ABSTRACT 
 
 
 
 
Title of Dissertation: SOCIOECONOMIC AND CLIMATE 
IMPACTS ON THE FUTURE OF WATER: 
AN INTEGRATED ASSESSMENT 
APPROACH TO DEMAND, SCARCITY, 
AND TRADE 
  
 Neal Thornton Graham,  
Doctor of Philosophy, 2019 
  
Dissertation directed by: Professor Fernando Miralles-Wilhelm,  
Department of Atmospheric and Oceanic 
Science 
 
 
Changes to socioeconomics and an evolving climate system are likely to play a vital 
role in how regions around the world use water into the future. Water projections for 
the future, while prolific, remain highly variable and dependent upon underlying 
scenario and model assumptions. In this study, the Global Change Assessment Model 
(GCAM) is used, where interactions between population, economic growth, energy, 
land, water, and climate systems interact dynamically within a market equilibrium 
economic modeling framework, to address how changing socioeconomic and climate 
conditions alter global water futures, and in turn, how water constrains the future of 
other systems. First, the impacts of efficiency changes are investigated with the 
addition of socioeconomically consistent water technologies across several sectors. 
Quantitative assumptions for the Shared Socioeconomic Pathways are extended to the 
water sector for the first time in a water constrained ? Integrated Assessment 
  
Modeling framework. It is found that significant water use reductions are possible 
under certain socioeconomic conditions, provided the ability to adopt appropriate 
technological advances in lower income regions.  Secondly, the relative contributions 
of climate and human systems on water scarcity are analyzed at global and basin 
scales under the Shared Socioeconomic Pathway-Representative Concentration 
Pathway (SSP-RCP) framework. Ninety scenarios are explored to determine how the 
coevolution of energy-water-land systems affects not only the driver behind water 
scarcity changes in different water basins, but how human and climate systems 
interact in tandem to alter water scarcity. Human systems are found to dominate water 
scarcity changes into the future, regardless of socioeconomic or climate future. 
However, the sign of these changes has a significant scenario dependence, with an 
increased number of basins experiencing improving water scarcity conditions due to 
human interventions in the sustainability focused scenario. Finally, the reliance on 
international agricultural trade is analyzed to understand how future socioeconomic 
growth and climatic change will impact the dependency on international water 
sources. The differentiation between renewable and nonrenewable water sources 
allow for the quantification of the various water sources needed to produce enough 
agricultural goods to meet global demands. The first Integrated Assessment Model 
projection of the evolution of external water sources to meet domestic agricultural 
demands show that there will be an increasing international dependencies. China, the 
United States, and portions of South America are pivotal in providing the necessary 
exports to meet demands in water scarce or high demand areas of the Middle East and 
Africa.  
 
  
  
  
 
 
 
 
 
 
 
 
 
 
 
SOCIOECONOMIC AND CLIMATE IMPACTS ON THE FUTURE OF 
WATER: AN INTEGRATED ASSESSMENT APPROACH TO DEMAND, 
SCARCITY, AND TRADE 
 
 
 
 
by 
 
 
Neal Thornton Graham 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy 
2019 
 
 
 
 
 
 
 
 
Advisory Committee: 
Professor Fernando Miralles-Wilhelm, Chair 
Professor Evan G. R. Davies 
Dr. Mohamad Hejazi 
Professor Nathan Hultman, Dean?s Representative 
Professor Xin-Zhong Liang 
Professor Raghu Murtugudde 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Neal Thornton Graham 
2019 
 
 
 
 
 
 
 
 
 
 
 
 
  
 
Dedication 
I would like to dedicate this thesis to my wife Ashley. The support you have 
given me through this process is completely and totally unrivaled. Picking up and 
moving your entire life to allow me to pursue this goal is a debt I will never be able to 
repay. We have literally been through the highest of highs and the lowest of lows 
during this journey, but I could not envision doing this with a better person. I love 
you. 
I would also like to dedicate this work to my mother Tracy. Being a single 
working mother while trying to raise a child is a task I cannot imagine undertaking. 
However, for so many years of my youth, you would have never known you were 
doing this all by yourself. You made my childhood memorable, even if it was the 
farthest thing from easy for you. You taught me everything I know, and there are no 
other words to say aside from I love you, mom. 
Additionally, I would like to dedicate this work to my grandparents, Carol and 
William Storms. I walked into the University of Maryland with both of you ready to 
watch this journey, but I leave with neither of you still watching. The influence that 
you have had in this journey cannot be described. I love you both and miss you 
dearly. 
 Finally, I would like to dedicate this work, all of the hours spent on campus, 
and the numerous late nights to you, Baby G. You may not be here yet, but you have 
provided the motivation for this final push that has allowed me to complete this 
project. I have not even met you yet, but I love you more than anything son, thank 
you for this final push. 
ii 
 
 
 
Acknowledgements 
The amount of influential people that I have been lucky enough to come 
across in my lifetime is not something I take for granted. The names listed here are 
just some of the many faces that have allowed me to reach this goal. 
 I would like to that Ms. Megan Piazza and Ms. Marianne DiRupo for their 
significant influences at Jefferson Township High School. Whether it be how to just 
sit back and laugh at some things, or to push yourself to become the best person one 
can be, you both instilled this in me during one of the most important stages of my 
educational career. 
 I would like to thank Dr. Tony Broccoli; without the introduction to this 
intriguing world of research, I am not sure if I would be here today. 
 I would like to thank my advisor Dr. Fernando Miralles-Wilhelm for 
providing me the necessary resources to make the absolute most of graduate school 
life. 
 Additionally, I would like to thank everyone at the Joint Global Change 
Research Institute. You all will never know how much influence you have had on me. 
Drs. Mohamad Hejazi, Evan Davies (U. of Alberta), Sonny Kim, Katherine Calvin, 
Stephanie Waldoff, and Abigail Snyder, you have either guided me throughout this 
process, or provided a word of advice whenever asked. While I was never an 
employee at JGCRI, I was always treated as an equal, and for that, I say thank you. 
 Not to be forgotten, I could not have made it through this graduate school 
experience without the friends I have made, KB, STH, AK, DN, AS, TS, thank you 
all. And Gina and Cory, words cannot describe how much more enjoyable you made 
this process, from the bottom of my heart, thank you both. 
iii 
 
 
 
Table of Contents 
Dedication ..................................................................................................................... ii 
Acknowledgements ...................................................................................................... iii 
Table of Contents ......................................................................................................... iv 
List of Tables ............................................................................................................... vi 
List of Abbreviations .................................................................................................. xii 
Chapter 1: Introduction ................................................................................................. 1 
1.1 Socioeconomics and Climate ? Present State and Future Projections .......... 2 
1.1.1 Current State of the World ........................................................................... 2 
1.1.2 Future Changes to the Socioeconomic and Climate Systems ...................... 4 
1.1.3 Scenarios for Future Socioeconomic Development (the SSPs) ................... 6 
1.1.4 Scenarios for Future Climatic Change (the RCPs) .................................... 10 
1.1.5 The IPCC and Scenarios of the Future ...................................................... 11 
1.2 Water Usage ? Supply, Demand, and Projections of Change ........................... 13 
1.2.1 Overview of Relevant Water Cycle Components ...................................... 13 
1.2.2 Current Use and Availability on Global Scales ......................................... 14 
1.2.3 Projected Changes to Water Use and Supply ............................................ 16 
1.3 The Global Change Assessment Model (GCAM) ............................................ 19 
1.3.1 What is Integrated Assessment Modeling? ................................................ 19 
1.3.2 The Global Change Assessment Model ..................................................... 20 
1.3.3 Modeling water in GCAM ......................................................................... 21 
1.3.3.1 Water Demand .................................................................................... 21 
1.3.3.2 Water Supply ...................................................................................... 23 
1.4 Noted Research Gaps and Objectives ............................................................... 26 
Chapter 2: Water Sector Assumptions for the Shared Socioeconomic Pathways in an 
Integrated Modeling framework (published as Graham et al., 2018) ......................... 31 
2.1 Introduction ....................................................................................................... 31 
2.2 Methods............................................................................................................. 33 
2.2.1 Water Technology Assumptions for the Agricultural Sector .................... 36 
2.2.2 Water Technology Assumptions for the Electricity Sector ....................... 38 
2.2.3 Water Technology Assumptions for the Manufacturing Sector ................ 39 
2.2.4 Water Technology Assumptions for the Municipal Sector ........................ 40 
2.2.5 Water Technology Assumptions for the Livestock and Primary Energy 
Sectors ................................................................................................................. 41 
2.2.6 Non-Water Sector SSP Assumptions ......................................................... 43 
2.2.7 Scenario Description .................................................................................. 43 
2.3 Results ............................................................................................................... 44 
2.3.1 Total Global Water Withdrawals ............................................................... 44 
2.3.2 Agriculture and Livestock .......................................................................... 46 
2.3.3 Electricity Generation ................................................................................ 49 
2.3.4 Manufacturing and Primary Energy Production ........................................ 51 
2.3.5 Municipal ................................................................................................... 54 
2.3.6 Income Region Differences ....................................................................... 55 
2.3.7 Impact of SSP Assumptions and Water Constraints .................................. 58 
2.4 Discussion ......................................................................................................... 61 
2.5 Conclusions ....................................................................................................... 66 
Chapter 3: Humans drive future water scarcity across all Shared Socioeconomic 
Pathways (Graham et al., 2019 ? submitted) .............................................................. 68 
iv 
 
 
 
3.1 Introduction ....................................................................................................... 68 
3.2 Methods............................................................................................................. 69 
3.2.1 Scenario Components ................................................................................ 71 
3.2.3 Climate Derived Impacts from General Circulation Models ..................... 72 
3.2.4 Human and Climate Contributions to Water Scarcity ............................... 73 
3.3 Results ............................................................................................................... 75 
3.3.1 Main Driver Behind Water Scarcity Changes ........................................... 76 
3.3.2 Co-Influence of Main and Secondary Drivers ........................................... 80 
3.3.3 Representation Across the SSP-RCP framework ...................................... 84 
3.4 Discussion ......................................................................................................... 87 
3.4.1 Limitations ................................................................................................. 89 
3.4.2 Future Recommendations .......................................................................... 89 
3.5 Conclusions ....................................................................................................... 90 
Chapter 4: Future changes in the flows of virtual water (Graham et al., in prep) ...... 91 
4.1 Introduction ....................................................................................................... 91 
4.2 Methods............................................................................................................. 91 
4.2.1 Scenario Components ................................................................................ 93 
4.2.2. Calculation of virtual water components in GCAM ................................. 94 
4.2.3 Virtual water assumptions for GCAM ....................................................... 97 
4.3 Results ............................................................................................................... 98 
4.3.1 Global estimates of all sources of virtual water ......................................... 98 
4.3.2. Regional total virtual water trade ............................................................ 103 
4.3.3. Basin level virtual water exports ............................................................. 108 
4.4. Discussion ...................................................................................................... 111 
4.4.1 Limitations ............................................................................................... 111 
4.4.2. Looking forward ..................................................................................... 112 
4.5 Conclusions ..................................................................................................... 113 
Chapter 5:  Conclusions and Future Work ................................................................ 114 
5.1 Addressing Research Questions Posed ........................................................... 114 
5.1.1. Water sector assumptions for the SSPs ................................................... 114 
5.1.2. Water scarcity drivers across the SSP-RCP scenario matrix .................. 116 
5.1.3. Future virtual water trade analysis .......................................................... 118 
5.2 Future Work .................................................................................................... 120 
5.3. Concluding Remarks ...................................................................................... 122 
Bibliography ............................................................................................................. 124 
  
 
 
v 
 
 
 
List of Tables 
Table 2.1 A comparison of previous studies which have looked at technological 
change with the SSPs and the resulting effects on future water demands .................. 32 
Table 2.2 Qualitative assumptions for the SSP scenarios within GCAM. All non-
water sector assumptions adopted from Calvin et al. (2017). ..................................... 35 
Table 2.3 Quantitative withdrawal changes applied to the various water sectors 
within GCAM ............................................................................................................. 42 
Table 2.4 Quantitative consumption changes made to various sectors within GCAM
..................................................................................................................................... 42 
Table 2.5 Scenario names and components added to each of the five SSP scenarios 44 
Table 2.6 Global water withdrawals by sector (bcm/year). Values are given both for 
total withdrawals in water constrained scenarios with and without technological 
change (TC). ............................................................................................................... 45 
Table 2.7 Income level classifications by GCAM region, as defined in Calvin et al. 
(2017). ......................................................................................................................... 57 
Table 4.1 Global physical water flows ..................................................................... 102 
Table 4.2 Comparison of nonrenewable and blue water withdrawals. Comparison to 
historical and future projected water withdrawals and nonrenewable groundwater 
depletion through the end of the century across various future socioeconomic 
analyses. .................................................................................................................... 103 
 
vi 
 
 
 
List of Figures 
 
Figure 1.1 (UN, 2017) Global historical population changes from 1950 to 2017, with 
projections and uncertainties through 2100. Differences in fertility and death rates 
result in increasing uncertainty through the end of the century. ................................... 3 
Figure 1.2 (From O?Neill et al., 2017) Representation of the five SSP scenarios and 
the characteristics of each with respect to climate change adaptation and mitigation. . 7 
Figure 1.3 Population and GDP projections for each of the 5 SSP scenarios as 
represented in GCAM. Population declines are observed in SSP1 and SSP5 after 2050 
while GDP continues to rise throughout the century. ................................................. 10 
Figure 1.4 (From O?Neill et al., 2016) Scenario matrix for the SSP-RCP framework 
set to be used in the upcoming IPCC AR6. Previous scenarios are shown to the right 
in green, with extensions to the SSPs shown as the white and blue boxes. All dark 
blue boxes, entitled Tier 1, are set to be the main, initial set of scenarios used by the 
IPCC in order to capture plausible future scenarios in which the socioeconomics and 
climate system combine to reach an end-of-century radiative forcing target. ............ 13 
Figure 1.5 (From IPCC, 2014) Temperature and precipitation changes based upon 
two RCP scenarios. Precipitation extremes are observed under RCP8.5 with increases 
in the northern high latitudes and along the tropics. Values for warming and 
precipitation are dampened under a RCP2.6 scenario showing the impact of future 
anthropogenic global warming. .................................................................................. 19 
Figure 1.6 Intensification of water supply from 2005 to 2100 as calculated from 
Xanthos (Section 1.3.3.2). Values consider the average water supply as calculated 
from five GCMs for each RCP scenario. Values below 1, depicted in red, represent 
decreases in available surface runoff from 2005 values, while values above 1, 
depicted in blue, represent relative increases in the amount of available water from 
2005 values. Intensification calculation provided as QR2100/QR2005????..25 
Figure 2.1 Global yearly water withdrawals by SSP scenario and sector (bcm/year). 
The top row shows scenarios run without technological change in the water sector, 
while the bottom row includes technological change in the SSP storylines. .............. 45 
Figure 2.2 Extent of irrigated land within GCAM across the five SSP scenarios.  
Values shown here represent the extent of irrigated land in the Tech_const scenario. 
The extent of irrigated land is determined endogenously in GCAM and depends 
solely on economics in our analysis, as explained in Calvin et al. (2017). In GCAM 
we allow the relative profitability of irrigated and rainfed crops to determine the 
irrigation shares and geographic location and extent of irrigated land. ...................... 47 
Figure 2.3 Global water withdrawals from irrigated agriculture and livestock 
production (bcm/year).   The top panel shows scenarios run without technological 
change in the water sector while the bottom panel includes technological change. ... 48 
Figure 2.4 A comparison of future irrigation water withdrawals (bcm/year) across 
several studies. The figure shows studies that have attempted to account for future 
socioeconomic changes on the agricultural sector; not all of these have provided 
future sectoral changes as applied in this study. The values from this study, shown as 
solid colored lines, account for future technological change and are the same values 
seen in the Tech_const panel of Figure 2.3. Studies shown in this comparison have 
several differences in underlying assumptions and are shown here to represent a 
comparison between different socioeconomic impacts on agricultural withdrawals. 
vii 
 
 
 
(Alcamo et al., 2007; Wada et al., 2016; Hanasaki et al., 2013b; Hejazi et al., 2014b; 
Shiklomanov, 2000; Shen et al., 2008) ....................................................................... 49 
Figure 2.5 Water withdrawals by power plant cooling technology (bcm/year). The 
top panel shows scenarios run without technological change in the water sector while 
the bottom panel includes technological change. ....................................................... 51 
Figure 2.6 Water withdrawals from manufacturing and primary energy sectors 
(bcm/year).  The top row shows scenarios run without technological change in the 
water sector, Ref_const, while the bottom row includes technological change in the 
SSP storylines, Tech_const. ........................................................................................ 53 
Figure 2.7 Comparison of industrial water withdrawals across several studies 
(bcm/year). GCAM does not explicitly model Industrial water demand, instead, a 
combination of water demands for electricity generation and manufacturing goods 
and services are used to allow for comparison to other studies. Values from this study, 
as solid colored lines, are shown for the Tech_const scenario. (Wada et al., 2016; 
Fujimori et al., 2016; Hanasaki et al., 2013a; Shiklomanov, 2000) ........................... 53 
Figure 2.8 Comparison of municipal water withdrawals across several studies 
(bcm/year). Studies shown in this comparison have offered the potential for 
socioeconomics to influence future demands within the municipal water sector. 
Values depicted from this study, shown in solid colored lines, are depicted for the 
Tech_const scenario. (Bijl et al., 2016; Wada et al., 2016; Hanasaki et al., 2013a; 
Hejazi et al., 2013b; Hejazi et al., 2014b; Shiklomanov, 2000) ................................. 55 
Figure 2.9 Global water withdrawal differences by income region and sector in each 
SSP scenario (bcm/year). High-income regions are shown in the top panel, medium-
income regions in the middle panel, and low-income regions in the bottom panel. 
Income regions are defined in Table 2.7. Differences shown represent reductions due 
to technological change and are calculated as Tech_const minus Ref_const. ............ 56 
Figure 2.10 Global water withdrawal changes (bcm/year) as a result of step-wise 
addition of SSP assumption components. The net impact on global water withdrawals 
is represented as the solid black line. .......................................................................... 59 
Figure 2.11 Change in global water withdrawals as a result of water constraints (top 
row) and water technologies (bottom row).  Top left figure represents the impact of 
adding water constraints to the SSP scenarios run with the existing set of SSP 
assumptions for GCAM (Calvin et al., 2017). The top right figure represents the 
impact of adding water constraints to the SSP scenarios run with the assumptions 
from this study. The bottom left panel represents the impact of adding the water 
technology changes outlined in this study to an unconstrained water set of scenarios. 
The bottom right figure represents scenarios with added water technologies while 
under water constraints. .............................................................................................. 60 
Figure 3.1 Scenario breakdown of 90 total scenarios and the SSP-RCP scenario 
matrix depicting the set of 15 scenarios (green shading) in which plausible solutions 
exist in GCAM and for which CMIP5 climate datasets are available. ....................... 71 
Figure 3.2 Spatial and temporal changes in the percentage of GCAM?s water basins 
in which the main drivers of water scarcity changes are attributed to humans (H) or 
climate (C) and whether the changes increase (+), decrease (-), or have negligible 
changes in water scarcity. Each scenario is aggregated into individual SSP scenarios 
and values are calculated from the total number of nonnegligible DWSI basins and 
total number of scenarios in each aggregation. The ?All? scenario represents the total 
of all basins in the suite of SSP-RCP-GCM scenarios ............................................... 77 
viii 
 
 
 
Figure 3.3 Main driver of water scarcity changes due to climate (green) or humans 
(red) in 2050 (top) and 2100 (bottom). Robustness of results shown as the degree of 
shading in which there is greater than (darker) or less than (lighter) 95% agreement 
across all 75 SSP-RCP GCM scenarios. Basins which observed negligible, ?Neg?, 
water scarcity changes are shaded as such if at least 95% of the scenarios agree for 
that particular basin, all basins with less confidence are shaded according to their 
main driver in the nonnegligible scenarios. ................................................................ 78 
Figure 3 4 Percentage of basins with climate (green) or humans (red) as main driver 
of water scarcity changes. All negligible basins have been removed and values 
represent the percent of remaining 75% of basins in which humans or climate 
dominated water scarcity changes. Box and whisker plots represent the uncertainty 
spread among GCM and RCP combinations. ............................................................. 78 
Figure 3.5 Main driver for water scarcity changes by SSP scenarios in 2100, as in 
Figure 3.2B. Classification is determined by the percentage of occurrence in each SSP 
scenario. SSP-based basins where water scarcity changes are deemed negligible are 
highlighted in gray. ..................................................................................................... 79 
Figure 3.6 Temporal changes in the percentage of GCAM?s water basins in which the 
simultaneous impact of human and climate systems on water scarcity changes are 
shown by component and sign of change. Each scenario is aggregated into individual 
SSP scenarios and values are calculated from the total number of nonnegligible DWSI 
basins and total number of scenarios in each aggregation. The ?All? scenario 
represents the total of all basins in the suite of SSP-RCP-GCM scenarios. ............... 81 
Figure 3.7 Water scarcity category in 2050 (top) and 2100 (bottom). Notation shown 
as (+) representing either human (H) or climate (C) increasing water scarcity and (-) 
decreasing water scarcity. (ex. H+C+ humans and climate both act to increase water 
scarcity in given basin. 95% agreement across all 75 SSP-RCP GCM scenarios. 
Basins which observed negligible water scarcity changes are shaded as such if at least 
95% of the scenarios agree for that particular basin, all basins with less confidence 
are shaded according to their scarcity category in the nonnegligible scenarios. ........ 82 
Figure 3.8 Water scarcity change category by SSP scenarios in 2100, as in Figure 
3.5B. Classification is determined by the percentage of occurrence in each SSP 
scenario. SSP-based basins where water scarcity changes are deemed negligible are 
highlighted in gray. ..................................................................................................... 83 
Figure 3.9 Human and climate impact quantifications across the 15 SSP-RCP 
scenarios in 2100. A, B, C, Numerical quantification of the relative human (IH, x-axis) 
and climate (IC, y-axis) impacts on water scarcity in 2100 across each of GCAM?s 
235 water basins. SSP-RCP combinations shown with no plots represent scenarios 
unattainable either through mitigation (i.e. SSP3-RCP2.6) or baseline assumptions 
fail to reach 2100 forcing levels (i.e. SSP1-4-RCP8.5). A, Probability distribution 
function of the human impact across each SSP scenario including the aggregates of 
all RCP combinations. B, Numerical representation of the 5 GCM mean human and 
climate relative impact across each basin (points). The WSI of each basin in 2100 is 
shown as the size of points. As GCM uncertainty increases, points move inwards 
from the mathematical limits of equations 3A and 3B. C, Probability distribution 
function of the climate impact across each RCP scenario including the aggregates of 
all SSP combinations. ................................................................................................. 86 
Figure 4.1 Annual water flows of green, blue, and groundwater embedded in 
agricultural trade for SSP2-RCP6.0. Range of virtual green and blue water exports 
ix 
 
 
 
and the amount of nonrenewable groundwater depletion embedded in agricultural 
trade for all SSP2-RCP6.0 scenarios including GCM uncertainty. VGE is shown on 
primary y-axis (left) while VBE and VGWE are represented on secondary y-axis 
(right). Solid lines represent the average for each water flow in SSP2-RCP6.0. Virtual 
green and blue water exports are shown to at least triple from starting 2010 values 
with increases due to population changes and global production shifts. VGE show 
steep increases towards 2050, with gains much smaller thereafter, reflective of the 
SSP2 population trajectory. VBE has a much steadier gain throughout the century 
with nearly consistent per-year increases. VGWE (purple) undergo a five-fold 
increase towards 2050 and gradual decreasing thereafter. ........................................ 100 
Figure 4.2 Per capita exports of each virtual water trade component by region. A, 
Virtual green water exports per capita. Argentina, Australia, and Canada represent the 
three highest per capita exporters of green water with general upwards trends 
throughout the century. B, Virtual blue water exports per capita. Significant upwards 
trends are shown in China and Southeast Asia, while declines occur in Pakistan and 
the Middle East after 2050. C, Virtual groundwater exports per capita. Peaks are 
observed in Pakistan and the Middle East as groundwater is extracted early in the 
century as the depletion lessens after 2050, the per capita values drop to zero. 
Increases are observed in Australia, Argentina, and Mexico. ................................... 101 
Figure 4.3 Virtual water trade fluxes by region and crop in 2010 and 2100. A, B, 
Global virtual green water trade (bcm) by crop and aggregate GCAM region in 2010, 
A, and 2100 for SSP2-RCP6.0, B. All green water traded is from rainfall that grows 
crops viable for consumption. Imports (negative values) are assumed to come 
proportionately from exporting regions dependent upon total regional exports of a 
crop type. Water intensities for these imports are then scaled dependent upon the 
proportionality of exporting region intensities. C, D, Global virtual blue water trade 
(bcm) by crop and aggregate GCAM region in 2010, C, and 2100 for SSP2-RCP6.0, 
D. Scaling of imports follows the same methods for green water. E, F, Global virtual 
groundwater trade (bcm) by crop and aggregate GCAM region in 2010, C, and 2100, 
D. Values for imports follow same logic as was used for VWT, with exporting 
regional nonrenewable to renewable water use calculated and applied. .................. 105 
Figure 4.4 Virtual water trade fluxes by region and crop in 2030 and 2050. A, B, 
Global virtual green water trade (bcm) by crop and aggregate GCAM region in 2030, 
A, and 2050 for SSP2-RCP6.0, B. All green water traded is from rainfall that grows 
crops viable for consumption. Imports (negative values) are assumed to come 
proportionately from exporting regions dependent upon total regional exports of a 
crop type. Water intensities for these imports are then scaled dependent upon the 
proportionality of exporting region intensities. C, D, Global virtual blue water trade 
(bcm) by crop and aggregate GCAM region in 2030, C, and 2100 for SSP2-RCP6.0, 
D. Scaling of imports follows the same methods for green water. E, F, Global virtual 
groundwater trade (bcm) by crop and aggregate GCAM region in 2030, C, and 2050, 
D. Values for imports follow same logic as was used for VWT, with exporting 
regional nonrenewable to renewable water use calculated and applied. .................. 106 
Figure 4.5 Crop breakdown of each virtual water trade component. A, Virtual green 
water exports by crop, from 2010 to 2100. An intensification of every crop type is 
seen throughout the century. B, Virtual blue water exports by crop and region. 
Increases in wheat and rice make up the largest portion of virtual blue water exports. 
x 
 
 
 
C, Virtual groundwater exports by crop. Quick increases in rice and miscellaneous 
crop trade is observed through 2050. ........................................................................ 107 
Figure 4.6 GCAM region breakdown of each virtual water export component. A, 
Virtual green water exports and from GCAM regions, from 2010 to 2100. Brazil and 
Northern Africa observe the largest increases in net exports. B, Virtual blue water 
exports by region. An intensification of exports from China is observed. C, Virtual 
groundwater exports by region. An intensification of exports from Pakistan and the 
Middle East prior to 2050. After 2060, the groundwater resources become increasing 
exhausted and more expensive, therefore the exports from these regions cease. ..... 108 
Figure 4.7 Basin level virtual water exports in 2050 and 2100 for all sources, Virtual 
green water exports (bcm) A and B, and blue water exports (bcm), C and D, in 2050 
and 2100 respectively for the average of five GCM runs for SSP2-RCP6.0. Virtual 
groundwater exports (bcm) in 2050 and 2100 for the same averaged GCM runs for 
SSP2-RCP.6.0 is shown in E and F. All values are considering the exports of 
agricultural crops only with additional, potentially necessary virtual water imports not 
considered. VGE are concentrated in much of China, central North America, and 
eastern South America. Exports of blue water come mainly from China, and the 
Missouri River basin in the United States. Virtual groundwater extraction in 
agricultural trade is largest in the California River basin, the Arkansas River basin, 
southwestern North America, the Nile River basin, western South America, and the 
Murray-Darling basin in Australia. ........................................................................... 110 
 
 
  
xi 
 
 
 
List of Abbreviations 
 
AEEI Autonomous energy efficiency improvement 
AR5/6 Assessment Report 5 or 6 
bcm Billion cubic meters 
BWC Blue water consumption 
BWW Blue water withdrawals 
CCS Carbon-capture-storage 
CH4 Methane 
CMIP Coupled Model Intercomparison Project 
CO2 Carbon dioxide 
EU European Union 
FAO Food and Agricultural Organization of the United 
Nations 
GCAM Global Change Assessment Model 
GCMs General circulation models 
GDP Gross Domestic Product 
GWC Green water consumption 
GWD Groundwater depletion 
IAM Integrated Assessment Model 
IAV Impact, Adaptation, and Vulnerability 
IC Relative climate impact on water scarcity 
IH Relative human impact on water scarcity 
IPCC Intergovernmental Panel on Climate Change 
ISIMIP Inter-Sectoral Impact Model Intercomparison Project  
LULCC land-use land-cover change 
N2O Nitrous oxide 
NDCs Nationally Determined Contributions 
OECD Organization for Economic Cooperation and 
Development 
PV Photovoltaics 
RCPs Representative Concentration Pathways 
Ref_const Refence SSP scenario run with water constraints 
SC Water scarcity changes due to climate 
SDGs Sustainable Development Goals 
SH Water scarcity changes due to humans 
SLCF Short-lived climate forcers 
SPAs Shared Policy Assumptions 
SSPs Shared Socioeconomic Pathways 
SX Water scarcity changes due to humans and climate 
TC Technological change 
Tech_const SSP scenario run with water sector technological 
assumptions and water constraints 
UN United Nations 
VBE Virtual blue water exports 
VBI Virtual blue water imports 
xii 
 
 
 
VBT Virtual blue water trade 
VGE Virtual green water exports 
VGI Virtual green water imports 
VGT Virtual green water trade 
VGWE Virtual groundwater exports 
VGWT Virtual groundwater trade 
VIC Variable Infiltration Capacity 
VWE Virtual water exports 
VWT Virtual water trade 
WMO World Meteorological Organization 
WSI Water scarcity index 
 
 
xiii 
 
 
 
Chapter 1: Introduction  
Water is essential to sustain life on Earth. Whether directly or indirectly, water 
is contained in everything that humans use on a day to day basis. The food and 
beverages that are consumed to fuel the human body contain water. Any form of 
transportation requires water to either manufacture the materials or to extract or create 
the fuel that runs the vehicle. The computer with which this thesis is being written 
required water to fabricate each of the components and requires water to provide the 
electricity allowing the battery to be charged each day. No matter where one looks, 
water surrounds them, but that water is becoming increasingly difficult for some 
people to access. Whether the demands are too high due to population increases, 
supplies are too low from climate change induced availability changes, the price is 
too high to afford clean water, water quality inhibits human consumption, or armed 
conflict has created competition between regions for water sources, the way in which 
humans use water is changing. It is extremely important to understand how the 
supplies and demands of water may change into the future in order to make policy-
relevant decisions to combat the potential negative consequences. In order to 
accomplish this, projections into the future must consider how the human system will 
change the use of water, spatially and temporally, by considering population growth 
changes, economic growth rates, and technological advancements that increase 
efficiencies. In addition, these projections must also consider how the climate system 
will impact the availability of water by shifting precipitation patterns as a result of 
anthropogenic global warming. 
1 
 
 
 
 Recent advances to select Integrated Assessment Models (IAMs) have 
allowed for the modeling of scenario-based ?what-if? statements to investigate the 
future of water. The studies that follow utilize the human-climate system feedback 
features of an IAM to gain an understanding of how the use of water might change 
into the future if human and climate systems evolve in predefined, yet dynamical 
ways. Investigations are conducted on how the demands of water change with 
socioeconomically viable efficiency improvements, how the evolving human and 
climate system interact to change the drivers of water scarcity, and how reliant the 
global trade market will be on the trading of water intensive agricultural goods across 
regional boundaries as a result of various socioeconomic and climate scenarios. These 
are assessed across a wide range of global futures which provide differences in the 
supply, demand, and access to water across global regions. These studies are 
conducted to provide a comprehensive first-step approximation to how, when 
dynamic human-climate feedbacks evolve in the future water use, scarcity, and the 
dependency on external sources of water change. Socioeconomic and climate system 
changes in the future are highly uncertain, but this thesis attempts to investigate a 
wide range of future outcomes to account for this uncertainty. 
1.1 Socioeconomics and Climate ? Present State and Future Projections 
1.1.1 Current State of the World 
 
Socioeconomic and climate systems feedback each other in an evolving way 
with resultant changes to one system affecting the other in both global and localized 
fashions. In order to make assessments for the future growth of both socioeconomic 
and climate systems, it is important to understand how the historical evolution of each 
has allowed the human-Earth system to arrive at its current state.  
2 
 
 
 
According to the United Nations, the global population has tripled since 1950 
to 7.5 billion people, recently climbing by 1 billion people between 2005 and 2017 
alone (Figure 1.1). This increase can be attributed to high fertility rates in Asia and 
increasing fertility rates in Africa, whereas much of the remainder of the world has 
observed minor changes in fertility (UN, 2017).  
In recent years the global economy has observed steady 3% per year growth in 
GDP, however smaller regions in developing countries are not observing such growth 
due to armed conflicts and an overall lack of diversification (UN 2019). This 
discontinuity between regions and income levels raises concerns about the ability to 
implement agreed upon nationally determined contributions (NDCs) in hopes of 
meeting the sustainable developments goals (SDGs) set forth in the Paris Agreement 
(Richards et al., 2016) 
 
Figure 1.1 (UN, 2017) Global historical population changes from 1950 to 2017, with projections and 
uncertainties through 2100. Differences in fertility and death rates result in increasing uncertainty 
through the end of the century. 
 The state of the climate system is reported yearly through the World 
Meteorological Organization (WMO) in order to assess how a suite of climatic 
change indicators have changed in the most recent year. In 2018, the global mean 
3 
 
 
 
temperature was the fourth highest on record, while atmospheric concentrations of 
carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), reached the highest 
observed values on record. The combination of these effects has a compounding 
impact on the climate system, resulting in, but not limited to, increased ocean heat 
content, sea level rise, ocean acidification, and increased atmospheric water vapor 
content (Vermeer and Rahmstorf, 2009; Held and Soden, 2006; Balmaseda et al. 
2013; IPCC, 2014). There were also several areas around the world that experienced 
higher than average rainfall, such as Eastern North America and much of northern 
India during the summer monsoon. However, the precipitation and resultant water 
availability was much lower than historical average in areas of Europe, Australia, and 
southern India (WMO, 2019). While several of these results may be attributed to 
natural interannual variability, the trends of increasing greenhouse gas concentrations 
and global mean temperature are virtually certain to continue (IPCC, 2014), are likely 
have large impacts into the future, and must be understood in order to properly assess 
the future of the human-Earth system. 
1.1.2 Future Changes to the Socioeconomic and Climate Systems 
 
Future projections of the socioeconomic system consist of assumptions of how 
the population and economy will evolve over the course of the century. The United 
Nations makes projections through the end of the century of how fertility and death 
rate will change regionally and globally. Current projections of global population 
growth (Figure 1.1) result in an increase from 7.5 billion people today, up to 11.2 
billion people in 2100 (UN, 2017).  This is due to continued increases in the fertility 
rate in Africa while the population growth slows in much of Asia after 2050.  
4 
 
 
 
Changes to global, regional, and country level economics through the end of 
the century require an understanding of current trends. The Organization for 
Economic Cooperation and Development (OECD) use the ENV-Growth modelling 
framework to project how the global economy will evolve into the future. However, 
the results are typically assumed, through the conditional convergence hypothesis, 
that GDP among differing income regions will converge as lower income regions 
experience quick growth, while higher income regions are near the peak of their GDP 
potentials (Chateau et al., 2014).  Other organizations make projections at smaller 
timescales based on current trends and without the conditional convergence 
hypothesis. Current UN projections have poor regions continuing to struggle in the 
global economy towards 2030 as trade policy disputes put the poorest regions at the 
forefront of the global economic struggle. Near term climate risks are also expected to 
create unfavorable economic prospects for island nations in the Caribbean, Indian 
Ocean, and Southwest Asia as precipitation extremes, sea-level rise, and decreases in 
freshwater availability, increase the dependence on imports, while potentially 
reducing the net exports and profitability from these nations (UN, 2019). 
 The climate system is nearly impossible to make projections for without the 
use of future scenarios and assumptions about how emissions, atmospheric 
concentrations of climate forcers, and land use land cover change (LULCC) will 
evolve into the future. The Intergovernmental Panel on Climate Change (IPCC) 
makes conditional statements on the likelihood of future changes to the climate 
system based upon the analysis of predetermined scenarios of the future (described in 
section 1.1.4). The agreement across studies allows for the IPCC to make assessments 
5 
 
 
 
of the future, however most forward-thinking assessments merely provide trajectories 
from present day conditions. 
1.1.3 Scenarios for Future Socioeconomic Development (the SSPs) 
Understanding how the human system will evolve into the future is not a 
trivial task and requires significant assumptions on both global and regional scales. 
UN population projections typically use a ?business-as-usual? methodology when 
projecting fertility and death rates based upon current observed trends, which result in 
a fairly small uncertainty range surrounding future projections (Figure 1.1). 
Combined with the OECD economic growth projections, these do not encompass a 
wide range of potential futures and leave the scientific community with nearly single 
projections. In order to account for a wide range of outcomes, scenarios are often 
developed with assumptions of how population, economic growth, and technological 
change will progress in the future by varying the projections made by such bodies as 
OECD and the UN. The most recent set of scenarios which have been adopted by the 
social and environmental science communities are the Shared Socioeconomic 
Pathways (SSPs) (O?Neill et al., 2017). The SSPs represent a set of five scenarios that 
are intended to span a set of futures to describe how socioeconomics result in making 
the mitigation of, or adaptation to, climate change harder or easier, with no direct 
consideration for climate change. Climate change considerations associated with the 
SSP framework are separated in order to account for human responses alone. The 
addition of climate change considerations are discussed in Section 1.1.5.  The five 
scenarios have been placed on a 2D plane to emphasize the low or high magnitude of 
these challenges by scenario (Figure 1.2). Each of these scenarios have storylines 
which have been expanded into quantitative values across each of the SSPs for 
6 
 
 
 
population (Samir and Lutz, 2017), economic growth (Dellink et al. 2017), and land 
use change (Popp et al., 2017). Further quantification of SSP assumptions has 
occurred within individual integrated assessment modeling teams based upon the 
applicability to each specific model. This has created an uncertainty range 
surrounding the potential outcomes of each SSP while also producing ?marker? 
scenarios by different modeling groups. ?Marker? scenarios were chosen based upon 
an individual model?s ability to represent the distinct characteristics of the SSP 
storyline (Riahi et al., 2017). Scenario storylines for each SSP are provided below.   
 
Figure 1.2 (From O?Neill et al., 2017) Representation of the five SSP scenarios and the 
characteristics of each with respect to climate change adaptation and mitigation. 
 SSP1 (Sustainability) assumes that throughout the century the population 
becomes more educated while shifting focus to meet development goals and adopting 
sustainable lifestyles through infrastructure efficiency improvements and diet 
changes. Population growth slows, leading to overall population declines in the 
second half of the century. Renewable energy becomes increasingly desirable while 
environmentally friendly technologies are adopted on a grand scale. Population 
7 
 
 
 
declines and global-first mentality allow from decreases in the demands for water 
intensive goods and the ability to freely trade in a global market. The focus on 
sustainability and willingness to invest in renewables allow for minimal challenges to 
future climate change mitigation and adaptation (O?Neill et al., 2017; van Vuuren et 
al., 2017).  
SSP2 (Middle of the Road) assumes that throughout the century technological 
change rates, population growth, and economic growth do not change drastically from 
historical values. Population levels off towards 2100 while educational and economic 
differences remain across regional boundaries. Global demands follow population 
projections and trading remains in the global market. As socioeconomic changes are 
small compared to today, this business-as-usual scenario represents a future with 
moderate challenges to both mitigation and adaptation (O?Neill et al., 2017; Fricko et 
al., 2017).  
SSP3 (Regional Rivalry) assumes that an increasing focus on meeting 
domestic needs, with little cooperation across regional boundaries, drives an 
increasingly fragile global economy. Limitations to future trade result in a need for an 
intensification of fossil fuel use, particularly in developing countries. Education is 
limited and population growth increases throughout the century, resulting in a global 
population over 12 billion people by 2100. The struggling economy, drastic increase 
in global demands of goods and services, a regional market and lack of a globally 
thinking society, results in an inability to invest in mitigation technologies and 
adaptation measures creating large challenges to both future climate mitigation and 
adaptation (O?Neill et al., 2017; Fujimori et al., 2017). 
8 
 
 
 
 SSP4 (Inequality) depicts a divergence in economic performance between 
regions as the century progresses, creating an increasing gap between the rich and 
poor. Higher income regions invest in efficient technologies and become increasingly 
educated, while in poor regions, people are forced to live in a labor-intensive 
environment and utilize a low technology economy. Global demands for goods and 
services remain highly dependent upon income level and trading with and between 
the poorest regions is almost non-existent. Therefore, the investment in, and adoption 
of, mitigation technologies is very high, due in large part to the wealthiest countries, 
but adaptation is more difficult particularly in the lower-income regions (O?Neill et 
al., 2017; Calvin et al., 2017).  
SSP5 (Fossil-fuel Development) includes strong economic development 
throughout the century, particularly in poor regions. This leads to a globalization of 
trade and an increasingly successful global economy. Education increases, yet the 
success of the economy leads to an energy-intensive lifestyle and a continued 
exploitation of fossil-fuel resources. However, population declines and educational 
increases lead to lowering demands while a booming economy allows for a global 
trade market. This dependency on fossil fuels makes mitigation efforts difficult, but 
high incomes lead to low challenges to adaptation (O?Neill et al., 2017; Kriegler et 
al., 2017).  
While important to understand, the qualitative assumptions are implemented 
in slightly different ways across modeling groups. For the purposes of the following 
studies, quantitative assumptions of each scenario are shown in Chapter 2 and in 
Calvin et al. (2017). These assumptions have been made for several non-water sectors 
(Table 2.2) and are expanded to the water sector to better represent socioeconomic 
9 
 
 
 
impacts on water technologies (Chapter 2 and Graham et al. 2018). Future projections 
of global population and GDP, as implemented in GCAM (Section 1.3), are shown in 
Figure 1.3. 
 
 
Figure 1.3 Population and GDP projections for each of the 5 SSP scenarios as represented in GCAM. 
Population declines are observed in SSP1 and SSP5 after 2050 while GDP continues to rise throughout 
the century.  
1.1.4 Scenarios for Future Climatic Change (the RCPs) 
The scientific community use climate change scenarios as a basis for 
modeling projections of the future, impact, adaptability, and vulnerability (IAV) 
10 
 
 
 
analysis, and to investigate mitigation strategies. The process of creating these 
scenarios follows a similar path as that of the socioeconomic scenarios described 
earlier. The International Panel on Climate Change (IPCC) has been using climate 
change scenarios since the release of The First Assessment Report (1990). Since this 
time, scenarios have evolved to include increasing levels of detail about future 
emissions to best estimate climatic impact. The most recent set of climate change 
scenarios are the Representative Concentration Pathways (van Vuuren et al., 2011). 
The RCPs consist of differing land-use change, atmospheric greenhouse gas 
emissions, and short-lived climate forcers (SLCF) concentrations derived from 
previous literature in the IAM community (Riahi et al., 2007; Fujino et al., 2006; 
Hijioka et al., 2008; Clarke et al., 2007; Smith and Wigley 2006; Wise et al., 2009; 
van Vuuren et al., 2007; van Vuuren et al., 2006). This data from the IAM community 
was harmonized for use in climate models in order to obtain a set of consistent futures 
for each RCP scenario (Lamarque et al., 2010; Hurtt et al., 2011; Meinshausen et al., 
2011). These scenarios span a set of four end-of-century increases in climate forcing 
from 2.6 W/m2 to 8.5 W/m2. The RCPs are prescribed to climate models in order to 
provide analysis of how the earth system will change by following these four 
emissions pathways and make up the basis for the IPCC Assessment Report 5 (AR5).  
1.1.5 The IPCC and Scenarios of the Future 
The IPCC is currently divided into three distinct Working Groups that have 
differing focuses with regards to climate change. Each Working Group creates a 
report on 1) the physical science behind the past, present, and future changes to the 
climate system, 2) the impacts, adaptation, and vulnerability associated with climate 
change, or 3) mitigation of climate change, respectively. The analysis that is 
11 
 
 
 
completed, and the literature reviews that make up the reports for the IPCC are based 
upon a set of scenarios consistent across modeling communities. In the most recent 
report, the RCPs provided the basis for analysis.  
For Assessment Report 6 (AR6) from the IPCC, a new set of scenarios will be 
used for future climatic change analyses. Following the work of Moss et al. (2010), 
the scientific community continues to evolve scenarios to include more detailed 
information about socioeconomics and climate change. For this reason, the SSPs and 
RCPs are being combined to form a set of future global change scenarios (Figure 1.4) 
that allow for comprehensive socioeconomic assumptions to be matched with future 
radiative forcing pathways to achieve future global warming targets (O?Neill et al., 
2016; Eyring et al., 2016). Each scenario is then matched with specific Shared Policy 
Assumptions (SPAs) in order to account for differences in adaptation and mitigation 
strategies across the SSPs (Kriegler et al., 2014; Calvin et al., 2017). These scenarios 
have been developed by the IAM community with greenhouse gas emissions, 
atmospheric concentrations of greenhouse gases, and land-use changes once again 
being provided to the climate modeling community for implementation into general 
circulation models (GCMs).  
12 
 
 
 
 
Figure 1.4 (From O?Neill et al., 2016) Scenario matrix for the SSP-RCP framework set to be used in 
the upcoming IPCC AR6. Previous scenarios are shown to the right in green, with extensions to the 
SSPs shown as the white and blue boxes. All dark blue boxes, entitled Tier 1, are set to be the main, 
initial set of scenarios used by the IPCC in order to capture plausible future scenarios in which the 
socioeconomics and climate system combine to reach an end-of-century radiative forcing target. 
 
1.2 Water Usage ? Supply, Demand, and Projections of Change 
1.2.1 Overview of Relevant Water Cycle Components 
The water cycle consists of several components that define the amount of 
water in the Earth system at any given time. These components are often represented 
in a water budget as shown in Equation 1.1, adopted and modified for these studies 
from Eagleson (1978), where P represents the precipitation, ET the 
evapotranspiration, QR the available renewable surface runoff and groundwater 
extraction, Wd the available water withdrawn for human use, SG the nonrenewable 
groundwater extracted from deep aquifer storage, and DST as the change in total water 
storage. 
 
? ? ?+ + (?/ ??1) ? ?4 = ??+ 1.1 
 
13 
 
 
 
For the purposes of this research, the focus will be placed on four components 
of Equation 1.1, explained hereafter with their GCAM (Section 1.3) definitions: ET, 
or the amount of evapotranspiration that occurs during crop growth, also referred to 
as green water consumption (Hoekstra et al., 2011). Wd, or the amount of surface 
runoff or groundwater recharge that is withdrawn in the production of all global 
goods and services, referred to as blue water (Hoekstra et al., 2011). SG, or the 
amount of nonrenewable groundwater that is extracted from deep aquifers in the 
production of global goods and services, an extension of blue water. Finally, QR is the 
total renewable runoff and groundwater recharge that is available for use in global 
production. QR represents the total amount of blue available water. Each of these 
values changes yearly dependent upon both human and climate systems and 
understanding how much, and which mechanisms cause these changes is the purpose 
of the following studies.   
1.2.2 Current Use and Availability on Global Scales 
The availability of water around the world relies heavily on the precipitation 
in a region providing enough surface runoff for human consumption. Recently, 
extreme precipitation events (i.e. floods and droughts) have been shown to be the 
worst since 1950 (Arndt et al., 2010). This has been attributed to increases in 
anthropogenic global warming (Zhang et al. 2007). Available surface water has seen 
decreases in low and mid-latitudes over the period 1940-2004, whereas areas where 
most winter precipitation has historically fallen as snow, are seeing increasing 
wintertime rainfall and increased water availability. This has been increasing the 
water availability at these locations, particularly in sub-seasonal timescales (Dai, 
2013). 
14 
 
 
 
 The global use of water has been increasing, largely due to increases in GDP 
and population in developing regions (Shiklomanov, 2000; Alcamo et al., 2007; 
WWAP 2015). It is estimated that the global demands for water are increasing at 1% 
per year due to population growth, economic development and technological 
advancements changing global consumptive patterns (WWAP, 2018). It is estimated 
that roughly 3850 km3 of freshwater are withdrawn for human use each year (FAO, 
2016). Of this, 69% is used for agricultural purposes, 19% for industrial uses, and 
12% for domestic needs. When considering the consumption of freshwater, nearly 
90% of the water that is not returned to streamflow is used for agricultural purposes 
(FAO, 2016). 
As a result of the drying of low to mid-latitude regions, water scarcity is 
increasing and causing increased stress on the ability to meet demands for water 
intensive products. In recent years, countries around the world have become 
increasingly reliant on imports of water intensive crops to meet demands. This 
process, known as virtual water trade (VWT) (Allan, 1998) has evolved as 
socioeconomic and climate conditions have created deficits in many countries around 
the world. Over the last three decades, the United States, Argentina, and Brazil 
represent some of the largest virtual water exporters (VWE) in the world, while 
China, through population increases and an increasing demand for soy products, has 
become the largest importing region for virtual water (Hoekstra and Chapagain, 2006; 
Dalin et al., 2012a; Carr et al., 2013). While country-to-country dynamics change 
with evolving socio-environmental conditions, global VWT over the 1986-2007 time 
period has more than doubled (Dalin et al., 2012a; Carr et al., 2013) to combat these 
human and clime influences.  
15 
 
 
 
1.2.3 Projected Changes to Water Use and Supply 
With projections of future GDP and population uncertain, future water 
demands remain highly uncertain. Several studies have investigated the effects of 
changing socioeconomics on future water demands with various results. 
Socioeconomic systems affect freshwater demands through population changes and 
economic growth by 1) increasing the number of people who require water and 2) 
affecting the affordability of reliable access to water and technological improvements 
to increase water sector efficiencies. In addition, socioeconomics affect what people 
demand (i.e., manufactured goods, food, domestic needs) which cause changes to the 
amount of water demands around the world.  Alcamo et al. (2007) found that water 
stress will increase as a direct result of increasing income and higher per capita water 
use in the future, while population acts as a secondary source of changes. Hanasaki et 
al. (2013 a, b) have estimated that across a set of pre-release SSPs, between 39% to 
55% of the population in the period 2071-2100 would be under severely water-
stressed conditions due to future population, demand changes in the municipal and 
industrial sectors, and changes in irrigated crop intensity, irrigation efficiency, and 
irrigated area. Hejazi et al. (2014b) found that future water demands will increase by 
31% to 242% by 2095 under a set of six future socioeconomic scenarios. Changes to 
water demands will alter the future stress placed upon the water availability of a 
location. Water scarcity is defined as the ratio of water demanded to the available 
water. Future changes to water scarcity have been attributed to population changes, 
economic growth, and resultant demand increases more so than to climate system 
impacts (V?r?smarty et al., 2000; Arnell, 2004; Alcamo et al., 2007; Hanasaki et al., 
2013b; Arnell and Lloyd-Hughes, 2014; Schewe et al., 2014; Schlosser et al., 2014; 
16 
 
 
 
Shen et al., 2014; Wada et al., 2014; Kiguchi et al., 2015). All results indicate that 
socioeconomic changes will be significant driving forces into the future, and that 
consequences will vary considerably by scenario.  
Water use is not just a local problem. Socioeconomic growth and future 
changes to the climate system lead to changes in food demand and shifts in global 
production causing changes to the global dependence on VWT (D?Odorico et al., 
2014; Distefano et al., 2018). An increasing emphasis is likely to be placed on the 
trading of goods across regional boundaries, particularly from water abundant regions 
to water stressed regions, as water availability decreases and demands cannot be met 
by domestic production alone (Hoekstra and Chapagain, 2008; Carr et al., 2013). 
Konar et al. (2013) found that total VWT is likely to decrease towards 2030 as 
climate change induced agricultural productivity increases cause the amount of water 
required to grow crops, or the virtual water content, to decline. Current VWT is 
determined to be unsustainable under worsening water scarcity into the future without 
shifts in global production (Orlowsky et al., 2014). Various future socioeconomic 
conditions are found to result in an increase of global VWT (Distefano and Kelly, 
2017; Ercin and Hoekstra, 2016), likely driven by the five main drivers of change 
identified by Ercin and Hoekstra (2014): population growth, economic growth, 
consumption patterns, global production and trade, and technological development. 
However, these analyses have thus far have not attempted to project future changes 
based upon the features in an IAM (Section 1.3).  
While future human dynamics are likely to cause the most significant impact 
to water use and scarcity around the world, the availability of water is likely to be 
altered by anthropogenic global warming driven hydrological changes. Climate 
17 
 
 
 
system warming is expected to cause precipitation shifts across regions (Figure 1.5) 
and therefore alter regional water availability, while increases in evapotranspiration 
and reductions in soil moisture will lead to an increased demand for irrigation water 
in certain areas of the world (Arnell 1999; D?ll 2002; Diaz et al. 2007; Fischer et al. 
2007; Kang et al, 2009; Wada et al., 2013). Surface runoff is expected to increase in 
the northern high latitudes and in current rainforests while decreasing in most dry 
tropical regions following the notion of ?wet get wetter, dry get drier?. The Xanthos 
(section 1.3.3.1) derived changes to water supply, or the amount of water available 
for human use, from 2005-2100 for each RCP is shown in Figure 1.6. Increases in 
water supply are observed in all RCP scenarios, particularly in northern high latitudes 
and northern Africa. Reductions in water availability increasingly occur within higher 
radiative forcing scenarios and are located over much of southwestern North 
America, northeastern South America, and southern Europe in RCP8.5. Future 
climate mitigation scenarios have been found to decrease the impact of climate 
factors (Blanc et al., 2014; Hanasaki et al., 2013; Wada et al., 2013; Arnell and 
Lloyd-Hughes 2014), but the increased demand for bioenergy to reach such targets 
will likely increase water demands (Hejazi et al. 2015; Yamagata 2018).  
18 
 
 
 
 
Figure 1.5 (From IPCC, 2014) Temperature and precipitation changes based upon two RCP 
scenarios. Precipitation extremes are observed under RCP8.5 with increases in the northern high 
latitudes and along the tropics. Values for warming and precipitation are dampened under a RCP2.6 
scenario showing the impact of future anthropogenic global warming. 
 
1.3 The Global Change Assessment Model (GCAM) 
1.3.1 What is Integrated Assessment Modeling? 
 Integrated Assessment Modeling is the process by which interdisciplinary sets 
of models are combined to analyze the interactions between human and Earth 
systems. In broad terms, this means linking agriculture, the economy, energy systems, 
and more recently the water cycle, to the carbon cycle in a consistent manor to allow 
these systems to coevolve and interact with one another (Calvin et al., 2019). This 
provides a unique opportunity to capture interactions between highly nonlinear 
systems outside of singular disciplinary framework. Historically IAMs have been 
19 
 
 
 
used to provide greenhouse gas emissions, concentrations of short-lived climate 
forcers, and LULCC to the suite of Coupled Model Intercomparison Project (CMIP) 
GCMs that are then forced by these variables to create comprehensive Earth-system 
analyses (As in Section 1.1.4). IAMs are also used to provide decision support for 
policy-makers by asking ?what if? questions about the future and providing potential 
outcomes in order to answer these theory questions. IAMs, much like climate models, 
are not meant to be used for fine temporal resolution decision making, rather these 
models are often used to analyze suites of scenarios in decadal time horizons. 
 While IAMs have several similarities amongst one another, there are also 
several differences. All IAMs have a global perspective that accounts for all 
anthropogenic sources of emissions and some representation of the climate system. 
However, across the family of IAM models, differences arise in the sophistication of 
the climate system, representation of land and agriculture, available spatial resolution, 
and sector specific representations (e.g. differences in economies, energy systems, 
etc.). For the purposes of the studies that follow, an IAM which has been used for the 
RCP and SSP scenarios (Wise et al., 2009; Calvin et al., 2017) has been chosen that 
has additional water sector capabilities, described in the sections that follow. 
1.3.2 The Global Change Assessment Model 
These investigations use the Global Change Assessment Model (GCAM) 
version 5.0, with updates to include water constraints and water technology 
assumptions (discussed below). GCAM is a global community Integrated Assessment 
Model that links socioeconomics, the energy system, land-use change, climate, and 
the water sector. GCAM has 32 energy-economy regions, 384 land regions, and 235 
global water basins (Calvin et al., 2019). GCAM uses the simple global system 
20 
 
 
 
carbon cycle climate model, hector (Hartin et al., 2015; Hartin et al., 2016) to track 
greenhouse gas emissions. GCAM tracks the emissions of 16 greenhouse gas that are 
then used as inputs into the CMIP suite of GCMs ex facto. As such, GCAM has been 
the primary model for the emissions suite for RCP4.5 (Thomson et al. 2011) and 
SSP4 (Calvin et al. 2017). GCAM has a historical spin-up period from 1975-2010 
solving at 15-year time steps through 2005 and then at 5-year time steps from 2005 
through 2100. GCAM is a market-equilibrium model that allows for prices to be 
adjusted within each time step to ensure that the supply and demand of goods and 
services remains equilibrated at each time step allowing for simultaneous market 
clearing across sectors. As mentioned earlier, GCAM is an RCP-class IAM which has 
been used for several previous international climate assessments (Edmonds and 
Reilly, 1985; Brenkert et al., 2003; Kim et al., 2006; Clarke et al., 2007; Thomson et 
al., 2011). 
1.3.3 Modeling water in GCAM 
 As discussed in Section 1.2.1, water within GCAM is modeled for available 
surface runoff, blue water withdrawals and consumption, green (biophysical) water 
consumption, nonrenewable groundwater supply and extraction, and to a lesser extent 
desalination. Sections 1.3.3.1 ? 1.3.3.2 describe the processes by which each are 
modeled and how historical supply values have been reconstructed and calibrated.  
 
1.3.3.1 Water Demand 
 
Water demand is modeled within GCAM across 235 individual basins. Six 
individual sectors generate water demands ? agriculture, electricity generation, 
manufacturing of goods and services, municipal (domestic), primary energy 
21 
 
 
 
extraction and processing, and livestock (Hejazi et al., 2014b) ? which are calculated 
based on cost and availability of supply (Kim et al., 2016, Turner et al. 2019a).  
Agricultural water demands in GCAM are modeled across each of the 235 major river 
basins for twelve crop commodities (Chaturvedi et al. 2013) and two biomass types 
and are disaggregated into rainfed and irrigated production. Agricultural blue water 
demands are driven by the extent of irrigated land, irrigated crop mix, irrigation 
efficiency, climate and weather, and several other lesser factors. The extent of 
irrigated land and irrigated crop mix are determined endogenously in GCAM and 
depend solely on economics, as explained in Calvin et al. (2017). Regionalized 
estimates of three main irrigation systems exist and are employed in GCAM (Rohwer 
et al. 2007). Electricity generation water demands are calculated based on cooling 
system types; once-through cooling systems, which account for 86% of the 
withdrawals in the energy sector in GCAM; recirculating cooling systems, which 
represent the largest water-consumption cooling process; and dry cooling systems, in 
which water usage is minimal. Manufacturing water demands are calculated based 
upon the energy required by the manufacturing sector. This value represents an 
aggregate of all goods and services produced in the manufacturing sector, as GCAM 
does not separate manufacturing into subsectors. Municipal water demands are 
calculated based upon population and per-capita income across all GCAM regions. 
Primary energy demands encompass oil production and refining, coal mining, gas 
production and processing, ethanol production and coal-to-liquids fuel production. 
Finally, water demands for the livestock sector represent the drinking water 
requirements for five livestock types (beef, dairy, sheep & goats, pigs, and poultry), 
as well as water for cleaning and servicing them. Detailed methods on water demand 
22 
 
 
 
modeling within GCAM are described in Chaturvedi et al. (2013), Hejazi et al. 
(2013b), and Hejazi et al. (2014b). 
 
1.3.3.2 Water Supply 
 
Water supplies in GCAM exist in three distinct categories: renewable water, 
non-renewable groundwater, and seawater. GCAM employs cost resource curves 
across all 235 basins that follow a logit formulation, a statistical representation of the 
competition between multiple objects (Clarke and Edmonds, 1993), to determine the 
share of each water source needed to meet the water demands within all basins (Kim 
et al., 2016, Turner et al., 2019a). As accessible water within a basin decreases, the 
price of water within that basin increases until higher cost options ? non-renewable 
groundwater or seawater ? are used to meet demands. Rising water prices in a basin 
lead to a compounding price increase on the goods and services that require higher-
priced water sources. In response, a corresponding change occurs in demands for 
agricultural production, energy resources, and all remaining goods and services 
within the basin. The inclusion of water constraints within GCAM has resulted in 
changes to regional agricultural production in water scarce regions, leading to 
increased dependency on international trading of goods to water scarce regions (Kim 
et al., 2016). As nonrenewable groundwater is increasingly depleted, expansions in 
rainfed agriculture and shifts of irrigated agriculture from scarce regions occurs 
(Turner et al. 2019b). 
Future available surface runoff supply is calculated by use of the global 
hydrologic model, Xanthos (Li et al. 2017, Liu et al., 2018; Vernon et al. 2019). This 
is completed by using bias-corrected hydrologic data derived from GCMs. With 
temperature and precipitation data, Xanthos calculates the potential 
23 
 
 
 
evapotranspiration using the FAO Penman-Monteith method (Monteith, 1965; Allan 
et al., 1998). From the potential evapotranspiration calculation, information regarding 
soil moisture capacity allows for runoff generation within a single grid cell. Within 
Xanthos, an accessible water module calculates the amount of water available for 
human use dependent upon the total runoff generated. This is aggregated to the 235-
basin level for implementation in GCAM. For the purposes of these studies, 5-year 
moving averages will be considered as the water supply (available runoff) to tease out 
interannual variability, while leaving the general trend of water supply evolution. 
Historical water supplies are also calculated by Xanthos for the 1975-2010 period. 
Xanthos has undergone extensive verification for the historical reconstruction of 
potential evapotranspiration and runoff generation. Xanthos has been calibrated to the 
Variable Infiltration Capacity (VIC) model runoff projections that have been forced 
with observation data in the WATCH dataset (Weedon et al., 2011; Vernon et al., 
2019). 
24 
 
 
 
 
 
Figure 1.6 Intensification of water supply from 2005 to 2100 as calculated from Xanthos (Section 
1.3.3.2). Values consider the average water supply as calculated from five GCMs for each RCP 
scenario. Values below 1, depicted in red, represent decreases in available surface runoff from 2005 
values, while values above 1, depicted in blue, represent relative increases in the amount of available 
Q
water from 2005 values. Intensification calculation provided as 789::<Q . 78::;
 
25 
 
 
 
1.4 Noted Research Gaps and Objectives 
 The evaluation of water futures has increased significantly in recent years by 
studying the effects of human systems, climate systems, and projecting how the use 
of water will change in the future. However, there remain significant research gaps 
which the studies that follow aim to address. Due to the recent advancements to the 
modeling of the water sector in GCAM, the following studies are able to capture the 
interactions between the human and climate system, and the resulting effects on water 
use and availability. A unique component of this research is the ability to constrain 
the amount of water that is available for use from all sources. This inclusion allows 
for more accurate representations of water use and are a critical aspect of these 
studies. 
 Three key areas of potential knowledge improvement have been found which 
will be the focus of the following studies. 1) There is currently a lack of 
understanding regarding the implications of adding socioeconomically viable water 
technologies to the set of SSP scenarios released by the IAM community and used for 
the upcoming IPCC report. Several studies have offered glances at how 
socioeconomics may change water use, but none have been able to capture both the 
human-energy-land feedbacks as well as account for limiting supplies of water. 2) 
Future water scarcity is widely believed to be driven by human interactions, however 
there remains a clear lack of knowledge regarding the codependence on the climate 
system as well as how the interactions between each may alter scarcity, positively and 
negatively, in the future. The use of the IAM framework allows for these interactions 
to be captured. 3) The concept of virtual water trade is gaining additional recognition 
in the sociohydrological community, however very little is known about how it will 
26 
 
 
 
change in the future. Even less is known about the reliance on nonrenewable 
groundwater use for meeting global demands. GCAM allows for trade evolution, 
specifically as a response to socioenvironmental forcings. As such, this provides a 
platform to evaluate, for the first time, the temporal evolution of virtual water trade. 
 In general, this work aims to gain a better understanding of the interactions 
between the human and climate systems on the future of water around the world. To 
do this, three research questions are posed to evaluate the noted research gaps: 
  
(i) How do quantitative assumptions for the water sector across the five 
Shared Socioeconomic Pathways scenarios change global water 
demands in a water constrained world?  
(ii) How does the coevolution of human-energy-water-land-climate 
systems affect the driver of water scarcity changes and how do human 
and climate systems interact to alter water scarcity attributions both 
spatially and temporally? 
(iii) How will future changes in socioeconomics and climate affect the 
amount of water embedded in international agricultural trade and what 
are the implications for nonrenewable groundwater extraction? 
 
 In order to answer these three questions, the following methods are used to 
examine the future of global water: 
(i) Establish qualitative and quantitative water demand assumptions for 
the Shared Socioeconomic Pathway scenarios and analyze the 
associated impacts 
27 
 
 
 
? Develop qualitative technological change assumptions for four 
water demand sectors, based upon previous literature and current 
advancement rates, consistent with the storylines of each SSP 
scenario. 
? Implement quantitative assumptions in GCAM and evaluate the 
impact on 21st century global water demands across each SSP 
scenario and individual sectors under basin level water constraints. 
? Document the reliance on different income regions towards 
reaching observed demand changes. 
? Demonstrate the differences between technology assumptions and 
water constraints on global water demands throughout the century. 
(ii) Analyze how future water scarcity will change spatially and 
temporally across a set of changing socioeconomic and climate 
conditions. 
? Establish GCM derived climate impacts on water supply, 
agricultural productivity, hydropower expansion, and building 
energy demands. 
? Combine climate impacts with each SSP scenario following the 
SSP-RCP modeling framework and use GCAM to establish water 
demands through 2100. 
? Document the relative contributions of both the human and climate 
systems to water scarcity changes at the GCAM basin level. 
28 
 
 
 
? Analyze in which basins human systems or climate systems have a 
larger impact and establish whether these systems are increasing or 
decreasing scarcity over time. 
? Document the dependence of the observed results on the 
socioeconomic and climate assumptions contained within each 
scenario. 
 
(iii) Establish, for the first time, a virtual water trade projection through the 
end of the century when considering feedbacks between the human 
and Earth system. 
? Using a single SSP-RCP combination, estimate, through 2100, the 
reliance on global trading of water intensive agricultural goods.  
? Estimate the amount of green, blue, and nonrenewable 
groundwater that must be consumed in the crop growth process for 
crops that are traded in the global market. 
? Downscale the virtual water exports to the GCAM basin to 
estimate location of exports. 
? Document how virtual water trading of all water sources 
progresses through 2100 by aggregating by crop type and GCAM 
energy-economy region. 
? Demonstrate locations of increased reliance on nonrenewable 
groundwater extraction to meet global agricultural demands. 
  
29 
 
 
 
 This thesis is organized as follows. Chapter 2 depicts the implementation of 
quantitative water sector assumptions in GCAM in the absence of climate change. 
Chapter 3 introduces the climate change impacts to the socioeconomic assumptions 
established in the previous chapter to investigate how human and climate systems 
drives future water scarcity changes. Chapter 4 establishes how both climate and 
socioeconomic impacts change the reliance on traded water intensive agricultural 
goods to meet changing demands by the end of the century. These chapters are 
organized to capture the results from Graham et al. (2018), Graham et al. (2019a, 
submitted), Graham et al. (2019b, in prep). Concluding remarks and future directions 
are provided in Chapter 5. 
30 
 
 
 
Chapter 2: Water Sector Assumptions for the Shared 
Socioeconomic Pathways in an Integrated Modeling framework 
(published as Graham et al., 2018) 
2.1 Introduction 
 In order to present results consistent with the next generation of 
socioeconomic scenarios, we employ the Shared Socioeconomic Pathways (SSP).   
These scenarios have been quantified using a set of global integrated assessment 
models (IAM), where each model has been used to produce the marker scenario 
(Riahi et al., 2017) for one SSP scenario (Calvin et al., 2017; Fricko et al., 2017; 
Fujimori et al., 2017; Kriegler et al., 2017; van Vuuren et al., 2017) and provide 
uncertainty ranges for the other SSPs. The IAM results provided alternative 
projections of the evolution of the economy, energy systems, land systems, emissions, 
and climate. 
Initially absent from the scenario descriptions have been assumptions for the 
water sector.  Therefore, studies have begun to investigate the complete water sector 
response under different socioeconomic scenarios.  Several studies (Hejazi et al., 
2014b; Hanasaki et al., 2013a, b) predate the multi-sector SSP assumption 
implementation in the global IAMs (Calvin et al., 2017; Fricko et al., 2017; Fujimori 
et al., 2017; Kriegler et al., 2017; van Vuuren et al., 2017); however, since their 
release, others have incorporated assumptions for water sectors within the SSP 
framework. For example, the Water Futures and Solutions initiative (WFaS) 
combined the hydro-economic classification system (Fischer et al., 2015) with the 
socioeconomic assumptions of the SSP scenarios to create varying degrees of 
technological change in the agricultural, municipal, and industrial sectors, which were 
then applied to a set of three global hydrological models (Wada et al., 2016).  Other 
31 
 
 
 
studies have focused on sector-specific assumptions (Fujimori et al., 2016; 
Mouratiadou et al., 2016) or on a subset of the five SSP scenarios (Bijl et al., 2016). 
Despite recent efforts, water demand changes as a response to both SSP storyline and 
limited water supply have not been addressed as of the submission of this thesis.  
Table 2.1 lists recent studies that have explored future technological changes within 
specific water sectors associated with changing socioeconomic scenarios. The use of 
an IAM is necessary for the inclusion of intersectoral and subsectoral feedbacks, 
incorporated through several sectoral water demand models, allowing for the 
endogenous calculations of water demand while considering both the cost and 
availability of various water sources (Hejazi et al., 2014a; Kim et al., 2016). All 
previous studies that modeled water in the SSPs have calculated water assuming 
unlimited supply and therefore have the potential to both overestimate water demands 
and exclude important feedbacks resulting from a constrained basin-level supply of 
water (Kim et al. 2016). Also absent from the discussion thus far is the magnitude of 
changes in water demands that result from the inclusion or exclusion of future water 
technology changes.  
 
Table 2.1 A comparison of previous studies which have looked at technological change with the SSPs 
and the resulting effects on future water demands 
 This Hanasaki et Wada Fujimori Mouratiadou  Bijl et 
Study al.   et al. et al. et al. (2016) al. 
(2013a, b) (2016) (2016) (2016) 
SSP Scenarios 1-5 1-5a 1-3 1-5 1, 2, 5 2 
Water Constraints X      
Use of IAM X   X X X 
Agriculture X X Xb  X  
Manufacturing X X X X  X 
Electricity X X X X X X 
Municipal X X X   X 
a Hanasaki et al. (2013a, b) predates the release of O?Neill et al. (2014) and the recently released SSP 
marker scenarios in which SSP assumptions were input into various IAMs. (Calvin et al., 2017; Fricko 
et al., 2017; Fujimori et al., 2017; Kriegler et al., 2017; van Vuuren et al., 2017)  
32 
 
 
 
b Wada et al. (2016) provide qualitative assumptions of agricultural changes that are not currently 
implemented into the Global Hydrologic Model used. 
 
 In this study, we employ GCAM to explore future water use and the effects of 
technological advancements in the water sector, following the SSP framework. 
Specifically, we explore a comprehensive set of water sector assumptions for each of 
the SSP scenarios that 1) quantifies the implications of technological change in the 
water sector and 2) considers the effects of limiting the supply of water across all 
sectors.  Changes that are consistent with each SSP scenario are simulated within the 
agricultural, manufacturing, municipal, and electricity generation sectors in terms of 
water withdrawals, as driven by technological advancements and associated 
efficiency changes. Further, we investigate the impact that these water technologies 
have on global water demands and note 1) when advanced technologies can result in 
substantial reductions in water use (SSP1 & SSP5) and 2) under which circumstances 
end-of-century water demands are reduced below base-year values. As different 
regions undergo technological changes at different rates, the resulting withdrawal 
changes vary by scenario and region. Finally, in combining socioeconomic and water 
sector assumptions with water constraints, we can both advance the SSP scenarios 
and assess their implications for other sectors, e.g., agriculture, energy, 
manufacturing, municipal, livestock and primary energy production. The effects of 
climate change on water are not explored in this study, as our focus is an exploration 
of demand changes by SSP scenario. 
 
2.2 Methods 
GCAM has developed a set of qualitative and quantitative assumptions for the 
five SSP scenarios after being chosen as the marker model for the SSP4 scenario 
33 
 
 
 
(Calvin et al., 2017; top portion of Table 2.2).  For this study we have developed a set 
of qualitative assumptions for each water sector, shown in the bottom portion of 
Table 2.2. The assumptions represent varying degrees of change in individual sectors 
based on future storylines of the SSP scenarios. The development of qualitative 
assumptions is intended to aid other modeling groups in exploring various 
quantitative changes while remaining true to the SSP storylines. We have categorized 
each sector and SSP using three levels of change, high, medium and low, for which 
quantitative assumptions are later derived. The specifics for all non-water sector 
assumptions are discussed in detail in Calvin et al. (2017). For future water demand 
changes, the assumptions developed here consider two factors, the desire for 
sustainable futures and the ability to invest in the technological advances required to 
produce varying degrees of sustainability. For the former, we consider the degree of 
sustainability of each SSP storyline in O?Neill et al. (2017). For the latter, we use the 
future global economic conditions (such as GDP per capita) as a proxy. SSP1 and 
SSP5 are thus assumed to have high degrees of progress for most water sector 
variables due to high income, as well as the sustainability focus of SSP1. SSP3 
experiences low degrees of progress because of low GDP growth. SSP2 includes a 
medium degree of progress in the water sector, consistent with its middle-of-the-road 
narrative. SSP4 is split between three income regions, where the high-income regions 
follow closely the assumptions from SSP1 and SSP5, the medium-income regions 
generally follow SSP2, and the low-income regions generally follow SSP3 
assumptions. From these assumptions, the quantitative responses to future 
technological change for each SSP scenario are derived and applied to the respective 
water sectors in GCAM, as described below. 
34 
 
 
 
 
Table 2.2 Qualitative assumptions for the SSP scenarios within GCAM. All non-water sector 
assumptions adopted from Calvin et al. (2017). 
  SSP1 SSP2 SSP3 SSP4 SSP5 
High- Medium- Low-
income income income 
SSP Prescribed Challenges to Low Medium High Low   High 
Challenges  Mitigation 
 Challenges to Low Medium High High   Low 
 Adaptation 
Socioeconomics Population in 6.9 billion 9 billion 12.7 billion 0.9 billion 2.0 billion 6.4 7.4 billion 
2100 billion 
 GDP per capita $46,306 $33,307 $12,092 $123,244 $30,937 $7,388 $83,496 
in 2100 
Fossil Resources Coal Med/Low Med/Med High/High Med/Low Med/Med Med/ High/High 
(Technological High 
Change/Acceptance) 
 Conventional Med/Med Med/Med Med/Med High/Low High/Low High/ High/High 
Gas & Oil Low 
 Unconventional Low/Med Med/Med Med/Med Med/Low Med/Low Med/ High/High 
Oil Low 
Electricity Nuclear High Med High Low Low Low Med 
(Technology Cost) 
 Renewables Low Med High Low Low Low Med 
 CCS High Med Med Low Low Low Low 
Fuel Preference Renewables High Med Med High High High Med 
 Traditional Low Low High Low Low High Low 
Biomass 
Energy Demand Buildings Low Med Low High Med Low High 
(Service Demands) 
 Transportation Low Med Low High Med Low High 
 Industry Low Med Low High Med Low High 
Agriculture & Land Food Demand High Med Low High Med Low High 
Use 
 Meat Demand Low Med High Med Med Med High 
 Productivity High Med Low High Med Low High 
Growth 
 Trade Global Global Global Regional Regional Local Global 
Water Use Irrigation  High Med Low High High Low High 
(Efficiency 
Improvements) 
 Manufacturing High Med Low High Med Low High 
 Energy High Med Low Med Med Med High 
Water (Technological Shift to High Med Low High Med Low High 
Change)  Recirculating 
and Dry 
Cooling 
 Municipal High Med Low High Med Low High 
 Municipal Med. Med High Low Med High Low 
Water Price 
35 
 
 
 
 
 
2.2.1 Water Technology Assumptions for the Agricultural Sector 
Agricultural withdrawals for irrigation purposes account for nearly 70% of all 
global water withdrawals (FAO, 2011). The efficiency with which water is delivered 
to crops and the extent and geographic location of irrigated land significantly affect 
withdrawal values. Because GCAM calculates both crop mix and extent of irrigated 
land endogenously, we do not implement any SSP-specific preferences for these 
factors and instead allow the relative profitability of irrigated and rainfed crops to 
determine the irrigation shares. According to the FAO, irrigation systems can be 
categorized as one of three types, each possessing different field application 
efficiencies. Surface irrigation (crop flooding) practices generally have a maximum 
efficiency of 60%, sprinkler irrigation systems (center pivot) have a maximum 
efficiency of 75%, and drip irrigation systems have the highest maximum efficiency 
at 90% (Brouwer, 1989). Unfortunately, historical values of global and regional 
irrigation efficiencies are not currently well documented. Country-level values are 
available from the FAO database (AQUASTAT, 2016); however, several countries 
have little to no data available. More recently, studies have incorporated increased 
biophysical dependencies as a result of spatially variable irrigation systems derived 
from AQUASTAT and applied based upon individual crop functional types 
(J?germeyr et al., 2015, 2016, 2017). Several recent studies have also investigated 
future changes in irrigation efficiency. Schmitz et al. (2013) calculated irrigation 
efficiency improvements of 2% to 12% from 2005-2045, using regional GDP per 
capita as the driving force. Fischer et al. (2007) assigned universal irrigation 
efficiency improvements of 10% per decade, and Chaturvedi et al. (2013) prescribed 
36 
 
 
 
field application efficiency changes through 2095 closely based on regional GDP and 
assumed a maximum field efficiency of 85%.  
Previous SSP studies have implemented changes in agricultural water 
demands through efficiency changes and crop yield improvements. Mouratiadou et al. 
(2016) addressed crop yield improvements, but these were only applied to bioenergy. 
Wada et al. (2016) held irrigation efficiency constant at base year values for the three 
SSP scenarios they analyzed, while Hanasaki et al. (2013a) assumed maximum 
irrigation efficiency of 100%. This study assumes differing rates of efficiency change 
(described here and shown in Graham et al., 2018) for future irrigation practices, 
using maximum efficiency values in line with the upper bounds shown in the 
literature (Rohwer et al. 2007; Sauer et al. 2010; J?germeyr et al. 2015). Specifically, 
we use current-generation irrigation technologies and the efficiencies that are 
currently obtainable as a basis for technological improvements.  
In the high technological-change scenarios of this study (SSP1, SSP5, and 
high- and medium-income regions of SSP4) GCAM regions are assumed to invest in 
the most efficient irrigation technology. The result is that, beginning in 2015, all 
regions begin linear yearly efficiency improvements with the goal of reaching drip 
irrigation efficiency, 90%, by 2100. The adoption of universal drip irrigation systems 
is assumed to require significant regulations, investments, and cultural changes 
(J?germeyr et al., 2015) at a global scale; therefore, we have assumed that field 
application efficiency will only reach a midway point between sprinkler and drip 
irrigation systems, 83%, in 2100.  The direct cost of adding drip irrigation systems 
across all GCAM regions has not been considered. In the medium technological-
change scenarios, regions will linearly adopt the next attainable technology from 
37 
 
 
 
2015 to 2100. This means that regions below the threshold of crop flooding will 
approach a field application efficiency of 60% by 2100. The same holds true for all 
other irrigation systems, with no individual region attaining field application 
efficiencies above 75%. Finally, in the low technological-change scenario, all regions 
are assumed not to invest in technological advances for irrigation efficiencies; 
therefore, all field application efficiencies remain at 2010 values throughout the 
century. 
2.2.2 Water Technology Assumptions for the Electricity Sector 
Three scenarios that combine changes in power-plant cooling system shares 
and efficiency improvements were adapted from Davies et al. (2013) for use in this 
study. The water efficiency improvements, originally described in Feeley et al. (2008) 
for coal and non-coal plants, result in up to a 48% reduction in water withdrawal and 
39% reduction in consumption, while the shifts in cooling systems vary regionally but 
converge to similar values in most regions by 2035, as described in Davies et al. 
(2013), and are then held constant through the remainder of the century. In this study, 
the medium technology change scenario assumes 50% of new power plants built after 
the year 2015 invest in the water savings technologies described in Feeley et al. 
(2008), and cooling system shares move from once-through cooling to recirculating 
and dry cooling at the same rate as in Davies et al. (2013); the assumptions therefore 
match a ?median_midtech? scenario in Davies et al. (2013). The high technology 
scenario differs from the medium technology case in assuming 100% adoption of 
water savings technologies (?hitech?) and a 10% increase in the share of dry-cooling 
power plants and a corresponding 10% decrease in the share of recirculating systems 
(?+10% Dry?). Finally, the low scenario is new for this study, and assumes no water 
38 
 
 
 
efficiency improvements and a 25% reduction in adoption of recirculating and dry 
cooling technologies, as compared with the other scenarios. Table 2.3 provides water 
efficiency and cooling system share changes in coal and non-coal power plant 
withdrawals. 
2.2.3 Water Technology Assumptions for the Manufacturing Sector 
In GCAM, water use for manufacturing goods and services is calculated 
separately from the water used to cool power plants during electricity generation. 
However, because the manufacturing sector is not broken into individual goods and 
services, GCAM uses an estimate of technological changes for the sum of all 
manufacturing branches. Therefore, following the aggregate manufacturing efficiency 
improvements found in Fujimori et al. (2016), we assume that the water withdrawals 
in the manufacturing sector decrease at a constant rate of 1.1% per year in the high 
technological-change scenarios, SSP1 and SSP5, by 0.55% per year in the medium 
technological-change scenario, SSP2 and by 0.275% per year in the low technological 
change scenario, SSP3. SSP4 has withdrawal decreases of 1.1% in high-income 
regions, 0.55% in medium-income regions, and 0.275% in low-income regions. 
 Along with withdrawal improvements, Bijl et al. (2016) investigated 
withdrawal to consumption ratio improvements in the manufacturing sector. They 
created three technological advancement scenarios with varying degrees of reductions 
in manufacturing water consumption. Here, we apply 0.5% consumption reduction 
per year (Bijl et al., 2016 medium scenario) to the scenarios with high qualitative 
assumptions, 0.25% per year to medium scenarios, and no change in consumption 
(Bijl et al., 2016 high scenario) to low qualitative assumption scenarios. A summary 
39 
 
 
 
of withdrawal changes is located in Table 2.3 and consumption changes can be found 
in Table 2.4. 
2.2.4 Water Technology Assumptions for the Municipal Sector 
Hejazi et al. (2013b) described the modeling of municipal water demand in 
GCAM. Changes made in the municipal sector for the SSP scenarios are based on 
technological advancement and the influence of price on water demand. 
Technological change in SSP2 uses the same technological change rate, represented 
by AEEI in Table 3, originally prescribed by Hejazi et al. (2013b), 5% per year, 
which accounts for efficiency improvements. In the high technological-change 
scenarios, SSP1 and SSP5, and the high-income regions of SSP4, technological 
change is assumed to progress 25% faster than the base value of 5% per year; in 
contrast, SSP3 experiences a 25% reduction in the rate of technological change, 
leading to a decrease in the adoption of water-efficient practices. 
 Price changes alter municipal water use, based on the GDP of a particular 
region, through their effects on the numbers of people with access to water-using 
appliances (Hejazi et al., 2013b). Here we base all future price elasticity changes 
within the municipal sector on GDP per capita characteristics (Table 2.2). In SSP1, 
SSP2, and the medium-income regions of SSP4, the price elasticity remains the same 
as in Hejazi et al. (2013b). In SSP5 and the high-income regions of SSP4, the price 
elasticity of municipal water withdrawals (percentage change in withdrawal for a 
percentage change in price) is decreased by 10%. SSP3 and low-income regions of 
SSP4 experience a 10% increase in price elasticity. Increases in price elasticity will 
result in an increase in adoption of water conservation strategies driven by higher 
40 
 
 
 
prices for municipal water, and lead to a decrease in overall municipal water demand. 
Decreases will allow people to use more water for municipal purposes. 
2.2.5 Water Technology Assumptions for the Livestock and Primary Energy Sectors 
In GCAM, water withdrawals for the livestock sector represent the drinking 
water requirements for five livestock types (beef, dairy, sheep & goats, pigs, and 
poultry), as well as water for cleaning and servicing them. The demand for livestock 
varies both by region and SSP (see Table 2.2). We have assumed that the livestock 
water demands (per head) into the future are unlikely to change greatly; therefore, the 
differences in livestock water withdrawals across SSPs are based on differences in 
livestock numbers alone and, within this study, include no future technological 
changes. 
 Finally, water use for the production and processing of primary energy was 
discussed in Hejazi et al. (2014b), who assumed no technological change would alter 
future water withdrawals in primary energy production in future periods, and that 
withdrawals would therefore scale with energy sector activities. In this study, we 
employ the same assumptions. 
  
41 
 
 
 
 
Table 2.3 Quantitative withdrawal changes applied to the various water sectors within GCAM 
  SSP1 SSP2 SSP3 SSP4 SSP5 
High- Medium- Low-
income income income 
Water Use 83% Field Increase by 83% Field 83% Field 83% Field 
(Efficiency Irrigation Efficiency by one efficiency Remain at Remain at 
2100 level 2010 Values 
Efficiency by Efficiency by 2010 Values Efficiency by 
Improvements) 2100 2100 2100 
1.1% per 0.55% per year 0.275% per 1.1% per 0.55% per 0.275% per 1.1% per 
 Manufacturing year year year year year year Withdrawal Withdrawal Reduction Withdrawal Withdrawal Withdrawal Withdrawal Withdrawal Reduction Reduction Reduction Reduction Reduction Reduction 
 Electricity (By Power 
Coal: 45.7% Coal: 22.9% Coal: 22.9% Coal: 22.9% Coal: 22.9% Coal: 45.7% 
Plant Type) Non-Coal: Non-Coal: No Change Non-Coal: Non-Coal: Non-Coal: Non-Coal: 47.7% 23.9% 23.9% 23.9% 23.9% 47.7% 
10% increase 25% 10% increase 25% 10% increase 
Water Electricity (Power in dry Davies et al. reduction in in dry Davies et al. reduction in in dry 
(Technological Plant Cooling cooling over (2013) median recirculating and dry cooling over 
(2013) recirculating 
median and dry cooling over 
Change) Systems) current midtech current 
values cooling from values midtech cooling from 
current 
current current values 
a
 Municipal AEEI  = AEEI = 0.05 AEEI = AEEI = 0.0625 0.0375 0.0625 AEEI = 0.05 
AEEI = AEEI = 
0.0375 0.0625 
 Municipal Water 
Price Price Elasticity Price Price Price Price Price 
Price Elasticity = = Elasticity = Elasticity = Elasticity = Elasticity = Elasticity = -0.32619 -0.32619 -0.35881 -0.29357 -0.32619 -0.35881 -0.29357 
 
aAEEI, or autonomous energy efficiency improvement, represents the technological change rate 
(%/year).  
 
Table 2.4 Quantitative consumption changes made to various sectors within GCAM 
  SSP1 SSP2 SSP3 SSP4 SSP5 
High- Medium- Low-
income income income 
Water Use Manufacturing 0.5% per 0.25% per 2010 Values 0.5% per 0.25%per 2010 0.5% per 
(Efficiency year year remain year year Values year 
Improvements) Consumption Consumption Consumpti Consumpti remain Consumpti
Reduction Reduction on on on 
Reduction Reduction Reduction 
 Energy (2035 Coal: 38.9% Coal: 19.5% No Change Coal: Coal: Coal: Coal: 
Coefficient Non-Coal: Non-Coal: 19.5% 19.5% 19.5% 38.9% 
Reductions) 41.2% 20.6% Non-Coal: Non-Coal: Non-Coal: Non-Coal: 
20.6% 20.6% 20.6% 41.2% 
 
42 
 
 
 
2.2.6 Non-Water Sector SSP Assumptions 
The non-water SSP-related assumptions in GCAM can be broken into five 
main components: socioeconomics, energy demand, agricultural productivity and 
growth, energy supply, and non-CO2 emission changes. These five components have 
qualitative assumptions shown in Table 2.2 and are described in detail in Calvin et al. 
(2017). Socioeconomic assumptions for the SSPs include population growth, GDP 
per capita changes, and associated income elasticities. Energy demand assumptions 
alter the cost, performance, and preference for energy demand technologies (e.g., 
vehicle efficiency or preference for walking/biking). Agricultural productivity and 
growth assumptions specify future crop yield improvements and assumptions 
about trade of crops and bioenergy. Energy supply includes assumptions about the 
cost and performance of energy generation technologies, the cost of energy 
extraction, and the cost and availability of carbon dioxide capture and storage. 
Finally, non-CO2 emission assumptions adjust the emissions factors for various 
technologies to reflect differing assumptions about air pollution controls. Unless 
otherwise noted, it is assumed that these five components are included within each 
SSP simultaneously; however, an assessment of their individual impacts on water 
demands across SSP scenarios is undertaken below. 
2.2.7 Scenario Description 
 In this study, we investigate the effects of both water constraints and water 
technologies on future global water demands across the SSP scenarios. To organize 
the results, we have developed a scenario naming system that differentiates SSP 
scenarios based on the assumptions used.  Scenario names therefore refer to both 
water technologies (Ref and Tech) and water constraints (const and unconst) and are 
43 
 
 
 
provided in Table 2.5. In these scenarios, Tech assumes that all water technological 
advances and efficiency changes shown in Table 2.3 are implemented. In the Ref 
scenarios, all non-water sector SSP assumptions are present and water sector values 
are based on Hejazi et al. (2014b) and are consistent across all SSPs. Water 
constraints introduce water markets as described in Kim et al. (2016), while 
unconstrained scenarios assume that water is freely available in each basin, and 
therefore the amount required to meet the demand of any good or service can be 
provided. Results shown below compare the effects of water sector technology 
changes under a constrained water supply (Ref_const and Tech_const), unless 
otherwise noted. 
 
 Table 2.5 Scenario names and components added to each of the five SSP scenarios 
 Ref_unconst Tech_unconst Ref_const Tech_const 
Existing SSP X X X X 
Assumptions a 
Water Technologies b  X  X 
Water Constraints c   X X 
a Existing assumptions for the SSP scenarios are adopted from Calvin et al. (2017) and are outlined in 
Table 2.2 which include all non-water sectors. 
b Assumptions for the water sector developed in this study and shown in Table 2.3. 
c Water supply is limited for renewable water only (Kim et al. 2016) 
2.3 Results 
2.3.1 Total Global Water Withdrawals 
Total global water withdrawals increase through 2050 in all scenarios; 
however, despite declines in water withdrawals in the second half of the century of 
several scenarios, only in SSP1 are withdrawals less than 2010 values by 2100. 
Across all scenarios, declines in withdrawals are due to the inclusion of large changes 
in irrigation practices and significant efficiency improvements across other sectors. 
44 
 
 
 
Changes in global withdrawals from 2010 to 2100 range from a decrease of 8% 
(SSP1) to an increase of 71% (SSP3). All results shown throughout Section 2.3 
include water constraints unless otherwise noted. 
 
 
Figure 2.1 Global yearly water withdrawals by SSP scenario and sector (bcm/year). The top row 
shows scenarios run without technological change in the water sector, while the bottom row includes 
technological change in the SSP storylines. 
Table 2.6 Global water withdrawals by sector (bcm/year). Values are given both for total withdrawals 
in water constrained scenarios with and without technological change (TC). 
Sector 2010 2050 with TC 2100 with TC 
  SSP1 SSP2 SSP3 SSP4 SSP5 SSP1 SSP2 SSP3 SSP4 SSP5 
Agricultural 2402 3297 3474 3596 3323 3234 2387 3145 4338 3180 2333 
Manufacturing 332 352 506 508 461 461 182 463 568 362 299 
Electricity 562 453 566 546 493 650 268 624 609 354 573 
Generation 
Municipal 474 735 742 730 710 816 604 774 811 638 808 
Total a 3861 4964 5446 5583 5142 5320 3560 5207 6600 4731 4168 
Sector 2010 2050 without TC 2100 without TC 
  SSP1 SSP2 SSP3 SSP4 SSP5 SSP1 SSP2 SSP3 SSP4 SSP5 
Agricultural 2402 4012 3749 3596 3416 3943 3073 3649 4338 3363 3345 
Manufacturing 332 548 631 568 583 718 493 760 728 578 808 
Electricity 562 546 627 526 540 905 392 750 574 408 942 
Generation 
Municipal 474 772 742 694 704 857 676 774 725 619 903 
Total a 3861 6004 5906 5543 5398 6583 5021 6134 6639 5166 6153 
a includes Livestock and Primary Energy withdrawals not depicted in table. 
 
45 
 
 
 
 While there are potentially significant scenario-based changes in future water 
withdrawals, the changes resulting from the inclusion of future water technologies are 
just as significant. Figure 2.1 shows total global withdrawals in scenarios with and 
without future water technological change. Significant reductions in post-2050 water 
withdrawals in SSP1 and SSP5 clearly result from technological change in the water 
sector, and stem largely from irrigation system improvements in low-income regions 
(see Section 2.3.6). All scenarios, aside from SSP3, show higher estimations of global 
water demand in 2100 in the absence of SSP-specific water technology adoption. 
Using 2100 with technological assumptions considered as a baseline, reductions in 
water withdrawal range from <1% in SSP3 to 32% in SSP5.  
2.3.2 Agriculture and Livestock 
Irrigation water withdrawals depend on the extent and geographic distribution 
of irrigated land, which in turn depends on agricultural demand, commodity prices, 
competition for land, etc. Irrigated land area differs across SSPs, even absent water-
related technological change as shown in Figure 2.2. Specifically, irrigation 
withdrawals increase steadily into the mid-century for all scenarios, at which point all 
scenarios excluding SSP3 begin to see either small decreases or a leveling-off of 
withdrawals (Figure 2.3, top panel, absent water technologies), or large decreases in 
withdrawals (Figure 2.3, bottom panel, including water technologies), depending on 
the exclusion or inclusion of technological change. Population declines in the second 
half of the century in SSP1 and SSP5, which leads, in combination with the newly-
implemented high-efficiency irrigation applications globally, to declines of more 900 
bcm (billion cubic meters) by 2100 from 2050 values. These results show that 
technological change may result in lower future irrigation water withdrawals in the 
46 
 
 
 
SSP scenarios than those observed by previous studies (Figure 2.4). The results here 
are contingent upon the extent of irrigated land (Figure 2.2) not changing drastically 
over the course of the century as implied in previous studies (Hanasaki et al. 2013a). 
In contrast, SSP3 does not invest in more efficient technologies, and therefore does 
not experience a decline after 2050.  
 
 
Figure 2.2 Extent of irrigated land within GCAM across the five SSP scenarios.  Values shown here 
represent the extent of irrigated land in the Tech_const scenario. The extent of irrigated land is 
determined endogenously in GCAM and depends solely on economics in our analysis, as explained in 
Calvin et al. (2017). In GCAM we allow the relative profitability of irrigated and rainfed crops to 
determine the irrigation shares and geographic location and extent of irrigated land.  
 
 
Table 2.6 shows that when water-efficiency technologies are included, global 
irrigation withdrawals in 2100 for SSP1 and SSP5 are lower than the 2010 values. In 
contrast, SSP3 shows a 81% increase in irrigation withdrawals by the end of the 
century. Agricultural water withdrawals have historically accounted for 70% of all 
water withdrawals globally. Incorporating the irrigation efficiency improvements of 
47 
 
 
 
SSP5, along with the emphasis on non-renewable energy sources, causes the share of 
agricultural water withdrawals to drop to 56% in 2100. In all other SSPs, agriculture 
represents between 60% (SSP2) and 67% (SSP1 & SSP4) of total global water 
withdrawals. Under our assumptions (no change in field efficiency between 2010 and 
2100 in SSP3), SSP3 has no differences in agricultural withdrawals with 
technological change. The significant decrease in the share of irrigation water 
withdrawal found in SSP5 illustrates the potential for future changes in irrigation 
practices to affect the global water withdrawal footprint dramatically, provided that 
socioeconomic conditions are viable and that there remains a high reliance on 
manufacturing and electricity water needs. 
 
 
 
 
Figure 2.3 Global water withdrawals from irrigated agriculture and livestock production (bcm/year).   
The top panel shows scenarios run without technological change in the water sector while the bottom 
panel includes technological change.  
  
48 
 
 
 
 
Figure 2.4 A comparison of future irrigation water withdrawals (bcm/year) across several studies. The 
figure shows studies that have attempted to account for future socioeconomic changes on the 
agricultural sector; not all of these have provided future sectoral changes as applied in this study. The 
values from this study, shown as solid colored lines, account for future technological change and are 
the same values seen in the Tech_const panel of Figure 2.3. Studies shown in this comparison have 
several differences in underlying assumptions and are shown here to represent a comparison between 
different socioeconomic impacts on agricultural withdrawals. (Alcamo et al., 2007; Wada et al., 2016; 
Hanasaki et al., 2013b; Hejazi et al., 2014b; Shiklomanov, 2000; Shen et al., 2008) 
 
 
Livestock withdrawals are based on dietary patterns in each SSP scenario and 
are therefore lowest in SSP1, where diet is expected to shift away from ruminants and 
towards other food sources. All remaining SSP scenarios have very similar 
withdrawal values for livestock. While SSP3 has the highest end-of-century 
population, livestock withdrawals are similar to the other scenarios because of the 
low per-capita meat demands associated with low income levels. 
2.3.3 Electricity Generation 
Water withdrawals for electricity generation depend largely on the fractional 
share of once-through cooling for power plants and the type of fuel used. Within the 
49 
 
 
 
Tech_const scenario, bottom of Figure 2.5, SSP1 experiences water withdrawals in 
2100 that are less than half of 2010 values, due to a shift away from coal, an increase 
in power plant water use efficiency, and a shift to less water-intensive cooling 
systems (e.g., recirculating and dry cooling). In SSP5, water efficiency improvements 
offset the large increase in non-renewable power sources, resulting in nearly identical 
withdrawal values in 2100 as in the base year. In SSP2, withdrawals do not change 
significantly throughout the century, because of smaller water-efficiency 
improvements and a lack of cooling-system share changes. All withdrawal increases 
due to electricity-demand increases are nearly negated by the assumed changes in 
water efficiency. SSP3 experiences an increase in withdrawals throughout the 
century, as efficiency changes are assumed not to occur. SSP4 is unique for the large 
decrease in water withdrawals in the second half of the century, which occurs as high-
income regions shift towards higher percentages of recirculating cooling systems and 
adopt water-efficient power plants as in SSP1 and SSP5. These regions typically have 
the highest electricity generation and the increase in recirculating cooling and 
efficiency results in a decrease in withdrawals. The differences between regional 
reductions are discussed further in Section 2.3.6. Figure 2.5 shows the influences of 
once-through cooling on water withdrawals and the movement away from this 
cooling technology.  
Comparison of scenarios with and without technological change reveals that 
the relatively small shift away from once-through cooling in combination with greater 
electricity generation increases water withdrawals in SSP3, while all other scenarios 
see decreases in water withdrawals. Reductions in withdrawals in SSP1, the high-
income regions of SSP4, and SSP5 are a result of increasing dry cooling and 
50 
 
 
 
efficiency changes in power plants, while lower withdrawals in SSP2 are a direct 
result of the efficiency changes. 
 
 
Figure 2.5 Water withdrawals by power plant cooling technology (bcm/year). The top panel shows 
scenarios run without technological change in the water sector while the bottom panel includes 
technological change. 
2.3.4 Manufacturing and Primary Energy Production 
In the base year (2010), manufacturing withdrawals account for 9% of the 
global water withdrawals. Withdrawals for manufacturing goods and services follow 
a similar pattern to the irrigation withdrawals, with most scenarios reaching a 
maximum near the middle of the century and decreasing in the second half of the 
century. In scenarios with high technological change (1.1% per year decrease in 
withdrawals), end-of-century demands are lower than the 2010 levels. For example, 
in Tech_const, SSP1 has a reduction of 45% from 2010 values by 2100, while SSP5 
sees a slight reduction (seen in the bottom of Figure 2.6).  SSP4 ends the century near 
the 2010 starting value because the high-income regions represent a large proportion 
51 
 
 
 
of the global manufacturing market and achieve the greatest technological change. 
Overall, the share of global water withdrawals from manufacturing decreases across 
all scenarios except for SSP3, with SSP1 reaching 5% of total global withdrawals by 
the end of century, the lowest share among scenarios. When technological change is 
included, all scenarios see reductions in water withdrawals, with magnitudes ranging 
from 22% in SSP3 to 63% in SSP1.  
 The largest amounts of water used in primary energy production and 
processing are in SSP5, due to its focus on future fossil fuel use; however, these 
withdrawals are still a rather small portion of the global total and these decrease 
below that of other scenarios in 2100 due to the high efficiency improvements 
prescribed. As described in Section 2.2.5, we have assumed that water demands for 
primary energy extraction will not change due to the SSP storylines, and rather will 
remain to be driven by the energy demand of each SSP scenario. Therefore, we 
observe no differences between Ref_const and Tech_const scenarios. 
 
52 
 
 
 
 
Figure 2.6 Water withdrawals from manufacturing and primary energy sectors (bcm/year).  The top 
row shows scenarios run without technological change in the water sector, Ref_const, while the bottom 
row includes technological change in the SSP storylines, Tech_const. 
 
 
Figure 2.7 Comparison of industrial water withdrawals across several studies (bcm/year). GCAM does 
not explicitly model Industrial water demand, instead, a combination of water demands for electricity 
generation and manufacturing goods and services are used to allow for comparison to other studies. 
Values from this study, as solid colored lines, are shown for the Tech_const scenario. (Wada et al., 
2016; Fujimori et al., 2016; Hanasaki et al., 2013a; Shiklomanov, 2000)  
53 
 
 
 
 
2.3.5 Municipal  
Municipal water demand was estimated to account for 12% of the total global 
water withdrawals in 2010 (FAO, 2011). Municipal demand in GCAM is driven by 
water price, the regional GDP per capita, and the technological improvement rate, 
which captures changes in the efficiency of appliances. The results in Table 2.6 show 
that scenarios such as SSP5 have high withdrawals due to a lower price elasticity, 
resulting in a higher demand of water for municipal purposes, such as for appliances 
and lawn watering; in contrast, withdrawals are high in SSP3 due to future population 
growth and a lack of efficiency improvements. All scenarios see smaller 
technologically-induced water changes than those in other sectors because of smaller 
expected efficiency improvement rates within the municipal sector when compared to 
the period-to-period changes in other sectors.  Changes to the municipal sector lead to 
less-drastic declines, or in some cases an increase, in post-2050 withdrawals. This 
leads to an overall change in the balance of global water withdrawals. For example, 
municipal water use in SSP1 increased from 12% of the total global water withdrawal 
in 2010 to 17% in 2100. 
 Differences between simulations with and without technological change are 
less than 20% for all SSPs. For SSP3 and SSP4, an increase in withdrawal values in 
2100 is observed due to slow technological progress (SSP3 and the low-income 
regions of SSP4), resulting in large percentage-based increases in water withdrawals. 
In SSP3, a difference of 86 km3 year-1 is observed globally in 2100.  
54 
 
 
 
 
Figure 2.8 Comparison of municipal water withdrawals across several studies (bcm/year). Studies 
shown in this comparison have offered the potential for socioeconomics to influence future demands 
within the municipal water sector. Values depicted from this study, shown in solid colored lines, are 
depicted for the Tech_const scenario. (Bijl et al., 2016; Wada et al., 2016; Hanasaki et al., 2013a; 
Hejazi et al., 2013b; Hejazi et al., 2014b; Shiklomanov, 2000)  
 
2.3.6 Income Region Differences 
In SSP4, technology adoption rates differ across regions with differing income 
levels. This means that in SSP4, differences in adoption rates of higher-efficiency 
technologies are based on the starting efficiencies in each region. Regional groupings, 
based on income levels in SSP4, were described in Calvin et al. (2017) and are 
included in Table 2.7.  
When all five SSP scenarios are analyzed in terms of the three income regions 
of SSP4, a pattern of technological improvements emerges (Figure 2.9). For most 
SSP scenarios, the largest-magnitude changes occur in low-income regions (bottom 
panel), where the irrigation systems are assumed to change by the largest amount. 
This results in savings of over 1000 billion cubic meters (bcm) in SSP5, more than 
55 
 
 
 
double the savings of the medium-income regions (middle panel) and triple the high-
income regions (top panel). The sector that experiences the largest reductions from 
technological change varies by income region. In high-income regions, in all 
scenarios but SSP3, a reduction in manufacturing and electricity withdrawals 
dominates annual withdrawal changes because of the relatively high industrialization 
and power generation in each of the high-income regions. For example, 78% of the 
total high-income SSP5 reductions in 2100 are found in the manufacturing and 
electricity sectors. In contrast, SSP3 experiences slight increases in mid-century water 
withdrawals because of higher electric-sector water withdrawals in high-income 
regions, which are associated with increases in once-through cooling.  
 
Figure 2.9 Global water withdrawal differences by income region and sector in each SSP scenario 
(bcm/year). High-income regions are shown in the top panel, medium-income regions in the middle 
panel, and low-income regions in the bottom panel. Income regions are defined in Table 2.7. 
Differences shown represent reductions due to technological change and are calculated as Tech_const 
minus Ref_const.  
 
 
 
56 
 
 
 
Table 2.7 Income level classifications by GCAM region, as defined in Calvin et al. (2017). 
Aggregate GCAM Regions Countries 
Region Included 
High Australia_NZ Australia, New Zealand 
Income Canada Canada 
EU-15 Andorra, Austria, Belgium, Denmark, Finland, France, 
Germany, Greece, Greenland, Ireland, Italy, Luxembourg, 
Monaco, Netherlands, Portugal, Sweden, Spain, United 
Kingdom 
European Free Iceland, Norway, Switzerland 
Trade Association 
Japan Japan 
South Korea South Korea 
Taiwan Taiwan 
USA United States 
Medium Argentina Argentina 
Income Brazil Brazil 
Central America and Aruba, Anguilla, Netherlands Antilles, Antigua & Barbuda, 
the Caribbean Bahamas, Belize, Bermuda, Barbados, Costa Rica, Cuba, 
Cayman Islands, Dominica, Dominican Republic, 
Guadeloupe, Grenada, Guatemala, Honduras, Haiti, Jamaica, 
Saint Kitts and Nevis, Saint Lucia, Montserrat, Martinique, 
Nicaragua, Panama, El Salvador, Trinidad and Tobago, Saint 
Vincent and the Grenadines 
China China 
Colombia Colombia 
EU-12 Bulgaria, Cyprus, Czech Republic, Estonia, Hungary, 
Lithuania, Latvia, Malta, Poland, Romania, Slovakia, Slovenia 
Europe_Eastern Belarus, Moldova, Ukraine 
Europe_Non_EU Albania, Bosnia and Herzegovina, Croatia, Macedonia, 
Montenegro, Serbia, Turkey 
Mexico Mexico 
Middle East United Arab Emirates, Bahrain, Iran, Iraq, Israel, Jordan, 
Kuwait, Lebanon, Oman, Palestine, Qatar, Saudi Arabia, 
Syria, Yemen 
Russia Russia 
South Africa South Africa 
South French Guiana, Guyana, Suriname, Venezuela 
America_Northern 
South Bolivia, Chile, Ecuador, Peru, Paraguay, Uruguay 
America_Southern 
Low Africa_Eastern Burundi, Comoros, Djibouti, Eritrea, Ethiopia, Kenya, 
Income Madagascar, Mauritius, Reunion, Rwanda, Sudan, Somalia, 
Uganda 
Africa_Northern Algeria, Egypt, Western Sahara, Libya, Morocco, Tunisia 
Africa_Southern Angola, Botswana, Lesotho, Mozambique, Malawi, Namibia, 
Swaziland, Tanzania, Zambia, Zimbabwe 
Africa_Western Benin, Burkina Faso, Central African Republic, Cote d?Ivoire, 
Cameroon, Democratic Republic of the Congo, Congo, Cape 
Verde, Gabon, Ghana, Guinea, Gambia, Guinea-Bissau, 
Equatorial Guinea, Liberia, Mali, Mauritania, Niger, Nigeria, 
Senegal, Sierra Leone, Sao Tome and Principe, Chad, Togo 
Central Asia Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, 
Mongolia, Tajikistan, Turkmenistan, Uzbekistan 
India India 
57 
 
 
 
Indonesia Indonesia 
Pakistan Pakistan 
South Asia Afghanistan, Bangladesh, Bhutan, Sri Lanka, Maldives, Nepal 
Southeast Asia American Samoa, Brunei Darussalam, Cocos (Keeling) 
Islands, Cook Islands, Christmas Island, Fiji, Federated States 
of Micronesia, Guam, Cambodia, Kiribati, Lao Peoples 
Democratic Republic, Marshall Islands, Myanmar, Northern 
Mariana Islands, Malaysia, Mayotte, New Caledonia, Norfolk 
Island, Niue, Nauru, Pacific Islands Trust Territory, Pitcairn 
Islands, Philippines, Palau, Papua New Guinea, Democratic 
Peoples Republic of Korea, French Polynesia, Singapore, 
Solomon Islands, Seychelles, Thailand, Tokelau, Timor Leste, 
Tonga, Tuvalu, Viet Nam, Vanuatu, Samoa 
 
 2.3.7 Impact of SSP Assumptions and Water Constraints 
In this section, we have added the five SSP assumptions discussed in Section 
2.2.6 and water technology changes in a step-wise manner to investigate the 
compounding impacts on global water demands. The assumptions have been added in 
this order: socioeconomics, energy demand, agricultural productivity and growth, 
energy supply, non-CO2 emissions, and water technologies. It is important to note 
that the following results are independent of the order of addition. Here we have 
assumed that the socioeconomic assumptions provide the minimal additions to be 
considered among the SSP scenarios; therefore, the changes in Figure 2.10 represent 
changes from the SSPs run with only socioeconomic assumptions. Values based on 
subsequent assumptions represent changes from the preceding assumption set. The 
solid black line represents net water demand change from the socioeconomic 
assumptions to the water technology assumptions.   
 
58 
 
 
 
 
Figure 2.10 Global water withdrawal changes (bcm/year) as a result of step-wise addition of SSP 
assumption components. The net impact on global water withdrawals is represented as the solid black 
line. 
 In particular, agricultural productivity growth and change is shown to have 
differing effects across the SSP scenarios, resulting in increases in water withdrawals 
in SSP1 and SSP5, and decreasing demands in SSP3 and SSP4. This is caused by 
changing crop yields and cropping intensities in each SSP (Calvin et al., 2017). 
Specifically, as crop yields improve and cropping intensity increases, water 
withdrawals increase (as in SSP1 and SSP5). Energy demand also has a clear, but 
opposite, effect on water demand in both SSP1 and SSP5. This is due to significantly 
different transportation practices; SSP5 experiences increases in the movement of 
goods by freight resulting in higher demands for fuel and water needed for fuel 
extraction, whereas in SSP1 investments into hybrid fuels and a reduction in fossil 
fuel consumption lead to decreases in demand. The introduction of water technologies 
has the largest impact on global water demand across all SSP scenarios as efficiency 
improvements, particularly within the agricultural sector, provide a large potential 
59 
 
 
 
demand reduction. It can be seen that when all SSP assumptions are combined, every 
SSP experiences a net decrease in global water demands, however the magnitude and 
sign of each component differ by SSP. The net changes in SSP2 are minimal aside 
from water technologies as the assumptions for SSP2 make up the base GCAM model 
scenario. 
 
 
Figure 2.11 Change in global water withdrawals as a result of water constraints (top row) and water 
technologies (bottom row).  Top left figure represents the impact of adding water constraints to the 
SSP scenarios run with the existing set of SSP assumptions for GCAM (Calvin et al., 2017). The top 
right figure represents the impact of adding water constraints to the SSP scenarios run with the 
assumptions from this study. The bottom left panel represents the impact of adding the water 
technology changes outlined in this study to an unconstrained water set of scenarios. The bottom right 
figure represents scenarios with added water technologies while under water constraints. 
 
 Water constraints in GCAM limit the amount of available water in river basins 
by using cost resource curves that increase the cost of water as the supply decreases 
(Kim et al., 2016).  Water sector technology advances result in various efficiency 
changes due to changing infrastructure and prices across the SSP scenarios (This 
study). Both of these factors affect water demands and are shown in Figure 2.11. The 
60 
 
 
 
impact of water constraints on the SSPs is shown in the top panel. SSP scenarios that 
do not consider advancements in water technologies (top left) experience water 
demand decreases of up to 400 km3 by the end of the century, in line with values 
shown in Kim et al. (2016). Constraints have far less of an impact on SSP scenarios 
that include future water technology changes (top right), as water demands decrease 
by around 100 km3 across all scenarios. The bottom panel analyzes the impact of 
water technologies in the SSP scenarios. Water technologies added to SSP scenarios 
in which water is assumed to be in unlimited supply (bottom left) have the largest 
water demand reduction of all scenarios with reductions reaching almost 2000 km3 by 
2100 in SSP5. When water technologies are added to scenarios with water constraints 
(bottom right), the differences are slightly smaller than those in unconstrained supply 
scenarios; however, technological advancements still clearly have a significant impact 
on future water demands.  These results demonstrate that advancements in water use 
efficiency, infrastructure changes across the energy sector, and price changes of all 
water and non-water markets and commodities will lead to lower water demands in 
future scenarios than the inclusion of water constraints alone. The reductions shown 
would act to decrease water scarcity across the globe. 
2.4 Discussion 
The incorporation of water sector futures into SSP scenarios relies on 
assumptions for each of the agricultural, electricity, manufacturing, and municipal 
sectors. The evolution of these individual sectors to the end of the century has the 
potential to be highly variable among scenarios, regions, and economic 
characteristics. In order to account for the high variability of future sectoral changes, 
this study has introduced varying technological changes based on the defined 
61 
 
 
 
characteristics of each SSP scenario. In all scenarios except for SSP4, we have 
assumed that the adoption of prescribed technological changes (Table 2.3) do not 
vary by region (2100 efficiency goal is uniform across regions, but adoption rates 
differ based upon base year starting efficiency). This results in a wide range of 
changes that depend on starting economic levels, as low-income regions transition to 
more-efficient technologies from lower base year values than is the case in the 
medium and high-income regions, which leads to advancement rates being very high 
for agriculture in particular. Results show similar patterns of water use across sectors 
due to the underlying assumptions of the SSPs within GCAM, expressed in Table 2.2 
and discussed in Calvin et al. (2017). Changes in population growth, particularly in 
SSP3, yield near-linear increases in water demands across all sectors, as there are 
minimal improvements in the water sector. In the SSP1 and SSP5 scenarios, similar 
water demands result in large part from the significant water sector improvements 
that have been discussed in this study and applied to GCAM, leading to noticeable 
changes in all sectors. Similar explanations apply for SSP2, as population levels off in 
the second half of the century and water sector improvements produce a leveling off 
of demands across most sectors.  
 The change rates and assumptions applied in this study attempt to follow 
values currently existing within the literature; however, in the scenarios in which 
economic conditions allow for sustainable improvement, they offer a new perspective 
on potential benefits of future changes in the water sector. High water savings that 
arise from increased agricultural and infrastructure efficiencies can provide scientists 
and policy-makers alike, global water savings potentials of sustainable futures. In 
sectors where historical data are limited, such as irrigation-efficiency improvements, 
62 
 
 
 
we have assumed that technological progress throughout the century will vary 
significantly based on the willingness to invest in efficient technologies, as 
represented in the assumptions of the different SSP storylines.  
 The introduction of water constraints in the SSPs presents a novel look at 
various global futures under a limited water supply. Our results showed that while 
technological change accounts for the largest impact on global water demands, the 
inclusion of water constraints also causes a noticeable decrease. The inclusion of 
water constraints also allows for better representation of global production in the 
SSPs and associated changes in trade and water demands caused by shifts in 
production to less water-scarce regions (Kim et al., 2016, Chapter 4). Water supplies 
are maintained at historical levels throughout the study period in order to isolate the 
impacts of technological change. 
 Since the introduction of the Shared Socioeconomic Pathways, several studies 
have assessed such assumptions (cf. Wada et al., 2016; Hanasaki et al., 2013a, b; 
Fujimori et al., 2016; Bijl et al., 2016; Mouratiadou et al., 2016). However, these 
studies have largely focused on individual sectors, predate the introduction of the SSP 
scenarios to the IAM community, and/or lack the ability to endogenously calculate 
water demands based upon cost and availability of water supplies. As a result, 
comparisons to other SSP studies of global water demands are difficult, although 
some sector-by-sector comparisons can be made. Agricultural sector withdrawals are 
on the lower side of existing projections (Figure 2.3).  For example, values provided 
by Hanasaki et al. (2013b) range from about 3100 bcm to 8400 bcm in 2085, 
depending on the scenario, while our study calculated withdrawals of less than 6248 
bcm in 2085 when technological advances are considered, or a range of 5350 bcm to 
63 
 
 
 
6622 bcm in 2085 without technological change. Such comparisons are affected by 
important differences between studies, including differences in the extent of irrigated 
land and cropping intensity, which have been determined endogenously in this study, 
while prescribed differences exist in Hanasaki et al. (2013b) and should be considered 
in future comparisons (Figure 2.2). Municipal demand in 2100 in this study ranges 
from 604 bcm to 811 bcm. These values are also within the lower range of previous 
studies? values of 450 bcm to 1400 bcm (Figure 2.8).  Finally, industrial (electricity 
generation + manufacturing) demand values in 2100 range from about 450 bcm to 
1177 bcm in this study. These values are comparable with previous estimates, which 
range from 250 bcm to 1500 bcm ? although note that mid-century values have a 
higher range (Figure 2.7).  
 The assumptions in this study represent different degrees of technological 
change within a single model (GCAM). The degree to which a given assumption 
affects the results is variable and each sector has different levels of uncertainty in 
future projections. However, based on the differences in agricultural water demand 
reductions across SSP scenarios, irrigation efficiency is the source of the largest 
uncertainty for future demand projections. In this study, regional irrigation efficiency 
values in 2100 vary from below 50% for cases without technological change, up to 
83% under the most ambitious technological change. The implications of this spread 
are clear in the 2100 values for agricultural water demands in Table 2.6. As discussed 
in Section 2.2.2, a comparison of percentage reductions in agricultural water 
withdrawals ? that reflect the varying degrees of technological change implemented ? 
can provide insight into the uncertainty across SSP scenarios and its connection to the 
chosen application efficiency. Reductions range from <1% in SSP3 to 32% in SSP5, 
64 
 
 
 
providing upper and lower limits on the effects of irrigation efficiency changes on 
future withdrawals. While each of the other sectors show significant changes in 
global water demand based on the assumed degree of technological change, both the 
impacts and uncertainties of these other sectors are far smaller when compared to the 
agricultural changes. Further analysis is needed to determine the multi-model spread 
of possible results. 
 This study highlights the importance of technological advancement in the 
water sector across various future global change scenarios, particularly when 
considering a limited supply of water. It is important to note that with these results we 
have found that across the SSP scenarios, differing assumptions made outside of the 
water sector can result in significant changes in future water demands (i.e. 
agricultural productivity growth and change). Although socio-economic scenarios 
tend to play a more important role than the climate scenarios in global water 
resources assessments (e.g. V?r?smarty et al. 2000; Hanasaki et al. 2013b), future 
SSP studies should consider effects of climate change on water supplies, energy, and 
land systems and how they may alter global and regional demands.  This study 
maintains historical water supply levels which may miss key changes in accessible 
water caused by climate change in the individual SSP scenarios. The absence of 
climate-change-driven water supply in this study has the potential to misrepresent 
water availability in future scenarios that experience significant warming; therefore, 
such changes will be considered in Chapter 3. Future studies should incorporate a 
broader set of water sector dynamics. For example, climate-induced changes to the 
crop-water intensity of plants, changes in crop breeding to produce more water-
efficient plants, and global shifts to more water-efficient crop mixes have the 
65 
 
 
 
potential to change agricultural water demands but have not been considered in this 
study. Water savings may also result from changes in water supply through expansion 
of desalination facilities and reservoirs, and water reuse. These factors also warrant 
future consideration to broaden the scope of water research in the SSP scenarios. 
2.5 Conclusions 
This study has focused on the development and implementation of a complete 
set of future water sector assumptions for the Shared Socioeconomic Pathways. It 
began with the development of assumptions for future technological change in the 
agricultural, electricity, manufacturing, and municipal sectors. These assumptions 
follow the storylines of each of the SSP scenarios and are added to GCAM to analyze 
global water demands through the end of the century. This work represents the first 
comprehensive set of water sector assumptions that have been applied to the SSP 
scenarios and have been run with an Integrated Assessment Model while including 
water constraints. The study has provided four key results.  
 First, while water constraints act to decrease water demands, future 
infrastructure changes in the water sector can increase water savings by up to 32% in 
2100 in the SSP scenarios, resulting in large potential changes in regional and global 
water scarcity. 
 Second, in SSP1, the focus on sustainability and the ability to invest in future 
water-efficiency improvements has the potential to lead to end-of-century water 
demands lower than present day demands despite a higher standard of living and 
similar global population.  
 Third, future water-demand changes in the SSPs depend strongly on adoption 
and implementation of water saving technologies in low-income regions. In SSPs 1, 
66 
 
 
 
2, and 5, more than half of the global water demand reductions result from adoption 
of more efficient technologies in low-income regions. 
 Fourth, when assumptions for the SSPs are broken down into their various 
components, there tend to be significant and differing impacts on water demands 
across SSP scenarios as a result of non-water sector assumptions (i.e. agricultural 
productivity changes and energy demands), proving that their inclusion is a necessity 
when analyzing future water demands in the SSPs.   
67 
 
 
 
Chapter 3: Humans drive future water scarcity across all Shared 
Socioeconomic Pathways (Graham et al., 2019 ? submitted) 
3.1 Introduction 
Although climate and socioeconomic systems affect water resources 
simultaneously, the individual contributions of these systems, and how they will 
change into the future is relatively unknown at global scales. Previous studies have 
generally attributed future changes in water scarcity to population changes, economic 
growth, and resultant demand increases more so than to climate system impacts 
(V?r?smarty et al., 2000; Arnell, 2004; Alcamo et al., 2007; Hanasaki et al., 2013b; 
Arnell and Lloyd-Hughes, 2014; Schewe et al., 2014; Schlosser et al., 2014; Shen et 
al., 2014; Wada et al., 2014; Kiguchi et al., 2015). Studies have also quantified the 
relative impacts of socioeconomic and climate changes and how they contribute to 
water scarcity at global and regional scales (Wada et al. 2011; Schewe et al., 2014; 
Veldkamp et al., 2015). Recent combinations of changing socioeconomic conditions 
and climatic change have brought to light the importance of the coevolution of human 
and climate systems on water scarcity estimates at sub-regional scales (Veldkamp et 
al., 2016; Rant et al., 2016). Despite recent advances, water scarcity attribution 
assessments often lack the representation of alternative sources of freshwater, 
constraints on water resources, alternative demand scenarios, and feedback linkages 
between hydrological and socioeconomic systems (Ligtvoet et al., 2018). 
This study aims to address this research gap by attributing the simultaneous 
relative contributions of both human and climate systems on water scarcity changes 
through the lens of GCAM. This analysis makes use of a wide range of 
socioeconomic and climate futures that are placed under constraints on sources of 
68 
 
 
 
freshwater, including nonrenewable groundwater, surface runoff, and desalinated 
seawater (Kim et al., 2016; Graham et al., 2018; Turner et al., 2019a). We use GCAM 
to quantify the relative effects of both systems on water scarcity at global and basin 
scales across 15 global futures that include five different socioeconomic conditions 
(the Shared Socioeconomic Pathways, SSPs) and four different climatic conditions 
(the Representative Concentration Pathways, RCPs). We use these 15 scenarios to 
first analyze a ?Human Alone? component to isolate the human impact on future 
scarcity and then apply general circulation model (GCM) derived climate impacts 
from five models to establish 75 ?Human and Climate? scenarios which allow for the 
quantification of climate impacts (Figure 3.1). The use of models that link energy, 
water, land, and climate in water scarcity assessments allows for the explicit 
representation of how climate-induced changes in water availability alter energy and 
land systems. The results are a novel analysis of how, with limited availability of 
water resources, humans and climate may drive future water scarcity across a range of 
socioeconomic and climate futures. Future spatial and temporal representations of 
human and climate system contributions to water scarcity changes are provided while 
considering the feedbacks between the human, energy, and land systems. 
3.2 Methods 
This analysis uses GCAM to make deterministic classifications of how, under 
water resource constraints, the human and climate systems interact to alter water 
scarcity across 15 socioeconomic and climate futures. Using GCAM 5.0 with 
inclusions of water constraints to both renewable and nonrenewable sources of water, 
we evaluate the drivers of future water scarcity by isolating the impacts that both 
humans and climate have while accounting for feedbacks between humans, energy, 
69 
 
 
 
and land. Secondly, we analyze the simultaneous impacts that human and climate 
systems have on water scarcity by determining whether each system is increasing or 
decreasing scarcity in all of GCAM?s 235 water basins. We begin with the 15 SSP-
RCP scenarios that are attainable in GCAM (see Figure 3.1) and then we derive how 
the human system alone change water scarcity towards 2100 by holding all climate 
variables to their 2005 levels while altering socioeconomic growth and technological 
change. Next, we analyze the same 15 scenarios while accounting for GCM derived 
climate impacts on 4 different sectors, water supply, agricultural productivity and 
change, hydropower expansion, and building energy expenditures. These impacts 
come from 5 different bias-corrected GCMs to make a suite of 75 climate runs. By 
subtracting the human derived impacts of the ?Human Alone? scenarios (Figure 3.1) 
from the ?Human and Climate? scenarios we isolate the impact that we define as the 
climate impact. Below we describe the GCAM model, the scenario components, 
climate derived impacts, and the calculations of scarcity changes and attribution. 
 
70 
 
 
 
 
Figure 3.1 Scenario breakdown of 90 total scenarios and the SSP-RCP scenario matrix depicting the 
set of 15 scenarios (green shading) in which plausible solutions exist in GCAM and for which CMIP5 
climate datasets are available. 
 
3.2.1 Scenario Components 
We use temporally varying socioeconomics and climate systems. These 
systems are represented by the Shared Socioeconomic Pathways (SSP), for 
socioeconomic change, in combination with the Representative Concentration 
Pathways (RCPs), for climatic change. The SSPs are a set of five future scenarios 
with varying changes to global population, the economy (Riahi et al., 2017), and land 
use (Popp et al., 2017), which were designed to explore varying degrees of challenges 
to climate change adaptation and mitigation (O?Neill et al. 2017). These scenarios 
have been defined, both qualitatively and quantitatively, using a set of global 
integrated human-Earth systems models, in which individual models have been used 
to produce a marker scenario (Riahi et al., 2017) for singular SSPs (Calvin et al., 
71 
 
 
 
2017; Fricko et al., 2017; Fujimori et al., 2017; Kriegler et al., 2017; van Vuuren et 
al., 2017) and provide uncertainty ranges across all other SSPs. The IAM results 
provide alternative projections of the evolution of the economy, energy systems, land 
systems, emissions, and climate. Within GCAM, quantitative assumptions of the 
SSPs have been made for the economy, energy sector, land use, agricultural sector 
(Calvin et al., 2017), and for the water sector (Graham et al., 2018). The RCPs are a 
set of four future climate scenarios in which end-of-century radiative forcing 
approaches four-levels by altering future greenhouse gas emissions and by changing 
underlying socioeconomic projections (van Vuuren et al., 2011).  
The SSPs and RCPs are combined to form a set of future global change 
scenarios which allows comprehensive socioeconomic assumptions to be matched 
with future radiative forcing pathways to achieve future global warming targets, 
creating the next set of scenarios that will provide the basis for future 
Intergovernmental Panel on Climate Change (IPCC) assessments (O?Neill et al., 
2016; Eyring et al., 2016, Section 1.1.5). Each is then matched with their individual 
Shared Policy Assumptions (SPAs) in order to account for differences in adaptation 
and mitigation strategies across the SSPs (Kriegler et al., 2014; Calvin et al., 2017). 
These combinations create a set of 15 potential global futures which are varied six 
times, for five GCMs and one set in which climate impacts across all sectors are 
neglected to produce 90 total scenarios. 
3.2.3 Climate Derived Impacts from General Circulation Models 
This study accounts for four different general circulation model (GCM) 
derived climatic impacts to water supply, agricultural productivity, hydropower 
expansion, and building energy expenditures. We calculate the impact on each aspect 
72 
 
 
 
at four different radiative forcing levels and apply these to the appropriate scenarios 
within the SSP-RCP scenario matrix. Future water supply is calculated by using bias-
corrected hydrologic data derived for four RCPs from five GCMs as part of the Inter-
Sectoral Impact Model Intercomparison Project [ISI-MIP; Warszawski et al. (2014)]. 
These values are entered into the global hydrologic model Xanthos (Li et al. 2017, 
Liu et al., 2018; Vernon et al. 2019), which calculates accessible water at the GCAM 
235-basin scale at five-year time steps. Climate derived impacts to crop yield changes 
(Rosenzweig et al. 2014), hydropower expansion (Turner et al., 2018), and building 
energy expenditures (Clarke et al., 2018) are calculated from the same set of ISI-MIP 
models and the climate varying impacts are added to their respective RCP scenarios. 
By including scenarios that both include and exclude climate impacts we can account 
for the compounding effects of changing hydrologic conditions, hydropower 
availability, crop yields, and energy demands on water scarcity, while also separating 
the impact of human activities from that of the climate system, which include 
changing water demands for agriculture, power generation, industry, and public 
supply. 
3.2.4 Human and Climate Contributions to Water Scarcity 
Water scarcity analyses can contain several different aspects of water use. For 
the purpose of this study, we will consider the water scarcity index, noted below as S, 
by comparing the amount of total water withdrawals, Wd from all sources (renewable 
surface runoff, nonrenewable groundwater extraction, and desalination) to the total 
GCM derived accessible surface runoff, QR, shown in Equation 3.1, where we 
calculate scarcity in time period, t, basin, b, and SSP-RCP-GCM scenario, s. 
 
73 
 
 
 
 WS D>,@,A = 	 Q  (3.1) /
The change in scarcity, for the ?Human Alone?, SH, and ?Human and Climate?, 
SX scenarios, is calculated in Equation 3.2a and b respectively, as we use 2005 as the 
base year of water scarcity and calculate changes throughout the century at any time 
period t. To isolate the effect of climate alone, SC, we subtract the change in water 
scarcity found in the ?Human Alone? scenarios from the ?Human and Climate? 
scenarios in Equation 3.2c. The ?Human and Climate? scenarios likely contain 
changes to the human system that are not captured by subtracting the ?Human Alone?, 
however these changes are assumed to be a result of the climate impacts and therefore 
are classified as a climate impact in this study. 
 
 ?SE(t) = 	 SEG ?	SE8::; (3.2A) 
 
 ?SH(t) = 	 SHG ?	SH8::; (3.2B) 
 
 ?SI(t) = 	DSH(t) ?	DSE(t) (3.2C) 
 
In order to calculate the impact that humans and climate have on water 
scarcity we adopt and modify Equations S1 and S2 from Veldkamp et al. (2015), 
which are shown as Equations 3.3a and b below. 
 
 ?S (t)
IE(t) = 	
E
|?SI(t)| + |?SE(t)|
 (3.3A) 
 
74 
 
 
 
 ?S (t)
II(t) = 	
I
|?SI(t)| + |?S (t)|
 (3.3B) 
E
 
IH, (human impact) and IC, (climate impact) represent the relative impact on 
water scarcity due to human or climate influences resulting from the change in water 
scarcity from 2005 values. In order to find the main driver of water scarcity we 
compare the two impacts as in Equations 3.4b and c. 
 
 |IE| + |II| = 1 (3.4A) 
   
 Human	Driven:	|II| < |IE| (3.4B) 
   
 Climate	Driven:	|II| > |IE| (3.4C) 
 
The sign of human and climate impacts allows for the classification of 
increasing or decreasing scarcity. 
3.3 Results 
Assessing the relative contributions of socioeconomic and climatic systems to 
the evolution of water scarcity allows for a deterministic classification of the main 
driver behind changes in scarcity. We quantify water scarcity using the water scarcity 
index (WSI), which is the ratio of water withdrawals to water availability. Water 
scarcity changes in extremely wet regions and regions with minimal demands are 
likely to produce minimal changes in WSI. To account for this, the lowest quartile of 
DWSI values are determined to be negligible in each scenario. Basins with negligible 
75 
 
 
 
scarcity changes are left unclassified and are referred to as ?Negligible? throughout 
the remainder of this study.  
3.3.1 Main Driver Behind Water Scarcity Changes 
Across all scenarios it is found that a majority of the basins that have 
negligible changes in water scarcity are located in the Northern Hemisphere high 
latitudes, Amazonian Rainforest, and parts of Australia, regardless of time period or 
scenario (Figure 3.3, Figure 3.5). In the basins which have a defined change in water 
scarcity, 78% experience water scarcity changes dominated by humans (red shading 
and Figure 3.4). This is true for SSP1, SSP2, and SSP4. Whereas in SSP3, extremely 
high population growth results in 90% of basins with water scarcity changes driven 
by humans, and SSP5, where only 71% are driven by humans due to the contributions 
from reaching RCP8.5 conditions (Figure 3.4). 
Throughout the century, the number of nonnegligible basins experiencing 
human driven water scarcity changes typically increases up to the year 2100. Human-
dominated water scarcity increases (H+) occur in 51% of basins while 27% of basins 
have decreasing scarcity due to humans (H-). Significant differences arise across 
SSPs as socioeconomic differences drive efficiency improvements, demand changes, 
and altered water dependency (Graham et al., 2018; Calvin et al., 2017). SSP1 
experiences a shift from H+ to H- in a large number of basins throughout the century, 
leading to just 26% of basins classified as H+ in 2100, while humans act to decrease 
water scarcity in 52% of basins. While not as extreme, similar shifts from H+ and H- 
occur in SSP4 and SSP5, whereas in SSP2 and SSP3 more than 65% of basins are 
H+. Although differences exist across each SSP scenario, the three scenarios that shift 
from H+ to H- experience global or regional GDP growth that allows for the adoption 
76 
 
 
 
of efficient water use technologies. Additional storyline features such as sustainability 
focus, fuel preference, and universal adoption rates also influence cross-SSP 
differences. 
Fig 3.3 shows a spatial representation of water scarcity drivers across the 
average of all future scenarios. Although socioeconomic scenario assumptions 
diverge in the second half of the century, the main driver remains humans in most 
basins. The robustness of the results decreases, and uncertainty increases by 2100 
(Figure 3.3), consistent with Fig 3.2. Basins in sub-Saharan Africa have significant 
agreement across scenarios of human driven scarcity changes in both 2050 and 2100 
riven by population growth increases throughout the century. 
 
Figure 3.2 Spatial and temporal changes in the percentage of GCAM?s water basins in which the main 
drivers of water scarcity changes are attributed to humans (H) or climate (C) and whether the changes 
increase (+), decrease (-), or have negligible changes in water scarcity. Each scenario is aggregated 
into individual SSP scenarios and values are calculated from the total number of nonnegligible DWSI 
basins and total number of scenarios in each aggregation. The ?All? scenario represents the total of all 
basins in the suite of SSP-RCP-GCM scenarios 
77 
 
 
 
 
Figure 3.3 Main driver of water scarcity changes due to climate (green) or humans (red) in 2050 (top) 
and 2100 (bottom). Robustness of results shown as the degree of shading in which there is greater than 
(darker) or less than (lighter) 95% agreement across all 75 SSP-RCP GCM scenarios. Basins which 
observed negligible, ?Neg?, water scarcity changes are shaded as such if at least 95% of the scenarios 
agree for that particular basin, all basins with less confidence are shaded according to their main driver 
in the nonnegligible scenarios. 
 
 
Figure 3 4 Percentage of basins with climate (green) or humans (red) as main driver of water scarcity 
changes. All negligible basins have been removed and values represent the percent of remaining 75% 
of basins in which humans or climate dominated water scarcity changes. Box and whisker plots 
represent the uncertainty spread among GCM and RCP combinations. 
78 
 
 
 
 
 
Figure 3.5 Main driver for water scarcity changes by SSP scenarios in 2100, as in Figure 3.2B. 
Classification is determined by the percentage of occurrence in each SSP scenario. SSP-based basins 
where water scarcity changes are deemed negligible are highlighted in gray. 
79 
 
 
 
3.3.2 Co-Influence of Main and Secondary Drivers 
Although the previous section focused on the dominant driver of water 
scarcity, the impacts by the secondary driver must be accounted. These impacts 
provide either compounding, where both systems change scarcity by the same sign, or 
counteracting, where human and climate systems have opposing signs, effects on 
water scarcity changes. To account for this, Figure 3.6 introduces four distinct water 
scarcity change categories to define how human (H) and climate (C) systems 
individually, yet simultaneously, increase (+) or decrease (-) water scarcity within a 
basin.  
 Expanding on the notion of differing signs of change and considering the 
simultaneous impacts of human and climate systems on future water scarcity. Figure 
3.6 shows the combination across all scenarios. When considering all scenarios 
(Figure 3.6, left), 52% of nonnegligible basins show a compounding effect (green and 
red) by the end of the century, whereas 48% of basins have counteracting effects 
(orange and blue). An increasing dependence on socioeconomic future arises 
throughout the century as SSP1 produces compounding effects in 60% of basins. This 
is due to a significant increase in H-C- basins and a shift away from H+C- conditions 
in parts of Africa and Eurasia (areas of decreasing robustness in Figure 3.7 and Figure 
3.8). SSP2 on the other hand has only 45% of basins with compounding impacts 
driven by 51% of global basins falling into the H+C- category. 
Figure 3.5B represents how water scarcity changes depend on geographic location, as 
distinct areas of H- and H+ arise while much of the world experiences climate 
induced scarcity decreases (C-, green and orange). Nearly all of Africa experiences 
counteracting water scarcity effects as population driven increases in demand lead to 
80 
 
 
 
increases in scarcity, while climate-driven increases in precipitation increase water 
supply and thus lower scarcity (Figure 3.8). As in Figure 3.2B, the robustness of the 
classification is again shown to decrease by the end of the century (right). This is due 
to differences in efficiency and the sustainability focus of the socioeconomic 
scenarios driving the evolution of water scarcity categories, particularly in central 
Eurasia and Africa where population change projections and GDP vary by SSP 
(Graham et al., 2018; Calvin et al., 2017). Areas of negligible changes remain the 
same as in Figure 3.3, by definition. 
 
 
Figure 3.6 Temporal changes in the percentage of GCAM?s water basins in which the simultaneous 
impact of human and climate systems on water scarcity changes are shown by component and sign of 
change. Each scenario is aggregated into individual SSP scenarios and values are calculated from the 
total number of nonnegligible DWSI basins and total number of scenarios in each aggregation. The 
?All? scenario represents the total of all basins in the suite of SSP-RCP-GCM scenarios.  
81 
 
 
 
 
Figure 3.7 Water scarcity category in 2050 (top) and 2100 (bottom). Notation shown as (+) 
representing either human (H) or climate (C) increasing water scarcity and (-) decreasing water 
scarcity. (ex. H+C+ humans and climate both act to increase water scarcity in given basin. 95% 
agreement across all 75 SSP-RCP GCM scenarios. Basins which observed negligible water scarcity 
changes are shaded as such if at least 95% of the scenarios agree for that particular basin, all basins 
with less confidence are shaded according to their scarcity category in the nonnegligible scenarios. 
 
 
82 
 
 
 
 
 
Figure 3.8 Water scarcity change category by SSP scenarios in 2100, as in Figure 3.5B. Classification 
is determined by the percentage of occurrence in each SSP scenario. SSP-based basins where water 
scarcity changes are deemed negligible are highlighted in gray. 
83 
 
 
 
 
3.3.3 Representation Across the SSP-RCP framework 
Results thus far have focused on differences across SSPs while aggregating all 
climate futures, but it is important to distinguish between impacts at multiple 
radiative forcing levels. The end of century distribution of human and climate impacts 
in a basin by scenario are displayed in Fig. 3.9B, where the size of each point 
represents the WSI of its basin in 2100. Applying Equations 3.3A and 3.3B, the 
human and climate impacts will fall on the drawn diamond. However, as each SSP-
RCP scenario is made up of five different GCM runs, the robustness of the relative 
contributions towards water scarcity changes of both human and climate for each 
basin can be captured. As points move towards the center of the diamond, the GCM 
agreement on human and climate impacts decreases.  
The distribution of human impacts is shown in Figure 3.9A with clear bimodal 
patterns across all SSPs. The peaks of all distributions fall where (-0.5 > IH > 0.5), 
where IH is the human impact, justifying the notion that humans are having a larger 
impact than climate on water scarcity in most basins globally independent of 
socioeconomic scenario. However, the distribution of these impacts vary greatly by 
SSP. Specifically, SSP1 is largely negatively favored (decreasing scarcity), whereas 
SSP2 and SSP3 lean significantly more positive. The spreads for SSP4 and SSP5 are 
much closer to a uniform bimodal distribution with close to equal numbers of basins 
with positive and negative human impacts.  
The distribution of climate impacts is shown in Figure 3.9C across each of the 
RCP radiative forcing levels. Independent of radiative forcing future, the relative 
impact of climate is slightly negative, shown as the peak frequency is below the zero 
84 
 
 
 
line in each distribution. The magnitude of this impact provides additional evidence 
for the relatively low impact of climate compared to humans across all SSP-RCP 
scenarios. The spread of climate impacts decreases as the end of century radiative 
forcing target decreases, leading to the conclusion that at lower radiative forcing 
futures, climate systems are likely to have smaller impacts on scarcity than humans in 
an increasing number of basins. The opposite result occurs in the RCP8.5 scenario as 
the distribution is more uniform across climate impacts, which emphasizes the 
important notion that at higher radiative forcings, water scarcity will likely be more 
susceptible to climate impacts. As RCP8.5 is currently only attainable under the SSP5 
scenario, additional research is needed to understand how different socioeconomic 
conditions alter future water scarcity at high radiative forcing levels. 
85 
 
 
 
 
Figure 3.9 Human and climate impact quantifications across the 15 SSP-RCP scenarios in 2100. A, B, 
C, Numerical quantification of the relative human (IH, x-axis) and climate (IC, y-axis) impacts on water 
scarcity in 2100 across each of GCAM?s 235 water basins. SSP-RCP combinations shown with no 
plots represent scenarios unattainable either through mitigation (i.e. SSP3-RCP2.6) or baseline 
assumptions fail to reach 2100 forcing levels (i.e. SSP1-4-RCP8.5). A, Probability distribution 
function of the human impact across each SSP scenario including the aggregates of all RCP 
combinations. B, Numerical representation of the 5 GCM mean human and climate relative impact 
across each basin (points). The WSI of each basin in 2100 is shown as the size of points. As GCM 
uncertainty increases, points move inwards from the mathematical limits of equations 3A and 3B. C, 
Probability distribution function of the climate impact across each RCP scenario including the 
aggregates of all SSP combinations. 
86 
 
 
 
3.4 Discussion 
Understanding the impact that human and climate systems have on water 
scarcity in the future is extremely important for policymakers to understand. We have 
expanded upon previous estimates of water scarcity drivers by quantifying how the 
coevolution of a dynamic socioeconomic system and changing climate system may 
alter water scarcity into the future. This has been done by using evolving and 
differing socioeconomic assumptions that account for changes across all sectors 
(Calvin et al., 2017; Graham et al., 2018), in combination with considerations for 
climate impacts to water availability, hydropower expansion (Turner et al. 2017), 
agricultural productivity changes (Rosenzweig et al. 2014), and building energy 
expenditures (Clarke et al., 2018). The use of these socioeconomic and climate 
assumptions in GCAM allow for feedback linkages between these systems, enabling 
price adjustments and resultant demand changes across sectors when supplies become 
increasingly depleted. 
While in line with previous estimates of the influence of human systems on 
water scarcity (V?r?smarty et al., 2000; Arnell, 2004; Alcamo et al., 2007; Hanasaki 
et al., 2013b; Arnell and Lloyd-Hughes, 2014; Schewe et al., 2014; Schlosser et al., 
2014; Shen et al., 2014; Wada et al., 2014; Kiguchi et al., 2015, Veldkamp et al., 
2015; Veldkamp et al., 2016), this study has also quantified the relative impacts 
across a wide range of potential futures and included combined variations to both the 
socioeconomic (SSP scenario assumptions) and climate (RCPs and GCM derived 
impacts) systems which have previously been unexplored. In addition, we have 
included the ability to not only withdraw, but constrain, local runoff and alternative 
87 
 
 
 
(groundwater and desalination) water sources (Kim et al., 2016; Graham et al., 2018; 
Turner et al., 2019a). 
These results have shown that human activities drive water scarcity changes 
through 2100 in 78% of basins that observed nonnegligible changes in water scarcity, 
and that these impacts may be increasing or decreasing dependent upon 
socioeconomic scenario and geographic location (Figures 3.2, 3.9A, and 3.9B). For 
example, throughout the century, SSP1 scenarios cause water scarcity to consistently 
move from H+ to H- (Figures 3.3 and 3.7), showing that increases in efficient water 
technologies (Graham et al., 2018), future GDP increases, and a focus on 
sustainability may help to alleviate some future water stresses particularly in Eurasia 
and the eastern United States (Figures 3.5 and 3.8). The counteracting impacts of 
human and climate systems are shown to depend on each socioeconomic scenario, 
since increases in basin numbers with counteracting impacts occur in SSP2, SSP3, 
and SSP5, while a shift to H-C- decreases the counteracting impacts in SSP1.  
We also find that when population growth continues at or exceeds present day values, 
GDP growth slows (Calvin et al., 2017), and there exists an inability to invest in 
efficient futures (Graham et al., 2018), increases in water demands lead to increases 
in human driven scarcity. The geographic location of water scarcity is also shown to 
have consistent locations, as the Amazon River basin and much of the Northern high 
latitudes experience negligible changes in WSI independent of future, while basins in 
the southwestern United States and much of Africa experience worsening water 
scarcity due to human activities, independent of socioeconomic or climate future 
(Figs 3.3 and 3.7). 
88 
 
 
 
3.4.1 Limitations 
This study quantified the relative impacts of both evolving human and climate 
systems on water scarcity. Future classifications of climate or human driven water 
scarcity changes would benefit from a distinct look into how the human components 
change between ?Human Alone? and ?Human and Climate? scenarios. This analysis 
assumes that the human system in both sets of scenarios are comparable. However, 
human-driven adaptation measures needed to address climate impacts in the ?Human 
and Climate? scenarios may result in differing econometric responses and slightly 
different human systems. We have included climate derived impacts for four distinct 
areas, but these results do not fully represent the impacts that a changing climate 
might have on the suite of energy-water-land sectors and thus the impact for the 
climate system may not be complete in select basins or scenarios. In addition, water 
sector technological advances have been prescribed across each SSP at scenario-
dependent rates, regardless of investment cost or cooperation (Graham et al., 2018). 
This may result in overestimations of potential water savings in SSP1 and SSP5 and 
therefore the results presented in this study should be taken as an initial analysis of 
potential water scarcity changes.  
3.4.2 Future Recommendations 
This study highlights the basins in which water scarcity is projected to change 
as climate and human systems evolve dynamically to 2100, and in particular 
identifies basins in which human activities may counteract climate to outweigh 
potential decreases due in large part to population driven demand increases. Future 
studies that account for the direction of water scarcity impacts may provide an 
89 
 
 
 
opportunity to analyze where human intervention may act to reduce future water 
stress and can in turn work together with the climate system to decrease water stress. 
3.5 Conclusions 
This study has focused on the assessment of how human and climate systems 
will change basin level water scarcity under various socioeconomic and climate 
scenarios. This work represents the first analysis of water scarcity drivers in the SSP-
RCP modeling framework while using an Integrated Assessment Model to account 
for cross-sectoral feedbacks while also including water constraints. The study has 
provided three key results.  
 First, global water scarcity is, on average, driven by human activities in 78% 
of non-negligible basins by 2100. This magnitude varies by socioeconomic scenario 
dependent upon the underlying storylines from O?Neill et al. (2017) providing the 
ability to decrease water use through technological improvements or increase through 
population growth.  
 Second, in more than half of global water basins human and climate activities 
compounding effects on water scarcity by 2100. The interaction between primary and 
secondary drivers show that under specific socioeconomic and climate storylines, the 
human and climate systems can work together to decrease water scarcity as seen in 
SSP1. 
 Third, the impact from climate change is shown to become increasingly 
concentrated as being minimally impactful, yet acting to decreasing water scarcity, as 
the radiative forcing target decreases. At higher radiative forcing levels, climate 
change is shown to have larger impacts and drive water scarcity changes in an 
increased number of global water basins by 2100. 
90 
 
 
 
Chapter 4: Future changes in the flows of virtual water (Graham 
et al., in prep) 
4.1 Introduction 
This study provides, for the first time, a future VWT analysis to account for 
changing socioeconomic conditions and general circulation model derived climate 
conditions. We use GCAM in the analysis of future VWT to allow for prices to 
respond to both the availability and profitability of growing and trading rainfed and 
irrigated agricultural goods across regional boundaries. This study also provides for 
the first time an estimate for future evolutions of the nonrenewable groundwater 
embedded in global virtual water trade. We provide global, regional, and basin level 
estimations of virtual water exports for all green and blue water and for nonrenewable 
groundwater, while also tracking the exported crops contributing to the VWT 
analysis. To better represent the availability of all water sources, we incorporate 
constraints to both renewable surface and groundwater recharge as well as to 
nonrenewable groundwater (Kim et al., 2016; Graham et al., 2018; Turner et al., 
2019a). This analysis utilizes the Shared Socioeconomic Pathways ? Representative 
Concentration Pathways framework by considering the evolution of VWT in the 
?middle of the road?, SSP2-RCP6.0 scenario. This study provides a novel estimation 
of the future evolution of the VWT network and the necessary contributions of 
nonrenewable groundwater to meet international agricultural demands. 
4.2 Methods 
This analysis uses GCAM to quantify how much water is embedded in the 
global trading of agricultural goods. This water, called virtual water (Allan 1996), is 
91 
 
 
 
calculated based upon how much water is consumed by the individual exported crop 
in the region where it was grown. In order to account for an evolving market and 
changing production conditions we use a defined future socioeconomic scenario, 
SSP2, under the set of Shared Socioeconomic Pathways (O?Neill et al., 2017; Calvin 
et al., 2017) matched with one global climate scenario RCP6.0, under the 
Representative Concentration Pathways (van Vuuren et al., 2011). A single future is 
utilized in order to gain an introductory analysis of a business-as-usual scenario 
without major deviations from present day trends.  This combination represents one 
example of the SSP-RCP framework (O?Neill et al., 2016), but can be extended to 
additional SSP-RCP combinations to analyze a wider range of potential 
socioenvironmental futures. We introduced climate derived impacts to allow for 
changing water needs based upon five general circulation models (GCMs). These 
impacts included water supply, crop yields, changes in hydropower availability, and 
building energy expenditures. We analyze the amount of green and blue water 
consumption that is embedded in global trade and differentiate between renewable 
surface and groundwater recharge, as well as nonrenewable groundwater to provide 
global estimates, regional contributions, and basin-level usage. Below we describe the 
GCAM model, scenario components, virtual water calculations in GCAM, and 
assumptions for the downscaling of exports and estimations of virtual water imports. 
This study uses the GCAM version 5.0 to investigate the relative contributions 
of climate and human systems on water scarcity regionally and globally under a wide 
range of scenarios. GCAM is a model that links energy, water, land, and climate 
systems to allow for the explicit representation of how climate-induced changes in 
water availability alter energy and land systems. GCAM is a market-equilibrium 
92 
 
 
 
model that allows for prices to be adjusted within each time step to ensure that the 
supply and demand of goods and services remains equilibrated at each time step 
allowing for simultaneous market clearing across sectors. This study accounts for a 
limited supply of water by employing cost resource curves across all 235 basins that 
follow a logit formulation to determine the share of each water source (renewable 
surface water, nonrenewable groundwater, and desalinated water) needed to meet the 
water demands within all basins (Kim et al., 2016; Turner et al., 2019a). As depletion 
of various water sources increases the extraction price increases, which leads to 
compounding price increases on the goods and services that require higher-priced 
water sources. 
4.2.1 Scenario Components 
We use a single temporally varying socioeconomic system represented by the 
Shared Socioeconomic Pathways (SSP), in particular SSP2, in combination with the 
Representative Concentration Pathways (RCPs), RCP6.0, for future climatic changes. 
The SSPs provide a set of five future scenarios with varying changes to global 
population, the economy (Riahi et al., 2017), and land use (Popp et al., 2017), which 
were designed to explore varying degrees of challenges to climate change adaptation 
and mitigation (O?Neill et al. 2017). This scenario uses SSP2, a ?middle of the road? 
scenario that has had qualitative and quantitative assumptions implemented into 
GCAM based upon the storylines outlined in O?Neill et al. 2017 (Calvin et al., 2017; 
Graham et al., 2018). SSP2 represents a world with steady population growth through 
the middle of the century, at which time the global population begins to equilibrate 
towards a 2100 value of 9 billion people. Economic growth continues at present-day 
values, and thus fuel and energy preferences remain very similar to what they are 
93 
 
 
 
today. For these reasons, this scenario represents one with medium challenges to both 
climate mitigation and adaptation (O?Neill et al., 2017). Combining these 
socioeconomic features with future climatic changes, we implement a future RCP6.0 
trajectory that results in end of century climate forcing of 6.0 W/m2.  
This study accounts for four different general circulation model (GCM) 
derived climatic impacts to water supply, agricultural productivity, hydropower 
expansion, and building energy expenditures. We calculate the impact on each aspect 
at four different radiative forcing levels and apply these to the appropriate scenarios 
within the SSP-RCP scenario matrix. Future water supply is calculated by using bias-
corrected hydrologic data derived for four RCPs from five GCMs as part of the Inter-
Sectoral Impact Model Intercomparison Project [ISI-MIP; Warszawski et al. (2014)]. 
These values are entered into the global hydrologic model Xanthos (Li et al. 2017, 
Liu et al., 2018; Vernon et al. 2019), which calculates accessible water at the GCAM 
235-basin scale at five-year time steps. Climate derived impacts to crop yield changes 
(Rosenzweig et al. 2014), hydropower expansion (Turner et al., 2018), and building 
energy expenditures (Clarke et al., 2018) are calculated from the same set of ISI-MIP 
models and the climate varying impacts are added to their respective RCP scenarios. 
4.2.2. Calculation of virtual water components in GCAM 
Virtual water calculations in GCAM require several assumptions to account 
for the fact that trading is done across 32 regions. Demands are calculated at the 
regional level while production occurs at the basin level, and the origin of importing 
goods is not traceable once exports are placed in the global market. In order to 
calculate the different components of virtual water trade, we must first calculate the 
regional and basin level trade. For this, it is impossible to calculate basin level 
94 
 
 
 
imports (Section 4.2.3), but all exports are trackable to the basin level, using the 
proportion of production as a proxy. The following calculations make use of 
agricultural production, demands, and water use of various sources as calculated by 
GCAM. These values are highly scenario dependent and are driven by socioeconomic 
and climate influences, as discussed in Chapters 1 and 2. By using the SSP2-RCP6.0 
scenario, we assume only one population trajectory along with one potential climate 
change outcome. The changes to either the human (SSPs) or climate (RCPs) system 
will have implications on all values and analyses.  
 The trade, T, of any crop, c, from any basin, b, using growth type, g, is 
calculated using the production of that crop in the basin, P, the regional demands D, 
and the proportion of basin level production to the total regional production. Growth 
types are classified as either rainfed, RFD, or irrigated, IRR. Positive values of T, 
represent exports, E, whereas negative values represent the need for imports, I. 
 
 ?
?\,],^(?) = ?
\,],^
\,],^ ? `?b ? d \ ij ?fgh ?
(4.1) 
\,],^
Virtual green water exports, VGE, are calculated by considering the green 
water consumption, GWC, the basin level rainfed crop production, and the rainfed 
exports, E. Virtual green imports, VGI, must consider the amount of virtual green 
water that is in the global market, VGE, the ratio of imports in a region, r, and total 
global imports of each crop, I. Finally, the total virtual green water trade (VGT) is 
calculated at the regional level as the combination of the exports and imports of 
virtual green water. 
 
95 
 
 
 
???
???\,](?) = 	
\,] ? 	?
? \,],/no
 
d \,],/no (4.2A) ?\fgh ?
i
\,],/no
  
\t
?
???q,](?) = rs??? u ?
q,],/no
\,] ?q ?  (4.2B) 
fgh fgh q,],/no
  
\t
???q,](?) = 	rs???\,]u + ???q,](?) (4.2C) 
fgh
 
Virtual blue water analysis follows the same process as for green water, with 
the slight adjustment of accounting for irrigated production and trade, as well as the 
blue water consumption, BWC. Here virtual blue water exports (VBE), virtual blue 
water imports (VBI), and virtual blue water trade (VBT) require the production of 
irrigated agriculture.  
???
???\,](?) = 	
\,] ? 	?  
? \,],w//
d \,],w// (4.3A) ?\ ? ifgh \,],w//
  
\t
?
???q,](?) = rs??? u ?
q,],w//
\,] ?qfgh ?
 (4.3B) 
fgh q,],w//
  
\t
???q,](?) = 	rs???\,]u + ???q,](?) (4.3C) 
fgh
96 
 
 
 
 
Finally, the calculation of virtual groundwater exports (VGWE) considers the 
ratio of groundwater depletion in a basin, GWD, to the total blue water withdrawals in 
the basin, BWW. Multiplying this proportion by the virtual blue water exports, yields 
the amount of the blue water exports that is from nonrenewable groundwater sources.  
 
???
????\,](?) = ???\,](?) ?
\
??? 	 (4.4) \
 
Total virtual groundwater trade (VGWT) and virtual groundwater imports 
(VGWI) are calculated in the same manner as 4.4, by considering the blue water 
imports and total trade as the first term on the right-hand side of the equation. 
4.2.3 Virtual water assumptions for GCAM 
Several assumptions must be made in order to analyze virtual water trade and 
differentiate between water sources being traded within GCAM. First, GCAM 
consists of 32 energy-economy regions between which trading can occur. This results 
in far fewer trade partners than there are when analyzing country-level data. There 
also exists an inability to track trading within each of the GCAM regions. 
Intraregional trade is aggregated in the calibration years to assess the trade between 
the 32 energy-economy regions, after which trade remains aggregated in future years 
due to spatial limitations within GCAM. This will lead to VWT values that are lower 
than analysis done at the country level but provide a valuable first approximation. 
Demands of goods in GCAM are determined at the regional level while production is 
calculated at the basin level, therefore in order to assess the basin-level trading of 
97 
 
 
 
goods, we assume that regional exports come from each basin proportionate to the 
amount of production of that good in the basin (Equation 4.1). This simplified 
downscaling measure likely results in some misrepresentations but is one of the best 
first-order downscaling methods to account for the lack of basin level demands. 
Additional assumptions must be made when considering the importing goods and 
virtual water in a region. After being exported, crops enter a global market and are 
distributed to regions based upon unmet demands, however there is no tracking to 
determine where these crops are coming from. The global market distributes crops 
based upon unmet regional demands, but does not keep track of source region. In 
order to attempt to estimate the virtual water imports, we assume that regions import 
crops and virtual water proportionate to the amount coming from exporting regions. 
For example, if 80% of the exports of corn come from the United States, any 
importing region will receive 80% of their imports of corn from the United States 
(Equation 4.3B). Finally, when differentiating between renewable water and 
nonrenewable water contained in the virtual water trade, we assume that the 
proportion of surface runoff withdrawals to nonrenewable withdrawals remain 
consistent for agricultural purposes. We use this to estimate the VGWT as a 
proportion of the VBE (Equation 4.4).  
4.3 Results 
4.3.1 Global estimates of all sources of virtual water 
Virtual water trade, for the purpose of this study, is the amount of water, 
green, blue, or nonrenewable groundwater, that is consumed in the production of an 
agricultural good that is then traded in the international market. The amount of virtual 
98 
 
 
 
water, from all sources, that is traded globally increases throughout the century 
(Figure 4.1). By definition, VWT, or (VWI + VWE), must equal zero globally, 
therefore we focus on the exports to assess the absolute amount of virtual water 
within the global market. Uncertainty of virtual water exports (VWE) increases due to 
differences in GCM-derived climate impacts for the RCP6.0 scenario. In total, VBE 
and VGE at least triple by the end of the century in response to both increases in 
population and resultant demand increases. While population increases drive initial 
increases in VWE, the export per capita experiences a more significant increase after 
2050 as global population begins to equilibrate (Figure 4.2). Throughout the century, 
a significant reliance emerges on trading from water-rich regions of North and South 
America or from regions that experience significant population dynamic changes such 
as China, to water scarce regions of India, the Middle East, and Pakistan. 
Nonrenewable groundwater is increasingly used in agricultural trade throughout the 
century with an observed doubling of VGWE by the end of the century, but with a 
definitive peak mid-century. 
99 
 
 
 
 
 
Figure 4.1 Annual water flows of green, blue, and groundwater embedded in agricultural trade for 
SSP2-RCP6.0. Range of virtual green and blue water exports and the amount of nonrenewable 
groundwater depletion embedded in agricultural trade for all SSP2-RCP6.0 scenarios including GCM 
uncertainty. VGE is shown on primary y-axis (left) while VBE and VGWE are represented on 
secondary y-axis (right). Solid lines represent the average for each water flow in SSP2-RCP6.0. Virtual 
green and blue water exports are shown to at least triple from starting 2010 values with increases due 
to population changes and global production shifts. VGE show steep increases towards 2050, with 
gains much smaller thereafter, reflective of the SSP2 population trajectory. VBE has a much steadier 
gain throughout the century with nearly consistent per-year increases. VGWE (purple) undergo a five-
fold increase towards 2050 and gradual decreasing thereafter.  
100 
 
 
 
 
Figure 4.2 Per capita exports of each virtual water trade component by region. A, Virtual green water 
exports per capita. Argentina, Australia, and Canada represent the three highest per capita exporters of 
green water with general upwards trends throughout the century. B, Virtual blue water exports per 
capita. Significant upwards trends are shown in China and Southeast Asia, while declines occur in 
Pakistan and the Middle East after 2050. C, Virtual groundwater exports per capita. Peaks are observed 
in Pakistan and the Middle East as groundwater is extracted early in the century as the depletion 
lessens after 2050, the per capita values drop to zero. Increases are observed in Australia, Argentina, 
and Mexico. 
 
101 
 
 
 
 Previous studies have reconstructed of historical VWT and are compared to 
the values found in this study along with future projections in Table 4.1. This study 
has used two methodologies to determine the VWT. We find when using export 
values and trade across all countries provided by FAO historical data (FAO, 2017), 
estimates of VWE, particularly VGE, are very similar to previous estimates. 
However, GCAM traces trade across 32 regions and therefore values provided in this 
study are likely to be lower than other studies due to the inability to track trade within 
a GCAM region. The same comparison between country level and GCAM regional 
level trade is completed for nonrenewable groundwater with similar comparisons to 
previous estimates of nonrenewable groundwater embedded in agricultural trade 
(Dalin et al., 2017). However, here, we consider the consumptive nonrenewable 
groundwater rather than just the extracted volume. While lower than previous 
estimates of nonrenewable water in trade, groundwater depletion values found in this 
study are in line with previous historical and future analyses (Table 4.2). 
 
Table 4.1 Global physical water flows 
Water Flows Annual flows (km3/year) Source 
 1996-2005 2010 2050 2100  
VGE 1352    Hoekstra and Mekonnen 
 2012 
  1239   This Study1 
  905 2745-30403 3222- This Study ? SSP2-RCP6.02 
37083 
VBE 255    Hoekstra and Mekonnen 
 2012 
  101   This Study1 
  56 122-1453 179-2083 This Study ? SSP2-RCP6.02 
VGWE  25   Dalin et al. (2017)4 
  17   This Study1 
  4 13.5-23.53 7.5-11.53 This Study ? SSP2-RCP6.02 
1Calculated using trade between each country from 2010 FAO country-level crop export data. 
2Calculated using trade between each of the 32 regions in GCAM. Does not include intraregional trade. 
3Range across the five GCM suite of SSP2-RCP6.0 model runs. 
4Calculated using groundwater depletion rather than groundwater consumption 
 
102 
 
 
 
Table 4.2 Comparison of nonrenewable and blue water withdrawals. Comparison to historical and 
future projected water withdrawals and nonrenewable groundwater depletion through the end of the 
century across various future socioeconomic analyses. 
Water Flows Annual flows (km3/year) Source 
 2000-2010 2050 2100  
Nonrenewable 280   Wada et al. (2010) 
Groundwater 
Depletion 
 
 140   Konnikow (2011) 
 292   Dalin et al. (2017) 
 332 775  Yoshikawa et al. (2014) 
 550 150-1750 60-1500 Kim et al. (2016) 
 300  510-680 Wada and Bierkens (2014) 
  320-910 110-480 Turner et al. (2019b) 
 218 520-873 314-401 This Study ? SSP2-RCP6.0 
Blue Water 3853   FAO (2016) 
Withdrawals 
 
 4000 5750 6000 Wada and Bierkens (2014) 
 3710 6195-8690 4869-12693 Hejazi et al. (2014b)  
 3250 3700-4200  Bijl et al. (2018) 
 3594 4931-5125  Alcamo et al. (2007) 
 3860 4875-5120 4490-4820 This Study ? SSP2-RCP6.0 
 
4.3.2. Regional total virtual water trade 
Changes to the components of the VWT network between 2010 and 2100 are 
shown in Figure 4.3. A large intensification of VGT is observed as the trading of oil 
crops represents a large proportion of initial trading in 2010 (Fig 4.3A). Increases in 
corn, oil crops, and wheat lead to significant VGE increases in 2100 (4.3B). The 
imports of VGE are concentrated in much of Africa, Europe, and India. VBT shows 
significant differences arising in China, Pakistan, India, and the Middle East as the 
availability of water for irrigation decreases and population changes throughout the 
century (Figure 4.3C and D). In 2100, China represents a large source of virtual water 
embedded in the exports of wheat and rice products. China represents a unique case 
of shifting from importer to exporter into the future, and this is due to steep declines 
in total population after 2030. Reduced demands allow for all excess production to be 
103 
 
 
 
used to meet international agricultural demands.  The United States represents another 
main source of future VBE by exporting several crops while needing to import 
miscellaneous crops (MiscCrops, e.g. fruits, vegetables, nuts) as parts of southwestern 
United States shifts production away from MiscCrops towards the end of the century. 
An important note is that virtual water trading does not necessarily mean that the 
trading of agricultural goods is actually increasing, rather it determines the amount of 
water required to grow the crops that are then traded globally. If the originating 
exports are extremely water intensive in a particular region, an intensification of 
VWT will be observed if this region is then trading these crops globally. We have 
found that this intensification does occur in the early part of the century as the Middle 
East and Pakistan contribute to the global market, while towards the end of the 
century, exports come from water-rich areas that require less water to grow (Figure 
4.6). 
 VGWE concentrate in several main regions, the United States, Mexico, 
western South America and Northern Africa. Following the same methodology of 
total virtual blue water trade, it is found that once again the water scarce regions of 
Pakistan, Middle East, and India represent some of the largest importers of 
nonrenewable groundwater in the form of agricultural goods. On a temporal scale, the 
water scarce regions are found to export nonrenewable groundwater early in the 
century but cease to do so after mid-century as climate induced water scarcity begins 
to hinder this ability (Figure 4.6). 
 
104 
 
 
 
 
Figure 4.3 Virtual water trade fluxes by region and crop in 2010 and 2100. A, B, Global virtual green 
water trade (bcm) by crop and aggregate GCAM region in 2010, A, and 2100 for SSP2-RCP6.0, B. All 
green water traded is from rainfall that grows crops viable for consumption. Imports (negative values) 
are assumed to come proportionately from exporting regions dependent upon total regional exports of a 
crop type. Water intensities for these imports are then scaled dependent upon the proportionality of 
exporting region intensities. C, D, Global virtual blue water trade (bcm) by crop and aggregate GCAM 
region in 2010, C, and 2100 for SSP2-RCP6.0, D. Scaling of imports follows the same methods for 
green water. E, F, Global virtual groundwater trade (bcm) by crop and aggregate GCAM region in 
2010, C, and 2100, D. Values for imports follow same logic as was used for VWT, with exporting 
regional nonrenewable to renewable water use calculated and applied. 
105 
 
 
 
 
Figure 4.4 Virtual water trade fluxes by region and crop in 2030 and 2050. A, B, Global virtual green 
water trade (bcm) by crop and aggregate GCAM region in 2030, A, and 2050 for SSP2-RCP6.0, B. All 
green water traded is from rainfall that grows crops viable for consumption. Imports (negative values) 
are assumed to come proportionately from exporting regions dependent upon total regional exports of a 
crop type. Water intensities for these imports are then scaled dependent upon the proportionality of 
exporting region intensities. C, D, Global virtual blue water trade (bcm) by crop and aggregate GCAM 
region in 2030, C, and 2100 for SSP2-RCP6.0, D. Scaling of imports follows the same methods for 
green water. E, F, Global virtual groundwater trade (bcm) by crop and aggregate GCAM region in 
2030, C, and 2050, D. Values for imports follow same logic as was used for VWT, with exporting 
regional nonrenewable to renewable water use calculated and applied. 
 
106 
 
 
 
 
Figure 4.5 Crop breakdown of each virtual water trade component. A, Virtual green water exports by 
crop, from 2010 to 2100. An intensification of every crop type is seen throughout the century. B, 
Virtual blue water exports by crop and region. Increases in wheat and rice make up the largest portion 
of virtual blue water exports. C, Virtual groundwater exports by crop. Quick increases in rice and 
miscellaneous crop trade is observed through 2050.  
107 
 
 
 
 
Figure 4.6 GCAM region breakdown of each virtual water export component. A, Virtual green water 
exports and from GCAM regions, from 2010 to 2100. Brazil and Northern Africa observe the largest 
increases in net exports. B, Virtual blue water exports by region. An intensification of exports from 
China is observed. C, Virtual groundwater exports by region. An intensification of exports from 
Pakistan and the Middle East prior to 2050. After 2060, the groundwater resources become increasing 
exhausted and more expensive, therefore the exports from these regions cease.  
 
4.3.3. Basin level virtual water exports 
While it is important to understand both global and regional changes to the 
components of VWT, expanding this to understand how specific basins contribute to 
the observed increases in exports of all virtual water variations provides an important 
localized analysis. Downscaling to the GCAM 235 water basin scale yields specific 
locations in which VGE and VBE originate. When comparing the 2050 to 2100 
values of VGE and VBE (Fig 4.7A-D), the intensification of exports is evident in 
much of the water basins in China. Blue and green exports also concentrate in the 
Missouri River basin, the La Plata basin in South America, and the Murray-Darling 
108 
 
 
 
basin in Australia. Each of these basins have large amounts of agricultural production 
and will be relied upon heavily to meet future demands.   
 Tracing the VGWE show a significant time evolution as basins in Saudi 
Arabia and the Indus River basin have large amounts of exported water in 2050, but 
do not contribute to the global VGWE in 2100. This is due to extraction in the first 
half of the century from the large underground aquifers in Saudi Arabia and India (Al 
Alawi,1994), causing additional pumping to become too expensive in these regions. 
Additionally, high amounts of VGWE come from the California River basin, the 
Arkansas River basin, and northwestern Mexican coast in North America, the Nile 
River, the La Plata basins, and the Murray-Darling basin in Australia. The Arkansas 
River basins resides on top of the southern portion of the Ogallala aquifer, which has 
the largest groundwater reserves in the United States. Exports are not shown from the 
Missouri River basin, on top of the deepest portion of this aquifer as groundwater 
extraction calibration (Turner et al. 2019a), has shown that groundwater recharge is 
greater than nonrenewable extraction in this basin (Scanlon et al., 2018), therefore the 
groundwater withdrawn in the Missouri River basin is classified as renewable and 
captured in the VBE. The Nile and La Plata basins are shown to be using 
nonrenewable groundwater for the production of rice, fibers, and corn that is 
demanded outside of regional boundaries (Fig 4.3E). 
109 
 
 
 
 
2050 2100 
 
 
Figure 4.7 Basin level virtual water exports in 2050 and 2100 for all sources, Virtual green water 
exports (bcm) A and B, and blue water exports (bcm), C and D, in 2050 and 2100 respectively for the 
average of five GCM runs for SSP2-RCP6.0. Virtual groundwater exports (bcm) in 2050 and 2100 for 
the same averaged GCM runs for SSP2-RCP.6.0 is shown in E and F. All values are considering the 
exports of agricultural crops only with additional, potentially necessary virtual water imports not 
considered. VGE are concentrated in much of China, central North America, and eastern South 
America. Exports of blue water come mainly from China, and the Missouri River basin in the United 
States. Virtual groundwater extraction in agricultural trade is largest in the California River basin, the 
Arkansas River basin, southwestern North America, the Nile River basin, western South America, and 
the Murray-Darling basin in Australia. 
110 
 
 
 
4.4. Discussion 
Using GCAM to account for the future evolutions of the global trade market 
dependent upon changing socioeconomic and climate conditions has allowed for a 
first look at how virtual water trade may evolve over the century. In this analysis, we 
have built upon previous advances in the reconstruction of the historical global virtual 
water trade network (Dalin et al. 2012, Mekennon and Hoekstra, 2012, Dalin et al., 
2017) by allowing future socioeconomic and climatic changes to alter the production 
of agricultural goods that causes resulting price fluctuations in the global trade market 
and potential restructuring of global agricultural trading. We find that as a result of 
changing socio-environmental conditions, the amount of virtual green and blue water 
in the global trade market will increase throughout the century from 905 bcm and 56 
bcm in 2010 to more than 3200 bcm and 170 bcm, respectively, by the end of the 
century. This time-forward look at virtual water trade has also provided the first 
analysis of how much nonrenewable groundwater is extracted from aquifers around 
the world to meet the international crop demand. An initial 500% increase in this 
extraction towards 2030 eventually levels off in the second half of the century as 
Middle Eastern regions move away from groundwater pumping. This results in a 
general doubling of nonrenewable groundwater in the global market by 2100, much 
of which originates from southwestern North America, the Murray-Darling and the 
Nile River basins. 
4.4.1 Limitations 
This first analysis of the evolution of virtual water trade and nonrenewable 
groundwater trade has yielded an initial set of results that can be used to understand 
111 
 
 
 
the dependency upon rainfall, renewable, and nonrenewable water usage in external 
regions to meet future global demands under evolving socioeconomic and climate 
conditions. However, future analysis of the virtual water trade network should expand 
upon the intraregional trade which was not captured in this study. Trading between all 
countries is likely to increase the VWT values shown and therefore a more extensive 
trade network is needed to fully understand how both renewable and nonrenewable 
water may be traded into the future. Along these lines, it is also important to better 
understand the virtual water flows into importing countries. Exports are able to be 
tracked explicitly in GCAM, however, once in the global market, the originating 
region of an imported good cannot be traced. It is important to understand where the 
imported virtual water is coming from to better understand interregional 
dependencies.  
4.4.2. Looking forward 
This study has focused only on water used to grow agricultural crops, and 
while nearly 90% of blue water consumption is used for agricultural purposes 
(Falkenmark and Rockstr?m, 2006), energy and industrial goods are extensively 
traded in the global market and it is important to understand international dependency 
on these different sectors and how this may change into the future. Understanding 
how trade changes into the future is not a trivial task, but obtaining estimates based 
upon differing climate and socioeconomic conditions can allow for a wide range of 
potential trade evolutions. This study has focused on one socioeconomic scenario 
(SSP2) and one climate scenario (RCP6.0), however, to obtain this range of potential 
outcomes it is important to analyze how the VWT network develops under different 
socioeconomic and climate conditions.  
112 
 
 
 
4.5 Conclusions 
This study has provided, for the first time, a projection of future virtual water 
trade. Using the trade feedbacks embedded in an Integrated Assessment Model, trade 
responses to price and availability of agricultural goods are possible. This work 
considers the amount of green, blue, and nonrenewable groundwater that must be 
used to meet international agricultural demands. The inclusion of water constraints 
and a consistent socioenvironmental scenario have allowed for the first estimations on 
the reliance on external sources of water to meet domestic demands. This work has 
provided three key results.  
 First, global virtual water trading is projected to increase by at least three 
times present day values by 2100. This includes at least a tripling of green water 
trade, and a doubling of blue water and nonrenewable groundwater trading. 
 Second, population dynamics allow for China to represent a main source of 
blue water exports in the future as population declines after 2030 create a surplus of 
production to demands. Slight changes to this specific projection would cause large 
changes in the current projections for the SSP2-RCP6.0 scenario. 
 Third, the contributions from nonrenewable groundwater show that continued 
extraction from California, the Nile River Basin, and the Murray-Darling basin will 
be needed to meet international agricultural demands. 
 
113 
 
 
 
Chapter 5:  Conclusions and Future Work 
 
 This thesis has provided a look at how the use and dependency of water 
resources might change in the future, dependent on changing socioeconomic and 
climate conditions. The use of an Integrated Assessment Model accounts for  
feedback linkages between the human, energy, and land systems while also placing 
constraints on the water resources accessible for use. Several key conclusions have 
emerged from this work and are presented in Section 5.1. Future investigative 
directions are provided in Section 5.2, with notes on how these studies can be 
improved and how they can be extended in the future. Finally, some concluding 
remarks about how the results here may be interpreted and potential uses for the 
integrated assessment of future water resources use in the future are provided in 
Section 5.3. 
5.1 Addressing Research Questions Posed 
In Section 1.4, three research questions were posed. The conclusions that these 
studies have yielded are presented below, with appropriate references to journal 
article submissions and sections of this text. 
5.1.1. Water sector assumptions for the SSPs 
What are the implications of quantitative assumptions for the water sector 
across the five Shared Socioeconomic Pathways scenarios on global water 
demands in a water constrained world?  
 
 In Chapter 2 and as published in Graham et al. (2018), quantitative water 
sector assumptions for each of the five SSPs were implemented into GCAM to 
114 
 
 
 
analyze how their inclusion would alter the water demands across six demanding 
sectors. Consistency with the framework laid out, first by O?Neill et al. (2017) and 
enhanced for GCAM by Calvin et al. (2017), was imperative. As the SSPs provide a 
wide range of potential socioeconomic futures, the impact of global water demands 
was highly dependent upon individual SSP assumptions. It was found that future 
infrastructure changes in the water sector in SSP5, due to significant increases to 
GDP across all regions, can decrease water demands by up to 32% in 2100. 
Additionally, in SSP1, the focus on sustainability and the ability to invest in future 
water-efficiency improvements has the potential to lead to end-of-century water 
demands lower than present day demands despite a higher standard of living and 
similar global population. SSP3 is the only scenario that does not decrease water 
demands, as several sectors maintain 2010 efficiency values and technological 
improvements are slowed in the municipal sector. As population growth is highest in 
this scenario, the lack of water demand reductions is likely to cause increased stresses 
on water resources in an effort to meet the increased global demands for goods and 
services.  
 Individual sector impacts of water technology inclusions have been analyzed 
and compared to previous studies. The results have shown that water demands from 
the five SSP scenarios run with socioeconomically viable water sector technology 
changes are within the range of previous sector assessments (Figures 2.4, 2.7, and 
2.8). 
 While cross-SSP reductions are shown to have a high dependency on the 
quantitative assumptions, the income level of regions in which the reductions occur is 
also shown to have a high dependency across the SSPs. Future water-demand changes 
115 
 
 
 
in the SSPs depend strongly on adoption and implementation of efficient water 
technologies in low-income regions (as defined in Calvin et al., 2017). In SSPs 1, 2, 
and 5, more than half of the global water demand reductions result from the adoption 
of more efficient technologies in low-income regions. These reductions stem from 
significant increases in irrigation technologies improving from crop flooding to 
universal drip irrigation by the end of the century. Sector specific reductions are also 
shown to have income level dependency as high-income regions observe reductions 
to the electricity and manufacturing sectors due to the highly industrialized state of 
these regions, whereas middle- and low-income regions see larger reductions with 
irrigation technology improvements (Figure 2.10).  
 Finally, it is shown that while the addition of water constraints is expected to 
decrease water demands due to limited supplies (Kim et al., 2016), the reductions 
with water technology improvements result in nearly an order of magnitude greater 
reduction (Figure 2.12). 
 This work has provided the first comprehensive analysis of water sector 
assumptions across the SSPs while accounting for limited supplies of water across 
global basins. This first-step IAM analysis of the various impacts that technological 
change can have on water demands across a range of socioeconomic futures provides 
a starting point for the analysis of future water savings brought upon by human 
interventions. 
5.1.2. Water scarcity drivers across the SSP-RCP scenario matrix 
How does the coevolution of human-energy-water-land-climate systems 
affect the driver of water scarcity changes and how do human and climate 
116 
 
 
 
systems interact to alter water scarcity attributions both spatially and 
temporally? 
 
 In Chapter 3 and Graham et al. submitted, the water sector technological 
assumptions from Chapter 2 are used in combination with climate impacts to four 
different sectors to analyze how the human and climate systems may act to alter water 
scarcity across a wide set of global futures. The use of various socioeconomic and 
climate futures allows for water use to be exposed to differing supply and demand 
scenarios. After accounting for basins which are deemed to have negligible changes 
in water scarcity, it was found that, across all scenarios, 78% of global basins have 
water scarcity changes that are driven by human systems. This value is in line with 
previous estimates of water scarcity drivers. However, dependent upon 
socioeconomic future, these changes may reduce water scarcity or enhance it. 
Throughout the century, SSP1 scenarios cause water scarcity to consistently move 
from human-driven increases to decreases (Figures 3.3 and 3.7), showing that with 
the sustainability focus and necessary means to do so, certain socioeconomic 
conditions can lead to up to 52% of basins experiencing human-driven water scarcity 
reductions from 2005 values. Counteracting water scarcity impacts are also found to 
have a scenario dependence, as SSP2, SSP3, and SSP5 all show increases throughout 
the century, however SSP1 experiences reductions due to human and climate driven 
reductions to water scarcity.  
When determining the relative impacts of human and climate systems on 
changes to water scarcity, it is found that regardless of SSP-RCP combination, the 
human system is likely to have a greater impact than the climate system. As radiative 
117 
 
 
 
forcing target decreases, there is increasing agreement that the climate system will act 
to decrease water scarcity around the world, whereas, this determination becomes less 
certain at higher radiative forcing levels (Figure 3.9C). Independent of socioeconomic 
and climate future, the human system is found to have a higher impact on water 
scarcity changes in a majority of basins around the world. 
 This work has expanded upon previous estimates of water scarcity drivers by 
quantifying how the coevolution of socioeconomic and climate systems may alter 
water scarcity into the future. The use of socioeconomic and climate assumptions in 
GCAM allow for feedback linkages between these systems, enabling price 
adjustments and resultant demand changes across sectors. It also allows, for the first 
time, the consideration of limited supplies of water, and alternative water sources to 
be included in an integrated assessment of water scarcity drivers. 
5.1.3. Future virtual water trade analysis 
How will future changes in socioeconomics and climate affect the amount 
of water embedded in international agricultural trade and what are the 
implications for nonrenewable groundwater extraction? 
 
 Chapter 4 and Graham et al. in prep  analyze socioeconomic and climate 
impacts on the future of water and the drivers behind water scarcity changes, and the 
dependency upon the trading of and for water intensive agricultural goods. Water 
scarcity is shown to increase in over 60% of basins in SSP2 (Figure 3.2). Due to this 
increase in scarcity, domestically grown food may not meet the increased future 
demands in some regions. Therefore, understanding the reliance on water intensive 
trade provides a unique analysis of regional dependencies in the future. Historical 
118 
 
 
 
reconstructions of virtual water trade networks have yielded several gross virtual 
water trade values. These values have been compared with two calculation 
methodologies (Section 4.2 and Table 4.1). It is shown that green water exports are 
comparable to previous studies while blue water export values are found to be lower 
than previous estimates. This is likely due to the aggregation of crop water intensities 
across regions in GCAM limiting the amount of water used for particularly intensive 
crops.  
Under SSP2-RCP6.0 conditions, it is found that the amount of green, blue, 
and nonrenewable groundwater used for the growth of internationally traded 
agricultural goods increases throughout the century. As climate change alters the 
future water availability around the world, regions in the Middle East become 
increasingly reliant on imports from water-rich regions. Regions in Africa also 
observe large increases of virtual imported water due to significant increases in 
demand from an ever-growing population. The reliance on China and the United 
States to meet international demands through blue water use is only possible if China 
follows the currently implemented SSP2 population growth projection. Outside of 
this projected population decrease in China, the global trade market may continue to 
shift to meeting evolving demands, specifically if the population dynamics do not 
occur as prescribed in SSP2. 
 Nonrenewable groundwater use for internationally traded agricultural goods is 
shown to also increase throughout the century. Areas of the southwestern United 
States, the Nile River basin, and the Murray-Darling basin in Australia use high 
amounts of groundwater to grow internationally traded agricultural goods. The notion 
of this potential reliance is an important note for policymakers. 
119 
 
 
 
 This analysis provides, for the first time, a projection of how reliant regions 
around the world will be on the trade of water intensive crops under a single 
socioeconomic and climate scenario. The quantification of the virtual green, blue, and 
nonrenewable groundwater trade provide a novel look at the water necessary to grow 
and trade the agricultural goods needed to meet global demands through the end of 
the century. 
5.2 Future Work 
  The results presented here represent a look at how water usage may change 
across various socioeconomic and climate futures while accounting for interactions 
between the human, energy, and land systems. This thesis has focuses on demands, 
scarcity, and the water embedded in global agricultural trade. However, there are 
several additional considerations that are worthy of exploration in the future. 
Throughout this work, no consideration for water quality was given and this is a 
limitation of GCAM as the modeling capability is currently not available. It is 
important to understand, not only the quantity of water that is need under various 
futures, but how the quality of water changes with implemented technological 
changes, socioeconomic evolution, and climate change.  
Chapter 2 provided an analysis of water sector assumptions across the SSPs; 
however, it was assumed that all technological advancements would be implemented, 
independent of cost. An analysis of the attainability, between cost and cooperation, 
would provide an important update to these assumptions. The cost of moving from 
crop flooding to drip irrigation may not allow these implementations to take place in 
certain regions and resultant water demand reductions may not be as great as 
presented. These assumptions are continued in each chapter of this work and 
120 
 
 
 
therefore the lack of cost consideration will be a limiting factor in future analyses 
until addressed. 
The virtual water trade analysis provided in Chapter 4 has limited the breadth 
of scenarios to just SSP2-RCP6.0. This scenario results in limited climate mitigation 
and provides one potential socioeconomic future. Analysis of the suite of SSP-RCP 
scenarios will provide a valuable assessment of how socioeconomic conditions, 
climate change mitigation, and the extensiveness of the global trade network might 
alter the dependencies of international trade to meet domestic demands. Additionally, 
the ability to account for intraregional trading in GCAM into the future will lead to 
increased accuracy in any virtual water assessment as current estimates are based on 
trade between 32 regions, rather than the over 200 countries that are tracked by the 
FAO.  
The analyses conducted in this thesis have utilized GCAM to investigate 
human and climate impacts on water usage in the future. However, this is not the only 
model that can be used in this type of study and the assumptions used for the SSP-
RCP modeling framework are not the only potential socioeconomic and climate 
futures. Using additional IAMs that can account for human-climate feedback linkages 
can help provide uncertainty ranges for each of the analyses from this thesis. Varying 
individual parameters in the SSP-RCP framework can also assist in the uncertainty 
analysis of these studies, similar to Lamontagne et al. (2018). While O?Neill et al. 
(2017) defined the storylines for the SSPs, the evolution of future socioeconomics is 
very much unknown. Therefore, it must be noted that a limiting factor of this research 
is brought upon by the lack of knowledge about the future evolution of the human and 
climate systems (Chapter 1). 
121 
 
 
 
Finally, the work compiled here has utilized the CMIP5 database for climate 
change impacts across the four previously mentioned RCPs. In the sixth installment 
of the CMIP, additional radiative forcing levels are being explored with inputs from 
IAMs (as described in Chapter 1). Expanding the climate impacts to additional 
radiative forcing levels to better match with the trajectories of the SSPs will improve 
this analysis. As shown in Figure 1.3, the baseline radiative forcing for SSP2, SSP3, 
and SSP4 is 7.0 W/m2, therefore the implementation of more appropriate baseline 
impacts will provide a more accurate representation of some of the SSP scenarios. 
5.3. Concluding Remarks 
 Wada et al. (2017) has stated that one of the key shortcomings of current 
water resource assessments, particularly in the future, is the necessity of modeling the 
co-evolution of human and systems while accounting for land use and climate 
interactions. Additionally, the Sustainable Development Goals have included in them, 
ambitious improvement to the water sector. Understanding both the positive and 
negative consequences of attaining these goals by 2030 are extremely important. In 
order to accomplish this, considerations for both human interventions and climate 
change must be taken into account. This body of work has included in it, not only the 
human-water interactions through and integrated modeling framework but has 
accounted for land use changes and climatic system evolution. This, by no means, is 
the final piece of this puzzle, and the necessity to further our understanding of the 
human-earth system interactions cannot be emphasized enough. Unfortunately, the 
uncertainty surrounding water use, availability, and the response to socioeconomic 
and climate forcings is unlikely to decrease significantly in the coming years. With 
the wide-ranging scenario base for the SSPs and RCPs, the ambiguity of projections 
122 
 
 
 
will remain high at long time scales. For these reasons, it will remain important to 
explore these ?what-if? scenarios in an IAM framework to explore how the human 
and climate systems will alter water on global and local scales. Projections, scenarios, 
and conditions will change and evolve over time, and it is important to react to these 
to understand the evolving consequences. Water availability is declining in many 
parts of the world, and the continued overuse will have lasting impacts in the future. 
Understanding how and why the use and availability of water may change in the 
future will allow for societies to take proper steps to avoid running out of water. 
Water sector analyses, particularly ones that include human interactions, need to 
continue and become increasingly sophisticated by including higher temporal 
resolution, increased spatial resolution, and accounting for the most up-to-date 
changes and projections in socioeconomic and climate systems in order to provide the 
most accurate projections to planners and policymakers. 
 
 
 
 
123 
 
 
 
Bibliography 
 
1. Al Alawi, Jamil. "Water in the Arabian Peninsula: problems and 
perspectives." Water in the Arab World: Perspective and Prognoses (1994): 
171-202. 
2. Alcamo, J., Fl?rke, M., & M?rker, M. (2007). Future long-term 
changes in global water resources driven by socio-economic and climatic 
changes. Hydrological Sciences Journal, 52(2), 247-275. 
3. Allan, John A. "Virtual water: a strategic resource." Ground water 36, 
no. 4 (1998): 545-547. 
4. Arndt, Derek S., Molly O. Baringer, and Michael R. Johnson. "State of 
the climate in 2009." Bulletin of the American Meteorological Society 91, no. 
7 (2010): s1-s222. 
5. Arnell, N. W. (1999). Climate change and global water 
resources. Global environmental change, 9, S31-S49. 
6. Arnell, N. W. (2004). Climate change and global water resources: 
SRES emissions and socio-economic scenarios. Global environmental 
change, 14(1), 31-52. 
7. Arnell, Nigel W., and Ben Lloyd-Hughes. "The global-scale impacts 
of climate change on water resources and flooding under new climate and 
socio-economic scenarios." Climatic Change 122, no. 1-2 (2014): 127-140. 
8. Balmaseda, M. A., Trenberth, K. E., & K?ll?n, E. (2013). Distinctive 
climate signals in reanalysis of global ocean heat content. Geophysical 
Research Letters, 40(9), 1754-1759. 
9. Bijl, D.L., Bogaart, P.W., Kram, T., de Vries, B.J. and van Vuuren, 
D.P., 2016. Long-term water demand for electricity, industry and 
households. Environmental Science & Policy, 55, pp.75-86. 
10. Bijl, David L., Hester Biemans, Patrick W. Bogaart, Stefan C. Dekker, 
Jonathan C. Doelman, Elke Stehfest, and Detlef P. van Vuuren. "A global 
analysis of future water deficit based on different allocation mechanisms." 
Water Resources Research 54, no. 8 (2018): 5803-5824. 
11. Blanc, Elodie, Kenneth Strzepek, Adam Schlosser, Henry Jacoby, 
Arthur Gueneau, Charles Fant, Sebastian Rausch, and John Reilly. "Modeling 
US water resources under climate change." Earth's Future 2, no. 4 (2014): 
197-224. 
12. Brenkert, Antoinette L., Steven J. Smith, Son H. Kim, and Hugh M. 
Pitcher. Model documentation for the MiniCAM. No. PNNL-14337. Pacific 
Northwest National Lab.(PNNL), Richland, WA (United States), 2003. 
13. Brouwer, C., Prins, K., and Heibloem, M.: Irrigation Water 
Management: Irrigation Scheduling. Training manual no. 4, Tech. Rep. 4, 
FAO Land and Water Development Division, Rome, Italy, 1989 
14. Calvin, K., Bond-Lamberty, B., Clarke, L., Edmonds, J., Eom, J., 
Hartin, C., Kim, S., Kyle, P., Link, R., Moss, R. and McJeon, H., 2017. The 
SSP4: A world of deepening inequality. Global Environmental Change, 42, 
pp.284-296. 
15. Calvin, Katherine, Pralit Patel, Leon Clarke, Ghassem Asrar, Ben 
Bond-Lamberty, Ryna Yiyun Cui, Alan Di Vittorio et al. "GCAM v5. 1: 
124 
 
 
 
representing the linkages between energy, water, land, climate, and economic 
systems." Geoscientific Model Development 12, no. 2 (2019): 677-698. 
16. Carr, Joel A., Paolo D?Odorico, Francesco Laio, and Luca Ridolfi. 
"Recent history and geography of virtual water trade." PloS one 8, no. 2 
(2013): e55825. 
17. Chapagain, Ashok K., and Arjen Y. Hoekstra. "The global component 
of freshwater demand and supply: an assessment of virtual water flows 
between nations as a result of trade in agricultural and industrial products." 
Water international 33, no. 1 (2008): 19-32. 
18. Chateau, J., Dellink, R., & Lanzi, E. (2014). An overview of the 
OECD ENV-Linkages model: version 3. OECD Environment Working 
Papers, (65), 0_1. 
19. Chaturvedi, V., Hejazi, M., Edmonds, J., Clarke, L., Kyle, P., Davies, 
E. and Wise, M., 2015. Climate mitigation policy implications for global 
irrigation water demand. Mitigation and Adaptation Strategies for Global 
Change, 20(3), pp.389-407. 
20. Clarke, John F., and Jae A. Edmonds. "Modelling energy technologies 
in a competitive market." Energy Economics 15, no. 2 (1993): 123-129. 
21. Clarke, L., Edmonds, J., Jacoby, H., Pitcher, H., Reilly, J., & Richels, 
R. (2007). Scenarios of greenhouse gas emissions and atmospheric 
concentrations. 
22. Clarke, L., Eom, J., Marten, E. H., Horowitz, R., Kyle, P., Link, R., ... 
& Zhou, Y. (2018). Effects of long-term climate change on global building 
energy expenditures. Energy Economics, 72, 667-677. 
23. Clarke, Leon, J. Lurz, M. Wise, J. Edmonds, S. Kim, S. Smith, and 
Hugh Pitcher. "Model documentation for the minicam climate change science 
program stabilization scenarios: Ccsp product 2.1 a." Pacific Northwest 
National Laboratory, PNNL-16735 (2007). 
24. D'Odorico, Paolo, Joel A. Carr, Francesco Laio, Luca Ridolfi, and 
Stefano Vandoni. "Feeding humanity through global food trade." Earth's 
Future 2, no. 9 (2014): 458-469. 
25. Dai, Aiguo. "Increasing drought under global warming in observations 
and models." Nature Climate Change 3, no. 1 (2013): 52. 
26. Dalin, Carole, Megan Konar, Naota Hanasaki, Andrea Rinaldo, and 
Ignacio Rodriguez-Iturbe. "Evolution of the global virtual water trade 
network." Proceedings of the National Academy of Sciences 109, no. 16 
(2012a): 5989-5994. 
27. Dalin, Carole, Samir Suweis, Megan Konar, Naota Hanasaki, and 
Ignacio Rodriguez?Iturbe. "Modeling past and future structure of the global 
virtual water trade network." Geophysical Research Letters 39, no. 24 
(2012b). 
28. Dalin, Carole, Yoshihide Wada, Thomas Kastner, and Michael J. 
Puma. "Groundwater depletion embedded in international food 
trade." Nature 543, no. 7647 (2017): 700.  
125 
 
 
 
29. Davies, E.G., Kyle, P. and Edmonds, J.A., 2013. An integrated 
assessment of global and regional water demands for electricity generation to 
2095. Advances in Water Resources, 52, pp.296-313. 
30. Dellink, R., Chateau, J., Lanzi, E., & Magn?, B. (2017). Long-term 
economic growth projections in the Shared Socioeconomic Pathways. Global 
Environmental Change, 42, 200-214. 
31. Diaz, JA Rodriguez, E. K. Weatherhead, J. W. Knox, and E. Camacho. 
"Climate change impacts on irrigation water requirements in the Guadalquivir 
river basin in Spain." Regional Environmental Change 7, no. 3 (2007): 149-
159. 
32. Distefano, Tiziano, and Scott Kelly. "Are we in deep water? Water 
scarcity and its limits to economic growth." Ecological Economics 142 
(2017): 130-147. 
33. Distefano, Tiziano, M. Riccaboni, and G. Marin. "Systemic risk in the 
global water input-output network." Water resources and economics 23 
(2018): 28-52. 
34. D?ll, P. (2002). Impact of climate change and variability on irrigation 
requirements: a global perspective. Climatic change, 54(3), 269-293. 
35. Eagleson, Peter S. "Climate, soil, and vegetation: 1. Introduction to 
water balance dynamics." Water Resources Research 14, no. 5 (1978): 705-
712. 
36. Edmonds, Jae, and J. M. Reiley. "Global energy-Assessing the future." 
(1985). 
37. Ercin, A. Ertug, and Arjen Y. Hoekstra. "Water footprint scenarios for 
2050: A global analysis." Environment international 64 (2014): 71-82. 
38. Ercin, A., and Arjen Hoekstra. "European water footprint scenarios for 
2050." Water 8, no. 6 (2016): 226. 
39. Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., 
Stouffer, R. J., & Taylor, K. E. (2016). Overview of the Coupled Model 
Intercomparison Project Phase 6 (CMIP6) experimental design and 
organization. Geoscientific Model Development (Online), 9(LLNL-JRNL-
736881). 
40. FAO (2017) Water Scarcity ? One of the greatest challenges of our 
time Document prepared by B.I. Nyoka. FAO, Rome 
41. FAO. 2011. The state of the world?s land and water resources for food 
and agriculture (SOLAW) ? Managing systems at risk. Food and Agriculture 
Organization of the United Nations, Rome and Earthscan, London 
42. FAO. 2016. AQUASTAT Main Database. Food and Agriculture 
Organization of the United Nations. 
http://www.fao.org/nr/water/aquastat/data/query/index.html?lang=en Database 
accessed on [2017/03/15] 
43. Feeley III TJ, Skone TJ, Stiegel Jr GJ, McNemar A, Nemeth M, 
Schimmoller B, et al. Water: a critical resource in the thermoelectric power 
industry. Energy 2008;33:1?11. 
44. Fischer, G., Hizsnyik, E., Tramberend, S., and Wiberg, D. 2015: 
Towards indicators for water security ? A global hydro-economic classi- 
fication of water challenges, IIASA Interim Report IR-15-013, IIASA, 
Laxenburg, Austria 
126 
 
 
 
45. Fischer, G., Tubiello, F.N., Van Velthuizen, H. and Wiberg, D.A., 
2007. Climate change impacts on irrigation water requirements: effects of 
mitigation, 1990?2080. Technological Forecasting and Social Change, 74(7), 
pp.1083-1107. 
46. Fischer, G?nther, Francesco N. Tubiello, Harrij Van Velthuizen, and 
David A. Wiberg. "Climate change impacts on irrigation water requirements: 
effects of mitigation, 1990?2080." Technological Forecasting and Social 
Change 74, no. 7 (2007): 1083-1107. 
47. Fricko, O., Havlik, P., Rogelj, J., Klimont, Z., Gusti, M., Johnson, N., 
Kolp, P., Strubegger, M., Valin, H., Amann, M. and Ermolieva, T., 2017. The 
marker quantification of the Shared Socioeconomic Pathway 2: A middle-of-
the-road scenario for the 21st century. Global Environmental Change, 42, 
pp.251-267. 
48. Fujimori, S., Hanasaki, N. and Masui, T., 2016. Projections of 
industrial water withdrawal under shared socioeconomic pathways and 
climate mitigation scenarios. Sustainability Science, pp.1-18. 
49. Fujimori, S., Hasegawa, T., Masui, T., Takahashi, K., Herran, D.S., 
Dai, H., Hijioka, Y. and Kainuma, M., 2017. SSP3: AIM implementation of 
shared socioeconomic pathways. Global Environmental Change, 42, pp.268-
283. 
50. Fujino, J., Nair, R., Kainuma, M., Masui, T., & Matsuoka, Y. (2006). 
Multi-gas mitigation analysis on stabilization scenarios using AIM global 
model. The Energy Journal, 343-353. 
51. Graham, Neal T., Evan GR Davies, Mohamad I. Hejazi, Katherine 
Calvin, Son H. Kim, Lauren Helinski, Fernando R. Miralles?Wilhelm et al. 
"Water Sector Assumptions for the Shared Socioeconomic Pathways in an 
Integrated Modeling framework." Water Resources Research 54, no. 9 (2018): 
6423-6440. 
52. Hanasaki, N., et al. (2013a), A global water scarcity assessment under 
Shared Socio-economic Pathways - Part 1: Water use, Hydrol. Earth Syst. 
Sci., 17(7), 2375-2391, doi: 10.5194hess-17-2375-2013. 
53. Hanasaki, N., et al. (2013b), A global water scarcity assessment under 
Shared Socio-economic Pathways - Part 2: Water availability and scarcity, 
Hydrol. Earth Syst. Sci., 17(7), 2393-2413, doi: 10.5194/hess-17-2393-2013. 
54. Hartin, Corinne A., Benjamin Bond-Lamberty, Pralit Patel, and 
Anupriya Mundra. "Ocean acidification over the next three centuries using a 
simple global climate carbon-cycle model: projections and sensitivities." 
Biogeosciences 13, no. 15 (2016): 4329-4342. 
55. Hartin, Corinne A., P. Patel, Adria Schwarber, Robert P. Link, and B. 
P. Bond-Lamberty. "A simple object-oriented and open-source model for 
scientific and policy analyses of the global climate system?Hector v1. 0." 
Geoscientific Model Development 8, no. 4 (2015): 939-955. 
56. Hejazi, M. I., Edmonds, J., Clarke, L., Kyle, P., Davies, E., 
Chaturvedi, V., ... & Calvin, K. (2013a). Integrated assessment of global water 
scarcity over the 21st century-Part 2: Climate change mitigation 
policies. Hydrology & Earth System Sciences Discussions, 10(3). 
57. Hejazi, M. I., Edmonds, J., Clarke, L., Kyle, P., Davies, E., 
Chaturvedi, V., ... & Calvin, K. (2014a). Integrated assessment of global water 
127 
 
 
 
scarcity over the 21st century under multiple climate change mitigation 
policies. Hydrology and Earth System Sciences, 18(8), 2859-2883. 
58. Hejazi, M., Edmonds, J., Chaturvedi, V., Davies, E. and Eom, J., 
(2013b). Scenarios of global municipal water-use demand projections over the 
21st century. Hydrological Sciences Journal, 58(3), pp.519-538. 
59. Hejazi, M., et al. (2014b), Long-term global water projections using 
six socioeconomic scenarios in an integrated assessment modeling framework, 
Technological Forecasting and Social Change, 81, 205-226, doi: 
http://dx.doi.org/10.1016/j.techfore.2013.05.006. 
60. Hejazi, Mohamad I., Nathalie Voisin, Lu Liu, Lisa M. Bramer, Daniel 
C. Fortin, John E. Hathaway, Maoyi Huang et al. "21st century United States 
emissions mitigation could increase water stress more than the climate change 
it is mitigating." Proceedings of the National Academy of Sciences 112, no. 
34 (2015): 10635-10640. 
61. Held, I. M., & Soden, B. J. (2006). Robust responses of the 
hydrological cycle to global warming. Journal of climate, 19(21), 5686-5699. 
62. Hijioka, Y., Matsuoka, Y., Nishimoto, H., Masui, T., & Kainuma, M. 
(2008). Global GHG emission scenarios under GHG concentration 
stabilization targets. Journal of global environment engineering, 13, 97-108. 
63. Hoekstra, Arjen Y., and Ashok K. Chapagain. "Water footprints of 
nations: water use by people as a function of their consumption pattern." In 
Integrated assessment of water resources and global change, pp. 35-48. 
Springer, Dordrecht, 2006. 
64. Hurtt, George C., Louise Parsons Chini, Steve Frolking, R. A. Betts, 
Johannes Feddema, G. Fischer, J. P. Fisk et al. "Harmonization of land-use 
scenarios for the period 1500?2100: 600 years of global gridded annual land-
use transitions, wood harvest, and resulting secondary lands." Climatic change 
109, no. 1-2 (2011): 117. 
65. IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of 
Working Groups I, II and III to the Fifth Assessment Report of the 
Intergovernmental Panel on Climate Change [Core Writing Team, R.K. 
Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 pp. 
66. J?germeyr, J., A. Pastor, H. Biemans, and D. Gerten (2017), 
Reconciling irrigated food production with environmental flows for 
Sustainable Development Goals implementation, 8, 15900, doi: 
10.1038/ncomms15900 
67. J?germeyr, J., D. Gerten, J. Heinke, S. Schaphoff, M. Kummu, and W. 
Lucht (2015), Water savings potentials of irrigation systems: global 
simulation of processes and linkages, Hydrol. Earth Syst. Sci., 19(7), 3073-
3091, doi: 10.5194/hess-19-3073-2015. 
68. J?germeyr, J., D. Gerten, S. Schaphoff, J. Heinke, W. Lucht, and J. 
Rockstr?m (2016), Integrated crop water management might sustainably halve 
the global food gap, Environ. Res. Lett., 11(2), 025002. 
69. Kang, Yinhong, Shahbaz Khan, and Xiaoyi Ma. "Climate change 
impacts on crop yield, crop water productivity and food security?A review." 
Progress in Natural Science 19, no. 12 (2009): 1665-1674. 
128 
 
 
 
70. Kiguchi, M., Shen, Y., Kanae, S., & Oki, T. (2015). Re-evaluation of 
future water stress due to socio-economic and climate factors under a 
warming climate. Hydrological Sciences Journal, 60(1), 14-29. 
71. Kim, S.H., Hejazi, M., Liu, L., Calvin, K., Clarke, L., Edmonds, J., 
Kyle, P., Patel, P., Wise, M. and Davies, E., 2016. Balancing global water 
availability and use at basin scale in an integrated assessment model. Climatic 
change, 136(2), pp.217-231. 
72. Kim, Son H., Jae Edmonds, Josh Lurz, S. Smith, and Marshall Wise. 
"The Object-oriented Energy Climate Technology Systems (ObjECTS) 
framework and hybrid modeling of transportation in the MiniCAM long-term, 
global integrated assessment model." Energy J 27 (2006): 63-91. 
73. Konar, Megan, Zekarias Hussein, Naota Hanasaki, Denise Leonore 
Mauzerall, and Ignacio Rodriguez-Iturbe. "Virtual water trade flows and 
savings under climate change." Hydrology and Earth System Sciences 17, no. 
8 (2013): 3219-3234. 
74. Kriegler, E., Bauer, N., Popp, A., Humpen?der, F., Leimbach, M., 
Strefler, J., Baumstark, L., Bodirsky, B.L., Hilaire, J., Klein, D. and 
Mouratiadou, I., 2017. Fossil-fueled development (SSP5): an energy and 
resource intensive scenario for the 21st century. Global Environmental 
Change, 42, pp.297-315. 
75. Kriegler, E., Edmonds, J., Hallegatte, S., Ebi, K. L., Kram, T., Riahi, 
K., ... & Van Vuuren, D. P. (2014). A new scenario framework for climate 
change research: the concept of shared climate policy assumptions. Climatic 
Change, 122(3), 401-414. 
76. Kyle, P., E. G. R. Davies, J. J. Dooley, S. J. Smith, L. E. Clarke, J. A. 
Edmonds, and M. Hejazi (2013), Influence of climate change mitigation 
technology on global demands of water for electricity generation, 
International Journal of Greenhouse Gas Control, 13, 112-123, doi: 
https://doi.org/10.1016/j.ijggc.2012.12.006. 
77. Lamarque, J-F., Tami C. Bond, Veronika Eyring, Claire Granier, 
Angelika Heil, Z. Klimont, D. Lee et al. "Historical (1850?2000) gridded 
anthropogenic and biomass burning emissions of reactive gases and aerosols: 
methodology and application." Atmospheric Chemistry and Physics 10, no. 15 
(2010): 7017-7039. 
78. Lamontagne, Jonathan R., Patrick M. Reed, Robert Link, Katherine V. 
Calvin, Leon E. Clarke, and James A. Edmonds. "Large ensemble analytic 
framework for consequence?driven discovery of climate change 
scenarios." Earth's Future 6, no. 3 (2018): 488-504. 
79. Li, X., Vernon, C. R., Hejazi, M. I., Link, R. P., Feng, L., Liu, Y., & 
Rauchenstein, L. T. (2017). Xanthos?A Global Hydrologic Model. Journal of 
Open Research Software, 5(1). 
80. Li, X., Vernon, C. R., Hejazi, M. I., Link, R. P., Huang, Z., Liu, L., & 
Feng, L. (2018). Tethys?A Python Package for Spatial and Temporal 
Downscaling of Global Water Withdrawals. Journal of Open Research 
Software, 6(1). 
81. Liu, Yaling, Mohamad Hejazi, Hongyi Li, Xuesong Zhang, and 
Guoyong Leng. "A hydrological emulator for global applications?HE v1. 
0.0." Geoscientific Model Development 11, no. 3 (2018): 1077-1092. 
129 
 
 
 
82. Meinshausen, Malte, Steven J. Smith, K. Calvin, John S. Daniel, M. L. 
T. Kainuma, Jean-Francois Lamarque, Km Matsumoto et al. "The RCP 
greenhouse gas concentrations and their extensions from 1765 to 2300." 
Climatic change 109, no. 1-2 (2011): 213. 
83. Monteith, John L. "Evaporation and environment, in the state and 
movement of water in living organisms." In Symp. Soc. Exp. Biol., pp. 205-
234. Academic Press, 1965. 
84. Mouratiadou, I., Biewald, A., Pehl, M., Bonsch, M., Baumstark, L., 
Klein, D., Popp, A., Luderer, G. and Kriegler, E., 2016. The impact of climate 
change mitigation on water demand for energy and food: an integrated 
analysis based on the shared socioeconomic pathways. Environmental Science 
& Policy, 64, pp.48-58. 
85. O?Neill, B. C., Kriegler, E., Ebi, K. L., Kemp-Benedict, E., Riahi, K., 
Rothman, D. S., ... & Levy, M. (2017). The roads ahead: Narratives for shared 
socioeconomic pathways describing world futures in the 21st century. Global 
Environmental Change, 42, 169-180. 
86. O?Neill, B.C., Kriegler, E., Riahi, K., Ebi, K.L., Hallegatte, S., Carter, 
T.R., Mathur, R. and van Vuuren, D.P., 2014. A new scenario framework for 
climate change research: the concept of shared socioeconomic 
pathways. Climatic Change, 122(3), pp.387-400. 
87. Orlowsky, Boris, Arjen Ysbert Hoekstra, Lukas Gudmundsson, and 
Sonia I. Seneviratne. "Today?s virtual water consumption and trade under 
future water scarcity." Environmental research letters 9, no. 7 (2014): 074007. 
88. Popp, A., Calvin, K., Fujimori, S., Havlik, P., Humpen?der, F., 
Stehfest, E., ... & Hasegawa, T. (2017). Land-use futures in the shared socio-
economic pathways. Global Environmental Change, 42, 331-345. 
89. Riahi, K., Gr?bler, A., & Nakicenovic, N. (2007). Scenarios of long-
term socio-economic and environmental development under climate 
stabilization. Technological Forecasting and Social Change, 74(7), 887-935. 
90. Riahi, K., Van Vuuren, D. P., Kriegler, E., Edmonds, J., O?neill, B. C., 
Fujimori, S., ... & Lutz, W. (2017). The shared socioeconomic pathways and 
their energy, land use, and greenhouse gas emissions implications: an 
overview. Global Environmental Change, 42, 153-168. 
91. Richards, M., Bruun, T. B., Campbell, B. M., Gregersen, L. E., Huyer, 
S., Kuntze, V., ... & Vasileiou, I. (2016). How countries plan to address 
agricultural adaptation and mitigation: An analysis of Intended Nationally 
Determined Contributions. CCAFS dataset. 
92. Rohwer, J., Gerten, D., and Lucht, W.: Development of functional 
irrigation types for improved global crop modelling, PIK report No. 104, 
Tech. Rep. 104, Potsdam Institute for Climate Impact Research, Potsdam, 
Germany, 2007 
93. Rosenzweig, Cynthia, Joshua Elliott, Delphine Deryng, Alex C. 
Ruane, Christoph M?ller, Almut Arneth, Kenneth J. Boote et al. "Assessing 
agricultural risks of climate change in the 21st century in a global gridded 
crop model intercomparison." Proceedings of the National Academy of 
Sciences 111, no. 9 (2014): 3268-3273. 
94. Samir, K. C., & Lutz, W. (2017). The human core of the shared 
socioeconomic pathways: Population scenarios by age, sex and level of 
130 
 
 
 
education for all countries to 2100. Global Environmental Change, 42, 181-
192. 
95. Schewe, J., Heinke, J., Gerten, D., Haddeland, I., Arnell, N. W., Clark, 
D. B., ... & Gosling, S. N. (2014). Multimodel assessment of water scarcity 
under climate change. Proceedings of the National Academy of 
Sciences, 111(9), 3245-3250. 
96. Schlosser, C. A., Strzepek, K., Gao, X., Fant, C., Blanc, ?., Paltsev, 
S., ... & Gueneau, A. (2014). The future of global water stress: An integrated 
assessment. Earth's Future, 2(8), 341-361. 
97. Schmitz, C., Lotze?Campen, H., Gerten, D., Dietrich, J.P., Bodirsky, 
B., Biewald, A. and Popp, A., 2013. Blue water scarcity and the economic 
impacts of future agricultural trade and demand. Water Resources 
Research, 49(6), pp.3601-3617. 
98. Shen, Y., Oki, T., Kanae, S., Hanasaki, N., Utsumi, N., & Kiguchi, M. 
(2014). Projection of future world water resources under SRES scenarios: an 
integrated assessment. Hydrological Sciences Journal, 59(10), 1775-1793. 
99. Shen, Y., Oki, T., Utsumi, N., Kanae, S. and Hanasaki, N., 2008. 
Projection of future world water resources under SRES scenarios: water 
withdrawal Hydrological Sciences Journal, 53(1), pp.11-33. 
100. Shiklomanov, I.A., 2000. Appraisal and assessment of world water 
resources. Water international, 25(1), pp.11-32. 
101. Smith, Steven J., and T. M. L. Wigley. "Multi-gas forcing stabilization 
with Minicam." The Energy Journal (2006): 373-391. 
102. Thomson, Allison M., Katherine V. Calvin, Steven J. Smith, G. Page 
Kyle, April Volke, Pralit Patel, Sabrina Delgado-Arias et al. "RCP4. 5: a 
pathway for stabilization of radiative forcing by 2100." Climatic change 109, 
no. 1-2 (2011): 77. 
103. Turner, S. W., Hejazi, M., Kim, S. H., Clarke, L., & Edmonds, J. 
(2017). Climate impacts on hydropower and consequences for global 
electricity supply investment needs. Energy, 141, 2081-2090. 
104. Turner, Sean WD, Mohamad Hejazi, Catherine Yonkofski, Son H. 
Kim, and Page Kyle. "Influence of Groundwater Extraction Costs and 
Resource Depletion Limits on Simulated Global Nonrenewable Water 
Withdrawals Over the Twenty?First Century." Earth's Future 7, no. 2 (2019a): 
123-135. 
105. Turner, Sean WD, Mohamad Hejazi, Katherine Calvin, Page Kyle, and 
Sonny Kim. "A pathway of global food supply adaptation in a world with 
increasingly constrained groundwater." Science of The Total Environment 
673 (2019b): 165-176. 
106. UN (2019), World Economic Situation and Prospects 2019, UN, New 
York, https://doi.org/10.18356/a97d12e3-en. 
107. United Nations, Department of Economic and Social Affairs, 
Population Division (2017). World Population Prospects: The 2017 Revision, 
Key Findings and Advance Tables. ESA/P/WP/248. 
108. Van Vuuren, D. P., Stehfest, E., Gernaat, D. E., Doelman, J. C., Van 
den Berg, M., Harmsen, M., ... & Girod, B. (2017). Energy, land-use and 
greenhouse gas emissions trajectories under a green growth paradigm. Global 
Environmental Change, 42, 237-250. 
131 
 
 
 
109. Van Vuuren, Detlef P., Jae Edmonds, Mikiko Kainuma, Keywan 
Riahi, Allison Thomson, Kathy Hibbard, George C. Hurtt et al. "The 
representative concentration pathways: an overview." Climatic change 109, 
no. 1-2 (2011): 5. 
110. van Vuuren, Detlef P., John Weyant, and Francisco de la Chesnaye. 
"Multi-gas scenarios to stabilize radiative forcing." Energy Economics 28, no. 
1 (2006): 102-120. 
111. Van Vuuren, Detlef P., Michel GJ Den Elzen, Paul L. Lucas, Bas 
Eickhout, Bart J. Strengers, Bas Van Ruijven, Steven Wonink, and Roy Van 
Houdt. "Stabilizing greenhouse gas concentrations at low levels: an 
assessment of reduction strategies and costs." Climatic change 81, no. 2 
(2007): 119-159. 
112. Veldkamp, T. I. E., Wada, Y., Aerts, J. C. J. H., & Ward, P. J. (2016). 
Towards a global water scarcity risk assessment framework: incorporation of 
probability distributions and hydro-climatic variability. Environmental 
research letters, 11(2), 024006. 
113. Veldkamp, T. I., Wada, Y., de Moel, H., Kummu, M., Eisner, S., 
Aerts, J. C., & Ward, P. J. (2015). Changing mechanism of global water 
scarcity events: Impacts of socioeconomic changes and inter-annual hydro-
climatic variability. Global Environmental Change, 32, 18-29. 
114. Vermeer, M., & Rahmstorf, S. (2009). Global sea level linked to 
global temperature. Proceedings of the national academy of sciences, 106(51), 
21527-21532. 
115. Vernon, C. R., Hejazi, M. I., Turner, S. W., Liu, Y., Braun, C. J., Li, 
X., & Link, R. P. (2019). A Global Hydrologic Framework to Accelerate 
Scientific Discovery. Journal of Open Research Software, 7(1). 
116. V?r?smarty, C. J., Green, P., Salisbury, J., & Lammers, R. B. (2000). 
Global water resources: vulnerability from climate change and population 
growth. Science, 289(5477), 284-288. 
117. Wada, Yoshihide, Ludovicus PH Van Beek, Cheryl M. Van Kempen, 
Josef WTM Reckman, Slavek Vasak, and Marc FP Bierkens. "Global 
depletion of groundwater resources." Geophysical research letters 37, no. 20 
(2010). 
118. Wada, Yoshihide, and Marc FP Bierkens. "Sustainability of global 
water use: past reconstruction and future projections." Environmental 
Research Letters 9, no. 10 (2014): 104003. 
119. Wada, Y., et al. (2013), Multi-model projections and uncertainties of 
irrigation water demand under climate change, Geophys. Res. Lett., 40(17), 
4626-4632, doi: 10.1002/grl.50686. 
120. Wada, Y., et al. (2016), Modeling global water use for the 21st 
century: The Water Futures and Solutions (WFaS) initiative and its 
approaches, Geosci. Model Dev., 9(1), 175-222, doi: 10.5194/gmd-9-175-
2016. 
121. Wada, Y., Gleeson, T., & Esnault, L. (2014). Wedge approach to water 
stress. Nature Geoscience, 7(9), 615. 
122. Wada, Y., Van Beek, L. P. H., Viviroli, D., D?rr, H. H., Weingartner, 
R., & Bierkens, M. F. (2011). Global monthly water stress: 2. Water demand 
and severity of water stress. Water Resources Research, 47(7). 
132 
 
 
 
123. Wada, Yoshihide, Marc FP Bierkens, Ad de Roo, Paul A. Dirmeyer, 
James S. Famiglietti, Naota Hanasaki, Megan Konar et al. "Human?water 
interface in hydrological modelling: current status and future 
directions." Hydrology and Earth System Sciences 21, no. 8 (2017): 4169-
4193. 
124. Warszawski, L., Frieler, K., Huber, V., Piontek, F., Serdeczny, O., & 
Schewe, J. (2014). The inter-sectoral impact model intercomparison project 
(ISI?MIP): project framework. Proceedings of the National Academy of 
Sciences, 111(9), 3228-3232. 
125. Weedon, G. P., S. Gomes, P. Viterbo, W. James Shuttleworth, E. 
Blyth, H. ?sterle, J. C. Adam, Nicolas Bellouin, O. Boucher, and M. Best. 
"Creation of the WATCH forcing data and its use to assess global and 
regional reference crop evaporation over land during the twentieth 
century." Journal of Hydrometeorology 12, no. 5 (2011): 823-848. 
126. Wise, M., Calvin, K., Thomson, A., Clarke, L., Bond-Lamberty, B., 
Sands, R., ... & Edmonds, J. (2009). Implications of limiting CO2 
concentrations for land use and energy. Science, 324(5931), 1183-1186. 
127. WMO, 2019; WMO Statement on the State of the Climate 2019, IBSN: 
978-92-63-11233-0  
128. WWAP (United Nations World Water Assessment Programme). 2015. 
The United Nations World Water Development Report 2015: Water for a 
Sustainable World. Paris, UNESCO. 
129. WWAP (United Nations World Water Assessment Programme). 2018. 
The United Nations World Water Development Report 2018: Nature-based 
Solutions. Paris, UNESCO. 
130. Yamagata, Yoshiki, Naota Hanasaki, Akihiko Ito, Tsuguki Kinoshita, 
Daisuke Murakami, and Qian Zhou. "Estimating water?food?ecosystem trade-
offs for the global negative emission scenario (IPCC-RCP2. 6)." Sustainability 
Science 13, no. 2 (2018): 301-313. 
131. Zhang, Xuebin, Francis W. Zwiers, Gabriele C. Hegerl, F. Hugo 
Lambert, Nathan P. Gillett, Susan Solomon, Peter A. Stott, and Toru Nozawa. 
"Detection of human influence on twentieth-century precipitation trends." 
Nature 448, no. 7152 (2007): 461. 
 
 
 
 
 
 
 
 
133 
 
 
 
 
134