Sustainability & Environment (Agri-Food Systems) · Sustainability Assessment, Carbon & Governance · also LCA, Life Cycle Thinking, Environmental Footprinting, Product Environmental Footprint
Quantifying the environmental impacts of a product from cradle to grave under ISO 14040 and 14044: goal and scope, inventory, impact assessment and interpretation, with the agricultural complications of multifunctionality, land use and biogenic carbon.
In the exams: GATE ES Section 9 (life cycle assessment); ACLCA LCACP; sustainability analyst and PhD research roles
Life cycle assessment (LCA) is the standard method for measuring the environmental burden of a product or service from the extraction of raw materials through production, use and disposal. It answers questions such as: is a litre of milk from a grazing farm better or worse for the climate than one from a housed herd; does a reusable crate beat a corrugated box; what share of the footprint of bread comes from the field, the mill, the bakery and the kitchen; and does a biogas plant reduce emissions once the digestate is spread? LCA is defined internationally by ISO 14040 and ISO 14044, which give four phases — goal and scope, inventory, impact assessment and interpretation — and the rules that make studies transparent and comparable.
Agri-food products are among the hardest to assess. Their largest emissions are biological (methane from cattle and rice, nitrous oxide from soils, ammonia and nitrate from fertilisers and manure) and must be modelled rather than metered. Almost every agricultural process yields several co-products, forcing choices about allocation. Land occupation, land-use change and soil carbon can outweigh everything else. This manual teaches the ISO framework in general terms and works through each of these complications with farm and food cases, computing every number shown.
LCA is examined in GATE Environmental Science and Engineering (Section 9, environmental management and sustainable development), within environmental assessment units of national eligibility tests, and in depth by the Life Cycle Assessment Certified Professional credential. Sustainability analysts, food companies, certification bodies and regulators use it daily for carbon footprints, environmental product declarations and the European Product Environmental Footprint. Methods and characterisation factors change with new IPCC reports and new versions of methods; the manual says which version it uses wherever a number depends on it.
Life cycle assessment (LCA) is the compilation and evaluation of the inputs, outputs and potential environmental impacts of a product system throughout its life cycle. That is the definition in ISO 14040, and every word matters. 'Product system' means the set of linked processes — called unit processes — that together deliver a product or service. 'Life cycle' means the consecutive stages from raw material acquisition or generation from natural resources to final disposal: cradle to grave. 'Potential' means that LCA does not predict actual damage at a place and time, as a risk assessment or an environmental impact assessment (EIA) does; it estimates the contribution of a product to environmental problems, scaled to a unit of function. Life cycle thinking is the wider habit behind the method: before praising a change, ask whether it merely moves the burden to another stage of the chain, another place or another environmental problem. Burden shifting is the core thing LCA is designed to catch. Replacing synthetic fertiliser with manure may lower the energy used to make nitrogen but raise ammonia emissions; growing tomatoes locally in a heated greenhouse may cut transport emissions but raise heating emissions far more; a biofuel may cut fossil carbon dioxide but raise eutrophication and land use. An LCA puts all of these on one table for the same function delivered.
The international framework is set by two standards. ISO 14040 (Principles and framework) describes the method in general terms; ISO 14044 (Requirements and guidelines) states what a study must contain. Both were issued in 2006, replacing the 1997–2000 series (ISO 14040, 14041, 14042 and 14043), and have since been amended without changing the core structure. They are written for practitioners and are not certifiable standards in the way ISO 14001 is. An LCA has four phases. Goal and scope definition states why the study is done, for whom, what function is compared and where the system boundary lies. Life cycle inventory analysis (LCI) collects and calculates the input and output flows of every process, scaled to the functional unit. Life cycle impact assessment (LCIA) translates the inventory into potential impacts on a set of impact categories such as climate change or eutrophication. Interpretation evaluates the results against the goal, checks completeness, sensitivity and consistency, and draws conclusions, limitations and recommendations. The phases are iterative: what is learned in inventory or impact assessment often sends the analyst back to sharpen the scope. Interpretation is linked to all three other phases rather than following them in a line. ISO 14040 also names the direct applications that lie outside the framework itself: product development and improvement, strategic planning, public policy making and marketing.
A family of standards builds on 14040/44. ISO 14067 sets requirements for the carbon footprint of products; ISO 14046 for the water footprint; ISO 14025 for Type III environmental declarations (environmental product declarations); ISO/TS 14071 for critical review; ISO/TS 14072 for organisational LCA; ISO 14045 for eco-efficiency assessment. Regional schemes build on the same base, most prominently the European Union's Product Environmental Footprint (PEF) method. LCA differs from its neighbours. EIA assesses a specific project at a specific site and its local significance; LCA assesses a product function wherever its processes occur. Material flow analysis tracks the mass of a substance through an economy without impact characterisation. Carbon accounting under the GHG Protocol Corporate Standard covers an organisation's scope 1, 2 and 3 emissions over a year, whereas a product carbon footprint covers one product's life cycle. Environmental risk assessment estimates the probability of harm from exposure at a site. In agriculture, LCA has become the standard way to compare farming systems, diets, feeds and food supply chains, because so much of food's impact lies upstream of the farm gate (fertiliser, feed, land conversion) or in field emissions that no factory chimney captures.
ISO 14044 requires the goal to state four things unambiguously: the intended application; the reasons for carrying out the study; the intended audience (to whom the results will be communicated); and whether the results are intended to be used in comparative assertions disclosed to the public. A comparative assertion is an environmental claim that one product is superior or equivalent to a competing product performing the same function. That last item has heavy consequences: a public comparative assertion triggers a critical review by a panel of interested parties and forbids weighting to a single score in the impact assessment. A study written for internal hotspot screening can be far lighter than one supporting an advertising claim that oat drink beats dairy milk. Being honest in the goal statement is therefore the first quality control in LCA.
The functional unit is the quantified performance of a product system for use as a reference unit. All inputs and outputs are related to it, and two systems can be compared only if they deliver the same function. In agri-food LCA the choice of functional unit often decides the answer. Per kilogram of product is simple but ignores quality: one kilogram of lettuce and one kilogram of beans do not feed a person equally. Per kilogram of protein, per unit of energy or per nutrient-density score reflects nutrition but each embeds a judgement. Milk is usually expressed per kilogram of fat- and protein-corrected milk (FPCM), which standardises milk to a set fat and protein content; the International Dairy Federation formula is FPCM = milk × (0.1226 × fat % + 0.0776 × true protein % + 0.2534), which standardises to 4.0 % fat and 3.3 % protein. One kilogram of milk at 4.2 % fat and 3.4 % protein is 1.032 kg FPCM. Land-based units (per hectare per year) reward extensive systems that use more land per unit of food, while mass-based units reward intensive systems; good studies report both and discuss the trade-off. The reference flow is the amount of product needed to fulfil the functional unit — for a functional unit of 'packaging 1,000 litres of milk to the shop', the reference flow is the number and mass of cartons or bottles required, which differs between a 1-litre carton and a returnable glass bottle used many times.
The system boundary defines which unit processes are included. Common boundaries are cradle-to-grave (raw material to disposal), cradle-to-gate (raw material to the factory or farm gate, used for intermediate products such as grain or fertiliser), gate-to-gate (one site) and cradle-to-cradle or closed-loop variants where materials are recycled. For food, cradle-to-farm-gate studies are common because the farm stage dominates most impact categories, but they cannot compare products that differ downstream (refrigerated versus ambient storage, cooking energy, food waste at home). The boundary decisions to state include capital goods (machinery, buildings, greenhouses), land-use change, human labour (usually excluded), and services. Cut-off criteria decide which minor flows can be left out. ISO 14044 names three bases: mass (exclude inputs below a set percentage of total mass), energy (below a set share of energy input) and environmental significance (an input small in mass may still matter, for example a pesticide or a refrigerant with high global warming potential). Using mass alone is a common mistake: the active ingredient of a pesticide is a tiny mass but can dominate freshwater ecotoxicity. Scope also fixes the impact categories and method, data quality requirements, assumptions, limitations, allocation procedures and the type and form of report and critical review.
The life cycle inventory is a table of elementary flows — flows drawn directly from the environment (crude oil, water, land occupation, carbon dioxide taken up by a crop) or released directly to it (methane to air, nitrate to groundwater, cadmium to soil) — per functional unit. Intermediate flows (electricity, diesel, fertiliser, feed) link one unit process to another and are not themselves impacts; they are traced back until only elementary flows remain. The work follows a standard sequence: draw a process flow diagram, prepare data collection sheets for each unit process, collect data, validate it with mass and energy balances, relate every process to the reference flow and functional unit, and aggregate. Mathematically, the inventory is solved with matrix algebra: a technology matrix of intermediate flows is inverted to find the scaling factor for each process needed to deliver the reference flow, and the scaling vector is multiplied by the intervention (environmental) matrix to give total elementary flows. This matrix method, set out by Heijungs and Suh, handles loops such as electricity used in steel used in power plants.
The foreground system comprises the processes the decision maker controls or can specify — on a dairy farm, the herd, feed rations, manure storage and field operations — and should be modelled with primary, site-specific data: measured fuel use, fertiliser records, herd registers, yields. The background system comprises processes supplied by generic markets — grid electricity, diesel refining, fertiliser manufacture, transport, packaging materials — and is modelled with secondary data from LCI databases, which hold thousands of unit process or aggregated datasets. In agri-food work the dividing line matters: purchased feed is background for a dairy farmer but foreground for a feed compounder. Primary data are not automatically better: a single year's farm records may be unrepresentative because of weather, and a measured emission at one site may not hold elsewhere. National agri-food databases now exist in several countries, alongside international commercial databases, and global datasets on crop production; their models, allocation rules and system boundaries differ, so mixing datasets from different databases is a classic source of inconsistency.
ISO 14044 lists data quality requirements that should be set in the scope: time-related coverage, geographical coverage, technology coverage, precision, completeness, representativeness, consistency, reproducibility, sources of data and uncertainty of information. A widely used semi-quantitative tool is the pedigree matrix (Weidema and Wesnæs), which scores each datum from 1 (best) to 5 on reliability, completeness, temporal correlation, geographical correlation and further technological correlation; the scores are converted into additional uncertainty factors in lognormal distributions. The PEF method uses a data quality rating built on technological, geographical and time representativeness and precision. Practical rules follow: use primary data for processes that dominate results; check that background datasets match the geography (a fertiliser made with coal-based ammonia in one country and natural gas in another can differ markedly in footprint); document every proxy; and keep the year consistent. A transport example shows how inventory arithmetic runs: 20 tonnes of grain moved 500 km is 10,000 tonne-kilometres, multiplied by a per-tonne-kilometre factor for the truck class and load factor in the database.
A multifunctional process delivers more than one product. Agriculture and food processing are dominated by them: a dairy cow gives milk and meat (cull cows and surplus calves); a sheep gives meat, wool and milk; wheat gives grain and straw; soybean crushing gives meal and oil; sugar-cane processing gives sugar, molasses, bagasse and electricity; slaughter gives meat, hides, offal and rendering products; a biogas plant treats manure and produces electricity, heat and digestate. The burdens of the shared process must be split between the co-products, or the problem must be avoided. The choice can change a product's footprint several-fold, which is why ISO sets a hierarchy and why sector guidelines (dairy, feed, livestock) prescribe a particular route.
ISO 14044 sets a stepwise procedure. Step 1: wherever possible, avoid allocation, either by dividing the unit process into sub-processes and collecting data for each (a dairy processing plant can meter butter and cheese lines separately), or by expanding the product system to include the additional functions of the co-products (system expansion). Step 2: where allocation cannot be avoided, partition inputs and outputs in a way that reflects the underlying physical relationships between them — the way inputs and outputs change with quantitative changes in the products delivered. Mass or energy content can serve when they reflect causality; a biophysical approach for milk and meat based on the feed energy needed to produce each is an example. Step 3: where physical relationships alone cannot be established, allocate in a way that reflects other relationships, commonly economic value. Whatever is chosen must be applied consistently to similar inputs and outputs, the sum of allocated flows must equal the unallocated total, and a sensitivity analysis must show the effect of alternative choices when several are reasonable. A worked soybean example: crushing 1 kg of beans yields about 0.79 kg meal and 0.19 kg oil (shares vary by crushing technology). By mass, meal carries 80.6 % of the crushing burden; if meal sells at 0.40 and oil at 1.00 currency units per kg, economic allocation gives meal 62.5 %. The burden on meal is lower under economic allocation because oil earns more per kg.
System expansion enlarges the function: instead of studying 'one kilogram of soybean meal', study 'one kilogram of meal plus 0.24 kg of oil' and compare it with an alternative system delivering the same basket. Its subtractive form, substitution or the avoided-burden method, credits the product system with the burden of the product the co-product displaces. If the oil from soybean crushing displaces palm oil in the market, the burden of producing that palm oil is subtracted from the soybean system; if biogas electricity displaces grid electricity, the grid emissions avoided are credited. Substitution needs a defensible answer to 'what is actually displaced?', which is a consequential question (chapter 5); credits can exceed burdens and give negative footprints that confuse readers. Recycling raises the same problem across life cycles. In the cut-off (recycled content) approach, burdens of recycling go to the product that uses the recycled material and the first life gets no credit. In the end-of-life or avoided-burden approach, the product sending material to recycling is credited for avoided virgin production. A 50:50 approach splits the burden and credit. The PEF method uses the Circular Footprint Formula, which shares burdens and benefits through a factor reflecting market supply and demand for the recycled material.
Attributional LCA (ALCA) asks: what share of the world's environmental burdens belongs to this product? It describes the flows within a chosen system as they are, using average data (the average grid mix, the average supplier of wheat), and deals with multifunctionality mainly by allocation. Its results are additive: if every product in the economy were assessed attributionally with consistent rules, the totals would add up to global emissions. Consequential LCA (CLCA) asks: how will global environmental flows change as a result of a decision? It models the system that responds to a change in demand, using marginal data (the supplier or technology that will actually increase or decrease output), and handles co-products by system expansion because a change in co-product supply displaces something in the market. Its results are not additive but answer 'what happens if' questions directly. The UNEP-SETAC Life Cycle Initiative and the ILCD Handbook describe the two as answering different questions rather than being better or worse; ISO 14040/44 accommodates both without using the terms.
The marginal supplier is the one whose output changes when demand changes. Weidema's widely taught procedure runs through five steps: define the time horizon and scale of the change; determine the market actually affected (local, regional, global); identify the trend in market volume (growing, stable or declining); identify which suppliers or technologies are constrained (by quotas, natural limits or policy) and which are flexible; and choose the most competitive flexible supplier — in a growing market the one with lowest long-run cost, in a declining market the least competitive, which closes first. In agriculture, constrained suppliers are common: milk under quotas, by-products whose output depends on the main product (straw output is fixed by grain demand, so straw cannot respond to straw demand), and land. A consequential model of extra demand for rapeseed oil may conclude that the marginal vegetable oil is palm oil from expanding plantations, which carries the land-use change consequences discussed in chapter 8.
Use attributional LCA to describe and report — product carbon footprints, EPDs, supply-chain hotspot identification and corporate scope 3 reporting are attributional by convention. Use consequential LCA to support decisions with market-wide effects — biofuel mandates, diet shifts across a population, a national livestock policy, or a new use for a by-product. Consequential results are more uncertain because they depend on economic modelling (price elasticities, partial or general equilibrium models) and on assumptions about the future. A classic agri-food illustration: in an attributional study, a dairy farm's beef co-product may receive a share of the farm's burdens by economic or biophysical allocation; in a consequential study, the meat from dairy culls displaces beef from suckler (beef-only) herds, and the dairy system is credited with the avoided suckler beef, which can greatly lower the footprint attributed to milk. Another: using a food-processing residue as animal feed is burden-free under cut-off attributional rules, but consequentially it diverts the residue from its previous use (perhaps another feed or energy), so the substitute for that previous use must be modelled. Indirect land-use change (iLUC) is a consequential concept and cannot be captured with average data.
ISO 14044 divides impact assessment into mandatory and optional elements. The mandatory elements are: selection of impact categories, category indicators and characterisation models; classification, the assignment of inventory results to impact categories (methane to climate change; ammonia to both acidification and eutrophication); and characterisation, the calculation of category indicator results by multiplying each inventory flow by its characterisation factor and summing. Optional elements are normalisation (dividing indicator results by a reference value, such as the total impact of a region or of an average person in a year, to judge relative magnitude), grouping (sorting or ranking categories), weighting (converting indicator results with value-based factors, often to a single score) and data quality analysis (gravity analysis, uncertainty and sensitivity analysis). Weighting rests on values, not natural science, and ISO 14044 states it shall not be used in LCA studies intended for comparative assertions disclosed to the public. Characterisation is simple arithmetic: 10 kg of methane with a characterisation factor of 27 kg CO2-eq per kg plus 2 kg of nitrous oxide at 273 gives 270 + 546 = 816 kg CO2-eq.
A cause-effect chain (environmental mechanism) runs from an emission to final damage. A midpoint indicator is chosen somewhere along that chain — radiative forcing for climate change, proton release for acidification, nutrient enrichment for eutrophication. Midpoints are closer to the inventory, so they have lower model uncertainty but are harder to interpret for a non-specialist and cannot be added across categories. An endpoint indicator sits at the end of the chain, at an area of protection: human health (in disability-adjusted life years, DALY, the sum of years of life lost and years lived with disability), ecosystem quality (in potentially disappeared fraction of species over area and time, or species·year) and natural resources (in surplus cost or energy). Endpoints are easier to communicate and allow aggregation across categories that damage the same area of protection, but carry more uncertainty and more value choices. Some methods offer both levels: ReCiPe provides a set of midpoint categories (18 in its 2008 version) linked to three endpoint areas of protection, under three cultural perspectives — individualist (short term, optimistic about technology), hierarchist (consensus default) and egalitarian (long term, precautionary) — which differ in time horizons and in which effects are included.
Methods package categories, models and factors. Problem-oriented midpoint methods trace back to CML (Leiden), which established equivalency units such as kg SO2-eq for acidification and kg PO4-eq for eutrophication. Damage-oriented methods trace back to Eco-indicator 99. Later methods include ReCiPe, IMPACT World+, the Japanese LIME method, the North American TRACI method, and the Environmental Footprint (EF) method that the European Commission requires for PEF, which has 16 impact categories with recommended models, normalisation factors per person and weighting factors. Consensus models have been built for specific categories: USEtox for human and freshwater ecotoxicity (in comparative toxic units, CTUh and CTUe) and AWARE for water scarcity. Because methods differ in models and factors, results from different methods cannot be mixed, and comparisons across studies must check the method and version. Normalisation can mislead: dividing by regional totals makes categories with poorly inventoried flows (toxicity, where many substances lack factors) look small or large for reasons unrelated to the product. A normalised result is calculated as indicator result divided by reference: a product with 2.0 kg CO2-eq against a per-person annual reference of 8,000 kg CO2-eq gives 0.00025 person-equivalents.
The climate change category characterises greenhouse gases by their global warming potential (GWP), the time-integrated radiative forcing of a pulse emission of 1 kg of gas relative to 1 kg of CO2, usually over 100 years (GWP100). Values are taken from the IPCC assessment reports and change between them, so every study must state which report it used. For the gases that dominate agriculture, the IPCC Fourth Assessment Report (AR4) gave 25 for methane and 298 for nitrous oxide; the Fifth (AR5) gave 28 for methane without climate-carbon feedbacks and 265 for nitrous oxide; the Sixth (AR6) gives 27.0 for methane of non-fossil origin, 29.8 for fossil methane (which includes the CO2 its carbon becomes) and 273 for nitrous oxide. GTP (global temperature change potential) and GWP20 are alternatives that weight short-lived methane differently. Agriculture is distinctive because its largest greenhouse gas emissions are often non-CO2: enteric methane from ruminants, methane from flooded rice and manure storage, and nitrous oxide from nitrogen applied to soils. A worked case: applying 120 kg N per hectare with a direct emission factor of 0.01 kg N2O-N per kg N gives 1.2 kg N2O-N, which is 1.89 kg N2O after multiplying by 44/28, or about 515 kg CO2-eq at a GWP of 273.
Eutrophication is the enrichment of water or soil with nutrients, mainly nitrogen and phosphorus, causing algal blooms, oxygen depletion and loss of species. Older methods (CML) expressed all nutrient emissions in a single category in kg PO4-eq, using stoichiometric factors from the Redfield ratio of algal biomass. Newer methods separate freshwater eutrophication (phosphorus-limited, in kg P-eq), marine eutrophication (nitrogen-limited, in kg N-eq) and terrestrial eutrophication (nitrogen deposition, in mol N-eq), and include fate: what fraction of an emission actually reaches the water body. Agricultural sources are nitrate leaching, phosphorus run-off and erosion, and ammonia volatilisation followed by deposition. Acidification arises from emissions of sulphur dioxide, nitrogen oxides and ammonia that release protons in soils and fresh water. In the stoichiometric approach, factors relate each gas's potential to release H+ to that of SO2: one mole of SO2 can yield two H+, one mole of NH3 one H+ (after nitrification) and one mole of NO2 one H+, giving about 1.88 kg SO2-eq per kg NH3 and 0.70 per kg NO2. Fate-based methods give different, region-specific factors, and the EF method uses accumulated exceedance in mol H+-eq. Ammonia from manure and urea is usually the main agricultural contributor to both acidification and terrestrial eutrophication.
Water inventories distinguish withdrawal (water taken from a source) from consumption (water not returned to the same catchment in the same period, because it is evaporated, transpired, incorporated into products or moved to another catchment). Impact assessment acts on consumption. ISO 14046 sets principles for water footprinting within LCA. The AWARE method, the consensus recommendation of the UNEP-SETAC Life Cycle Initiative's WULCA working group, characterises consumption by the water remaining in a catchment after human and aquatic ecosystem demands are met, relative to the world average; factors run from 0.1 to 100, and the result is in cubic metres world-equivalent. A cubic metre of irrigation water consumed where the factor is 40 counts forty times as heavily as one where it is 1. The 'water footprint' used in water-resources studies (green, blue and grey water volumes) is a related but different volumetric concept and should not be confused with an impact-assessed water scarcity footprint. Land use is inventoried as occupation (area × time, m²·year) and transformation (area converted from one state to another, m²). Impact models express effects on soil quality (soil organic carbon, erosion resistance, filtration, biotic production) and on biodiversity (potential species loss relative to a reference state). One hectare producing 8 tonnes of grain a year implies 10,000 m²·yr ÷ 8,000 kg = 1.25 m²·yr per kg.
Biogenic carbon is carbon taken up from the atmosphere by plants through photosynthesis and later released when biomass decomposes, is eaten or burned. Two conventions are used. In the 0/0 approach (carbon neutrality), uptake and release of biogenic CO2 are both given a characterisation factor of zero, on the assumption that they balance within the life cycle. In the −1/+1 approach, uptake is counted as −1 kg CO2-eq per kg CO2 and every release as +1, so the balance becomes explicit and any carbon stored at the end of the study period, or released as methane rather than CO2, is visible. The difference matters for food. In cattle, carbon in feed is released partly as CO2 (respiration, neutral) and partly as methane; methane's GWP must be counted in both approaches, and AR6 provides a non-fossil methane GWP that excludes the CO2 its carbon returns to, avoiding double counting. For biomass with long life — timber in buildings, carbon stored in biochar — temporary storage delays emission; whether delayed emission deserves a credit is contested, and ISO 14067 requires biogenic emissions and removals to be documented separately rather than netted silently. A kilogram of dry straw containing 45 % carbon embodies 0.45 × 44/12 = 1.65 kg CO2 taken from the air.
Direct land-use change (dLUC) is the conversion of land on which the product is grown from one use to another — forest or grassland to cropland — within a defined period. The conversion releases carbon from vegetation and soil. Product footprint standards (ISO 14067, the GHG Protocol Product Standard, PAS 2050 and PEF) include dLUC and use the IPCC convention of spreading the emission over 20 years: if land was converted within the previous 20 years, the emission is amortised in equal annual shares over that period. A worked example: clearing forest that loses 100 t C per hectare in biomass and soil releases 100 × 44/12 = 366.7 t CO2 per hectare; amortised over 20 years this is 18.3 t CO2 per hectare per year; with a crop yield of 3 t per hectare, about 6.1 kg CO2 per kg of product from dLUC alone — often more than all other farm emissions. When the previous land use is unknown, standards provide fallback rules based on national statistics on land-use change in the producing country.
Indirect land-use change (iLUC) is the conversion elsewhere caused by displacing existing production: if cropland in one country is switched to bioenergy crops, food demand still has to be met and some other land, perhaps forest in another continent, is converted. iLUC cannot be observed on the farm and is estimated with economic equilibrium models or deterministic land-market models; estimates vary widely between models. ISO 14067 does not require iLUC and asks that it be documented separately if calculated. Soil organic carbon (SOC) change from management (tillage, residues, cover crops, manure, grazing) without land-use change is another area of active debate. IPCC Tier 1 methods use stock change factors applied over a 20-year default period; measured changes are small relative to the stock and hard to detect, sequestration saturates as soils approach a new equilibrium, and gains are reversible if practice stops. Product standards differ on whether and how SOC changes from management may be included; where included, they are usually reported separately with evidence of permanence. Analysts should be explicit about all three items — biogenic CO2 convention, dLUC method and SOC treatment — because each can swing the climate result of an agricultural product more than any farm-level efficiency gain.
ISO 14044 sets out interpretation as three elements. First, identification of significant issues: the inventory flows, processes, life cycle stages and impact categories that drive results, found by contribution analysis (what percentage of each indicator comes from each process), dominance analysis and anomaly checks. Second, evaluation, through three checks: a completeness check (is all relevant information available?), a sensitivity check (how much do results depend on uncertain data, methodological choices such as allocation and boundary, and assumptions?), and a consistency check (were assumptions, methods and data applied consistently to all compared systems — same data age, same geography, same allocation?). Third, conclusions, limitations and recommendations that follow from the evaluation and match the goal. Interpretation should not overreach: a cradle-to-farm-gate study cannot conclude anything about packaging; a study using one method cannot claim robustness across methods.
Sensitivity analysis tests how a result responds to changes in an input or choice. One-at-a-time perturbation changes a parameter by a set fraction (for example 10 %) and records the change in the result; the ratio of relative change in output to relative change in input is the sensitivity coefficient or elasticity. If a 10 % increase in the nitrous oxide emission factor raises the carbon footprint of wheat by 4 %, the sensitivity coefficient is 0.4 — a sign that the emission factor deserves better data. Scenario analysis replaces whole sets of assumptions: mass versus economic allocation, a 2030 grid mix versus today's, cut-off versus avoided-burden recycling, with and without dLUC. Global sensitivity analysis methods apportion output variance to inputs and are used in research. ISO 14044 requires sensitivity analysis on allocation choices when more than one approach is reasonable, and expects it on any key assumption that could change a comparative conclusion.
Uncertainty in LCA comes from parameters (measurement error and natural variability in fuel use, yields, emission factors), from models (characterisation and field emission models), from scenarios and choices (functional unit, allocation, method), and from spatial and temporal variability. Parameter uncertainty is propagated by Monte Carlo simulation: each uncertain input is sampled from its distribution — commonly lognormal in LCI databases — and the model is run thousands of times to produce a distribution of results. For a lognormal parameter described by its median and geometric standard deviation (GSD), about 95 % of values lie between median ÷ GSD² and median × GSD²; a median of 2.0 kg with GSD 1.5 gives roughly 0.89 to 4.5 kg. Comparisons should use dependent (paired) sampling, in which shared background processes take the same random value in both systems in each run, and report the share of runs in which system A outperforms system B, rather than overlapping error bars. A difference between two products smaller than the combined uncertainty is not a robust finding. Field emissions from agriculture, especially nitrous oxide, are among the most uncertain flows in all of LCA; IPCC itself gives an uncertainty range of several-fold around its default direct emission factor.
Agricultural systems are open, biological and variable. Their emissions are diffuse, coming from soils, animals and manure rather than chimneys; they depend on climate, soil, weather in a particular year and management; and many are not measured but modelled. Yields vary from year to year, so a single-year inventory can mislead and multi-year averages (three to five years) are preferred. Crop rotations raise allocation questions of their own: nitrogen fixed by a legume and residual phosphorus benefit the following crop, and analysts must decide how to attribute inputs across the rotation. Farm inputs that are trivial in mass, like pesticides and trace metals in fertilisers and feed additives, can dominate toxicity categories. The International Dairy Federation, the FAO Livestock Environmental Assessment and Performance (LEAP) Partnership and PEF category rules for several food products give sector-specific guidance on functional units, allocation and emission modelling.
Nitrous oxide is estimated with IPCC methods. At Tier 1, direct emissions are a fraction of nitrogen inputs (synthetic fertiliser, manure, crop residues, mineralised soil organic matter): the 2006 Guidelines use EF1 = 0.01 kg N2O-N per kg N applied, and the 2019 Refinement keeps 0.01 as the aggregated default but disaggregates it by climate and fertiliser type. Indirect emissions arise from nitrogen volatilised as ammonia and NOx and later deposited, and from nitrogen leached or run off; each fraction carries its own emission factor, and the default fractions and factors differ between the 2006 Guidelines and the 2019 Refinement, so the version must be stated. Tier 2 uses country-specific factors and Tier 3 process models. Ammonia from fertilisers and manure is commonly estimated with the European air pollutant emission inventory guidebook's factors or national inventory methods, varying with fertiliser type, soil pH, temperature and application technique. Nitrate leaching uses simple fractions or models driven by drainage, soil and crop uptake. Phosphorus losses are modelled from erosion, run-off and leaching. Pesticide emissions are partitioned between the field (soil), air (drift and volatilisation) and water using pesticide emission models, or by default assigned wholly to agricultural soil. Heavy metals are balanced as input (fertiliser, manure, deposition) minus output (harvest), the remainder accumulating in soil or leaching. Enteric methane at Tier 2 depends on gross energy intake and the methane conversion factor (Ym); manure methane and nitrous oxide depend on the management system.
Background data for fertilisers, energy, machinery, buildings, transport, packaging and feed come from LCI databases. Several countries now publish public agri-food databases, built with agreed national rules, and international commercial databases cover thousands of processes worldwide; global crop and feed datasets have been built under FAO and UNEP-SETAC initiatives. Their allocation conventions and emission models differ, so mixing them needs care. Three agri-food cases show common findings, each reported as a range because results vary by region and system. Bread: wheat cultivation, especially fertiliser production and field N2O, usually dominates climate impact, with baking energy, home toasting and wasted bread significant downstream. Milk: enteric methane, feed production and manure management typically dominate, and the footprint depends heavily on milk yield per cow, feed conversion and how burdens are allocated between milk and meat. Rice: methane from flooded paddies is often the largest source; water regime (continuous flooding versus alternate wetting and drying) and straw management change it. Across global syntheses, ruminant meat has the highest greenhouse gas and land-use footprint per unit of protein, with large overlaps between the best and worst producers of each food.
A carbon footprint of a product (CFP) is the sum of greenhouse gas emissions and removals in a product system, expressed as CO2-equivalents and based on an LCA using the single impact category of climate change. ISO 14067:2018 (which replaced the 2013 technical specification) sets requirements consistent with ISO 14040/44. A partial CFP covers selected stages, such as cradle-to-gate. Key provisions: GWP100 from the most recent IPCC report should be used; fossil and biogenic emissions and removals are calculated and documented separately; emissions from direct land-use change are included, and indirect land-use change, if calculated, is documented separately; emissions from aircraft transport are documented separately; and offsetting is not part of the CFP — a product cannot reduce its footprint by buying credits. Because a CFP covers one category only, it cannot reveal burden shifting to eutrophication or water, which is a known weakness when it is used alone for food. Other frameworks for product carbon footprints include the GHG Protocol Product Life Cycle Accounting and Reporting Standard and PAS 2050.
The European Commission's Product Environmental Footprint (PEF) method, first issued as a recommendation in 2013 and revised in 2021, is a multi-criteria LCA method with fixed choices to make results comparable. It uses the EF impact assessment method with 16 impact categories (including climate change with sub-indicators for fossil, biogenic and land-use change emissions, freshwater and marine eutrophication, acidification, water use, land use, particulate matter and toxicity categories), normalisation and weighting to a single score, a data quality rating, the Circular Footprint Formula for recycling and energy recovery, and product-specific PEF Category Rules (PEFCRs) that set the functional unit, representative product, benchmark and the most relevant impact categories, processes and life cycle stages for each product group. The companion Organisation Environmental Footprint (OEF) applies the same approach to organisations. Unlike ISO 14044 for public comparisons, PEF uses weighting by design, because comparability within product groups is its aim.
ISO's environmental labels come in three types. Type I (ISO 14024) are third-party certified ecolabels awarded against criteria. Type II (ISO 14021) are self-declared environmental claims such as 'compostable' or 'recycled content'. Type III (ISO 14025) are environmental product declarations (EPDs): quantified LCA-based information presented in a set format, developed under a programme run by a programme operator, following Product Category Rules (PCRs) that fix the functional or declared unit, boundary, allocation and indicators for a product category. EPDs are verified, and for business-to-consumer communication independent third-party verification is required. EPDs are comparable only when based on the same PCR and functional unit. In construction, EN 15804 sets core PCRs organised in modules: A1–A3 product stage, A4–A5 construction, B1–B7 use, C1–C4 end of life, and module D for benefits and loads beyond the system boundary. Critical review is the process that checks consistency of an LCA with the ISO standards. ISO/TS 14071 sets out three types: review by an internal expert, by an external expert, or by a panel of interested parties with a chair and at least two further members. For studies supporting comparative assertions disclosed to the public, ISO 14044 requires a panel review. The reviewer checks that methods are consistent with the standard, scientifically and technically valid, data are appropriate to the goal, interpretation reflects limitations, and the report is transparent.
Social life cycle assessment (S-LCA) assesses the social and socio-economic aspects of products and their potential positive and negative impacts along the life cycle. Its reference is the UNEP-SETAC Guidelines for Social Life Cycle Assessment of Products, first published in 2009 and updated by UNEP in 2020 to cover products and organisations. It follows the four ISO phases, but most social impacts are linked to the conduct of the companies in the chain rather than to physical flows of a product, so inventory data are collected per organisation or per sector-country and attributed to the product through activity variables such as worker-hours. The guidelines organise indicators by stakeholder category — workers, local community, society, consumers, value chain actors and, from 2020, children — and by subcategory: for workers, freedom of association, child labour, fair salary, working hours, forced labour, equal opportunities, health and safety and social benefits. Two families of impact assessment exist: reference scale approaches (formerly Type I), which score performance against reference points such as international conventions or local norms, and impact pathway approaches (formerly Type II), which model cause-effect chains to outcomes such as health. Generic sector-country risk databases are used for screening hotspots before site-specific audits. In agri-food chains the most frequent hotspots are child and forced labour, wages below living wage, smallholder income, land tenure and occupational exposure to pesticides.
Life cycle costing (LCC) sums all costs associated with a product over its life cycle that are borne by one or more actors. The SETAC code of practice distinguishes three types. Conventional LCC covers the private costs of one actor, often the owner or user (purchase, operation, maintenance, disposal), and may ignore costs borne by others. Environmental LCC covers costs borne by all actors in the life cycle, uses the same system boundary and functional unit as the parallel environmental LCA, and is intended to be read alongside it; it should not double count environmental impacts that are also monetised. Societal LCC adds monetised external costs, such as damage from emissions, to the costs of all actors. Costs occurring in different years are brought to a common basis by discounting: the present value of a cost C in year t at discount rate r is C ÷ (1 + r)^t. A drip irrigation system costing 1,000 now and saving 150 a year for 10 years at a 6 % discount rate has a net present value of savings minus investment of about 104, so it pays back within its life under those assumptions.
Life cycle sustainability assessment (LCSA) combines the three: environmental LCA, LCC and S-LCA on a consistent functional unit and system boundary, often summarised as LCSA = LCA + LCC + S-LCA. It responds to the criticism that environmental LCA alone may recommend options that harm workers or smallholders or that are unaffordable. The difficulties are real: social data are qualitative and organisation-level, costs are borne by different actors, and trade-offs between the three pillars need value judgements that no method can remove. For a practitioner in agri-food chains, LCSA is most useful as a structured way to put an environmental gain — say, lower nitrogen losses from precision fertiliser application — next to its costs to farmers and its effects on farm labour, before a buyer imposes it on suppliers. For a researcher, open problems include weighting across pillars, dynamic and prospective LCA for technologies still at pilot scale, regionalised characterisation for water, land and nutrients, and the treatment of biodiversity, which current land-use indicators capture only partly.
A student must be able to state the four ISO phases and what each contains, define functional unit, reference flow, system boundary and cut-off, and tell mandatory LCIA elements from optional ones. The numerical core is short but examined: characterisation sums with stated GWPs, converting N2O-N to N2O and C to CO2, functional-unit conversions (per hectare to per kilogram, FPCM), mass versus economic allocation shares, and land occupation per kilogram. Practise explaining in two sentences why per-kilogram and per-hectare results rank farms differently, and why attributional and consequential studies of the same product give different answers. These are the questions a viva or interview panel asks first.
A sustainability analyst in a food company, consultancy or certification body builds carbon footprints and LCAs to ISO 14067, ISO 14044 or a PEF category rule. The daily decisions are: which functional unit and boundary meet the client's goal; where primary farm data are essential and where database data will do; which allocation rule the sector guidance prescribes; which IPCC version and field emission models to use; how to handle dLUC with an uncertain land history; and how to show uncertainty honestly. The practitioner also prepares EPDs under a programme operator's PCR, answers critical reviewers' comments, and refuses to let a client publish a comparative claim that the data cannot support.
A head of sustainability or consultancy partner uses LCA to set product and supply-chain strategy: which commodities and suppliers drive the footprint, which interventions (feed additives, low-emission fertiliser, reduced food waste, packaging change) deliver real reductions, and how to report them without greenwashing. Decisions include investing in primary supplier data, choosing between PEF and ISO-based approaches in different markets, managing the risk of public comparative claims, commissioning panel reviews, and reconciling product footprints with corporate scope 3 inventories and science-based targets. The senior professional also judges when a single carbon footprint hides trade-offs in water, nutrients or social conditions.
A professor or principal scientist works on the open problems: regionalised characterisation for nutrients, water and land; biodiversity indicators that capture farming intensity; field emission models validated across climates; dynamic and prospective LCA for emerging technologies; consequential models linked to economic equilibrium models for land-use change; metrics for short-lived climate pollutants; integration of LCA, LCC and social LCA; and uncertainty methods that support decisions. They write method guidance for sector bodies and governments, sit on critical review panels, supervise theses and set examination questions that test reasoning about choices rather than recall of standard numbers.
This manual describes ISO 14040/44 in general terms; it does not reproduce clause text, and readers preparing EPDs or PEF studies must work from the current standard, PCR or PEFCR. Characterisation factors other than the IPCC GWPs and stoichiometric acidification factors are not tabulated, because they differ between methods and versions. It does not give product footprint values for specific foods or countries, because published values vary widely by system, region and method; it gives the hotspots instead. Not yet covered in depth: dynamic and prospective LCA, input-output and hybrid LCA, regionalised characterisation, detailed toxicity modelling, organisational LCA (ISO/TS 14072), the detailed formula of the Circular Footprint Formula, and building-level LCA under EN 15978. The IDF biophysical allocation formula for milk and meat is described in principle but its coefficients are not quoted because they have been revised between editions.
Original practice questions written to the syllabus and pattern of the exams named in each question's examTags. They are not past papers and do not reproduce any board's questions. Every numerical answer was computed in a script; where a value depends on a method version (IPCC GWPs, emission factors, characterisation factors) the stem states the value to use.
These notes are free to read here, with no account.
What the Laisha app adds:
Free for ten days, then a subscription. We would rather write that here than have you find it later.