Carbon and water footprint of coffee consumed in Finland—life cycle assessment

The system studied included the coffee supply chain from coffee cultivation to the use stage (Fig. 1). Inputs included into the coffee cultivation system were fertilizers, pesticides, fuels, lime, irrigation water and coffee plants. Fuel, electricity and water for primary processing (dry milling and wet processes) were included as well, but waste and side flow management (i.e. wastewater treatment or composting of processing side flows) were not. The transportation of all cultivation inputs, as well as coffee cherry transportation to primary processing and green coffee transportation to Vuosaari harbour in Finland, was included. The transportation from the harbour to the roasting facility in Finland was less than 1 km and was excluded. District heating and electricity used in the roasting was included. For the package production, the package material production for items such as granulates was included, as well as conversion processes including energy use, water use and material efficiency. The end-of-life was included by using a scenario approach which involved an incineration process with energy recovery. Electricity, water, cups, filters for coffee making and washing detergent production were all within the system boundaries in the coffee making and washing stage, as well as emissions from municipal biowaste treatment related to coffee brewing waste. Infrastructure, i.e. coffee machine production, was excluded in all stages.

The functional unit of the study is 1 l of consumed coffee. In addition, some results are presented per kg of green coffee and per cup (140 ml) of coffee.

Characterization factors for greenhouse gases were used according to IPCC 2019 (Myhre et al. 2013, Table 8.7) with climate carbon feedback: 1 for carbon dioxide, 34 for biogenic methane, 36.75 for fossil methane and 298 for nitrous oxide emissions .

The water scarcity impact was calculated according to the AWARE method by Boulay et al. (2018). The indicator quantifies the potential of water deprivation to either humans or ecosystems and it is based on the available water remaining per unit of surface in a given watershed relative to the world average, after human and aquatic ecosystem demands have been met. Values range from 0.1 to 100. The country-specific characterization factors used are presented in Table 1.

Both characterization methods are used also in European Commission Product Environmental Footprint methodology (Fazio et al. 2018).

For most relevant processes, primary data was obtained from the suppliers of Paulig Ltd. and its own operations. The consumer stage is based on literature data, as well as processes with only minor impacts. The data sources are explained in more detail below.

The greenhouse gas emissions due to changes in land use stem from a change in the carbon stocks on the land. According to the most recent international Life Cycle Assessment guidelines, such as the Product Environmental Footprint (PEF) (European Commission 2013) and PAS2050 (BSI 2011), greenhouse gas emissions related to changes in land use shall be assessed but need to be kept separate when communicating the results.

Greenhouse gas emissions due to land use changes in coffee cultivation were assessed according to the PEF (European Commission 2018) and PAS 2050:2011 (BSI 2011) and the supplementary document PAS2050-1:2012 (BSI 2012) on a national level as previous land use unknown. Changes in four land use categories were considered, namely cropland and perennial crops, grassland and forest land. The carbon stock factors are from IPCC (IPCC 2006), FAO’s Global Forest Resource Assessment (2010) and European Commission (2010). Worst case of weighted and normal average has been used as required in PAS2050.

The calculations were made using Direct Land Use Change Assessment Tool (Version 2013.1) by Blonk Consultants. The tool was updated with most recent data during the time of the study regarding the land use area of all crops, grassland and forest areas (1993–2015) from FAOSTAT (2018), and the replacement of other land use to cultivation area of coffee was assumed accordingly. The land use changes were estimated using a 3 years’ average for the three most recent years (2013–2015) and a 3-year average from 20 years back (1993–1995).

In Brazil, the cultivation area had increased in the observed period only modestly. In fact, for more than the last 10 years, the cultivation area has decreased, see Fig. 3. Thus, the additional greenhouse gas emissions were only 0.04 tCO₂-eq./ha/year.

In Nicaragua, the cultivation area had increased significantly at the beginning of the observed period (see Fig. 4). However, in the last 10 years, there has not been a significant increase, only annual variation. The additional greenhouse gas emission from land use changes was large and amounted to 5.2 tCO₂-eq./ha/year.

In Honduras, the cultivation area had increased significantly and pretty constantly in the observed period (see Fig. 5). The emissions from land use change were though not so high as in Nicaragua as part of the increased area came from land used for annual crops in addition to forest areas. However, the additional greenhouse gas emissions from changes in land use were still large at 3.1 tCO₂-eq./ha/year.

In Colombia instead, the cultivation area has decreased in the past 20 years, even though it has been increasing in the last few years, and thus, no emissions related to land use changes were allocated to coffee cultivation (see Fig. 6).

Making coffee at home included water use, coffee beans, filters (if used) and electricity used in two types of coffee machines; traditional coffee machine with a filter (drip-brew) and a French press (see inventory data in Supplementary material Table S2). For drip-brewing coffee makers, the average standby time (plate kept hot) is assumed to be 37 min (Humbert et al. 2009). There is a lack of information on liquid food waste in households in Finland, but according to a single survey by Luke/Hanna Hartikainen (personal comm.), as a baseline, it is assumed that on average, 1.25% of coffee is wasted by the consumer.

Some scenarios with different consumer behaviour at home were calculated (Supplementary material Table S3). For drip-brewed coffee, scenarios with increasing heating standby time up to 120 min instead of 37 min were calculated. For using a French press, some extra hot water may be used to warm up the French press and “pot heating” scenarios included this option. In the baseline scenario, food-waste was assumed to be 1.25%. However, much higher food waste assumptions are made in the literature (Chayer and Kicak 2015). Increasing the food waste rates up to 30% was calculated for both coffee machines.

Coffee making using automatic office coffee machines was also calculated. A comparison of automatic office coffee machines included three types of machines from two companies: two professional automatic coffee machines with a fridge and one without a fridge. Data on the consumption of coffee beans, energy, water and cleaning detergents was obtained from the manufacturing companies. Inventory data for basic black coffee is presented in Table S2 in the supplementary material. Coffee drinks with milk and sugar were also studied. Three scenarios were formulated:

The inventory data for these coffee drink options is presented in Table 4. In terms of the water scarcity impact, it was assumed that Finnish milk (Usva et al. 2019) and Danish sugar (EcoInvent) were used.

Dish washing data is presented in Table S2. Different usage times have been presented in different studies (Lighart and Ansems 2007; Refiller 2018). In this study, the usage time for a mug was assumed to be 3000 times, according to Lighart and Ansems (2007).

In the case of drip-brew coffee, the coffee grounds and the paper filter are disposed of after use. It is assumed that in Finland, 32% of kitchen biowaste is collected separately and 68% together with mixed waste. Mixed waste is incinerated and separately collected biowaste is managed by composting (50%) or used in biogas reactors (50%) (Silvennoinen et al. 2017).

The end-of-life scenario for package waste management was incineration, which was assumed to take place at the Vantaa Energia plant in the capital area of Finland. The energy recovery has been taken into account as recommended in the PEF instructions. The heat values for polyethylene (43 MJ/kg) and nylon (32 MJ/kg) (Shibasaki 2017; Tsiamis et al. 2016; Walters et al. 2000) were used to assess the amount of energy. According to personal communications with Laura Ikäheimo and Kalle Patomeri from Vantaan Energia, the recovery rate is 88%. It was assumed that the replaced energy consisted of:

The amount of produced carbon dioxide emissions was calculated based on the carbon contents of nylon and polyethylene.

The carbon footprint results for 1 l of drip-brewed black coffee without sugar originating from eight farms were calculated and the characterization was applied according to the IPCC 2013 (Myhre et al. 2013). The results are presented in Fig. 7 and they vary from 0.27 to 0.70 kg CO₂ eq/l coffee. Coffee cultivation accounts for the vast majority of the impacts and the significance of processing, packaging and transportation are negligible. The consumer stage also makes a significant contribution to the impacts of coffee.

The water scarcity footprint of 1 l of drip-brewed black coffee without sugar is illustrated in Fig. 8. Irrigation was the largest contributor to the results. For this reason, the irrigated coffee chains are presented separately from the non-irrigated one. The total amount of water consumed in the non-irrigated systems studied was about 8 l, and in irrigated systems, it came to 60 and 110 l per litre coffee, corresponding a water scarcity impact about 0.02 m³ eq/litre coffee for non-irrigated systems and from 0.15 to 0.27 m³ eq/litre coffee for irrigated systems.

The fertilization rate was the largest contributor to the carbon footprint (Fig. 7). Fertilizer manufacturing has a significant carbon footprint; in addition, nutrient use causes direct nitrous oxide (N₂O) emissions from soils. This can be seen especially in the carbon footprints of Brazil/Minas Gerais and Colombia farms having high N-fertilization rate compared with the yield.

A noticeable CO₂ emission source is liming, especially in the Brazil Cerrado and Minas Gerais areas where high levels of lime are applied annually (see Section 2.2.2). Other agricultural inputs include fuel in cultivation, if used, seedling production and pesticide production. The farms in Brazil had the highest impact on “other agricultural inputs” due to the higher number of machines and the diesel consumption compared with the other case farms.

The climate and water scarcity impacts of coffee roasting were less than 1% of the total impact of the value chain. The carbon footprint of packaging is less than 2% and the water scarcity impact is less than 0.5% of the total impact of the value chain. The reductions of emission in the end-of-life amounted to 3% of total carbon footprint, taking into account credits for obtained energy and emitted amounts of carbon dioxide from plastics incineration.

The water scarcity impact is naturally higher for the irrigating farms in Brazil (Fig. 8). The amount of irrigation water exceeds all other water consumption. Fertilizer manufacturing and wet processes in primary processing comprise most of the impacts in addition to irrigation.

Carbon footprints due to land use changes are presented on a national level for Nicaragua, Honduras, Colombia and Brazil in Fig. 9. In Colombia, in the observed period, no land use changes had occurred. In Brazil, they were very small at about 0.5%. For Honduran and Nicaraguan green coffee, the emissions from land use changes contributed 60–75% of the total carbon footprint.

The climate and water scarcity impacts of coffee making and consumption both in the office and at home per 1 l of black coffee without sugar are presented in Fig. 10. Green coffee from the Colombian case farm is selected here to represent the primary production.

When increasing the food waste rate to 30%, the total carbon footprint of 1 l of consumed coffee increases by about 27%. A scenario with 120 min of stand-by instead of 37 min increases the total carbon footprint of drip-brewed coffee by about 4%. The extra hot water to heat the French press increases the total impact by about 3% as well.

The office coffee machines varied only slightly from each other. The automatic coffee machine models with a fridge had higher electricity consumption which resulted in a higher carbon footprint in the coffee-making stage. The carbon footprints of drip-brewed coffee and French press coffee for consumers at home are lower due to smaller amount of ground coffee consumed per litre of coffee (see Table 4). The electricity consumption for dish washing accounted for the majority of the carbon footprint, and the water consumption for dish washing accounted for the majority of water scarcity impact in the consumer stage. The consumer scenario results are presented in detail in Figure S1 in the supplementary material.

The results of coffee drinks with milk and sugar are presented in Figures S2 and S3 in the supplementary material. Coffee drinks with milk have a higher carbon footprint because milk itself has a higher impact, and also because, in the case of automatic coffee machines, more coffee beans are used for the coffee drinks with milk, especially latte and cappuccino. In terms of the water scarcity impact, it is lower for Finnish milk than for coffee, and therefore, the homemade coffee drinks without milk, but more coffee instead, have higher water scarcity impacts than coffee drinks with milk.

Primary production is the hotspot of the coffee chain. In the study by Killian et al. (Killian et al. 2013), the share of farming was 1.02 kg (21%) CO₂ eq/kg green coffee. In our study the climate impact of farming varied from 1.1 to 6.6 kg (from 32 to 78%), indicating also a lot of variation between farming systems.

In contrast to many previous papers (e.g. Büsser and Jungbluth 2009; Hicks 2017; Humbert et al. 2009; Killian et al. 2013; Salomone 2003), the contribution of the consumption stage to the carbon footprint was smaller in this study. According to Killian et al. (Killian et al. 2013), the share of consumption was 2.15 kg (45%) CO₂ eq/kg of green coffee, and in our study, the consumption stage was 1.6 kg (from 19 to 49%). The differences concerning the consumption stage are mostly caused by the difference in electricity production types between Finland and other countries in Europe. In 2015, as much as 79% of the electricity produced in Finland was produced by renewables and nuclear. The share of renewables was 45% (Official Statistics of Finland2016). For example, the EU in 2016, renewable electricity was 29% and nuclear energy sources contributed 26% of all gross electricity generation (European Environment Agency (2018)).

Transport and retail packaging were of minor importance and the studies by Büsser and Jungbluth (2009) support that. The roastery used renewable electricity and waste-derived biogas partly for heat production. Due to very low emission rates of these energy sources, the share of the roastery is only 0.1–0.2% of the carbon footprint for roasted and packed coffee.

Irrigation dominates the water scarcity impact results. In the two irrigating farms, the irrigation was 750 and 2425 m³/ha (602 and 1233 l/kg green coffee, respectively). Pfister and Bayer (2014) assessed the irrigation level for Brazilian coffee 1104 m³/ha. We calculated the total amount of water consumed per litre coffee about 8 l for non-irrigated and 60 and 110 l for irrigated systems. Humbert et al. (2009) calculated the water consumption in the whole coffee chain (drip-brew) 40 to 400 l of non-turbined water per litre of coffee, depending on whether the coffee cherries were irrigated or not.

Coltro et al. (2006) studied water consumption for coffee processing. The range in their study was from 0.072 to 60 l water per kg of green coffee and the weighted average was 11.437 l. In our study, the range is from 0 (no wet process) to up to 6.235 l per kg of green coffee.

As explained in Section 2.2.2, especially in Nicaragua and Honduras, the coffee cultivation area has grown strongly causing a significant increase in the total carbon footprint of Nicaraguan and Honduran coffee.

In Brazil, the cultivation area had increased in the observed period only modestly. In the next 2 years, if the cultivation area does not increase again, there will not be any additional emissions from land use change to be considered.

In Nicaragua, the cultivation area had increased significantly at the beginning of the observed period, but in the last 10 years, there has not been a significant increase. For now, the additional greenhouse gas emissions from land use changes are large, but in the next years, if the cultivation area does not expand again, the additional greenhouse gas emissions will decrease.

In Honduras, the cultivation area has increased steadily and significantly throughout the observed period. In the last few years, the increase has accelerated even further, and if continues, in the future, an even larger share of greenhouse gas emissions should be allocated to coffee cultivation.

In Colombia instead, no emissions related to land use changes are allocated to coffee cultivation. However, in the last 5 years, the area has been increasing, and thus, if the trend continues, in few years’ time, there may be significant additional emissions to be considered.

In addition to the climate impact, land use changes may also have their own implications on the water balance in the area. Both changes in the surface runoff and river discharge as well as water quality degradation are common consequences when natural vegetation is changed into an agricultural land (Foley et al. 2005).

Only the climate and water scarcity impacts were the focus of this study. In the literature, it has been recommended concerning the milling process that the focus should be specifically on the proper management of wastewater (Killian et al. 2013). There may be unstudied methane emissions related to wastewater and sludge treatment, and eutrophication impact related to wastewater treatment and cultivation should also be studied further. Biodiversity loss is globally one of the most serious environmental problems and in terms of coffee, biodiversity impact especially due to land use change should be studied in the future. In addition to environmental impacts, social issues should also be incorporated as part of the overall system management (Adams and Ghaly 2007).

In this study, we recorded variation between the farms. More representative samples are needed if one wants to study the average environmental impacts of coffee production in each of the producer countries. The variation between farms also means that there are still many opportunities to optimize coffee production. Improving the processes in cultivation and fertilization is key to achieve lower GHG emissions.

As explained in Section 1, nitrogen use is reported as the most important factor in the coffee chain in terms of the carbon footprint. In this study, the share of fertilizers (incl. fertilizer production and emissions from use) contributed altogether 60 to 99% of the total carbon footprint in the farming stage. However, there was notable variation between the case farms.

Both the Brazil/Minas Gerais and Colombian case farms applied relatively high levels of nitrogen fertilizers compared with the yield and in general, and there was a large variation in N fertilization (see Section 2.2.2). Coltro et al. (2006) concluded that although the use of fertilizers and pesticides depends on the specific needs of each agricultural field; these great differences evidence a clear opportunity for the reduction of these inputs. A sensitivity analysis for the carbon footprint with fertilization rates of − 20% and + 20% is presented in Supplementary S3.

In this study, we saw that irrigation dominates the water scarcity impact results. Due to the high water consumption on irrigated farms, optimization of farming practices is needed. In Brazil, the irrigated area was 10% of the total coffee plantation area in 2007 and provided 22% of the yield (de Assis et al. 2014). In a study by de Assis et al. (de Assis et al. 2014), irrigation increased the mean yield of coffee by almost 50% compared with non-irrigated cultivation (plant density10,000 or 20,000 plants per hectare). However, Eriyagama et al. (Eriyagama et al. 2014) concluded that the coffee-producing countries Brazil, Nicaragua, Colombia and Honduras have higher water consumption (if irrigated) and slightly lower yields than they potentially could by implementing better farming practices.

All Central American production sites, Nicaragua, Honduras and Colombia, apply wet processing but there are some differences in the amount of water: the Nicaragua/San Jose de la Quilali 1 farm used 3.3 l and the Colombian farm used 2 l per kg of green coffee during the wet process. This indicates differences in the process technologies applied on these case sites.

However, actions to reduce climate and water scarcity impacts (at least) should be executed side by side to avoid negative side effects.

Only few water scarcity impact assessment studies have been executed before. Irrigation dominates the results, but mostly modelled irrigation data is available, no primary data. In the future, more primary data on the actual amount of irrigation water should be collected.

In terms of water scarcity impact, it might be important to find out the origin of some inputs; i.e., we found out that fertilizer production had an relatively high water scarcity impact. The manufacturing country of the fertilizers was not known and therefore, global characterization factors were used for fertilizer production. All four countries concerned in this study: Brazil, Honduras, Colombia and Nicaragua, have much more abundant water resources than the world average (see Table 1). If the fertilizer manufacturing area were known to be some of those countries, the impact would have been lower.