWG summarizes the relative carbon footprints of food commodities in terms of car miles driven.

Eat Smart Chart. Eat smart your food choices affect the climate

Detailed report

Case Study: Cycling Jerseys

Project summary

Peloton de Paris is Belgium’s first cycling café and a small business that is thoughtful about their impact on the planet. They have created their own line of cycling apparel that they sell worldwide, and asked us to conduct a comparative life-cycle assessment (LCA) of two of their cycling jerseys. Both are polyester based, but one uses virgin polyester and the other is made from recycled polyester.

Systems modeled in the study

The diagrams below illustrate the supply chains for the two jersey product systems modeled in this study. The functional unit for the LCAs is one cycling jersey. The system boundary is cradle-to-warehouse in both cases.

Life Cycle Inventory and System Boundaries for Anibal/Matrix Jersey

Life Cycle Inventory and System Boundaries for Fez/GreenFly Jersey

LCA tool and LCI database

We used our new carbon modeling tool, CarbonScope, to conduct the LCAs in this project. The life-cycle inventory (LCI) database underlying the analysis is CarbonScopeData.


The three life-cycle impact categories considered in this study are embodied carbon (Kg CO2e), embodied energy (MJ) and embodied water (L). The table below summarizes the LCA results and shows the significantly lower environmental impacts from using recycled materials. The study also found that switching to ground transportation for delivering the jerseys to the warehouse in Belgium would further reduce the environmental footprints of the jerseys. 

Detailed report