You are on page 1of 51

0DQXVFULSW &OLFNKHUHWRGRZQORDG0DQXVFULSW:DWHU5HVRXUFHVB&ULWLFDOLW\

B8WDKBZLWKRXW)LJXUHV 7DEOHVGRF[

1 Water Resources Criticality Due to Future Climate Change and Population

2 Growth: Case of River Basins in Utah, USA

3 [Krishna B. Khatri, Ph.D., A.M.ASCE1, Courtenay Strong, Ph.D.2, Adam K. Kochanski, Ph.D.3,

4 Steven Burian, Ph.D., M.ASCE, P.E. 4, Craig Miller, P.E.5 , Candice Hasenyager, P.E.6 ]

5 Abstract

6 This study investigates annual river basin-level water criticality (ratio of available water to withdrawals)

7 considering effects of climate change on supply and effects of future population change on demand. A

8 steady state water balance model was developed to estimate the water mass budget and disaggregate the

9 internal and external water supply sources at a river basin level. Future precipitation and evapotranspiration

10 were dynamically downscaled under a moderate greenhouse gas emission scenario to 4-km horizontal

11 resolution using a regional climate model for a decade centered on 2090. The climate data were also

12 statistically downscaled via Bias Correction Spatial Disaggregation (BCSD) method applied to the CMIP5

13 (the fifth phase of the Coupled Model Intercomparison Project) archives for four emission scenarios for

14 decades centered on 2040 and 2090. Bootstrapping and k-nearest-neighbor (k-NN) algorithms were applied

15 to simulate future water demand and external basin supply with uncertainty. Water stress is classified into

16 four levels: (i) “very high-water stress” when the water criticality ratio < 1.25, (ii) “high water stress” when

17 criticality ranges between 1.25 to 2.5, (iii) “moderate water stress” when criticality ranges between 2.5 to

18 10, and (iii) “no water stress” when criticality > 10. A basin with criticality ratio of one indicates that the

19 basin demand has been exactly met by the available supply sources. Most river basins have current water

20 criticality less than two, and are dependent on inflow from other basins (i.e., are not self-sustaining). Future

21 projections indicate modest increases in net available water for Utah through the end of the current century

22 from climate change, with increasing vulnerability largely driven by population growth. Sevan basins out

23 of eleven achieve a high (4 basins) and a very high (3 basins) water stress status by the 2040s. Four basins

1
24 achieve a very high-water stress status by the 2090s compared to only two basins in a very high-water stress

25 status in 2010s.

26 Authors keywords: Water Resources Planning; River Basin; Water Balance Model; Climate Change;

27 Dynamically and Statistically Downscaling; Population Growth; k-NN, Bootstrapping; Uncertainty.

28 1
Research Assistant Professor, Dept. of Civil and Environmental Engineering, University of Utah, Salt

29 Lake City, UT 84112

30 2
Assocaite Professor, Dept. of Atmospheric Sciences, University of Utah, Salt Lake City, UT 84112

31 3
Research Assistant Professor, Dept. of Atmospheric Sciences, University of Utah, Salt Lake City, UT

32 84112

33 4
Professor, Dept. of Civil and Environmental Engineering, University of Utah, Salt Lake City, UT 84112

34 5
Engineer Specialist IV. Utah Division of Water Resources, Salt Lake City, UT84116

35 6
Engineering Manager. Utah Division of Water Resources, Salt Lake City, UT84116

36

37

2
38 Introduction

39 Analysis of future water availability and water demand is a key first step for sustainable water

40 resources management (Gleick 2000; Johnson et al. 2001; Veldkamp et al. 2015; Vörösmarty et

41 al. 2000). Researchers, planners, and policy makers need to know whether the total water supply

42 sources can meet future water demand at various spatial scales such as a single watershed, a

43 municipal boundary, a basin, or a state. However, an accurate estimation of future water

44 availability and demand, at a basin level, is challenging due to the complex dynamics of

45 hydrological systems. The complexity is exacerbated by uncertainty in external drivers including

46 climate change, population growth, economic growth, urbanization, ageing infrastructure systems,

47 policy change, unknown interaction effects, and limited data availability (Barnett et al. 2005;

48 Beniston 2003; Beven 2007; Khatri 2013; Palmer et al. 2008; Rijsberman 2006).

49 Water scarcity methods, which analyze how much water of a specified quality is needed versus

50 how much water is available (or can be made available in a specific region and time), are a

51 commonly utilized framework in water resources planning. Water scarcity studies at basin to

52 national spatial scales generally use water index terms. For example, “water stress index”

53 (Falkenmark et al. 1989; Shiklomanov 1991), “water scarcity” (Seckler at al. 1998), “water

54 criticality” (Alcamo et al. 1997), “water vulnerability index” (Raskin et al. 1997; Vörösmarty et

55 al. 2000), and “water sustainability” (Sandoval-Solis et al. 2010) are indices used to compare

56 available water resources to water demand. Table 1 presents the most widely used water scarcity

57 indices.

58 [Please insert Table 1 here]

3
59 The methodology used for measuring water scarcity has evolved over the past twenty years. Earlier

60 studies presented (see Table 1) simple and easy to understand indicators and methods to assess

61 water availability and water demand; however, they do not comprehensively cover the multiple

62 facets of water resources and water demand assessment. For example, the Falkenmark indicator

63 (Falkenmark et al. 1989) quantifies a total annual renewable water resources per capita, but does

64 not consider changes in water availability due to climate change, socio-economic change, and

65 investment made in infrastructure development. The criticality ratio (Alcamo et al. 1997) divides

66 a water use by an availability but does not disaggregate different water demand and supply sources,

67 such as water available to meet human water demand, consumptive water use and return flows

68 (Rijsberman 2006). The Seckler at al. (1998) approach considers the cost of required infrastructure,

69 but is not always easy to apply due to complex and subjective analysis of socioeconomic dimension

70 and different water users. The water poverty index [e.g., (Sullivan et al., 2003)] combines multiple

71 dimensions of water resources management, however this index does not represent a unique

72 physical status of a basin as the index is the aggregated result of multiple indicators. Studies based

73 on hydrologic modeling [e.g., (Scherer et al. 2015)] are often preferable, if the scale of analysis is

74 manageable, data are available, and the simulations are possible with an acceptable computational

75 expense. Existing scarcity analysis frameworks are now often applied in a way that includes an

76 accounting for uncertainty in the analysis stemming from specification of multiple drivers of

77 changes. Detailed discussion on the methods to derive the indices and their applicability are

78 available elsewhere (Alcamo et al. 2007; Brown and Matlock 2011; Goharian et al. 2016; Liu et

79 al. 2016; Pedro-Monzonís et al. 2015; Roy et al. 2012; Veldkamp et al. 2015).

80 Previous research established methods to analyze water stress levels at different temporal and

81 spatial scales. What is unknown is how to analyze whether a river basin is self-sufficient to support

4
82 potential demand, and to what extent the basin is dependent on other basins for sustainability of

83 water resources. This study investigates current and future (i.e., 2040s and 2090s) water stress

84 levels of eleven river basins in the state of Utah, USA. The water criticality index, used to measure

85 the water stress level, compares the water availability and withdrawals in a basin. The objective

86 was to assess if a river basin can support potential water demand internally, and to what extent the

87 basin is dependent on other basins considering the effects of climate change on water supply and

88 effects of future population change on demand. As detailed in the next section, the analysis

89 framework uses a water balance model driven by dynamically and statistically downscaled future

90 precipitation and evapotranspiration, population-driven municipal water demand modeling, and

91 bootstrapped estimates of external supply and agricultural demand.

92 Methods

93 Study Area and Data Sources

94 Utah features complex terrain representative of much of the Great Basin of the western US and

95 other arid regions of the world. Although largely arid, regions of orographic precipitation provide

96 a water supply that supports residential and agricultural uses. The largest water supply resource in

97 Utah is mountain snowpack; the statewide average annual precipitation is approximately 345 mm,

98 and varies from 250 to 595 mm between the river basins. Utah’s irrigated agriculture area covers

99 4,590 km2, or approximately 2.10% of the state’s area (United States Department of Agriculture

100 2014). In 2005, about 90% of Utah’s fresh water diversions were for irrigation. Of the water

101 diverted for irrigation, most (approximately 82%) was taken from surface water with the balance

102 (approximately 18%) coming from groundwater (Maupin et al. 2014).

5
103 Utah has been one of the top five fastest growing states since 2008 and this rapid population growth

104 is projected to continue in the future (Utah Foundation 2014). The recent report of the Utah

105 Governor’s Office of Management and Budget highlights that Utah has the nationally highest

106 fertility rate (2.33 versus national average of 1.86) and largest household size in 2014 (3.16 versus

107 national average of 2.65), and the state’s life expectancy in 2010 ranked 10th in the nation (Utah

108 Economic Council 2016). In 2010, 91% of the population lived in urban areas compared to 80%

109 in 1970. The majority of the urban population (more than 75%) lives along the Wasatch Front (Salt

110 Lake, Davis, Utah, and Weber counties) (Utah Foundation 2014).

111 The State of Utah plans, manages, and regulates water resources of twelve river basins. Fig. 1

112 presents the river basins, main rivers, and lakes in the state of Utah. This study covers all basin

113 regions within the state border except for the small portion of the Columbia River basin in the

114 extreme northwest corner of the state, which has the least geographical coverage and an almost

115 negligible water contribution to the other river basins. The Utah Division of Water Resources

116 (DWRe) is responsible for a comprehensive water plan that describes water resource development

117 opportunities and problems in the basins, identifies options, and makes recommendations for

118 future actions (DWRe 2016). The DWRe is the provider of most of the data used for this study,

119 such as demographic information within the basins, Municipal and Industrial (MAI) water

120 consumption rate, water conservation policy, agricultural demand, groundwater, basin transfers,

121 and diverted flows. Historical and future climate fields were derived using dynamical downscaling

122 with a regional climate model along with statistical downscaling as detailed below.

123 [Please insert Fig. 1 here]

6
124 Water Resources Criticality Analysis

125 Water critically is defined as the ratio of annual water availability to withdrawals at a basin level.

126 This definition is conceptually similar to quantities used in prior studies to represent water stress

127 in large geographical areas, mainly at a basin, national, or global level (Alcamo et al. 1997; Raskin

128 et al. 1997; Rijsberman 2006; Vörösmarty et al. 2005). Criticality was calculated for each of the

129 river basins for future time (t), distinguishing between an internal criticality

ܳூ ሺ‫ݐ‬ሻ
130 ߰ூ ሺ‫ݐ‬ሻ ൌ ሺͳሻ
‫ܦ‬௕ ሺ‫ݐ‬ሻ

131 and an external criticality

ܳா ሺ‫ݐ‬ሻ
132 ߰ா ሺ‫ݐ‬ሻ ൌ Ǥሺʹሻ
‫ܦ‬௕ ሺ‫ݐ‬ሻ

133 A basin’s internal water criticality,\ I (t ) , quantifies the water stress condition in a basin as the

134 ratio of gross excess precipitation QI (i.e., precipitation minus evapotranspiration) to total basin-

135 level water demand, DB at time, t. The internal criticality of the basin therefore assesses whether

136 the basin is self-sustained in meeting the total basin level water demand, by gross excess

137 precipitation now or in the future (i.e., with climate change). The external water criticality, \ E (t )

138 , of a basin is the ratio of the external yield, QE to the total basin water demand at time t. The QE

139 external yield term includes water resources available from gross excess precipitation (internal

140 yield), as well as external supply sources including water imported from other basins by pipes and

141 canals, potential groundwater sources, and transboundary river flows. The external water criticality

142 index incorporates information about a basin’s dependencies on other basins and external supply

7
143 sources. Methods for quantifying the internal yield, QI (t ) , external yield, QE (t ) , and total demand,

144 DB (t ) , are discussed in the following sections.

145 The water criticality ratios (equations 1 & 2) collectively indicate the level of water stress condition

146 in a basin. Earlier studies (e.g., Alcamo et al. 2000; Raskin et al. 1997) explained the level of water

147 stress with reference to the water criticality ratio. With reference to the Shiklomanov (1991) and

148 Alcamo et al. (2000) studies, this study classifies water stress using four levels: (i) “very high-

149 water stress” when the water criticality ratio is less than 1.25, (ii) “high water stress” when

150 criticality ranges between 1.25 to 2.5, (iii) “moderate water stress” when criticality ranges between

151 2.5 to 10, and (iii) “no water stress” when greater than 10. More discussion on indicators, water

152 stress analysis, stress level classifications, and breakpoints are available in the publication listed

153 on Table 1.

154 Internal and External Yield

155 We consider a steady state water balance model to estimate the water mass budget and disaggregate

156 the internal and external water supply sources over a basin. The standard water budget model is

157 given by (Rasmussen et al. 2014; Tekleab et al. 2011; Zhang et al. 2008)

dS
158 P  ET  QR (3)
dt

159 where dS / dt is the time rate of change of water storage in and above ground (S), P is

160 precipitation, ET is evapotranspiration, and ܳோ is total runoff (i.e., sum of surface runoff, interflow,

161 and base flow) representing the flux of water that is not taken up by ET or stored as soil moisture.

8
162 Averaging or integrating over at least one year, the change in storage term is small relative to the

163 fluxes, and Equation (3) simplifies to

164 ܳோ ൌ ܲ െ ‫ܶܧ‬. (4)

165 It is noted that symbols, such as ܲ, hereafter denote integrations over at least one year. Equation

166 (4) is driven by climate parameters in a basin, and is used here as the internal yield. The historical

167 and future ܲ and ET within the basin boundary were based on dynamical and statistical

168 downscaling as detailed below.

169 A basin may receive water from multiple external supply sources: (i) transboundary basin supply,

170 ܳ஻ ; (ii) water imported by diversion from other basins via water supply pipes and irrigation canal

171 systems, ܳ஽ ; and (iii) extractable groundwater resources, ܳீ . In this case, the water balance within

172 the basin can be written as

173 ܳா ൌ ܲ െ ‫ ܶܧ‬൅ ܳ஻ ൅ ܳ஽ ൅ ܳீ , (5)

174 where ܳா quantifies total water available in a basin, defined as the external yield. The external

175 yield thus represents the potential water supply sources available in a basin due to excess of

176 precipitation over evapotranspiration climatic fluxes, and other available supply sources including

177 transboundary basin transfers, ܳ஻ , water imported from other basins, ܳ஽ , and potential sources of

178 groundwater available within the basin, ܳீ . It is noted that quantity of water exported to other

179 basins, either by irrigation canal and drains or by municipal supply pipes, is subtracted from the

180 supply while calculating ܳ஽ and external yield. This means that ܳ஽ represents a net external

181 imported supply.

9
182 The historical time series data of the transboundary basin transfers, water imported by pipes and

183 canals, and groundwater extraction in monthly time steps for years 1989 to 2010 and for all the

184 eleven river basins were obtained from DWRe. For the future analysis, we assume that all the

185 basins will continue to import and export the existing diverted water supply from their built

186 infrastructure systems, and all the natural river basin supply will continue in the future. A 95%

187 confidence interval of all the future external water supply sources (i.e., groundwater, extraction,

188 diverted water, basin supply) was derived by bootstrapping from corresponding historical time

189 series records as detailed below.

190 Total Water Demand Estimation

191 The total annual water demand at the basin level, ‫ܦ‬஻ ǡcan be written

192 ‫ܦ‬஻ ൌ ‫ܦ‬ெ஺ூ ൅ ‫ܦ‬஺௚ , (6)

193 where ‫ܦ‬ெ஺ூ is municipal and industrial water demand and ‫ܦ‬஺௚ is the agricultural water demand in

194 a basin.

195 The MAI water demand for the future was determined by calculating the per capita water demand

196 and the future population. Per capita water demand in any river basin depends on multiple

197 uncertain factors including climatic condition, house occupancy rate, residential lot size, outdoor

198 water uses, water price, water saving technologies, and human behavior. The effect of water price

199 is likely very small, and there are moderate economies of scale associated with family size

200 (Abdallah and Rosenberg, 2012, DeOreo et al., 2016). DWRe provided targeted average daily

201 water demand for all the river basins. DWRe calculates the average basin level per capita water

202 demand by averaging the daily per capita demand records available from all the water suppliers

10
203 within a river basin boundary. It is noted that the domestic water consumption data does not have

204 disaggregated micro demand components, including outdoor irrigation demand of turfgrass. The

205 turfgrass irrigation demand is one of the largest components of municipal water use in Utah and

206 will have effects on future climate change (e.g., increased ET by turfgrass). The goal of the Utah

207 state water conservation plan is to reduce the 2000 per capita water demand from public

208 community systems by at least 25% by 2025 (DWRe 2010). DWRe monitors the water uses within

209 the basins and sets a goal to meet the 25% objective. We applied the k-NN algorithm (see next

210 section) to generate 21-nearest neighbor samples for each of the basins from the historical water

211 use records. The selected samples were bootstrapped (see next section) 10,000 times to generate a

212 95% confidence interval of daily water demand for each of the basins. For simplicity, per capita

213 water use values are assumed stationary after the state goal is reached in 2025 on through the 2040s

214 and 2090s simulations.

215 DWRe provided the future population data (up to 2060s) for each of the basins based on the

216 population forecast of the Utah Governor’s Office. An annual population growth rate for each of

217 the basins was calculated from the decadal population projection of the Governor’s Office.

218 Multiple sources of uncertainty exist in future population growth rates such as unpredictability of

219 future birth rate, death rate, fertility rate, net migration and economic growth. The scope of this

220 study was to consider the population projection data available from the Governor’s office. The

221 annual growth rate calculated from the Governor’s Office projection was considered to apply a k-

222 NN algorithm and select 21 values of the nearest growth rate. Historical population growth rate

223 (years 1989 to 2010) in each of the basins was also gathered from the US census Bureau (U.S.

224 Census Bureau 2011). Those growth rates were bootstrapped 10,000 times to generate a 95%

225 confidence interval of future population for each of the eleven basins. Similarly, the population of

11
226 2090s was projected considering the population growth rate of 2060s (after k-NN) and projected

227 population of 2060. This means that each basin has a different growth rate for each decade, and

228 the growth rate of 2060 was assumed to continue into the 2090s. The total MAI demand,‫ܦ‬ெ஺ூ , for

229 each basin was calculated by multiplication of per capita MAI demand and forecasted total

230 population.

231 The historical agricultural water use data from years 1989 to 2010 for all the basins were provided

232 by DWRe. Agricultural land will likely be reduced in the future due to ongoing urbanization over

233 the state (Utah Foundation 2014), and various factors such as types of crops, agricultural practices,

234 future irrigation technologies, climatic variables, and policies will have further impacts on future

235 agricultural demand. In addition, global trends indicate that technology improvements will save

236 water, and advances in agricultural practices will improve irrigation water efficiency and

237 agricultural water productivity in the future (Gleick 2002). We simulate ‫ܦ‬஺௚ by bootstrapping 95%

238 confidence intervals from total historical agricultural water demand.

239 Quantifying multiple sources of uncertainty in total water demand estimation stemming from the

240 climate change, population growth, per capita water use, and agricultural water demand, and

241 interacting effects is nontrivial, and is beyond the scope of this study. Readers can find detailed

242 discussion on the sources and methods of uncertainty analysis elsewhere (Buizza 1997; O'Neill

243 2004; Kabir et al. 2015; Mannina et al. 2012; Parker and Wilby 2013). This study analyzes the

244 central tendencies and the bounds of variation within 95% confidence interval applying the k-NN

245 algorithm, bootstrapping technique, and scenarios analysis as discussed in the following section.

246 k-NN Classification Method

12
247 We applied k-NN classification method to select possible population growth rates for each of the

248 river basins with reference to the historical growth rates available from the census records. The k-

249 NN method is used in many supervised learning classification problems (Kataria and Singh 2013).

250 In the classification phase, k is a user-defined constant, and an unlabeled vector is classified by

251 assigning the label, which is most frequent among the k training samples nearest to that query point

252 (Wu et al. 2008). More discussion on k-NN methods and their application in hydrology and water

253 resources areas can be found elsewhere (Bhatia 2010; Rajagopalan and Lall 1999; Wu et al. 2008).

254 We constructed a vector of all the historical population growth rates within several counties of

255 each river basin that were available from the census records (U.S. Census Bureau 2011), and the k

256 nearest neighbors of the Governor’s decadal growth rate were selected based on Euclidean

257 distance. Let R ^r1, r2 ,..., r3` be the set of population growth rate with n training points, and let

258 X test be the test point. The k training points from the training set, R, were found that have closest

259 distance to X test which is measured by the Euclidean distance, ‫ܦ‬௜

260 Di || Ri  X test || , (7)

261 where ݅ is the specific index of the training set. The distribution of population growth rates for

262 particular decades was then developed by bootstrapping the set of the nearest neighbors (10,000

263 samples).

264 Application of Bootstrapping Technique

265 Bootstrapping was used to derive confidence bounds on future projections of population, total

266 water demand (also MAI and agricultural demand), and other water supply sources (i.e., basin

267 transfers, diverted flows, and groundwater). The bootstrapping is a simple non-parametric
13
268 technique and provides valid statistical results even if the standard statistical assumption of

269 normality does not hold. The bootstrap method has an advantage over analytical methods in that it

270 can provide solutions for confidence intervals in situations where exact analytical solutions may

271 be unavailable, or in which approximate analytical solutions are inadequate (Efron and Tibshirani

272 1994; Härdle et al. 2003).

273 We considered a bootstrapping sample size of 10,000 in drawing from the historical records.

274 Further discussions on the methods of bootstrapping techniques and hydrological applications are

275 available elsewhere, such as analysis of time series data (Härdle et al. 2003), specification of

276 bootstrap sample size (Tibshirani 1985), applications for urban water demand forecasting (Tiwari

277 and Adamowski 2013), forecasting of reservoir inflow (Kumar et al. 2015), estimation of

278 uncertainty in the total impact of climate change, and assessment of the performance of commonly

279 used impact functions (Tol 2015).

280 Scenarios Analysis

281 The internal and external criticality estimation (equations: 1 and 2) are associated with multiple

282 sources of uncertainty due to unknown drivers of future climate change, population growth, per

283 capita MAI water demand, agricultural water demand, basin transfer, imported water supply,

284 groundwater recharge rate, and more. For example, major sources of uncertainty in climate change

285 include estimating radiative forcing (emission scenario), knowledge of physical processes and

286 boundary condition, non-linear interactions, future development policy, modeling the natural

287 system, and downscaling. As presented, multiple sources of uncertainty exist in the projection of

288 future population growth, and prediction of water demand. In this case, it is difficult to describe,

14
289 quantify, and propagate uncertainty in the modeling process, and often almost impossible to

290 represent the uncertainties in probabilistic terms.

291 Scenario analysis is one of the methods of uncertainty analysis commonly used for analyzing long

292 term uncertainties that are not readily quantifiable (Arnell 2004; Hall and Solomatine 2008; Moss

293 et al. 2010). A scenario is a story that describes a possible future state; it is not a forecast and each

294 scenario is one alternative image of how the future can unfold (IPCC 2011). For simplicity, this

295 study considers three scenarios defined by 95% confidence intervals values of the water supply

296 and demand analysis: (i)low supply (2.5th percentile) paired with high demand (97.5th percentile),

297 (ii) ensemble mean supply paired with ensemble mean demand, and (iii) high supply (97.5th

298 percentile) paired with low and minimum demand.

299 WRF Regional Climate Model

300 The Weather Research and Forecasting (WRF) model (Skamarock et al. 2005) was used to

301 dynamically downscale P and ET to 4-km horizontal resolution grid covering Utah for historical

302 years 1989-2010, as well as two future decades: 2035-2044 and 2085-2094 under the

303 Representative Concentration Pathway (RCP) 6.0 greenhouse gas emission scenario. To capture

304 lake impacts on precipitation, we incorporated a thermodynamic slab model of the Great Salt Lake.

305 Additionally, the surface physics scheme has been modified to account for high salinity of the lake

306 and the impact of urban irrigation on evaporative fluxes in Salt Lake valley (Strong et al. 2014).

307 The model utilized NOAH land surface model, which estimates evapotranspiration using canopy

308 resistances and surface exchange coefficients (Chen and Dudhia 2001).

309 This modeling framework uses a nested domain configuration with 36-km, 12-km, 4-km

310 resolutions, and has been deployed in order to downscale historical reanalysis data and to provide

15
311 future regional clime predictions. In the historical simulation performed for years 1985-2010, the

312 initial and lateral boundary conditions were derived from Climate Forecast System Reanalysis

313 (CFSR) data (Saha et al. 2010) obtained 6-hourly at 38-km horizontal resolution. This historical

314 WRF simulation validated well against observed daily precipitation from snowpack telemetry

315 (SNOTEL) stations with a small overall bias (less than 0. 5mm) and annual totals within 20% of

316 SNOTEL (Scalzitti et al. 2016a), consistent with findings for similar WRF configurations used in

317 the neighboring Colorado Headwaters (Rasmussen et al. 2011). Also, using the historical WRF

318 fields to calculate American Society of Civil Engineers (ASCE) Standard Reference

319 Evapotranspiration (ASCE-ET; Allen et al. 2005) yields results that compare well with

320 observationally-based hourly ASCE-ET over Utah (Strong et al. 2017).

321 As detailed in Scalzitti et al. (2016b), the climate forcing for the future runs performed for years

322 2085-2094 were derived from climate simulations performed using the National Center for

323 Atmospheric Research (NCAR) Community Climate System Model Version 4 (CCSM4) forced

324 by RCP 6.0 (Masui et al. 2011). CCSM4 was chosen because it best captures historical

325 relationships between Great Basin precipitation and Pacific modes of variability (Smith et al.

326 2015). The development of the boundary conditions followed the so-called pseudo-global warming

327 method (e.g., Rasmussen et al., 2011). First, the climate perturbations in the temperature and

328 moisture fields were computed as monthly differences between the future CCSM4 RCP6.0 climate

329 (for years 2085-2094), and historical CCSM simulations for years 1985-2004. Then, the historical

330 CFSR initial and boundary conditions were perturbed with these offsets, representing the expected

331 monthly changes in the temperature and moisture due to climate change. The pseudo global

332 warming technique applies perturbations to initial and boundary conditions generated from

333 historical reanalysis fields derived from meteorological surface and upper air observations,

16
334 meaning future climate is simulated with minimal alteration of the historical sub-monthly storm

335 track characteristics and frequency. This approach isolates robust climate change effects (i.e.

336 warming and moistening) from storm track-related precipitation changes, which may vary

337 markedly among climate models (Zhao et al. 2015). As described in (Scalzitti et al. 2016b),

338 historical reference climate was simulated for the two decades 1985-2004, and end of the current

339 century climate was simulated for the decade centered on 2090. Longer analysis periods would be

340 useful to provide visibility to natural variability effects but were not feasible given the

341 computational expense of WRF, so we incorporate results from statistical downscaling of the entire

342 CMIP5 ensemble corresponding to RCP 6.0 as described below.

343 The dynamically downscaled P and ET results were spatiotemporally integrated to annual volumes

344 within the boundaries of each of the eleven river basins. Average annual P and ET values for the

345 historical reference period were derived based on the hourly downscaled P and ET historical results

346 for years 1989 to 2010 (a subset of the longer 26-year period discussed in Scalzitti et al. 2016a).

347 Similarly, average annual P and ET values for 2090s were based on the climate simulations

348 corresponding to years 2085-2094. For each basin, changes in P, ET, and P-ET were calculated

349 for 2090s relative to the 1989-2010 historical reference period.

350 Statistical Downscaling

351 In addition to the dynamical downscaling with WRF, we consider the range of climate change

352 outcomes captured in the Coupled Model Intercomparison Project phase 5 (CMIP5) (Taylor 2011).

353 Specifically, we use precipitation and temperature results from CMIP5 which have been

354 statistically downscaled via Bias Correction Spatial Disaggregation (BCSD) method (Reclamation

355 2013) for RCP 6.0. The complete BCSD archive has simulations from approximately 20 global

17
356 climate models with several of the models providing multiple runs to make visible uncertainties in

357 model physics, internal variability, and initial conditions (there are 231 downscaled simulations in

358 total). Here we focus on the 36 simulations from RCP 6.0 to correspond with the WRF downscaling

359 results. Using the 1989-2010 base period of each simulation as a historical reference, we calculated

360 precipitation change as a ratio (fP) and temperature change as a difference (߂ܶሻ for each simulation,

361 and recorded the mean and 2.5th and 97.5th percentiles of these climate change factors for each

362 basin. To investigate how warmer/drier and cooler/wetter climate outcomes could alter

363 sensitivities calculated based on the WRF dynamical downscaling results, we scaled historical

364 WRF P by the fP to generate alternative WRF-based realizations of future precipitation. This

365 approach follows so called changed factor or delta methods (e.g., Zahmatkesh et al., 2015)

366 conventionally applied to historical observations; here, we apply the climate change factors to

367 historical WRF output to facilitate comparison to the main criticality analysis results, which are

368 based on dynamically downscaled historical and future WRF.

369 ET is not provided by the BCSD statistical downscaling method. To calculate ET values responsive

370 to the various ߂ܶ climate offset values applied in our sensitivity analysis, we use

ET f  ETh
371 ET ' ETh  'T (8)
T f  Th

372 where subscript h indicates data averaged over the historical WRF simulation (1989-2010) and

373 subscript f indicates data averaged over the end-of-century WRF simulation (2085-2094). While

374 simplistic, this formulation for ET’ incorporates some of the physically-based, century-scale

375 climate change signals captured by WRF which are relevant to ET and consistent with overall

376 patterns in the CMIP5 ensemble, namely the tendency for a warmer atmosphere to be moister, and

18
377 the tendency for precipitation, and hence soil moisture, to increase with warming over the study

378 region.

379 Analysis and Results

380 Population Growth and Changes in Total Water Demand

381 The forecasted population with a 95% confidence interval for the Jordan River basin is shown in

382 Fig. S1. The results show that Jordan River basin had the highest population in the state in 2010,

383 and the existing population will be nearly doubled by 2060. Southeast Colorado River basin had

384 the smallest population of 17,700 in 2010, and is expected to have doubled its population by 2090

385 (not shown). Basins with the highest rate of population growth were Utah Lake, Kanab, and Cedar.

386 At the state level, total population is expected to almost double by the end of 2050 (i.e., population

387 of 5,339,600 in 2050 compared to 2,726,700 in 2010, excluding Columbia River basin). Similarly,

388 Fig. S2 presents the daily per capita water demand, resulting total MAI water demand, annul

389 agriculture demand, and total water demand in Jordan River basin.

390 The total forecasted water demand in each of the eleven river basins is presented in Fig. 2. Basins

391 with high ‫ܦ‬ெ஺ூ (e.g., Weber, Jordan, Utah Lake, and Kanab) have a rapid increase in future total

392 water demand compared to basins with higher‫ܦ‬஺௚ (e.g., Bear River, Sevier, Uintah, and West

393 Colorado). Percent water demand increases for 2060 relative to 2010 are substantial in the four

394 basins with large MAI demand: Jordan (75%), Kanab (106%), Utah Lake (57%), and Weber

395 (52%). In contrast, basins with the smallest populations have relatively small (less than 10%)

396 increases in demand by 2060: Bear (5%), Sevier (6%), Uintah (2%), and Cedar (7%). The small

397 demand increases in these four basins reflect their predominantly agricultural use of water: Bear

19
398 (94%), Sevier (97%), Uintah (98%), and Cedar (94%), which is not expected to rapidly increase

399 in the future.

400 [Please insert Fig. 2 here]

401 Climate Change at the River Basin Level

402 Based on the ensemble of statistically downscaled coupled modeling results, we show annual mean

403 historical ܲand ET values in Figs. 3a and 3d. ܲ is highest in the northeastern portion of the state

404 where the storms are most active and mountainous terrain generates orographic precipitation. ET

405 tends to be larger where higher precipitation values support elevated soil moisture, but ET is also

406 large in the West Desert basin, which encompasses the Great Salt Lake. Changes in average P and

407 ET for 2040s and 2090s indicate modest precipitation increases for all basins (Figs. 3b and c).

408 Future decades also have elevated ET for all basins (Figs. 3e and f) but these changes are smaller

409 than precipitation increases, and the net effect of climate change is an increase in P-ET for all

410 basins during both future periods.

411 [Please insert Fig. 3 here]

412 For each basin, we consider climate outcomes corresponding to the 2.5th percentile, mean, and

413 97.5th percentile of the CMIP5 ensemble downscaled via BCSD. For 2040s, the precipitation

414 change factors were centered close to 1.0, with the 2.5th percentiles as small as 0.77 and 97.5th

415 percentiles as large as 1.25 (Fig. 4a). The 2090s precipitation change factors were larger overall,

416 centered near 1.07 with 2.5th percentiles as small as 0.81 and 97.5th percentiles as large as 1.41

417 (Fig. 4b). For 2040s, temperature delta values were centered close to 1.22, with the 2.5th percentiles

418 as small as 0.19 and 97.5th percentiles as large as 2.56 (Fig. 4c). The 2090s temperature delta values

20
419 were larger, centered near 3.30 with 2.5th percentiles as small as 1.19 and 97.5th percentiles as large

420 as 5.58 (Fig. 4d).

421 The ensemble mean precipitation change factor for each basin was similar to the results from WRF

422 downscaling of CCSM4 (Fig.4b, dashed vertical red line). The WRF dynamical downscaling of

423 CCSM yielded precipitation factors averaging 1.09 across the 11 basins for 2090. This is similar

424 to the 1.08 factor obtained by averaging across the five statistically downscaled CCSM members.

425 This agreement is consistent with the bias correction used in the statistical downscaling and the

426 small bias present in the WRF dynamical downscaling (Scalzitti et al., 2016b).

427 [Please insert Fig. 4 here]

428 Internal and External Water Criticality

429 The internal and external water criticality of the eleven river basins for 2010s, 2040s, and 2090s

430 were analyzed using equations (1) and (2). As an example, Figure S3 presents the internal and

431 external yields and total demand for Jordan River basin in the 2010s, 2040s, and 2090s decades.

432 The external yield in Jordan River basin results from water imported from Utah’s federal water

433 project, groundwater, and transboundary basin water from Utah Lake. The five basins without any

434 natural transboundary basin transfers include Utah Lake, Weber, Kanab, Sevier, and Cedar. For

435 these basins, the external supply sources are only the groundwater in the basin and water supply

436 imported from other basins by pipes and canals.

437 The minimum, mean, and maximum values of internal and external water criticality indices based

438 on the ensemble CMIP5 climate data (with 95% confidence interval) for each of the eleven river

439 basins are shown in Table S1. Note that a basin with a negative internal yield has a negative

21
440 criticality, implying that there is no excess precipitation to contribute to recharging groundwater

441 or generating runoff and evapotranspiration loses exceeds the total precipitation. Exceptionally,

442 mean internal criticality of the West Desert basin was (-20.79) in 2010s, where the basin has a very

443 high negative internal yield due to substantial evaporation from the Great Salt Lake, which is

444 approximately 10% of the total river basin area. Therefore, while calculating the external yields

445 and criticality for the West Desert basin, internal yield was not included. Persistent negative

446 external criticality in this basin would be consistent with declines in the area of the Great Salt Lake

447 since the early 1980s (Wurtsbaugh et al. 2016).

448 Five of the basins (i.e., Kanab, Southeast Colorado, Uintah, West Colorado, and Weber) in Table

449 S1 had mean internal criticality ratios greater than 2.00 in 2010s. The other six basins had mean

450 internal criticality ratios less than one for the historical period, meaning demand exceeded gross

451 effective precipitation. A very high historical water stress status (criticality < 1.25) was found for

452 Sevier and Cedar, and high water stress status (criticality < 2.5) was found for Bear, Jordan, Utah

453 Lake, and Weber.

454 Three basins will experience very high water stress in 2040s as the mean external criticality will

455 be less than 2.50. At the same time, four basins (Uintah, West Colorado, West Desert, and

456 Southeast Colorado) avoid water stress because gross effective precipitation will increase and

457 population growth will be modest. The remaining four basins (Bear, Jordan, Kanab, and Weber)

458 reach mean external criticality less than 2.50 in the 2040s because population growth outpaces

459 gains in gross effective precipitation. By 2090s, seven river basins reach very high internal water

460 stress. Three basins (Bear, Jordan and Weber) escape the very high level due to substantial external

461 basin transfer. Only Southeast Colorado has abundant supply and stays at no water stress level

462 through the 2090s.

22
463 In addition to the ensemble mean results, Table S1 also presents the uncertainty on the criticality

464 analysis based on two additional scenarios defined in Methods Section “Scenarios Analysis”: low

465 supply paired with high demand and high supply paired with low demand. It is noted that criticality

466 results are driven by the climate results (precipitation and evapotranspiration), water supply

467 variability, and water demand. Therefore, total uncertainty varies from basin to basin, and rages

468 from as low of 33% to more than 100%, driven largely by uncertainty in precipitation change

469 projected by climate models (i.e., projected precipitation as illustrated by whiskers in Fig. S3) and

470 water demand.

471 Discussion

472 Climate Change and Impacts on Water Resources

473 The climate change results presented here are similar in magnitude to the 0-10% precipitation

474 increases projected by other studies for the Rocky Mountain headwaters of Colorado River

475 (Christensen et al. 2004; McCabe Jr and Hay 1995; Nash and Gleick 1993). The results presented

476 here are based on analysis of annual totals or averages, and it is noted that month-to-month or

477 seasonal variations in climate change amplitude may be additionally important considerations for

478 water resources planning and management.

479 Population Growth and Water Demand

480 By 2060s, most of the river basins will have almost doubled their populations, with substantial

481 accompanying increases in water demand; however, the proportions of MAI versus agricultural

482 water demand vary markedly among the basins. The population forecasting was based on data

483 available from the Utah’s Governor’s Office, and the population projection results are consistent

484 with the forecast by the Utah Foundation (Utah Foundation 2014). The direct implication of the

23
485 growing population is increased MAI demand in all the basins. For example, by 2060s, Jordan

486 basin will have an approximately 71% increase in MAI demand that will result in a 75% increase

487 in total water demand; Utah Lake will have an approximately 76% increase in MAI demand and

488 resulting 57% increase in water demand. Due to the population growth, MAI water demand for

489 the state overall will be approximately doubled (increased by 116%) by 2060s.

490 Future agricultural demand was forecasted by statistical resampling of the historical records. A

491 few of the basins, such as Southeast Colorado, Uintah and West Colorado, will have predominantly

492 agricultural demand (>90%). As discussed, it is uncertain how agricultural water demand will

493 change in the future; however, earlier studies indicate that irrigation efficiency and agricultural

494 productivity improvement will result in less irrigation water requirements in the future (Gleick

495 2002; Wallace 2000), while at the same time climate change will make water demand highly

496 variable (Tukimat et al. 2017). Based on the results presented here, total water demand averaged

497 across the eleven basins will increase by approximately 23% by 2060s, with changes for individual

498 basins ranging from 19% to 28%.

499 Consideration of direct human impacts on global water supply remains a poorly articulated but

500 potentially important facet of water demand and supply analysis (Alcamo et al. 2007; Roy et al.

501 2012; Vörösmarty et al. 2000). Results here considered only the population drivers, but Utah’s

502 future water vulnerability will likely be driven not only by population growth, but also by other

503 socio-economic changes.

504 Water Supply Sources and Water Criticality at a River Basin Level

505 The gross effective precipitation (P-ET) accounts for approximately 24% of water supply source

506 at the state level. Four river basins receive transboundary supply from other states: Southeast

24
507 Colorado (about 97% of basin supply from the Colorado River basin), Uintah (about 72% from

508 Wyoming), and Bear (about 56% from Wyoming and Idaho). Groundwater meets about 4.5% of

509 total supply at the state level and ranges from 0.1 to 55.2% in various basins (e.g., Uintah 0.40%,

510 and Cedar: 55.2%). Six basins have substantial basin transfer supplies (i.e., components of the

511 transboundary basin transfers, ܳ஻ , and water imported from other basins, ܳ஽ ) that range from

512 46% to 97% (Bear, Jordan, Southeast Colorado, Uintah, West Colorado, and West Desert), and

513 the other river basins have no external basin supply. The preceding indicates that none of the basins

514 is supported by gross precipitation, and future water balance will be governed by overall external

515 water supply dependencies, which exceed 60%.

516 Results in Figure 5 showed that the internal and external criticality of several basins will be

517 decreased (become more critical) despite increases in average gross effective precipitation (P-ET).

518 This means the future water criticality or water stress situation will be driven by increasing water

519 demand with modest adjustment by average increase in annual precipitation.

520 [Please insert Fig. 5 here]

521 The uncertainty in criticality results varies from basin to basin and decade to decade (Figure 5).

522 The whiskers bars indicate how results differ in three scenarios mainly driven by climate change

523 outcomes corresponding to the 2.5th percentile, mean, and 97.5th percentile of the BCSD

524 statistically downscaled CMIP5 ensemble. It is noted that the criticality results are very sensitive

525 to both the supply sources and demand (equations 1 and 2). Any change in supply (equation 5) or

526 demand (equation 6) will of course produce a proportional change in water criticality.

527 Uncertainty stemming from the climate model outcomes is large, and accounts for most (about

528 90%) of the uncertainty envelope in the criticality results. The WRF 2090 mean precipitation

25
529 results are within the 95% confidence bounds of the statistically downscaled CMIP5 simulations

530 as expected because WRF was driven by boundary conditions derived from one of the ensemble’s

531 global climate models. The value of including the WRF dynamical downscaling in this study is

532 twofold. First, WRF is the basis for calculating all ET fields, even for the statistically downscaled

533 CMIP5 ensemble results via equation (8). Second, dynamical downscaling is an important

534 complement to statistical downscaling because the latter can fail if historical statistical

535 relationships are not stationary into the future. In the data analyzed here, the assumptions

536 underlying the statistical downscaling appear reasonable because the WRF dynamical downscaling

537 results are near the center of the precipitation change distribution from statistical downscaling, and

538 they in particular are very close to the CCSM4 members of the statistical downscaling.

539 As discussed above, a basin with criticality ratio of one indicates that the basin demand has been

540 exactly met by the available supply sources. As a result, a basin that has criticality ratio less than

541 one will have serious water stress conditions as defined in previous studies (Alcamo et al. 2000;

542 Raskin et al. 1997). Also, it is noted that future agricultural demand was bootstrapped based on

543 past records, so the criticality issues identified here may be changed if there are substantial

544 reductions in agricultural demand by, for example, reductions in agriculture area or increases in

545 irrigation efficiency. The overall average annual internal criticality ratio of the state is 1.90 (-0.46

546 to 5.05) in 2040s, and 1.12 (-0.78 to 3.83) in 2090s. Similarly, annual external criticality ratios

547 average 5.50 (2.81 to 8.97) in the 2040s, and 4.02 (1.89 to 6.96) in the 2090s.

548 Results collectively indicate that Utah is water stressed inside its state boundary, but external

549 supply meets the water demand. There is also marked variability on water availability and water

550 demand between the basins. In principle, moving water from a water surplus basin to a water scarce

551 basin could provide a solution to criticality issues, but there may be substantial engineering

26
552 challenges and legal challenges related to water rights and water policy. Utah’s water rights,

553 similar to other western US States, govern water transfers.

554 Conclusions

555 This study concludes that the criticality of Utah’s water resources will be driven mainly by

556 population growth eclipsing modest increases in gross effective precipitation from climate change.

557 Due to changes in water demand, more river basins will be water stressed in the future (2040s and

558 2090s) compared to 2010s, although there is a wide range of uncertainty due to climate change.

559 The 2040s results show three basins at “very high water stress”, four basins with “high water

560 stress”, three basins with “moderate water stress”, and only one basin with “water surplus”. In

561 2090s, four basins reach “very high water stress”, and three basins reach “high water stress” status.

562 At the state level, criticality analysis indicates a supply that is just sufficient to balance the recharge

563 rate with nontrivial external dependence (meaning demand is not sustained by the internal supply

564 sources alone). Basin transfers may be one of the options to manage demand, but transferred water

565 supply sources will need to meet high drinking water standards, and the state level MAI water

566 demand will be approximately doubled by 2060s. The results indicate the need for serious water

567 resources planning in the state of Utah, as most basins depend on external supply sources and are

568 not self-sufficient.

569 The results of this study provide quantitative and easily interpreted guidance for strategic planning

570 of water resources. The criticality results will inform water managers and planners on the current

571 and future water stress level of all the river basins in the state of Utah. The results will also inform

572 about river basins that are self-sustaining to meet the water demand from the water resources

573 available within their boundary, and river basins which have high levels of external dependencies

27
574 in meeting water demands. The internal and external critically ratios explicitly indicate internal

575 and external dependencies of a basin, which were not presented in earlier studies. Moreover, the

576 results on the worst and best case scenarios will help to understand the implication of extreme

577 scenarios of water supply and demand in a river basin. The results will, therefore, assist engineers

578 and planners not only in understanding the effects of climate change and population growth on

579 water resources and demand but will also be useful for planning water storage systems for specific

580 basins, in identifying candidate basins for water diversions to meet growing stresses elsewhere,

581 and in developing climate adaptation plans.

582 The study assessed annual water criticality including minimum and maximum values for eleven

583 river basins over the state of Utah for 2010s, 2040s, and 2090s using publicly available data

584 (mainly from DWRe), downscaled climate data from a regional climate model, and publicly

585 available statistically downscaled climate model output. Due to the geographical extent of the

586 study areas, the framework neglects detailed internal heterogeneity and potential limitations of

587 water routing, although population and climate are spatially explicit prior to spatial integration.

588 Other limitations to be addressed in future research include more detailed analysis of incoming

589 transboundary river flow, groundwater availability, future agricultural demand, and other water

590 management aspects such as conjunctive groundwater management and reuse/recycle

591 opportunities. External criticality of certain basins depends crucially on the external water supply

592 sources and demand. Accurate analysis of incoming transboundary river flow will need detailed

593 hydrologic modeling responsive to upstream imbalances. Groundwater availability will depend on

594 the recharging rate of groundwater, and that will be governed by several factors including changes

595 in land use, climate change, groundwater withdrawal practices, and changes in surface

596 characteristics that are not considered in this study. The future agricultural demand will depend on

28
597 several factors including types of crop, agricultural policy, irrigation technology, irrigation

598 practices, and future agricultural market. Those factors carry substantial uncertainty and warrant

599 further detailed analysis. Finally, the formulations used here assume that time rates of change of

600 storage are negligible, thus the water balance effects of water reservoirs as well as infrastructure

601 expansion in the future are also not considered.

602 Acknowledgements

603 This work was supported by the National Science Foundation cooperative agreement EPSCoR

604 IIA-1208732. Any opinions, findings, and conclusions or recommendations expressed in this

605 material are those of the authors and do not necessarily reflect the views of the National Science

606 Foundation. We would like to acknowledge high-performance computing support from

607 Yellowstone (award UUSL006) provided by NCAR's Computational and Information Systems

608 Laboratory, sponsored by the National Science Foundation, as well as from the University of Utah

609 Center for High Performance Computing (CHPC).

610 [Insert supplementary section here]

611 References
612 Abdallah, A. M. & Rosenberg, D. E. (2012). “Heterogeneous residential water and energy linkages

613 and implications for conservation and management”. Journal of Water Resources Planning

614 and Management, 140, 288-297.

615 Alcamo, J., Döll, P., Kaspar, F., and Siebert, S. (1997). "Global change and global scenarios of

616 water use and availability: an application of WaterGAP 1.0." Center for Environmental

617 Systems Research (CESR), University of Kassel, Germany, 1720.

29
618 Alcamo, J., Henrichs, T., and Rosch, T. (2000). "World water in 2025." World Water Series

619 Report, 2.

620 Alcamo, J., Flörke, M., and Märker, M. (2007). "Future long-term changes in global water

621 resources driven by socio-economic and climatic changes." Hydrological Sciences

622 Journal, 52(2), 247-275.

623 Allen, R., Walter, I., Elliott, R., Howell, T., Itenfisu, D., Jensen, M., and Snyder, R. (2005). "The

624 ASCE standardized reference evapotranspiration equation. 2005." Technical Committee

625 report to the Environmental and Water Resources Institute of the American Society of Civil

626 Engineers from the Task Committee on Standardization of Reference Evapotranspiration,

627 ASCE.

628 Arnell (2004). “Climate change and global water resources: SRES emissions and socio-economic

629 scenarios.” Global Environmental Change, 14, 31-52.

630 Barnett, T. P., Adam, J. C., and Lettenmaier, D. P. (2005). "Potential impacts of a warming climate

631 on water availability in snow-dominated regions." Nature, 438(7066), 303-309.

632 Beniston, M. (2003). "Climatic change in mountain regions: a review of possible impacts." Climate

633 variability and change in high elevation regions: Past, present & future, Springer, 5-31.

634 Beven, K. (2007). "Towards integrated environmental models of everywhere: uncertainty, data

635 and modelling as a learning process." Hydrology and Earth System Sciences Discussions,

636 11(1), 460-467.

637 Bhatia, N. (2010). "Survey of nearest neighbor techniques." International Journal of Computer

638 Science and Information Security,, 8(2), 302-305.

639 Brown, A., and Matlock, M. D. (2011). "A review of water scarcity indices and methodologies."

640 White Paper, 106, 19.

30
641 Buizza, Roberto. (1997). “Potential forecast skill of ensemble prediction and spread and skill

642 distributions of the ecmwf ensemble prediction system.” Monthly Weather Review, 125(1),

643 99-119.

644 Chen, F., and Dudhia, J. (2001). "Coupling an advanced land surface–hydrology model with the

645 Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity."

646 JoAMS Monthly Weather Review, 129(4), 569-585.

647 Christensen, N. S., Wood, A. W., Voisin, N., Lettenmaier, D. P., and Palmer, R. N. (2004). "The

648 effects of climate change on the hydrology and water resources of the Colorado River

649 basin." Climatic Change, 62(1-3), 337-363.

650 Deoreo, W. B., Mayer, P. W., Dziegielewski, B. & Kiefer, J. 2016. “Residential end uses of

651 water, version 2”. Water Research Foundation.

652 http://www.waterrf.org/Pages/Projects.aspx?PID=4309 [December 22, 2017].

653 Division of Water Resources (2010). "Municipal and Industrial Water Use in Utah : "why do we

654 use so much water when we live in a desert?" ", Todd Adams, Eric Klotz, Gregory

655 Williams, Eric Jones, and L. Summers, eds.Salt Lake City, Utah, 36.

656 Efron, B., and Tibshirani, R. J. (1994). An introduction to the bootstrap, CRC press.

657 Falkenmark, M., Lundqvist, J., and Widstrand, C. (1989). "Macro-scale water scarcity requires

658 micro-scale approaches." Proc., Natural resources forum, Wiley Online Library, 258-267.

659 Gleick, P. H. (2000). "A look at twenty-first century water resources development." Water

660 International, 25(1), 127-138.

661 Gleick, P. H. (2002). "Water management: Soft water paths." Nature, 418(6896), 373-373.

31
662 Goharian, E., Burian, S. J., Lillywhite, J., and Hile, R. (2016). "Vulnerability assessment to support

663 integrated water resources management of metropolitan water supply systems." Journal of

664 Water Resources Planning and Management, 04016080.

665 Hall, J. and D. Solomatine (2008). "A framework for uncertainty analysis in flood risk

666 management decisions." Intl. J. River Basin Management, 6(2): 85–98.

667 Härdle, W., Horowitz, J., and Kreiss, J. P. (2003). "Bootstrap methods for time series."

668 International Statistical Review, 71(2), 435-459.

669 IPCC (International Panel on Climate Change) (2011). “IPCC Data Distribution Center -

670 Definition of Terms Used Within the DDC Pages.” http://www.ipcc-

671 data.org/ddc_definitions.html [Acess:Sept12,017].

672 Johnson, N., Revenga, C., and Echeverria, J. (2001). "Managing water for people and nature."

673 Science, 292(5519), 1071-1072.

674 Kabir, G., et al. (2015). "Integrating Bayesian Linear Regression with Ordered Weighted

675 Averaging: Uncertainty Analysis for Predicting Water Main Failures." ASCE-ASME

676 Journal of Risk and Uncertainty in Engineering Systems, Part A: Civil Engineering.

677 Kataria, A., and Singh, M. (2013). "A Review of Data Classification Using K-Nearest Neighbour

678 Algorithm." International Journal of Emerging Technology and Advanced Engineering,

679 3(6), 354-360.

680 Khatri, K. B. (2013). Risk and Uncertainty Analysis for Sustainable Urban Water Systems, CRC

681 Press

682 Kumar, S., Tiwari, M. K., Chatterjee, C., and Mishra, A. (2015). "Reservoir inflow forecasting

683 using ensemble models based on neural networks, wavelet analysis and bootstrap method."

684 Water Resources Management, 29(13), 4863-4883.

32
685 Liu, J., Liu, Q., and Yang, H. (2016). "Assessing water scarcity by simultaneously considering

686 environmental flow requirements, water quantity, and water quality." Ecological

687 Indicators, 60, 434-441.

688 Masui, T., Matsumoto, K., Hijioka, Y., Kinoshita, T., Nozawa, T., Ishiwatari, S., Kato, E., Shukla,

689 P., Yamagata, Y., and Kainuma, M. (2011). "An emission pathway for stabilization at 6

690 Wm− 2 radiative forcing." Climatic Change, 109(1-2), 59-76.

691 Maupin, M. A., Kenny, J. F., Hutson, S. S., Lovelace, J. K., Barber, N. L., and Linsey, K. S. (2014).

692 "Estimated use of water in the United States in 2010." US Geological Survey.

693 McCabe Jr, G. J., and Hay, L. E. (1995). "Hydrological effects of hypothetical climate change in

694 the East River basin, Colorado, USA." Hydrological Sciences Journal, 40(3), 303-318.

695 Moss, Edmonds, Hibbard, Manning, Rose, Van Vuuren, Carter, Emori, Kainuma and Kram

696 (2010). “The next generation of scenarios for climate change research and assessment.”

697 Nature, 463(7282), 747-756.

698 Nash, L. L., and Gleick, P. H. (1993). "Colorado River basin and climatic change. The sensitivity

699 of streamflow and water supply to variations in temperature and precipitation." Pacific Inst.

700 for Studies in Development, Environment and Security, Oakland, CA ,USA.

701 New, Mark, & Hulme, Mike. (2000). “Representing uncertainty in climate change scenarios: A

702 monte-carlo approach”. Integrated assessment, 1(3), 203-213.

703 O'Neill, Brian C. (2004). “Conditional probabilistic population projections: An application to

704 climate change.” International statistical review, 72(2), 167-184.

705

33
706 Palmer, M. A., Reidy Liermann, C. A., Nilsson, C., Flörke, M., Alcamo, J., Lake, P. S., and Bond,

707 N. (2008). "Climate change and the world's river basins: anticipating management

708 options." Frontiers in Ecology and the Environment, 6(2), 81-89.

709 Parker, J. M. and R. L. Wilby (2013). "Quantifying household water demand: a review of theory

710 and practice in the UK." Water Resources Management, 27(4): 981-1011.

711 Pedro-Monzonís, M., Solera, A., Ferrer, J., Estrela, T., and Paredes-Arquiola, J. (2015). "A review

712 of water scarcity and drought indexes in water resources planning and management."

713 Journal of Hydrology, 527, 482-493.

714 Rajagopalan, B., and Lall, U. (1999). "A k-nearest-neighbor simulator for daily precipitation and

715 other weather variables." Water Resources Research, 35(10), 3089-3101.

716 Raskin, P., Gleick, P., Kirshen, P., Pontius, G., and Strzepek, K. (1997). Water futures: Assessment

717 of long-range patterns and problems. Comprehensive assessment of the freshwater

718 resources of the world, SEI.

719 Rasmussen, R., Liu, C., Ikeda, K., Gochis, D., Yates, D., Chen, F., Tewari, M., Barlage, M.,

720 Dudhia, J., and Yu, W. (2011). "High-resolution coupled climate runoff simulations of

721 seasonal snowfall over Colorado: a process study of current and warmer climate." Journal

722 of Climate, 24(12), 3015-3048.

723 Rasmussen, R., Ikeda, K., Liu, C., Gochis, D., Clark, M., Dai, A., Gutmann, E., Dudhia, J., Chen,

724 F., and Barlage, M. (2014). "Climate change impacts on the water balance of the Colorado

725 headwaters: high-resolution regional climate model simulations." Journal of

726 Hydrometeorology, 15(3), 1091-1116.

727 Reclamation (2013). Downscaled CMIP3 and CMIP5 Climate Projections: Release of

728 Downscaled CMIP5 Climate Projections, Comparison with Preceding Information, and

34
729 Summary of User Needs, U.S. Department of the Interior, Bureau of Reclamation,

730 Technical Service Center, Denver, Colorado, 116 p.

731 Rijsberman, F. R. (2006). "Water scarcity: fact or fiction?" Agricultural water management, 80(1),

732 5-22.

733 Roy, S. B., Chen, L., Girvetz, E. H., Maurer, E. P., Mills, W. B., and Grieb, T. M. (2012).

734 "Projecting water withdrawal and supply for future decades in the US under climate change

735 scenarios." Environmental science & technology, 46(5), 2545-2556.

736 Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., Tripp, P., Kistler, R., Woollen, J.,

737 Behringer, D., Liu, H., Stokes, D., Grumbine, R., Gayno, G., Wang, J., Hou, Y.-T., Chuang,

738 H.-Y., Juang, H.-M. H., Sela, J., Iredell, M., Treadon, R., Kleist, D., Van Delst, P., Keyser,

739 D., Derber, J., Ek, M., Meng, J., Wei, H., Yang, R., Lord, S., Van Den Dool, H., Kumar,

740 A., Wang, W., Long, C., Chelliah, M., Xue, Y., Huang, B., Schemm, J.-K., Ebisuzaki, W.,

741 Lin, R., Xie, P., Chen, M., Zhou, S., Higgins, W., Zou, C.-Z., Liu, Q., Chen, Y., Han, Y.,

742 Cucurull, L., Reynolds, R. W., Rutledge, G., and Goldberg, M. (2010). "The NCEP Climate

743 Forecast System Reanalysis." Bulletin of the American Meteorological Society, 91(8),

744 1015-1057.

745 Sandoval-Solis, S., McKinney, D., and Loucks, D. (2010). "Sustainability index for water

746 resources planning and management." Journal of Water Resources Planning and

747 Management, 137(5), 381-390.

748 Scalzitti, J., Strong, C., and Kochanski, A. (2016a). "Climate change impact on the roles of

749 temperature and precipitation in western U.S. snowpack variability." Geophys Res Lett.

35
750 Scalzitti, J., Strong, C., and Kochanski, A. K. (2016b). "A 26 year high-resolution dynamical

751 downscaling over the Wasatch Mountains: Synoptic effects on winter precipitation

752 performance." Journal of Geophysical Research(121), 3224–3240.

753 Scherer, L., Venkatesh, A., Karuppiah, R., and Pfister, S. (2015). "Large-scale hydrological

754 modeling for calculating water stress indices: implications of improved spatiotemporal

755 resolution, surface-groundwater differentiation, and uncertainty characterization."

756 Environmental science & technology, 49(8), 4971-4979.

757 Seckler, D. W. (1998). World water demand and supply, 1990 to 2025: Scenarios and issues,

758 Iwmi.

759 Shiklomanov, I. "The world’s water resources." Proc., Proceedings of the international

760 symposium to commemorate, 93-126.

761 Skamarock, W. C., Klemp, J. B., Dudhia, J., Gill, D. O., Barker, D. M., Wang, W., and Powers, J.

762 G. (2005). "A Description of the Advanced Research WRF Version 2. NCAR Tech. Note

763 TN-468 STR, 88pp." Available from NCAR; P.O. BOX 3000; Boulder, CO, 7-25.

764 Smith, K., Strong, C., and Wang, S.-Y. (2015). "Connectivity between historical Great Basin

765 precipitation and Pacific Ocean variability: A CMIP5 model evaluation." Journal of

766 Climate, 28(15), 6096-6112

767 Strong, C., Kochanski, A. K., and Crosman, E. T. (2014). "A slab model of the Great Salt Lake for

768 regional climate simulation." Journal of Advances in Modeling Earth Systems, 6(3), 602-

769 615.

770 Strong, C., Khatri, K. B., Kochanski, A. K., Lewis, C. S., and Allen, L. N. (2017). "Reference

771 evapotranspiration from coarse-scale and dynamically downscaled data in complex terrain:

772 Sensitivity to interpolation and resolution." Journal of Hydrology, 548, 406-418.

36
773 Sullivan, C. A., Meigh, J. R., and Giacomello, A. M. (2003). "The water poverty index:

774 development and application at the community scale." Natural resources forum, 27(3),

775 189-199.

776 Taylor, K. E., Stouffer, R. J., and Meehl, G. A. (2012). "An overview of CMIP5 and the experiment

777 design." Bulletin of the American Meteorological Society, 93(4), 485-498.

778 Tekleab, S., Uhlenbrook, S., Mohamed, Y., Savenije, H., Temesgen, M., and Wenninger, J. (2011).

779 "Water balance modeling of Upper Blue Nile catchments using a top-down approach."

780 Hydrology and Earth System Sciences, 15,(7).

781 Tibshirani, R. (1985). "How many bootstraps?" Department of Statistivs, Standford University,

782 DTIC Document. Technical Report No. 362.

783 Tiwari, M. K., and Adamowski, J. (2013). "Urban water demand forecasting and uncertainty

784 assessment using ensemble wavelet-bootstrap-neural network models." Water Resources

785 Research, 49(10), 6486-6507.

786 Tol, R. S. (2015). "Bootstraps for meta-analysis with an application to the impact of climate

787 change." Computational Economics, 46(2), 287-303.

788 Tukimat, N., Harun, S., and Shahid, S. (2017). "Modeling Irrigation Water Demand in a Tropical

789 Paddy Cultivated Area in the Context of Climate Change." Journal of Water Resources

790 Planning and Management, 143(7), 05017003.

791 U.S. Census Bureau (2011). "Population estimates: The state of Utah and counties 1940-2011."

792 U.S. Census Bureau Population Estimates.

793 United States Department of Agriculture, U. (2014). "Census of Agrciculture, farm and Ranch

794 Irrgation Survery (2013)." special studies Part1, V. a. J. T. R. Tom ed.

37
795 Utah Division of Water Resources (2016). "Water for Utah, A Review of Duties and Funding

796 Programs of the Division and Board of Water Resources."

797 Utah Economic Council (2016). "2016 Economic Report to the Governor, Demographic and

798 Economic Analysis." A collaborative endeavor of the David Eccles School of Business and

799 Governor’s Office of Management and Budget.

800 Utah Foundation (2014). "Flowing Toward 2050, Utah's Water Outlook." Report Number 723, D.

801 T. Harbeke, C. B. Garbett, D. Matsumori, and S. J. H. Kroes, eds.

802 Utah Foundation (2014). "A Snapshot of 2050, An Analysis of Projected Population Change in

803 Utah." Daniel T. Harbeke, Bryson Garbett, Douglas Matsumori, and S. J. H. Kroes, eds.

804 Veldkamp, T. I., Wada, Y., de Moel, H., Kummu, M., Eisner, S., Aerts, J. C., and Ward, P. J.

805 (2015). "Changing mechanism of global water scarcity events: Impacts of socioeconomic

806 changes and inter-annual hydro-climatic variability." Global Environmental Change, 32,

807 18-29.

808 Vörösmarty, C. J., Douglas, E. M., Green, P. A., and Revenga, C. (2005). "Geospatial indicators

809 of emerging water stress: an application to Africa." AMBIO: A Journal of the Human

810 Environment, 34(3), 230-236.

811 Vörösmarty, C. J., Green, P., Salisbury, J., and Lammers, R. B. (2000). "Global water resources:

812 vulnerability from climate change and population growth." science, 289(5477), 284-288.

813 Wallace, J. (2000). "Increasing agricultural water use efficiency to meet future food production."

814 Agriculture, Ecosystems & Environment, 82(1), 105-119.

815 Wurtsbaugh, W., Miller, C., Null, S., Wilcock, P., Hahnenberger, M., and Howe, F. (2016).

816 "Impacts of Water Development on Great Salt Lake and the Wasatch Front."

817 http://digitalcommons.usu.edu/wats_facpub/875/ [May 12, 2016]

38
818 Wu, X., Kumar, V., Quinlan, J. R., Ghosh, J., Yang, Q., Motoda, H., McLachlan, G. J., Ng, A.,

819 Liu, B., and Philip, S. Y. (2008). "Top 10 algorithms in data mining." Knowledge and

820 information systems, 14(1), 1-37.

821 Zahmatkesh, Z., Karamouz, M., Goharian, E., and Burian, S. J. (2014). "Analysis of the effects of

822 climate change on urban storm water runoff using statistically downscaled precipitation

823 data and a change factor approach." Journal of Hydrologic Engineering, 20(7), 05014022

824 Zhang, L., Potter, N., Hickel, K., Zhang, Y., and Shao, Q. (2008). "Water balance modeling over

825 variable time scales based on the Budyko framework–Model development and testing."

826 Journal of Hydrology, 360(1), 117-131.

827 Zhao, L., Xu, J., Powell, A. M., and Jiang, Z. (2015). "Uncertainties of the global-to-regional

828 temperature and precipitation simulations in CMIP5 models for past and future 100 years."

829 Theoretical and Applied Climatology, 122(1-2), 259-270.

39
7DEOH &OLFNKHUHWRGRZQORDG7DEOH7DEOHVDOOGRF[

Table 1: Major Water Stress Indices used in Water Scarcity Analysis

Water Stress Indices Water Stress Level Classification


Falkenmark et al. (1989): Falkenmark indicator or A region is considered water stressed, if renewable
“water stress index” represents water scarcity as a water supplies fall below 1,700 m3, water scarce
relationship between water availability and human when supply falls below 1,000 m3, and absolute
population (i.e., water availability per capita per scarce when supply is below 500 m3.
year).
Shiklomanov (1991): Water stress index based on Water stress level is classified into four types based
the ratio of annual water withdrawals to the annual on the water index values: (i) no water stress: 0 to
water availability. 0.1, (ii) low water stress: 0.1 to 0.2, (iii) moderate
water stress: 0.2 to 0.4 and (iv) high water stress:
>0.4.
Raskin et al. (1997): Water resources vulnerability A basin is considered water scarce, if annual
index as the total annual withdrawals as a percent withdrawals are between 20-40% of annual supply,
of available water resources. and severely water scarce if this value exceeds 40%.
Alcamo et al. (1997): Water “criticality ratio” - the Classifies water stress into four levels: (i) “very high
ratio of water withdrawals for human use to total water stress” when the water criticality ratio is less
renewable water resources. than 0.80, (ii) “high water stress” when criticality
ranges between 0.8 to 0.4, (iii) “moderate water
stress” when criticality ranges between 0.4 to 0.1,
and (iii) “no water stress” when greater than 0.1.
Seckler at al. (1998): Physically water scarce and Physically water scarce means a region will not be
economically water scarce indicators comparing able to meet the estimated water demands during a
the share of the renewable water resources given time in the future, even after accounting for
available for human needs, and water demand future adaptive capacity. Economically water scarce
analysis based on consumptive use and return regions have sufficient renewable resources, but
flows. would have to make significant investment in water
infrastructure to make these resources available to
people.
Vorosmarty et al. (2000): Water scarcity index Classified a water stress level into four classes: (i)
comparing water available and water withdrawals. low: <0.1, (ii) moderate: 0.1 to 0.2, (iii) medium-
high: 0.2 to0.4, and (iv) high: >0.4.
Sullivan et al. (2003): The water poverty index Each of the components is first standardized in the
generated by aggregating the multiple indicators of range of 0 to 100. The resulting water poverty index
the five key components of water resources and value will vary between 0 and 100. The highest value,
management that include water resources, access, 100, is considered the best situation, while 0 is the
capacity, water use, and environment. worst.
Scherer et al. (2015): Water stress index was Water stress index was calculated for four aspects of
calculated after a detailed water availability water resources: (i) annual total water resources, (ii)
analysis applying a hydrological model, and monthly water resources, (iii) monthly surface water
consideration of seasonal fluctuations and spatial resources, and (iv) monthly groundwater with
heterogeneity within a watershed. uncertainty.

1
Table S1: Criticality of River Basins for Decades 2010s, 2040s, and 2090s (For the future periods, the bracketed values correspond to the three scenarios
defined in Methods Section “Scenarios Analysis”: low supply paired with high demand, mean supply paired with mean demand, and high supply paired
with low demand.

Basin Criticality with Uncertainty (within a 95% Confidence Interval)


Water Stress Level Years: 1989 to 2010 Years: 2035 to 2044 Years: 2085 to 2094
Internal External Internal External Internal External
1. Very high-water Bear: 0.70 Sevier: 1.10 Bear: [0.06, 0.69, 1.41] Sevier: [0.12, 0.86, 2.10] Bear: [0.05, 0.78, 1.58] Kanab: [0.80, 1.05, 2.65]
stress Jordan: 0.61 Cedar: 1.25 Jordan: [0.11, 0.45, 0.85] Utah L.: [0.39, 1.06, 1.87] Jordan: [0.07, 0.35, 0.68] Sevier: [0.17, 0.96, 2.65]
(criticality≤1.25) Sevier: 0.84 Utah L.:[-0.07, 0.56, 1.33] Cedar: [0.60, 1.16, 2.80] Utah L.: [-0.08, 0.45, 1.17] Cedar: [0.35, 1.25, 2.60]
Utah L.: 0.80 Sevier: [-0.07, 0.66, 1.89] Kanab: [0.73, 0.96, 2.55] Utah L.: [0.23, 0.79, 1.53]
Cedar: 0.46 Cedar: [-0.07, 0.44, 2.04] Sevier: [-0.01, 0.76, 2.44]
West D.: West D.: Cedar: [-0.23, 0.61, 1.92]
(-20.79) [-24.54, -18.96, -13.01] West D.: [-18.62, -13.96, -6.82]
2. High water stress Uintah: 2.47 Bear: 1.77 Kanab: [0.48, 1.93, 3.18] Bear: [0.96, 1.77, 2.61] Weber: [0.78 1.59, 2.59] Bear: [0.88, 1.78, 2.76]
(1.25<criticality≤2.50) Weber: 2.12 Jordan: 2.95 Uintah: [0.44, 2.38, 5.32] Jordan: [1.48, 1.98, 2.61] Jordan: [0.93, 1.33, 1.81]
Utah L.: 1.47 Weber: [0.78, 1.59, 2.59] Kanab: [-0.61, 2.12, 5.08] Weber: [0.65, 1.53, 2.33]
Weber: 1.59 Weber: [0.99, 1.79, 2.78]
3. Moderate water Kanab: 2.55 Kanab: 2.93 West Col.: Uintah: [5.90,8.79, 12.78] Uintah: [-0.06, 2.59, 5.63] Uintah: [5.94, 9.01, 13.96]
stress S.E. Col.: 4.33 Uintah: 8.73 [-0.26, 2.46, 3.91] West D.: [6.28,7.72, 9.31] West Col.: West D.: [5.50, 6.29, 7.06]
(2.50<criticality≤10) West Col.: 2.74 West Col.: 3.94 S.E. Col.: West Col.: [-0.10, 3.22, 6.18] West Col.:
West D.: 7.24 [-24.54, -18.96, -13.00] [0.65, 3.57, 5.25] S.E. Col.: [0.80, 4.31, 7.50]
[-13.74, 5.68, 33.53]
4. No water stress S.E. Col.: S.E. Col.: S.E. Col.:
None 141.48 None None
(criticality>10) [127.19, 167.48, 219.48] [114.86, 158.61, 213.34]

2
)LJ &OLFNKHUHWRGRZQORDG)LJXUH)LJSGI



)LJ &OLFNKHUHWRGRZQORDG)LJXUH)LJSGI

WKWKSHUFHQWLOHV

WKWKSHUFHQWLOHV

 PHDQ
%HDU5LYHU

   


 

  -RUGDQ5LYHU


:HEHU5LYHU
   
   



 

8WDK/DNH 6HYLHU5LYHU

       




 

&HGDU%HDYHU 8LQWDK
       




 
:HVW&RORUDGR5LYHU  6RXWKHDVW&RORUDGR5LYHU
       





 .DQDE9LUJLQ5LYHU :HVW'HVHUW
       
<HDU <HDU




)LJ &OLFNKHUHWRGRZQORDG)LJXUH)LJSGI


)LJ &OLFNKHUHWRGRZQORDG)LJXUH)LJSGI


)LJ &OLFNKHUHWRGRZQORDG)LJXUH)LJSGI


)LJ6 &OLFNKHUHWRGRZQORDG)LJXUH)LJ6SGI



)LJ6 &OLFNKHUHWRGRZQORDG)LJXUH)LJ6SGI



)LJ6 &OLFNKHUHWRGRZQORDG)LJXUH)LJ6SGI


)LJXUH&DSWLRQ/LVW

Figure Captions

Fig. 1: Study area: major river basins, rivers, and lakes in the state of Utah.

Fig. 2: Total water demand for each river basin (bold curve) with the shading indicating of 95%

and 75% confidence intervals.

Fig. 3: (a) Mean annual precipitation areally averaged over each basin for the historical period

(1989 to 2010), (b) for a mid-century period (2035 to 2044) minus historical, and (c) for a late-

century period (2085 to 2094) minus historical. (d-f) Same as (a-c) but for annual mean

evapotranspiration. Future results are based on 36 statistically downscaled members of the CMIP5

ensemble for RCP 6.0.

Fig. 4. Histograms of climate change factors relative to 1989-2010 based on 36 statistically

downscaled CMP5 simulations corresponding to RCP 6.0: change in average precipitation for (a)

2035-2044 and (b) 2085-2094 given as ratios; change in average temperature for (c) 2035-2044

and (d) 2085-2094 given as differences. Each curve corresponds to one of the eleven basins as

indicated in the legend, and vertical lines indicate the 2.5th percentile, mean, and 97.5th percentile

averaged across the eleven basins.

Fig. 5. Internal and external water criticality indices for each basin over the state of Utah for 2010s,

2040s, and 2090s. For each variable in each panel, the bar indicates mean criticality across all

years in the period and red line indicates that supply just balances demand. Whiskers on future

decade bars indicate how results differ when driven by climate change outcomes corresponding to

the 2.5th and 97.5th percentiles of the statistically downscaled CMIP5 ensemble for RCP 6.0.

Crossed ellipse indicates mean criticality in 2090s based on the dynamically downscaled climate

data.

1
Fig. S1: Example of population forecast (with a 95% confidence interval) for Jordan River Basin.

The top inset panels show population distribution functions for decades starting from 2040s to

2090s generated by k-NN algorithm and bootstrapping with sample size 10,000

Fig. S2: Future water demand and basin transfer using Jordan River basin as example. (a) Daily

average per capita MAI demand generated by bootstrapping with sample size 10,000 and

consideration of the proposed water conservation plan for Utah (conserve 25% by 2025). (b) Total

MAI demand forecasted by simulating per capita water demand and total basin population. (c)

Annual agricultural demand generated by bootstrapping (10,000 samples) historical agricultural

water uses from years 1989 to 2010. (d) Total future water demand forecasted considering MAI

and agricultural demand.

Fig. S3: Jordan River yields for three decades (2010s is 1989-2010, 2040s is 2035-2044, and 2090s

is 2085-2094). In each panel, P is precipitation, ET is evapotranspiration, GW is Groundwater,

Div is water imported by pipes and canal diversions, Basin is transboundary basin transfer, D is

total water demand, Int.Y is internal yield, and Ext.Y is external yield. The whiskers indicate the

range of uncertainty for each of the variables. (a) Basin level yields and demand for 2010s. (b)

Basin level yields and demand for 2040s. P & ET were derived from CMIP5 and GW, D, & Basin

by bootstrapping. (c) Basin level yields and demand for 2090s as in (b).

You might also like