The inventory scheme depends on the impact assessment method
The ISO 14046 specifies that the inventory of water elementary flows shall include inputs and outputs from each unit process being part of the system to be studied, while respecting a balance. Information on each water elementary flow should include quantity, source, quality, form of water use, geographical location, and temporal aspects. But in practice, the water inventory requirements differ amongst methods. On the one hand, there are methods whose inventory is based solely upon the water consumed (Hospido et al. 2012; Milà i Canals et al. 2008; Pfister et al. 2009). Water consumption is water removed from, but not returned to, the same drainage basin (e.g.: evaporation, transpiration, integration into a product). Berger and Finkbeiner (2012) pointed that the continental evaporation recycling rate is important in some areas and advocated to account for evaporative water returned to the basin (via precipitation) in his method (Berger et al. 2014). On the other hand, there are methods relying on the water withdrawal (removal of water from any water body, permanently or temporarily) and the water released (returned to the same catchment area where it was withdrawn during the same period of time) like Boulay et al. (2011a, b) (Fig. 1.3). Indeed, a method accounting for water quality degradation cannot rely on the single water consumption flow since the quality of the withdrawn and released water flows have to be compared. The recommendations for LCA practitioners and researchers are to inventory both withdrawn and released waters (Bayart et al. 2010; Kounina et al. 2013).
Rainwater has a special and complex status because it relates both to land use and water use in the water cycle: this water flow is only accessible though the soil and for plants (except in the case of rainwater harvesting). The consumption of rainwater stored in the soil profile has received poor attention in LCA because it is considered less environmentally relevant from a pure water consumption perspective (Núñez et al. 2013). Most methods consider that the use of rainwater does not have direct effect on water scarcity/availability. Yet, soil water consumption has an influence on water availability in rivers and aquifers. Only Milà i Canals et al. (2008) and Núñez et al. (2013) proposed a method to account for rainwater use through the land-use eﬀects on the water cycle. In Nuñez et al. (2013), the inventory flow is the net change in soil-water availability under the production system compared to the natural reference situation. But this approach has to cope with two issues: first, the definition of the potential natural vegetation and its water consumption, second, the fact that natural vegetation always consumes more water than an agricultural production system, thus leading to positive impact of the production system on the water availability. This shows the complexity to account for the water hydrological cycle and water flow redistribution within the LCA framework.
The water sources must be distinguished as they might face different scarcities/availabilities
In theory, “the inventory flows represent a set of water types each representing an elementary flow with its own characterisation factors” (Bayart et al. 2010). Indeed, each water type (source of water in the environment) has different renewability rates and functionalities. For example, differentiating groundwater from surface water in CF calculation is important because some regions suffer much more from groundwater scarcity and others more from surface water scarcity. Such a distinction is not made in Pfister’s method where surface and ground waters are weighted with a unique CF. Only a few methods differentiate water sources (except fossil water): Hospido et al. (2012) and Boulay et al. (2011a & b). Hospido and colleagues (2012) proposed to associate a specific CF for each water type of the irrigation profile: surface, ground, desalinated and non-conventional water (based on Milà i Canals et al. 2008 method). However, in practice, they allocate the same CF for surface and ground waters, and do not consider the water that may be released to the environment. Boulay et al. (2011a) not only proposed an inventory distinguishing the water sources, but also accounted for the quality of input and output waters through water categories.
Water quality has to be inventoried as quality degradation may contribute to water deprivation
The WULCA group recommended the “use [of] water quality parameters to characterize freshwater flows”. Boulay and colleagues (2011a) developed an inventory method whereby water quality is related to a function assessing to which users the waters withdrawn and released are functional (useful) (Bayart et al. 2010). This functionality-based water inventory considers that water quality degradation can lead to water deprivation if not suitable anymore for specific users (Boulay et al. 2011a). Thus, it allows for an impact assessment associated with water quality consumption and degradation.
Midpoint impact assessment: water scarcity or availability?
At midpoint, most methods are characterizing a water deprivation impact, in cubic meter equivalent, i.e. a volumetric use is adjusted against the water scarcity/availability conditions that prevail at the place of consumption or withdrawal.
Water indices are used as characterisation factors
The Characterisation Factors (CF) used to convert the water inventory flows into impacts are based on water indices and represent the actual pressure on water resources. These water indices are originally non-LCA-based indicators and are recognized as proxies for water scarcity/availability (see supplementary information). They consist of a ratio of water use by different sectors to the water available, but vary depending on what is considered as “water use” (numerator) and water available (denominator). Indeed, they are either based on a withdrawal-to-availability ratio (Milà i Canals et al. 2008; Pfister et al. 2009; Ridoutt and Pfister 2010) or a consumption-to-availability ratio (Boulay et al. 2011b; Hoekstra et al. 2012). The CFs rely on existing water availability data from hydrological models at global scale such as WaterGAP or UNH/GRDC (Alcamo et al. 2003; Fekete et al. 2002). An alternative of global hydrological model was recently tested, using a large-scale hydrological modelling with the Soil Water Assessment Tool (SWAT)(Neitsch et al. 2009) (Scherer et al. 2015). This attempt showed that although the SWAT model outperformed the global models at large watershed scale, its use on a global scale is unlikely because of the high calibration eﬀorts required (Scherer et al. 2015).
The related CF for each approach is then multiplied with the inventory elementary flows of water consumption (Milà i Canals et al. 2008; Pfister et al. 2009) or water withdrawal and release (Boulay et al. 2011b) (Fig. 1.3). See supplementary information for a detailed description of characterisation factors and the original water indices they rely on.
Endpoint impact assessment: gaps and overlapping
Endpoint methods assess potential damages from water use on the AoP Human Health, Ecosystems and Resources (Fig. 1.2). Some cause–effect chains (pathways) are not yet covered by any methods, in particular pathways related to water quality degradation. Regarding the AoP Human Health, methods assess the impacts from a water deprivation on different sectors: agricultural users (Motoshita et al. 2010; Pfister et al. 2009), domestic users (Motoshita et al. 2011), and fisheries (Boulay et al. 2011b). The methods vary in terms of data sources for the definition of CF, socio-economic parameters accounted, and which users are considered affected (sectors are more or less sensitive to a water deprivation). A comparison of human health indicators showed that the results are greatly influenced by two model assumptions: the inclusion or not of trade effects (how food supply shortage in a country will spread to other countries through international trade) and the inclusion or not of the domestic sector as an affected user (for example, Pfister et al. (2009) considered that water deprivation generally did not affect domestic users) (Boulay et al. (2015a and 2015c)). The AoP Ecosystem received much attention with many methods addressing the damages of water consumption or degradation on aquatic and/or terrestrial ecosystems. However, these methods do not cover all cause-effect chains, and cannot be used in a complementary way because they are incompatible in their current forms. Their integration into one consistent indicator would require a harmonization process (Núñez et al. 2015).
Conversely, the AoP Resource received little attention and is not sufficiently developed (Kounina et al. 2013).
The framework proposed by the WULCA group considers that only fossil water use or overuse of renewable water can affect the Resources for future generations (Fig. 1.4). But there is a lack of clear definition of what a renewable use of water is? What is the threshold above which a water body should be considered as overused? Since the renewability rate of a water body depends on many local-specific factors, defining such a threshold is complex and will be limited by a lack of data on the state of groundwater resources. Another important aspect regarding the AoP Resource is that a loss of water quality is not considered as affecting the Resource (Fig. 1.4). Yet, some situations exist where a water body may be polluted almost irreversibly (e.g: a permanently salinised deep aquifer due to a saline intrusion). In contrast, the impacts of water quality decrease on human health and ecosystem quality are considered as existing cause-effect chain (Fig. 1.4). It has been argued that accounting for water quality in water use impact assessment could lead to double counting with indicators of water pollution (Berger and Finkbeiner 2012). But this is rarely the case since: (i) when water is not drinkable, human may not drink water so the ingestion route of exposure to the contaminant do not occur, (ii) the pathway leading to human health damages from water use refers to biological contamination and hygiene rather than toxicity (Boulay et al. 2015c). In the situation where toxic water was drunk, human would not suffer from water deprivation but from toxicity. Nevertheless, we cannot make the exact same reasoning for the AoP Ecosystems. Contrary to human who can (more or less) decide to use a type of water, ecosystems have to endure a type of water quality, in term of their living environment (water as a compartment), and in term of drinkable water (water as a resource). Thus, the boundary between water-compartment and water-resource is thinner for Ecosystems than for human. As a result, the double counting risk between indicators of water pollution and indicators of water availability is real regarding the impacts on Ecosystems.
To conclude, a comprehensive assessment of water use damages at endpoint level is not possible so far, but the WULCA group is working in this direction.
Salinisation associated with brine disposal
Many activities generate saline wastewater: e.g. mining, pumping of shallow saline aquifer and seawater desalination, therefore the problem of brine disposal is raised (Williams 2001). This is a topical question to address while many countries need to complement their water supply with seawater desalination (Zhou et al. 2013b). In coastland desalination plants, brine may be discharged in seawater (impacting the marine ecosystem), whereas in inland areas, brine discharge is more problematic because diluting brine in a water stream or discharging it directly in the soil may lead to water and soil salinisation (Sánchez et al. 2015). New alternatives are studied, such as the use of brine water for agricultural use, in combined scheme (e.g. microalgae cultivation, fish production and halophyte forage scrub irrigation). But these alternatives do not prevent the gradual salinisation of land (Sánchez et al. 2015). Brine disposal is a major cause of aquatic ecotoxic impact, and the subject is of growing interest in research (Zhou et al. 2014). This salinisation type highly depends on the salts composition of the brine, and the discharge location. Salinisation due to brine disposal will be driven by many biophysical factors relying on the geographical features of the discharge context, actually all biophysical factors identified for the other salinisation types.
Salinisation associated with overuse of a water body
In many coastal areas, excessive withdrawal of groundwater and/or rivers streams leads to seawater intrusion: the decrease of the coastal aquifer table level induces seawater inflow in the aquifer, leading groundwater to long-term salinisation (Flowers 1999; Scanlon et al. 2007; FAO 2011). The depth of the interface between freshwater and seawater is reduced when the aquifer table is decreased as illustrated by the Ghyben-Herzberg formula, a linear relationship often used to simulate seawater intrusion (for a review of methods investigating seawater intrusion processes, see Werner et al. (2013) and Sreekanth and Datta (2015)). In the estuaries and deltas, seawater intrusion happens when the freshwater flow of the river is reduced because of excessive water withdrawal upstream or the construction of impoundments (Williams 2001; FAO 2011). Sea-level rise induced by climate change is an aggravating factor of seawater intrusion (FAO 2011). In non-coastal areas, saline intrusion may result from saline water transfer from a saline aquifer to an overused aquifer. This type of salinisation happens when too much water is withdrawn from a water body, independently of the usage. However, irrigation is the principal cause because 70% of all water extraction worldwide is devoted to agricultural use (World Water Assessment Program 2009). Salinisation associated with saline intrusion involves mechanisms at the regional (e.g. fluctuating sea level) and local (e.g. well) scales (Werner et al. 2013). The biophysical factors involved are the distance to the coast or estuary, and the presence of saline aquifer. The management factors are the volume of freshwater withdrawal and the exploitation rate of the water body (river or aquifer) (Table 2.1).
Table of contents :
WHY DOING LIFE CYCLE ASSESSMENT OF AGRICULTURAL SYSTEMS?
1.1. FEEDING THE PLANET WITHOUT DESTROYING IT
1.1.1. Agriculture is feeding the planet, but has many impacts on the environment
1.1.2. An increasing pressure…
1.1.3. Identify the environmental hot spots and mitigation options
1.2. LCA OF AGRICULTURAL SYSTEMS: CHALLENGES
1.2.1. LCA methodology
1.2.2. The cause and effect chain or environmental “pathway”
1.2.3. Why applying LCA to agricultural systems is relevant?
1.2.4. The inventory: a crucial LCA stage for agricultural systems
1.2.5. One limitation of LCA relates to the modelling of freshwater use impacts.
1.3.2. The importance of scales
CHAPTER 1 HOW TO ASSESS THE IMPACTS ASSOCIATED WITH WATER USE IN AGRICULTURAL LCA?
1.1. WATER FOOTPRINTS TERMINOLOGIES
1.2. AN OVERVIEW OF THE DIFFERENT METHODS
1.3. INVENTORY SCHEMES: WATER QUANTITY, QUALITY AND SOURCE
1.3.1. The inventory scheme depends on the impact assessment method
1.3.2. The water sources must be distinguished as they might face different scarcities/availabilities
1.3.3. Water quality has to be inventoried as quality degradation may contribute to water deprivation
1.4. MIDPOINT IMPACT ASSESSMENT: WATER SCARCITY OR AVAILABILITY?
1.4.1. Water indices are used as characterisation factors
1.4.2. Water scarcity indicators vs. water availability indicators
1.4.3. Open questions on characterisation factors
1.4.4. Spatial and temporal scales: consistency with the goal and scope of the study
1.4.5. Water indices versus fate and effect modelling
1.5. ENDPOINT IMPACT ASSESSMENT: GAPS AND OVERLAPPING
1.7. WATER IS A RESOURCE, BUT ALSO A VECTOR OF POLLUTANTS, NUTRIENTS AND SALTS
1.8. THESIS SPECIFIC OBJECTIVES
CHAPTER 2 SALINISATION IMPACTS IN LIFE CYCLE ASSESSMENT: A REVIEW OF CHALLENGES AND OPTIONS TOWARDS THEIR CONSISTENT INTEGRATION
2.2. SALINISATION ENVIRONMENTAL MECHANISMS
2.2.2. Human interventions causing soil and water salinisation
2.2.3. Water and soil salinisation damages to Ecosystems, Human health and Resources
2.2.4. Complexities related with salinisation in space and time
2.3. CRITICAL ANALYSIS OF SALINISATION IMPACT ASSESSMENT METHODS IN LCA
2.3.1. Salinisation associated with irrigation: Feitz and Lundie (2002)
2.3.2. Salinisation associated with overuse of a water body: Amores et al. (2013)
2.3.3. Salinisation associated with brine disposal: Zhou et al. (2013b)
2.3.4. Salinisation associated with salt release: Leske and Buckley (2003; 2004a; 2004b)
2.3.5. Lack of consistent frameworks
2.4. TOWARDS A CONSISTENT FRAMEWORK FOR SALINISATION IMPACTS ASSESSMENT IN LCA: METHODOLOGICAL ISSUES AND RECOMMENDATIONS
2.4.1. Context of LCIA for assessing salinisation impacts
2.4.2. Modelling options for the different salinisation types
2.4.3. Toward operationalisation
CHAPTER 3 INVENTORY OF FIELD WATER FLOWS FOR AGRI-FOOD LCA: CRITICAL REVIEW AND RECOMMENDATIONS OF MODELLING OPTIONS
3.2. CRITICAL ANALYSIS OF WATER INVENTORY DATABASES
3.2.1. Water inventory and agri-food LCA databases
3.2.2. Limitations of water inventory and agri-food LCA databases
3.3. MODELLING OPTIONS FOR FIELD WATER FLOWS INVENTORY IN AGRICULTURAL LCA STUDIES
3.3.1. Model specifications description
3.3.2. Modelling approach selection
3.3.3. Further developments of models, tools and databases
CHAPTER 4 E.T.: AN OPERATIONAL FIELD WATER AND SALT FLOWS MODEL FOR AGRICULTURAL LCA ILLUSTRATED ON CITRUS
4.2. MATERIAL AND METHOD: FIELD WATER AND SALT FLOWS MODEL PRESENTATION
4.2.1. Model specifications and general principles
4.2.2. E.T. Model description
4.2.3. Validity domain of E.T.
4.2.4. Model testing
4.3. RESULTS AND DISCUSSION: MODEL TESTING
4.3.1. Comparison of E.T. outputs to other model formalisms
4.3.2. Sensitivity of E.T. model outputs to parameters’ variations
4.3.3. Testing of E.T. model for different scenarios of practice
4.3.4. Model with degraded data
4.3.5. Model limitations and improvement perspectives
4.3.6. Model usage recommendations
4.3.7. Model outputs comparison with databases
4.3.8. Model usage within a LCA study
REFERENCES CHAPTER 4
CHAPTER 5 LIFE CYCLE ASSESSMENT OF A PERENNIAL CROP INCLUDING AN IN-DEPTH ASSESSMENT OF WATER USE IMPACTS: THE CASE OF A MANDARIN IN MOROCCO
5.2. MATERIALS AND METHODS
5.2.1. Geographical context
5.2.2. LCA goal and scope
5.2.3. Inventory of Moroccan mandarin production: from cradle-to-farm-gate
5.2.4. Life cycle impact and damage assessment
5.2.5. COMPARISON WITH PUBLISHED LCA STUDIES ON CITRUS
5.3.1. Market gate – midpoint
5.3.2. Farm gate – midpoint
5.3.3. Farm gate – endpoint
5.4.1. Comparison with published references on citrus
5.4.2. Water – energy nexus
5.4.3. Water use impacts
REFERENCES CHAPTER 5
DISCUSSION AND PERSPECTIVES (IN FRENCH)
COMMENT MIEUX EVALUER LES IMPACTS ASSOCIES AUX FLUX D’EAU ET DE SELS ?
COMMENT REALISER UN INVENTAIRE PERTINENT DES FLUX D’EAU ET DE SELS MOBILISES DANS LES SYSTEMES AGRICOLES ? .
EST-IL POSSIBLE D’APPLIQUER LE MODELE D’INVENTAIRE DES FLUX D’EAU ET LES INDICATEURS ASSOCIES POUR EVALUER DES
PRATIQUES AGRICOLES ?
GENERAL CONCLUSION (IN FRENCH)
PRESENTATIONS IN CONFERENCES
PARTICIPATION IN BOOKS AND REPORTS
2. MATERIALS AND METHODS
2.1. Geographical context
2.2. LCA goal and scope
2.3. Inventory of Moroccan tomato production: from cradle-to-farm-gate
2.4. Life cycle impact and damage assessment
2.5. LCA comparison of Moroccan and French off-season tomato production
3. RESULTS AND DISCUSSION
3.1. Environmental impacts of the Moroccan off-season tomato production and delivery
3.2. LCA comparison of imported Moroccan and local French production systems
3.3. The need for a reliable inventory for accurately modelling the impacts of freshwater use