The second group of models includes sectoral or multi-sectoral global economic models that are designed to investigate possible strategies for responding to future socioeconomic and climatic change. Outputs from GHMs and GGCs previously discussed are sometimes aggregated and soft-linked to economic models, used as inputs to provide physical resource constraints and supply and demand curves. However, the economic models add the important capability of evaluating economic trade-offs and synergies among multiple strategies for both mitigating environmental degradation and ensuring the provision of adequate and affordable food, energy, and water to meet future anthropogenic demands. Consequently, these models provide insights into possible regional and global solutions to nexus challenges and thus are useful for informing long-term policy and planning decisions. The inclusion of environmental objectives is particularly relevant to the identification of synergistic strategies for meeting multiple SDGs. The SDG targets seek not only to eradicate human poverty and hunger but also a range of other objectives, including the mitigation of climate change and biodiversity loss.
Both partial equilibrium (PE) and computable general equilibrium (CGE) models have been developed at the global-scale for exploring the response of supply systems to socioeconomic and climatic changes. PE models include: (1) agro-economic, models which are designed to identify land use, management, and demand responses in the agricultural and forestry sectors; (2) energy-economic models, which evaluate transformations of the energy sector; and (3) hydro-economic models, which evaluate management and demand responses in the water sector. Integrated nexus frameworks have also been developed for evaluating interactions among water, energy, and land management, either through the linkage of PE models or the extension of CGE models, which endogenously capture interactions among global economic sectors.
3.1. Agro-Economic Models
Agro-economic models are useful for exploring the interactions between water and land management as they quantify how future food, fiber, and bioenergy demands can be met under water, land, and climate constraints and economic impacts. The main response strategies that are evaluated by these models are spatial re-allocation of crops and livestock production to locations with higher productivity (including inter-regional trade), expansion of managed land, agricultural intensification, and demand response. Many agro-economic models are linked to global gridded crop models, which provide the potential yields, suitable areas, and fertilizer and irrigation water requirements for particular crops as inputs.
Global agro-economic models incorporating water constraints on agricultural production include GLOBIOM (GLObal BIOsphere Management model [36
]; linked to EPIC and LPJmL), MAgPIE (Model of Agricultural Production and its Impact on the Environment [35
]; linked to LPJmL), GCAM (Global Change Assessment Model [81
], IMAGE (Integrated Model to Assess the Global Environment [83
]; linked to MAGNET and LPJmL), and IMPACT-WATER (International Model for Policy Analysis of Agricultural Commodities and Trade [84
]). Crop yields and irrigation are constrained by water availability within each agricultural production unit, which can be represented on a grid as small as 30-arcminutes. GLOBIOM and MAgPIE do not include explicitly a representation of water allocation between up and downstream spatial units. However, it should be noted that although MAgPIE, GLOBIOM and IMAGE make production decisions at high resolution, commodity demands, prices, and trade are calculated within coarser economic regions (up to 57 regions in GLOBIOM, 10 regions in MAgPIE, and 26 regions in IMAGE) (Figure 1
). The fact that demands need only to be met by the aggregated regional production tends to concentrate agricultural expansion in the most profitable and productive areas, regardless of their proximity to demands [35
]. Notable exceptions occur when internal transportation cost is used for the distance to internal markets such as in regional GLOBIOM implementations [86
]. This assumption may be appropriate in developed regions with sophisticated food distribution infrastructure, but may misrepresent cropland expansion and overestimate the potential for adaptation in developing regions with poor infrastructure and a reliance on subsistence farming [87
In terms of land use conversion, most global agro-economic models project a 20–25% increase of cropland by 2050 relative to 2005 when climate change is not considered [88
]. Most of the expansion occurs in South America and sub-Saharan Africa. However, the results across models are quite diverse as a result of differences in the potential for endogenous productivity responses, availability of cropland, and the ease of land conversion and trade (see [88
] and [89
] for descriptions of model differences). Further research is needed to resolve some of the underlying uncertainties that drive these model differences. Furthermore, developments are needed to improve the representation of localized water constraints, governance, and the heterogeneity of both agricultural producers and consumers in order to better account for financial and institutional constraints on adaptation [87
]. Finally, most global agro-economic models do not track the energy requirements associated with agricultural production, which will be necessary for examining the land-energy nexus including the response of farmers to varying energy prices [90
Global agro-economic models have been applied to examine several nexus challenges, including: (1) the role of trade in reducing regional water scarcity [91
]; (2) the potential for reduced consumption of livestock products and reduced food waste to decrease irrigation water requirements [79
]; (3) the implications of bioenergy expansion for irrigation water requirements, land use, and terrestrial ecosystem impacts [36
]; and (4) the implications of environmental flow requirements (EFRs) for food production [28
]. Within AgMIP, harmonized scenarios were conducted with several global agro-economic models to examine robust solutions for meeting future food demand under climatic and socioeconomic change [89
]. The resulting studies explore the roles of trade [102
] land use conversion [88
], and land intensification in meeting future demands for major crops (wheat, rice, soybeans, and grains) [100
Although climate change under a high GHG emissions scenario (RCP 8.5) is expected to reduce global yields of major crops by an average of 17% by 2050, economic responses, such as cropland expansion and intensification, can greatly reduce productivity impacts [100
]. However, intensification may exacerbate other environmental challenges, such as non-CO2
GHG emissions and water scarcity if irrigation and fertilizer use are expanded [97
]. Studies using GLOBIOM and MAgPIE confirm the value of response strategies, but indicate a much stronger role for trade in compensating for productivity losses and regional water scarcity since food production can be shifted to regions with more abundant water and better growing conditions [91
]. The importance of trade highlights the need for models with a global scope as they can explicitly account for the disparate growing conditions among regions and deploy trade when economically favorable. Reduced consumption of livestock-based products (e.g., meat, eggs, and milk) and reduced food waste have also been demonstrated as effective strategies for alleviating future agricultural land and water requirements, especially in conjunction with liberalized trade [79
]. There are also the impacts of higher CO2
concentrations on the quality of the nutrients in food [106
], indicating a need to link food impacts and mitigation strategies at the global-scale.
The introduction of climate policy is expected to alter global land use patterns as land-based mitigation strategies, such as bioenergy cultivation and conservation/expansion of forests, will change the relative value of land uses [3
]. The negative effects of stringent land-based climate mitigation on global hunger and food consumption was shown in one recent study to be greater than the equivalent impacts of climate change [107
]. Global models agree that targeted policies can help avoid tradeoffs with food pricing [9
]. Of particular relevance to the nexus is the deployment of bioenergy since its cultivation may have significant implications for both land and water resources, particularly when coupled with CCS to achieve negative emissions [108
]. Several studies examine the implications of bioenergy expansion by introducing the bioenergy demands projected by energy-economic models into agro-economic models to assess land use change under land and water constraints [36
]. Agro-economic models have also been applied to explore the deforestation implications of biofuel expansion across a range of bioenergy conversion pathways as well as the required adaptation responses to support bioenergy expansion under forest protection policies [35
]. These studies suggest that large-scale bioenergy expansion consistent with limiting global mean temperature change to 2°C above pre-industrial levels could roughly double agricultural water requirements by the end of the century [99
]. Moreover, the land required for this expansion would result in extensive conversion of forest and pasture and land conversion could increase by 41% if irrigation of bioenergy crops is prohibited [99
]. These findings suggest that large-scale bioenergy deployment could be counterproductive for land-based GHG emissions, terrestrial ecosystems, and water stress.
EFRs have only recently begun to be incorporated into global agro-economic models to explore the response of the agricultural sector to reduced water availability stemming from environmental considerations [28
]. These studies find that the introduction of EFRs substantially reduces agricultural water withdrawals and irrigated cropland globally, but that food demands can still be met through cropland expansion, intensification, and increased trade [68
]. While this finding implies that trade-offs may exist between aquatic and terrestrial ecosystem protection, increased food demand due to socioeconomic change requires 5-9 times more conversion of unmanaged land than EFRs, suggesting that EFRs have only a moderate impact on terrestrial ecosystems at the global scale [28
Finally, while agro-economic models have been successful in quantifying the implications of food and bioenergy expansion for the magnitude, or quantity, of land use change (e.g., deforestation), they have yet to relate the impacts of the resulting land use change and specific management practices to ecosystem quality [112
]. This is particularly true for land management practices that do not alter the land cover classification, but may still degrade terrestrial ecosystems, such as the use of forest residues for bioenergy. Consequently, next-generation models are needed that spatially disaggregate changes in land use and management practices and relate these changes using empirical studies to the health and biodiversity of specific ecosystems. The Global Biodiversity Model (GLOBIO) is an example of an attempt to address this shortcoming by relating human-induced land use pressures to biodiversity loss at the global scale [113
]. Although GLOBIO presents a novel methodology for linking land use pressures to biodiversity loss, the value of the findings is inherently constrained by limitations at the global scale associated with: (1) modeling environmental drivers of biodiversity loss; (2) disaggregating land use change across ecosystems; and (3) establishing robust relationships between environmental drivers and biodiversity loss for broad land use categories. Thus, there is significant scope for improving the representation of how land conversion and management decisions impact terrestrial ecosystem quality, diversity, and function.
3.2. Energy-Economic Models
The energy sector accounts for approximately 15% of global total water withdrawals with the majority withdrawn for the cooling of thermoelectric power plants [114
]. Meanwhile, in the “middle-of-the-road” shared socioeconomic pathway (SSP2), global electricity generation is expected to triple from 2005 to 2050 and increase more than six-fold in Africa and Asia, according to the average of six global energy-economic models [115
]. Thus, without changes in the water-use intensity of electricity generation, it is expected that the share of energy sector water use will increase, especially in developing countries. Strategies for reducing energy-related water use include shifting to more water-efficient cooling technologies (e.g., recirculating and dry cooling), using alternative water resources (e.g., wastewater or seawater), constructing less water-intensive energy transformation technologies (e.g., gas instead of coal; wind and solar PV), and using distribution infrastructure (e.g., transmission lines) to import energy from other, possibly more water-abundant, regions. However, adapting to water scarcity is expected to increase energy prices and may significantly alter future energy transitions [13
]. Meanwhile, regional studies suggest that the adoption of energy-intensive water supply technologies, such as water conveyance and desalination, could substantially increase energy demands and the cost of water supply while exacerbating the climate change mitigation challenges already faced by the energy sector in water-scarce regions [13
]. Thus, it is important that energy-economic models improve their representation of water-energy trade-offs in assessing energy transition pathways.
However, global energy-economic models typically operate at a coarse spatial scale with energy production and demand represented within macro-regions (Figure 1
). Thus, it is challenging to incorporate meaningful water constraints, as there is typically sufficient water when assessed at the macro-regional scale, even though constraints might occur in reality at the asset-scale. As a first step, several global energy-economic models have begun to track the water consumption and withdrawal of the electricity sector and entire energy sector as a post-processing exercise [58
]. These studies apply exogenous assumptions about cooling technology transitions to pre-existing energy system transformation pathways and thus do not explore how mitigation pathways respond to water scarcity. In scenarios where cooling technologies are fixed at present shares, the studies indicate that water withdrawals are expected to increase between 0% and 150% over the 21st century, depending on the future electricity generation portfolio [58
Both global and regional assessments generally suggest that increasing energy demands and electrification will likely lead to growing energy-related water consumption [58
]. Global mitigation pathways that rely more heavily on wind and solar PV technologies are expected to consume less water than pathways that rely on hydropower or low-carbon thermal power technologies that require water for cooling, such as nuclear, concentrating solar power (CSP), and fossil technologies with carbon capture and storage (CCS) [58
]. Moreover, the adoption of air and seawater cooling can help to maintain greater mitigation flexibility by reducing the water consumption of thermal power generation.
More recently, the economic impacts of cooling system choices were incorporated into a global energy-economic model to study the interactions between the SDG for clean water (SDG6) and the Paris Agreement 1.5 °C target [13
]. The results indicate that combined policies drive power systems towards water-efficient low-carbon generation technologies (e.g., wind and solar) faster than if each policy was applied on its own. This is because wind and solar are usually expanded more aggressively later in the 21st century in future 1.5 °C pathways simulated with global energy-economic models. However, when the climate policy is layered with the SDG6 water efficiency policy, it is better to expand wind and solar systems before 2030 to reduce energy sector water use in the SDG timeline. Less water-efficient power generation choices pose the risk of stranded assets after 2030 due to potential water efficiency standards driven by the SDGs.
Technologies that convert biomass to transportation fuels, electricity, and heat are also widely deployed in energy-economic models to mitigate climate change, especially when coupled with CCS to achieve negative emissions [107
]. The cultivation of biomass requires both land and water resources, but these nexus trade-offs are underrepresented in global energy-economic models. Regional biomass potentials are typically represented by exogenously-derived supply curves that are not accounting for dynamic water and land trade-offs associated with their exploitation. However, several studies have recently coupled agro- and energy-economic models in an effort to better account for the emissions, land use, and water trade-offs [3
]. Studies using these frameworks suggest that water constraints are expected to increase development costs in some regions due to water stress [99
Energy- and hydro-economic models need to be coupled over multiple decision scales to reflect the complex interactions between systems at different locations. An additional benefit of linking hydro and energy-economic models is the ability to represent innovative technologies at the interface between the two sectors. For example, the application of combined heat and power (CHP) to reduce cooling loads by providing waste heat to nearby industries (including to support desalination [123
]), the recovery of energy, fertilizer, and even nutrient recovery from wastewater [124
], and the use of treated wastewater as cooling water for thermoelectric power plants [125
]. Finally, terrestrial ecosystems are not only affected by food, fiber, and bioenergy production, but also from land conversion and air pollution associated with energy supply systems. Examples include ecosystem destruction associated with coal mining (e.g., open pit and mountaintop removal) and flooding for hydroelectric reservoirs as well as ecosystem degradation from acid rain associated with coal-based power plants [126
]. Spatially-explicit energy-economic models would improve the assessment of such localized ecosystem impacts.
3.4. Global Integrated Nexus Solution Frameworks
Global integrated nexus solution frameworks bridge the gap across all three resources and have been developed by the coupling and extension of existing global models. Frameworks are including transformational changes in the WEL resource supply-chain based on economic decision-criteria, and are generally classified as either partial equilibrium (PE) or computable general equilibrium (CGE) type models. CGE-based frameworks have been developed to improve the representation of the land and water impacts associated with bioenergy deployment and to capture the trade-offs between land-based and energy-based mitigation strategies. In contrast to PE models, which include representation of a limited number of economic sectors (e.g., agriculture, livestock, and forestry in agro-economic PE models), CGE models include representation of all global economic sectors and thus can provide insight into the macro-economic implications of future policy and resource developments. Many global nexus solution frameworks exist including GCAM [81
]; AIM/CGE (Asia Pacific Integrated Model/Computable General Equilibrium; [135
]), GTAP-BIO-W (Global Trade Analysis Project model with Biofuels and Water; [137
]), IGSM-WRS (Integrated Global System Model-Water Resource System; [138
]), ANEMI [139
]; and MuSIASEM (Multi-scale integrated analysis of societal and ecosystem metabolism; [140
While global PE models include very detailed representations of the commodities, demands, and transformation technologies in their respective sectors, global CGE models tend to be less resolved. However, sectoral detail in CGEs is often tailored for answering specific research questions. For example, GTAP-BIO-W is an extension of the single-period GTAP model [141
] in which biofuel production and irrigated agriculture have been disaggregated and irrigation water has been added as a production input for which crops compete at the scale of major river basins [137
]. Studies using GTAP-BIO-W have applied exogenous assumptions regarding future biofuel expansion and irrigation water availability to examine how these future conditions/shocks impact global welfare loss, land use change, and indirect land use change (ILUC) emissions [142
]. However, GTAP-BIO-W has limited representation of energy and water supply technologies that could help to mitigate water constraints (e.g., interbasin transfers) and/or bioenergy-induced land use change (e.g., alternative low-carbon transport fuels). Thus, the evaluation of nexus trade-offs is hampered by the limited representation of technological solutions. Moreover, as a single-period model, it is not designed to identify long-term transition pathways for adapting to and mitigating global socioeconomic and climatic changes.
In contrast, EPPA (Emissions Prediction and Policy Analysis; [144
]), which is the CGE model used in the IGSM-WRS framework, and AIM/CGE are recursive dynamic models that are explicitly designed to evaluate long-term transition pathways [146
]. To account for trade-offs among energy- and land-based mitigation strategies, both models include disaggregated representations of energy production technologies and distinguish the agriculture, livestock, and forestry sectors [145
]. AIM/CGE also disaggregates several agricultural crops and energy end-use technologies. In addition, both models have been coupled to biophysical land use models to exploit spatially-explicit information on land productivity and to downscale the macro-regional land demands determined by the CGE models [150
]. Downscaled land use allocation is important for capturing the spatial distribution of carbon stock density and tends to indicate larger ILUC emissions associated with bioenergy expansion than macro-regional estimates [142
]. Yet, land allocation is driven primarily by biophysical conditions (e.g., land productivity) and does not consider other pertinent factors (e.g., infrastructure and institutions).
Although land-energy trade-offs are well-represented through endogenous competition among food production, bioenergy cultivation, and afforestation [149
], water linkages are less developed and the implications of water constraints for energy and agricultural decisions have not been fully explored. Like the energy-economic PE models, AIM/CGE has quantified industrial water withdrawals as a post-processing exercise in which energy-related water use is a function of the future energy technology portfolio [135
]. In contrast, in the IGSM-WRS framework, EPPA generates scenario-based GDP and population projections, which are used by the water resource model to determine future municipal and industrial water demands [138
]. Thus, future energy-related water use is not explicitly tied to the portfolio of technologies deployed, but rather driven by GDP and population projections.
The IGSM-WRS framework includes a linkage with the MIT Earth System Model (MESM), which projects changes in temperature, precipitation, and runoff. A unique feature of the IGSM-WRS is its incorporation of the water module of the IMPACT-WATER model, which explicitly accounts for water routing among river basins as well as water storage associated with existing reservoirs. Thus, given changes in anthropogenic water demands and natural runoff, IGSM-WRS can assess future water stress globally. However, the link between EPPA/MESM and the water resource model is unidirectional meaning that water constraints do not influence agricultural and energy supply decisions. Moreover, the water management module only considers the existing built environment (e.g., reservoirs) and thus does not assess adaptation responses to water scarcity. Finally, while CGE models often include modules to downscale model outputs (e.g., water and land demands), all decisions and trade-offs are assessed at the macro-regional scale, as seen in Figure 1
Global nexus assessment frameworks have also been developed by coupling agro- and energy-economic PE models to examine cross-sectoral climate change mitigation strategies across a range of socioeconomic scenarios. Whereas, the dynamics of the energy- and agro-economic systems are fully coupled in GCAM, agro-economic trade-offs, such as land-based emissions, deforestation, and water use, are emulated through bioenergy supply curves in the energy modules of IMAGE, REMIND-MAgPIE, and MESSAGE-GLOBIOM [3
]. Along with AIM/CGE, these PE models have been used to quantify uncertainties across the SSP marker scenarios [15
]. Multi-model analysis provides a framework for assessing the scale of the uncertainties and to identify robust trends across different modeling assumptions. The multi-model results have also provided the basis for global pathways communicated in recent reports by the Intergovernmental Panel on Climate Change, which are informing policy-makers of impacts, uncertainties and required actions and solutions [154
]. Table 1
highlights the strengths and weaknesses of the SSP scenario modeling frameworks in terms of their assessed ability to incorporate nexus trade-offs. The main weaknesses of the models can be attributed to: (1) high data and computational requirements; (2) inconsistent spatial and temporal definitions across the models and the historical datasets used for calibration; and (3) diversity in the representation of future technology and policy solutions over a multi-decadal simulation period.
The multi-model analysis of the SSPs revealed important WEL resource trends under future policies. In a stringent climate change mitigation scenario (RCP2.6), cropland is expected to increase across all SSPs because of the expansion of bioenergy cultivation [3
]. Taking the average across models for each SSP, cropland increases 5-53% between 2005 and 2100. Moreover, climate change mitigation incentivizes the conservation/expansion of forests, which increase on average 3-16% globally. The increase in both cropland and forest is achieved through the reduction of pasture (6-20% loss) and other natural land (1-18% loss), which includes unmanaged grassland and savannah. Thus, reduced impacts on forest ecosystems may come at the expense of impacts to other terrestrial ecosystems. To improve the protection of both forest and non-forest ecosystems, a universal carbon price that applies uniformly to all sectors, including land use change and other terrestrial emissions can promote the most cost-effective land-based mitigation [80
]. While bioenergy expansion puts additional pressure on natural land by increasing the land required for crops, differences in deforestation trends across the models are driven primarily by the availability of land-based mitigation options that increase forested land, such as afforestation and reforestation [3
]. Variation in the land use implications associated with bioenergy expansion is also driven by differences in the structural features and assumptions of the models and, particularly, how they represent land conversion, yield improvements, food demand elasticity, trade, and bioenergy feedstocks [95
Although the land-energy and land-water linkages have been improved in many models, including IMAGE and AIM/CGE [149
], there has been less progress in capturing the linkages between the water and energy supply sectors. For example, in these frameworks, energy supply decisions (e.g, cooling technology choice) and energy sector water use are not responsive to water constraints. Rather, energy sector water use is provided as an exogenous water demand that impacts the water available for irrigation. GCAM and MESSAGE-GLOBIOM have added the capability to assess the endogenous responses of the agricultural, energy, and water sectors to water scarcity by adding water constraints at the scale of major global river basins [13
]. Initial applications of these integrated models indicate that water withdrawals are reduced when water constraints are included. This is because sectors are incentivized to reduce water demand through various response strategies. In GCAM, the agricultural sector exhibits a large response with a 20% reduction in withdrawals by the end of the century. Water constraints have a smaller impact on the energy sector with a transition to less water-intensive production and cooling technologies.
Although the inclusion of water constraints and the simultaneous assessment of solutions across land, energy, and water represent a significant step forward, there are opportunities for improvement. First, many response options for addressing water scarcity are not explicitly included. For example, a portfolio of efficient irrigation technologies, interbasin transfer options, and expansion of water storage can provide flexibility in meeting water supply targets. Finally, global frameworks are operating with annual resolution and thus do not explicitly account for the seasonal and intra-annual variability of water supply and use. Instead, the frameworks are introducing the concept of accessible water, which in GCAM is a function of baseflow and storage in existing reservoirs [82