Accepted Manuscript

Research papers

Identifying Water Price and Population Criteria for Meeting Future Urban Water
Demand Targets

Negin Ashoori, David A. Dzombak, Mitchell J. Small

PII: S0022-1694(17)30723-0
DOI: https://doi.org/10.1016/j.jhydrol.2017.10.047
Reference: HYDROL 22327

To appear in: Journal of Hydrology

Received Date: 30 August 2017

Revised Date: 20 October 2017
Accepted Date: 22 October 2017

Please cite this article as: Ashoori, N., Dzombak, D.A., Small, M.J., Identifying Water Price and Population Criteria
for Meeting Future Urban Water Demand Targets, Journal of Hydrology (2017), doi: https://doi.org/10.1016/
j.jhydrol.2017.10.047

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Identifying Water Price and Population Criteria for Meeting Future

Urban Water Demand Targets

Negin Ashoori1, David A. Dzombak2, Mitchell J. Small3

1
Corresponding author: Negin Ashoori, Postdoctoral Scholar, Civil and Environmental
Engineering, Stanford University
473 Via Ortega, Stanford, CA 94305, USA
Email: nashoori@stanford.edu
2
David A. Dzombak, Hamerschlag University Professor and Department Head, Civil and
Environmental Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA
3
Mitchell J. Small, H. John Heinz Professor, Civil and Environmental Engineering, Engineering
Public Policy, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

Abstract

Predictive models for urban water demand can help identify the set of factors that must be

satisfied in order to meet future targets for water demand. Some of the explanatory variables

used in such models, such as service area population and changing temperature and rainfall rates,

are outside the immediate control of water planners and managers. Others, such as water pricing

and the intensity of voluntary water conservation efforts, are subject to decisions and programs

implemented by the water utility. In order to understand this relationship, a multiple regression

model fit to 44 years of monthly demand data (1970-2014) for Los Angeles, California was

applied to predict possible future demand through 2050 under alternative scenarios for the

explanatory variables: population, price, voluntary conservation efforts, and temperature and

precipitation outcomes predicted by four global climate models with two CO2 emission

scenarios. Future residential water demand in Los Angeles is projected to be largely driven by

price and population rather than climate change and conservation. A median projection for the

year 2050 indicates that residential water demand in Los Angeles will increase by approximately

1
36 percent, to a level of 620 million m3 per year. The Monte Carlo simulations of the fitted

model for water demand were then used to find the set of conditions in the future for which water

demand is predicted to be above or below the Los Angeles Department of Water and Power 2035

goal to reduce residential water demand by 25%. Results indicate that increases in price can not

ensure that the 2035 water demand target can be met when population increases. Los Angeles

must rely on furthering their conservation initiatives and increasing their use of stormwater

capture, recycled water, and expanding their groundwater storage. The forecasting approach

developed in this study can be utilized by other cities to understand the future of water demand

in water-stressed areas. Improving water demand forecasts will help planners understand and

optimize future investments in water supply infrastructure and related programs.

Keywords Water demand forecasting; Urban water management; Regional climate impact;

Multivariate; Monte Carlo Simulations

1. Introduction

Long-term forecasting of water demand is critical to the planning and management of a

water supply system (Zhou et al., 2002; Herrera et al., 2010; Cai and Rosegrant, 2002). Effective

demand forecasting requires an understanding of the influence of population and climate on

water use (Ruth et al., 2007; Zhou et al., 2000; Milano et al., 2013), as well as how water

consumers react to changes in water prices and conservation efforts. Effective water demand

forecasting leads to better informed water management decisions (Bougadis et al., 2005;

Grouillet et al., 2015).

2
 Water demand forecasting can be especially beneficial for semi-arid regions that are

impacted by frequent droughts (Zhou et al., 2000). Globally, increases in water withdrawal in

addition to competing demands for water resources allow for an environment in which future

drought impacts are exacerbated.

California experienced one of the worst droughts in its history between 2012-2017 with

record high temperatures and record low precipitation (Hanak et al., 2016). Over 80% of water

consumption in California is for the agricultural sector, whereas 17.6% is for residential,

commercial, and industrial use (Wilson et al., 2015). The continuous five years of drought

impacted all water sectors from urban to agriculture. The agricultural sector was greatly

impacted by the drought as farmers had a 50% decrease in surface water allocation in 2015,

which forced them to rely on limited groundwater. In the urban sector, water use was cut by

almost 25% from 2013 to 2016 (Hanak et al, 2016).

Drought conditions in California are a concern in Los Angeles, since almost 90% of its

water is imported (LADWP, 2010) and a majority of the water supply is from snowmelt runoff in

Sierra Nevada, where warming climate and changes in precipitation patterns have decreased

supply reliability (NOAA, 2015; Jeton et.al., 1996; Roy et al., 2012). The city of Los Angeles

obtains its water from five sources: Los Angeles Aqueduct (originating from Eastern Sierra

Nevada), Colorado River Aqueduct, California State Water Project (SWP), local groundwater,

and recycled water (LADWP, 2015). Further information on the water sources supplying Los

Angeles is provided in Table A2 of the Appendix. The Los Angeles Department of Water and

Power (LADWP), which is solely responsible for providing water to over 4 million Los Angeles

residents, purchases water from the Metropolitan Water District of Southern California (MWD)

when supplies are limited (MWD, 2010). The population served by LADWP has increased by an

3
annual growth rate of approximately 1.3 percent from 1980 to 2010. The population of Los

Angeles is expected to grow by 0.4 percent annually in the next 25 years and reach 4.5 million in

2035 (U.S. Census Bureau 2010). Increases in population will exacerbate the effects of limited

water supply reliability in Los Angeles.

Significant progress has been made in recent years in understanding the causes, severity,

impact and trends in the global distribution of drought events (Mishra and Singh, 2010; Wada et

al. 2011; Wilhite, 2016). As in other cities and regions, the capacity to anticipate and respond to

drought events in Los Angeles has evolved over time, with intertwining trajectories of population

growth, water law, water supply acquisition, treatment, pricing, and voluntary or mandatory

conservation (Buurman et al. 2017; Sullivan et al., 2017).

Variations in climate have an impact on both the sources of water and water demand by

consumers (McFarlane et al., 2012; Pulido-Calvo et al., 2012). Prior studies on its effect on water

demand show that increases in temperature and decreases in precipitation lead to increases in

water demand (Chang et al., 2014; Parandvash and Chang, 2016; Guo and Shen, 2016; Balling

and Gober, 2006; Polebitski et al., 2011; Jain et al., 2002; Rodrguez-Daz et al., 2007).

Reductions in precipitation increase water usage for outdoor landscaping, whereas increased

temperatures increase outdoor usage associated with swimming pools. Additionally, increases in

temperature increase evapotranspiration rates. For example, Los Angeles loses 128 cm per year

due to evapotranspiration, which can decrease recharge rates (LADWP, 2010). For Los Angeles,

the projected increase in temperatures from 2030-2050 vs. 1950-1999 using an average model

developed through the World Climate Research Programs Coupled Model Intercomparison

Project Phase 5 (CMIP5), which uses 34 global climate models (GCMs), ranges from 1.12-

2.52C (LADWP, 2015). With climate change likely to intensify drought severity in Southern

4
California (Cheng et al. 2016; Berg and Hall, 2017) and other regions, integrated assessments

that consider a broad range of scientific, engineering, economic and social processes are needed.

Additionally, with anticipated increase in competing demand of water resources (Fuller

and Harhay, 2010), water withdrawals in California are projected to exceed 100% of the

precipitation that is available in the state by 2050 (Roy et al., 2012). This will impact the

allocated water Los Angeles receives from the Los Angeles Aqueduct, California State Water

Project, California State Water Project, and local groundwater. For example, Luo et al. (2017)

used a VIC hydrologic model to simulate the hydrology of California, Nevada, and parts of Utah

to show how precipitation and temperature anomalies impacted the 2012-2015 California

drought. Results indicated that precipitation deficits produced extreme agricultural drought

whereas warmer temperatures decreased overall snowpack in high-elevation regions of

California. The multi-year drought which started in 2012 reduced the amount of water supplied

from the Los Angeles Aqueduct. Therefore, LADWP relied on other imported water from MWD,

which is the most expensive water source for LADWP. These price changes increased residential

water prices in Los Angeles, which influenced users to use less water. Increases in price, as well

as the mandatory conservation measures, resulted in a 22% reduction in water demand between

2007 to 2015 (LADWP, 2015).

There are various methods currently used to reduce water demand in Los Angeles,

including water pricing and conservation programs (LADWP, 2010). Previous studies have

analyzed the relationship of conservation and water pricing and have seen that people are

responsive to price increases (Rinaudo et al., 2012; Renwick and Green, 2000). For example,

Renwick and Green (2000) showed that a 10% increase in pricing policies in eight different

water utility regions in California decreased demand by 1.6%. The study also showed that

5
households were more likely to respond to price increases in the summer months due to the

capability of reducing outdoor water use. For Los Angeles, alterations in pricing began in 1993

when tiered water pricing was introduced which involved an increased fee for water use above a

certain allocated Tier 1 level. The tiered water price structure in Los Angeles has been shown to

exert downward pressure on water demand (Ashoori et al., 2015). Tiered water pricing provides

a monetary incentive to customers to reduce their water use. As of April 2016, single-family

households were charged through a four-tiered structure to further increase conservation,

although multi-family households still have a two-tiered rate structure. These tiered pricing

structures were implemented to reduce the impact of water supply variability and population

growth (LADWP, 2015). In addition to water pricing structures, conservation methods have been

implemented in Los Angeles since 1990 to reduce water demand (LADWP, 2010). During

previous drought years in Los Angeles, such as 2012-2017, consumers lowered demand through

voluntary measures. Between 2013 and early 2016, water demand decreased by nearly 25%

(Hanak et al., 2016). Through investing in various conservation programs and measures, such as

installation of water efficient household appliances and mandatory outdoor watering restrictions,

LADWP has been able to keep water demand levels low despite a 1.3% annual growth in

population since 1980 (LADWP, 2010).

To prepare for future impacts of drought and population growth, the City of Los Angeles

has examined ways to reduce per capita water use by an additional 25 percent by 2035 through

conservation incentives, education, and action campaigns (Sustainable City pLAn, 2015). The

implementation of the Sustainable City pLAn has led LADWP to set target water demand goals.

The citys water efficiency goals are mainly to reduce dependency on imported water supplies

and to increase future water sustainability.

6
 The objective of this study was to apply a multiple linear regression model for water

demand in Los Angeles (Ashoori et al., 2016) to project residential water demand until 2050

under various scenarios for the governing factors (climate change, population growth, pricing

structures, and conservation levels). The study uses the projections to assess the potential to meet

2035 water use targets considering potential changes in water price and population. The multiple

linear regression model for water demand was calibrated to data for the period 1970-2000, cross-

validated for the years 2001-2014, then refit for the full period 1970-2014 to provide maximally-

informed parameter estimates for the future projection period (2015-2050). The study analyzed

the likelihood of reaching the Sustainable City pLAn water demand target 2035 goals under

changing population and price, which are the main factors impacting water demand in Los

Angeles. Preceding studies have focused on modeling factors that impact water supply and

demand (Babel et al., 2007, Gaudin, 2006), but none have projected future water demand under

various future scenarios in Los Angeles to meet water demand targets. Contributing factors to

urban water demand, such as climate change, population growth, pricing and conservation are

difficult to project accurately. Therefore, a systematic set of scenarios was developed to

determine an envelope of residential water demand possibilities for the future as well as to

analyze the probability that future water demand targets can be met. Through water demand

forecasting, water planners can make more informed decisions regarding supply management in

addition to resiliency to future droughts. By understanding the underlying factors influencing

residential water demand in the service area of LADWP and forecasting them, approaches for

sustainable management of the supply system can be identified. This model forecasting approach

can be applied to other water stressed regions that are impacted by climate change and

population growth.

7
2. Methodology

2.1 Multiple Linear Regression Equation

The projections of water demand in this paper build on our previous analysis of the main factors

influencing water demand in Los Angeles (Ashoori et al., 2016). A multiple linear regression

(MLR) model for Los Angeles water demand was developed which includes conservation level,

population, price, temperature and precipitation as the explanatory variables and total monthly

water demand as the dependent variable. Further information on model development can be

found in Ashoori et al. (2016). Data on monthly residential water demand, Tier 1 pricing, and

conservation estimates from 1970 to 2014 were obtained from LADWP. The MLR model was

calibrated and cross-validated with these data. In the analysis that follows, the MLR model is

used to project water demand under alternative future scenarios of population, climate, pricing,

and conservation.

The MLR model for monthly water demand, with Tier 1 price, population, conservation

levels, temperature, and precipitation as explanatory variables, was as follows:

Mt =

(1)

where Mt, the monthly water demand in Los Angeles (m3) is the dependent variable, and the

other coefficients and variables are as given in Table 1. Other candidate variables, such as

median household income, Tier 2 water price, and drought index values that included

8
evapotranspiration rates, were investigated but their influence on fitting of historical water

demand data were either not statistically significant or were removed from the regression due to

collinearity. Cross validation of the model was performed for model evaluation. The cross

validation model was developed using water demand data from 1970-2000 and tested on the

dataset from 2001-2014. Results from the cross validation and calibration statistics from the

model are provided in the Appendix. There are various methods for adjusting multivariate

regression analysis results for the presence of autocorrelation (Bence, 1995). In this paper, the

use of cross validation, and the good performance of the model under it, provides assessment to

ensure that the estimates of the model are not significantly biased. As seen in Table 1, five

variables were ultimately used in the MLR model and to project under various scenarios for

water demand from 2015-2050.

Table 1 Descriptions of the explanatory variables in the MLR model of Los Angeles, California
residential water demand
Variables in Description Units Abbreviation Fitted P-values Units of values
the Model values+
Precipitation Total monthly cm RM 0.077* m3/cm
precipitation

Temperature Average monthly C TM <0.001** m3/ oC
temperature
Price Tier 1 price $ WP1 <0.001** m3/$

Population Total estimated # people PM <0.001** m3/person

monthly population
Conservation Conservation Level C - -

Low^ CL <0.001** m3

Medium^ CM <0.001** m3

High^ CH <0.001** m3
+ fitted beta values determined from historical data (1970-2014) (Ashoori et al., 2016)
** Indicates the coefficient is significant at the p=0.01 level
*Indicates the coefficient is significant at the p= 0.10 level
^
Indicates categorical dummy variables. Each conservation level is individually compared to very low levels of
conservation

9
Substitution of the fitted values of Table 1 into Equation 1 yields:

Mt (m3) =

(2)

The coefficients in the model are all statistically significant, and can be interpreted as

follows. For climate variables, a one-centimeter increase in precipitation would decrease water

demand by 74,600 m3 per month and a one-degree Celsius increase would increase residential

water demand by 1.36 million m3 each month. As for population, each additional person

contributes 27.8 m3 to the Los Angeles monthly water demand. For price increases, a one-dollar

per m3 increase in Tier 1 water price would decrease water demand by 12.3 million m3 per

month. For conservation, the three conservation levels yielded separate water demand

coefficients. Compared to a baseline of very low levels of conservation, a low conservation level

decreases water demand by 5.49 million m3 per month, a medium conservation level decreases

monthly water demand by 8.76 million m3, and a high conservation level decreases monthly

water demand by 10.3 million m3 compared to very low levels of conservation.

2.2 Scenario Development

The model represented in Equation (2) was used to estimate water demand for different scenarios

of climate change, population growth, pricing structures, and conservation methods from 2015-

2050. A summary of the scenarios developed is presented in Table 2. In this study, three

scenarios were developed for each of the four factors in the MLR model for water demand:

population, Tier 1 water price, precipitation, and temperature. Additionally, four scenarios for

conservation level were developed. As the climate-related variables precipitation and

10
temperature co-vary, combined scenarios were developed for these two variables. In total, 108

scenarios were generated for the model inputs and water demand forecasting for Los Angeles.

As seen in Table 2, three bounding scenarios (low-bound, baseline, high-bound) were

selected considering certain combinations of the explanatory variables. The low-bound scenario

estimates water demand for the case in which population and climate remain constant while price

and conservation efforts are maximized. The low-bound scenario projects water demand under

the best case scenario for minimization of water demand. For the baseline scenario, projections

for the variables are estimated to be in line with what has been predicted for Los Angeles by

LADWP (2010). Finally, a high-bound scenario for water demand was developed to analyze

water demand increases given maximum population growth and climate change (with increased

temperature and decreased precipitation), while keeping Tier 1 price and conservation at a

minimum. Although the high-bound scenario shows the upper level of water demand for the

various conditions of population, Tier 1 price, climate, and conservation, water availability

limitations in the future are likely to stimulate additional mandatory water conservation methods

that would reduce the likelihood of such an extreme outcome. Possible feedbacks between future

available water supply, climate change, population change, and pricing and conservation

decisions provide a high priority for ongoing research, but remain highly speculative (Dawadi

and Ahmad, 2013; Larson et al., 2015; Sampson et al., 2016; Gober et al., 2017). Our effort to

identify the critical factors for future outcomes and the manner in which they could co-occur in

meeting, or failing to meet, water demand objectives, represents an important first step for

formulating and benchmarking such behaviorally-predictive models.

11
Table 2 Summary of bounding scenarios for projecting residential water demand in Los Angeles,
California from 2015-2050
Low-bound Scenario Baseline Scenario High-bound Scenario

Population Constant baseline Medium growth High growth

(2014 level) (20% growth)1 (40% growth)

Tier 1 Price High Increase Medium increase Constant baseline

(200% growth) (50% growth) (2014 level)

Climate Constant baseline B1 for precipitation and A2 for precipitation and

(2014 level) temperature2 temperature3

Conservation High Medium Low

1
Data for projections of baseline population were obtained from the Los Angeles Department of Water and Power (2010)
2,3
Data for projections of climate variables were obtained from the California Energy Commission, Cal-Adapt data source
All other scenarios were chosen to reflect historic and potential future bounds

It is acknowledged that the influence of the explanatory variables on water demand in the

future could be altered by structural changes in water supply or housing infrastructure, evolving

perceptions and behavior, or other factors. Other investigators have developed water demand

models with higher geographic resolution for residential demand, with consideration of factors

such as sociodemographic profile and landscape use (Panagopoulos, 2014; Baerenklau et al.,

2014; House-Peters et al., 2011). However, the projections that follow are based on data

available and provide a baseline against which other modeling assumptions can be compared.

2.2.1 Population

Los Angeles is the second largest city in the United States, with the LADWP service population

being 3.9 million in 2015 (LADWP, 2015). Since 1980, the population of Los Angeles has

12
increased at an average annual rate of 1.3% (LADWP, 2010). With a projected population of 4.5

million by 2035 (US Census Bureau, 2010), it is imperative to estimate quantitatively the

increase in water demand due to population growth, recognizing that other factors, such as

conservation efforts and pricing strategies, can alleviate the impacts of population and climate on

water demand (Babel et al., 2007). The three scenarios selected for population projections result

from varying assumptions of growth in Los Angeles. The low-bound scenario population

projection assumes that population stays constant at 3.9 million after 2014. The baseline scenario

provides for an incremental increase of 20% in population by 2050, meaning the population

would reach roughly 4.7 million by 2050. The high-bound population scenario assumes

population will increase by 40% during the projection period, giving a population of 5.4 million

by 2050. These three scenarios encompass a broad range in the spectrum of potential population

growth influencing Los Angeles water demand.

2.2.2 Tier 1 Price

With the likelihood of future increases in water prices in Los Angeles, it is useful to investigate

how consumers will react and how pricing will affect total water demand (Billings and Jones,

2011). Various price scenarios were developed in order to represent possible changes in Los

Angeles water price in the future. Three scenarios were established using 2014 Tier 1 LADWP

prices as the starting reference price. The 2014 Tier 1 water price was $1.56 per m3 (LADWP,

2013). Therefore, in the low-bound scenario the Tier 1 rate was maintained at this constant value

from 2015 to 2050. The baseline scenario for price provides for a 50% incremental increase in

price from 2015 to 2050, whereas the high-bound scenario increases the Tier 1 water price by

13
100% from 2015 to 2050 (i.e., by a factor of 2). All values for water prices were readjusted for

inflation to the year 2014.

2.2.3 Climate

The climatic variables temperature and precipitation have an effect on short-term seasonal

changes in water demand. Los Angeles residents, which include single and multi-family

households use over 30% of their water for outdoor purposes (LADWP, 2010). That use

fluctuates depending on the temperature and amount of precipitation in the region (LADWP,

2010). Projections of precipitation in the southwestern United States in 2050, made from an

ensemble of 16 global climate models, indicated decreases in precipitation for most of the

models (Roy et al., 2012). As parametrized in the MLR model, this leads to an increase in water

demand, most likely for outdoor uses such as landscaping, as noted above. Climate change

effects on water supply are also likely to be crucial to the sustainability of the Los Angeles water

system (Ashoori et al., 2015), but this study focuses solely on the impacts of climate change on

water demand.

In order to develop three scenarios for climate encompassing both temperature and

precipitation, two different methods were employed. A bootstrap technique was initially used to

generate temperature and precipitation inputs for a climate scenario in which there is no change

in the current climate (a low change scenario). The bootstrap procedure involved re-sampling the

joint monthly temperature and precipitation dataset with replacement. The dataset used included

the 1970-2014 monthly values for temperature and precipitation in Los Angeles. For the two

14
additional scenarios, four different climate models were used (California Energy Commission,

2015):

Parallel Climate Model (PCM1)

Community Climate System Model (CCSM3)

Geophysical Fluid Dynamics Laboratory (GFDL)

National Centre for Meteorological Research (CNRM)

The climate models were analyzed using two different climate scenarios (A2, B1) for

temperature and precipitation in Los Angeles. The A2 scenario assumes continuously increasing

global population and economic growth whereas the B1 scenario assumes the introduction of

resource-efficient technologies accompanying the increase in global population (Nakicenovic et

al., 2000). Predictions for each model and scenario were obtained from the California Energy

Commission, Cal-Adapt data source (California Energy Commission, 2015). For each climate

scenario the average of the four different models was calculated and used in the projection.

Further information on the climate models can be observed in Table A3 and A9 in the Appendix.

Together, the three scenarios provide a broad range of estimates for the effect of climate on

water demand.

2.2.4 Conservation

Los Angeles water conservation methods can be categorized into voluntary, mandatory, and

market-based strategies (Maggioni, 2015). Since the 1990s, LADWP has implemented various

conservation methods and analyzed their impact on water demand. These conservation tools

have helped reduce per capita water use in Los Angeles (LADWP, 2010). The amount of water
15
demand in Los Angeles is influenced by consumer engagement in conservation, and modeling

that relationship is critical to future water planning management. LADWP is expected to increase

conservation activities in the coming decades (LADWP, 2010). In order to show its impact on

the overall water demand, four scenarios were established. The very low level conservation

scenario estimates zero water conservation after 2014. In the regression analysis, the other three

conservation levels (low, medium, high) were individually compared against the very low level

of conservation. The coefficient of each conservation level was then used to project water

demand under changing scenarios of conservation.

2.3 MLR Variables for 2015 to 2050

Figure 1 depicts each of the explanatory variables in the MLR model and their assumed or

calculated values from 2015 to 2050.

[FIGURE 1]

All projections were modeled in Tableau (Chabot et al., 2003), a software tool used for

interactive data visualization. Users are able to choose from the various model scenarios for

price, population, climate models, climate scenarios, and conservation levels in order to visualize

individual scenarios of water demand.

2.4 Meeting 2035 water demand targets

In order to analyze if projected water demand will meet LADWPs 2035 target reduction of 25%,

a Monte Carlo simulation was performed using the probability distribution of the variables.

16
Three values for population and price were fixed in the multiple linear regression equation for

2035. A Monte Carlo simulation of all other inputs: conservation, temperature, and precipitation,

as well as the regression residual, was run with 10,000 simulations. Values generated were

inputted into the multiple regression. The probability of simulations in which water demand did

not meet 2035 target levels was plotted.

3. Results and Discussion

3.1 Projecting water demand 2015-2050

Projections of total monthly urban water demand in Los Angeles for 2015-2050 were developed

using the various scenarios for population, water pricing, conservation methods, and climate. The

MLR model forecasting results indicate that price, population, temperature, precipitation, and

conservation levels all were significant in impacting projected water demand. As shown in

Figure 2, the monthly MLR model, used to project water demand from 2015 to 2050 (Figures 3-

5), closely follows the actual monthly water demand in Los Angeles during the period for which

the model was fit, 1970-2014. The R2 for the model was 0.72 indicating that 72% of the

variability in the historical monthly water demand is explained by the explanatory variables.

Further assessment of the performance of the forecasting models measures of accuracy can be

seen in the Appendix (Legates and McCabe, 1999). The figure also shows the expected monthly

variation in water demand. High demands are seen in hot, dry summer months, whereas low

demands are typical for cooler, wetter winter months, as seen in the fluctuation of temperature

and precipitation from historical data in Figure A1 in the Appendix. Historically, Los Angeles

receives on average 36 cm of precipitation each year, with most of it occurring between late

December to early March (LADWP, 2015). Additionally, highest temperatures are observed in
17
the months of July-September. This high and consistent degree of monthly variation is one

reason why a strong weather-climate signal can be identified in the historical water demand

record for Los Angeles, allowing relatively confident prediction of the effect of possible future

changes in temperature and rainfall on its water demand.

[FIGURE 2]

Figure 3 shows projections of water demand from 2015-2050 with a focus on the effects

of different population scenarios. As seen in Figure 3, the water demand under high population

growth, is estimated to vary from 710 million m3 to 1.10 billion m3 in 2050 depending on the

other variables in the analysis. The solid lines in Figure 3 indicate the 50 th percentile of the

forecasting, whereas the dashed lines depict the 25th and 75th percentile ranges of the forecasting

scenario. The medium population growth yielded an estimate of 450-840 million m3 in 2050. For

the case of no population growth the range of values for 2050 water demand in Los Angeles is

estimated at 200-580 million m3.

[FIGURE 3]

Similar to population, price is predicted to have a significant influence on future water

demand. Figure 4 shows the varying scenarios of water demand in Los Angeles under changes in

Tier 1 water pricing. The results for each of the three water price scenarios indicated with

different colors fall into three distinct groups. The groupings are caused by the somewhat larger

impact of population in the model. The model results indicate that population stability and
18
increases in price can drastically reduce water demand, by as much as 230 million m3 per year in

the future (i.e., nearly half of current use). Without the regulation of Tier 1 price, the potential

range of water demand shifts upward from 200-870 million m3 in 2050 to 360 million to 1.10

billion m3 in 2050.

[FIGURE 4]

The effect of climate on water demand in the projections is not as significant as that of

population and price. In Figure 5, water demand projections for the three scenarios of climate

change are shown with consideration of baseline scenarios of population, price, and

conservation. What can be observed is a small increase in water demand under the A2 climate

scenario compared to the B1 and no change scenarios. The difference between no change in

climate and the A2 climate scenario in 2050 is 28 million m3 per year. Therefore, a 5% increase

in water demand could be attributable to climate change by 2050, under baseline scenarios of

population, price, and conservation. Although it appears that changes in climate will not have a

major effect on future water demand in Los Angeles, climate change is likely to have a strong

influence on the overall water supply system, which relies primarily on water from outside the

city of Los Angeles (Ashoori et al., 2015). Further information on the climate, population, and

price scenarios can be observed in Table A8 in the Appendix.

[FIGURE 5]

19
 Shown in Table 3 are the projected values for residential water demand in Los Angeles

between 2015 and 2050 for bounding scenarios representing the extreme low and extreme high

outcomes, with a middle (baseline) scenario included for comparison. The total monthly water

demand projected for Los Angeles for 2015 is in the range of 450-600 million m3. However, the

plausible range of forecasted water demand expands considerably, extending from 196 million

m3 to 1.09 billion m3 for 2050 under changing scenario values of population, price, climate, and

conservation.

Table 3 Total projected residential water demand* in Los Angeles for the period 2015-2050

Year Low-Bound Baseline High-Bound

Scenario** Scenario** Scenario**
2015 450 470 600
2020 400 490 680
2025 360 520 740
2030 330 530 800
2035 300 550 880
2040 270 570 950
2045 230 590 1010
2050 200 620 1090
*In million m3 per year
** Scenario definitions
Low Bound Scenario: projected residential water demand with no population growth, high increases in water pricing and
conservation efforts, and no climate change.
Baseline Scenario: medium increase in population growth, water pricing, and conservation efforts from 2015-2050, and B1
climate change scenario.
High Bound Scenario: highest increase in population, no increase in price and no conservation with A2 climate change scenario.

The baseline scenario, which is perhaps the most likely scenario due to LADWP price,

conservation, and population projections (LADWP, 2015), yields an increase in water demand

from 460 million m3 to 620 million m3 per year by 2050. This is equivalent to an increase of

36%. From this analysis, it is seen that future water demand is expected to be largely driven by

price and population rather than climate change. Conservation can decrease water demand and

20
climate change can cause modest increases, but their long-term effects are small compared to the

large potential impacts of population growth and pricing policies in the LADWP service region.

However, during periods of extreme drought water consumers are capable of conserving an

increased amount of water due to mandatory restrictions (LADWP, 2010). Additionally, new

conservation methods implemented in Los Angeles can reduce water demand stress that is due to

multi-year droughts. For example, in 2015, the mayor set new conservation goals for Los

Angeles residents to reduce water use by 20% by 2017 and they have already reduced water

demand by 22% from 2007-2015 (Sustainable City pLAn, 2015). Exploring the long-term and

short-term changes in behavior to drought and water demand changes is important to investigate.

Such information will lead to improved forecasts and better solutions for future water demand

management.

3.2 Meeting future water demand targets

LADWP is investing in water conservation and local water supplies to reduce dependency on

imported water supplies and to increase future water sustainability. One main target for LADWP

is to reduce residential water demand by 25% by 2035. As shown in Figure 6, the probability of

meeting the 2035 LADWP water demand target decreases significantly with increased

population.

[FIGURE 6]

Without population growth, the probability that projected water demand for Los Angeles will

meet the 2035 water demand target for LADWP is 84.1%. Increases in price will ensure a 100%
21
chance of meeting the water demand target. However, with a 20% increase in population, the

probability of reaching the target drops. Without the influence of an increased price per m3 of

water, there is only a 3.1% chance of meeting the 2035 target with a population of 4.7 million.

LADWP predicts that their service area population will reach 4.4 million by 2035 indicating the

probability of reaching their water demand targets are difficult without increases in Tier 1

pricing, conservation, and alternative local water sources (LADWP, 2015). Even with doubling

of price, the impact of a 20% increase in population will not ensure a 100% probability of

reaching the target. That probability drops significantly to 0% once population increases by 40%.

In order to prepare for increasing population, as well as for unreliability of imported water

supply, LADWP and Bureau of Sanitation are investing more resources into conservation

initiatives, stormwater capture, expanding groundwater storage, as well as recycled water

(Hagekhalil et al., 2014). With approximately 39% of total water demand being for outdoor

water use (LADWP, 2015), LADWP can better reach their 2035 target through limitations on

outdoor water consumption. These include less water intensive landscapes, restricting times of

watering, and increasing the usage of rainwater tanks. Other cities have greatly reduced water

use through outdoor water restrictions. For example, in Melbourne, Australia, outdoor watering

is prohibited between the hours of 10am to 8pm and has contributed to reducing total water use

by 35% from 2000-2009 (Cahill and Lund, 2011).

The outcomes of this forecasting will assist decision makers in Los Angeles to plan for

future impacts of water demand. The methodology employed to forecast residential water

demand in Los Angeles can be applied to other water supply systems as well.

22
4. Conclusions

Management of water demand is important to the future sustainability of the Los Angeles water

supply system and other systems in arid regions. As uncertainty is present in future climate

forecasts, population estimates, pricing structures, and conservation methods, water planners

must direct their attention to long-term strategies that are robust under a range of scenarios. In

this study, the total monthly residential water demand in Los Angeles during the period of 2015-

2050 was projected across various scenarios of population, water pricing, climate, and

conservation levels. Three future scenarios of climate, population, and price as well as four

scenarios of conservation levels were developed based on historical trends. The maximum total

water demand is projected to occur when population increases, climate becomes drier and hotter

under a A2 climate model, and price and conservation efforts remain stable. For this case the

projected maximum residential water demand in 2050 is estimated to be 1.09 billion m3, a 139%

increase from the 2014 water demand of 460 million m3. However, in the more likely baseline

scenario, water demand is estimated to be 620 million m3 per year in 2050 indicating that water

demand will increase by 36% from 2014. The study results indicate that changes in population

and price in Los Angeles are expected to play a bigger part in influencing future water demand

than climate change and conservation programs. Understanding the importance and future

pathways of the factors affecting water demand will be a key step in ongoing planning for a

reliable and sustainable water supply for Los Angeles.

The study also examined the probability of reaching the 2035 LADWP water demand

target reduction of 25% under changing population and price, which are the main factors

impacting water demand in Los Angeles. The Monte Carlos simulation results showed that any

increase in population will significantly decrease the likelihood of reaching the 2035 water

23
demand target, even with increasing water prices. For example, without the influence of an

increased price per m3 of water, there is only a 3.1% chance of meeting the target with a

population increase of 20%. In order to prepare for increased water demand, Los Angeles must

expand their conservation initiatives, and increase their local water supply, such as stormwater

capture, recycled water, and expanding groundwater storage, and further restrict outdoor water

use.

This study is novel in analyzing how forecasted scenarios of population growth, climate

change, price, and conservation methods, can impact water demand in Los Angeles as well as

understand their capacity to enable reaching LADWP water demand target. However, future

research on water demand differences between single and multi-family households, as well as the

impact of socioeconomic status, can be beneficial in understanding the role of land use and

income on forecasted water demand.

The modeling approach in this study can be utilized by other cities to understand the

future of water demand in water-stressed areas. Demand forecasting requires an understanding of

the influence of population, climate, and consumer behaviors due to water prices and

conservation efforts in order to make informed water management decisions. The approaches

used to develop the water demand forecasting model and its application to evaluate future water

demand targets can be used in decision-support tools to provide information that will lead to

better solutions for future water demand management in water-stressed areas such as Los

Angeles.

Acknowledgements

24
The authors thank Simon Hsu and Priscilla Gonzalez of the Los Angeles Department of Water

and Power for providing data used in this study. Support for this work was provided by a

Carnegie Mellon University College of Engineering Deans Fellowship and a Steinbrenner

Institute U.S. Environmental Sustainability Fellowship to Negin Ashoori. The Fellowship was

supported by a grant from the Colcom Foundation, and by the Steinbrenner Institute for

Environmental Education and Research. The work was also supported by the Hamerschlag

University Professorship of David Dzombak, and the H. John Heinz Professorship of Mitchell

Small.

References

Ashoori N., Dzombak D.A., Small M.J. 2015. Sustainability Review of Water Supply Options in

the Los Angeles Region. Journal of Water Resources Planning and Management. doi:

10.1061/(ASCE)WR.1943-5452.0000541, A4015005

Ashoori N., Dzombak D.A., Small M.J., 2016. Modeling the Effects of Conservation,

Demographics, Price, and Climate on Urban Water Demand in Los Angeles, California.

Water Resources Management, 1-16. doi: 10.1007/s11269-016-1483-7.

Babel M., Gupta A., Pradhan P., 2007. A Multivariate Econometric Approach for Domestic

Water Demand Modeling: an Application to Kathmandu, Nepal. Water Resources

Management 21, 573-589. doi: 10.1007/s11269-006-9030-6.

Baerenklau K.A., Schwabe K.A., Dinar A., 2014. The Residential Water Demand Effect and

Increasing Block Rate Water Budgets. Land Economics 90(4), 683-699

Balling R.C., Gober, P., 2006. Climate Variability and Residential Water Use in the City of

Phoenix, Arizona. Journal of Applied Meteorology and Climatology 46, 1130-1137

25
Bence J.R., 1995. Analysis of Short Time Series: Correcting for Autocorrelation. Ecology 76(2),

628-639

Berg, N. and Hall, A., 2017. Anthropogenic Warming Impacts on California Snowpack During

Drought. Geophysical Research Letters 44(5), 2511-2518.

Billings R.B., Jones C.V., 2008. Forecasting Urban Water Demand. American Water Works

Association. Denver, Colorado.

Bougadis J., Adamowski K., Diduch R., 2005. Short-Term Municipal Water Demand

Forecasting. Hydrological Processes 19, 137-148. doi: 10.1002/hyp.5763

Buurman, J., Mens, M. J. and Dahm, R. J., 2017. Strategies for Urban Drought Risk

Management: A Comparison of 10 Large Cities. International Journal of Water

Resources Development 33(1), 31-50.

Cahill, R., and Lund, J., 2011. Residential Water Conservation in Australia and California.

https://watershed.ucdavis.edu/pdf/Cahill_Residential%20Water%20Conservation%20in

%20Australia

Cai, X., and Rosegrant, M. W., 2002. Global Water Demand and Supply Projections. Part 1. A

Modeling Approach 27(2), 159-169.

California Energy Commission, 2015. Cal-Adapt. http://cal-adapt.org/data/tabular/

Chabot C., Stolte C., Hanrahan P., 2003. Tableau (computer software)

Chang H., Praskievicz S., Parandvash H., 2014. Sensitivity of Urban Water Consumption to

Weather Climate Variability at Multiple Temporal Scales: The Case of Portland.

International Journal of Geospatial and Environmental Research 1(7).

26
Cheng, L., Hoerling, M., AghaKouchak, A., Livneh, B., Quan, X. W. and Eischeid, J., 2016.

How has Human-Induced Climate Change Affected California Drought Risk? Journal of

Climate 29(1), 111-120.

Dawadi, S. and Ahmad, S., 2013. Evaluating the Impact of Demand-Side Management on Water

Resources Under Changing Climatic Conditions and Increasing Population. Journal of

Environmental Management 114, 261-275.

Fuller A.C., Harhay M.O., 2010. Population Growth, Climate Change and Water Scarcity in the

Southwestern United States. American Journal of Environmental Science 6(3), 249-252.

doi: 10.3844/ajessp.2010.249.252

Gaudin S., 2006. Effect of Price Information on Residential Water Demand. Applied Economics

38, 383-393. doi: 10.1080/00036840500397499

Gober, P., White, D.D., Quay, R., Sampson, D.A. and Kirkwood, C.W., 2017. Socio-Hydrology

Modelling for an Uncertain Future, with Examples from the USA and

Canada. Geological Society, London, Special Publications 408(1), 183-199.

Grouillet, B., Fabre, J., Ruelland, D., and Dezetter, A., 2015. Historical Reconstruction and 2050

Projections of Water Demand Under Anthropogenic and Climate Changes in Two

Contrasted Mediterranean Catchments. Journal of Hydrology 522, 684-696.

Guo, Y., and Shen, Y., 2016. Agricultural Water Supply/Demand Changes Under Projected

Future Climate Change in the Arid Region of Northwestern China. Journal of Hydrology

540, 257-273.

Hagekhalil, A., Kharaghani, S., Tam, W., Haimann, R., Susilo, K., 2014. City of Los Angeles

the Green Blue City One Water Program. ICSI 2014: Creating Infrastructure for a

Sustainable World: 482-501. http://ascelibrary.org/doi/pdf/10.1061/9780784478745.043.

27
Hanak, E., Mount, J., and Chappelle, C., 2016. Californias Latest Drought. PPIC Water Policy

Center. Public Policy Institute of California.

Herrera, M., Torgo, L., Izquierdo, J., Perez-Garcia, R., 2010. Predictive Models for Forecasting

Hourly Urban Water Demand. Journal of Hydrology 387, 141-150.

House-Peters L., Pratt B., Chang H., 2010. Effects of Urban Spatial Structure,

Sociodemographics, and Climate on Residential Water Consumption in Hillsboro,

Oregon. Journal of American Water Resources Association 46(3), 461-472

Jain A., Varshney A.K., Joshi U.C., 2002. Short-Term Water Demand Forecast Modelling at IIT

Kanpur Using Artificial Neural Networks. Water Resources Management 15, 299-321.

Jeton, A. E., Dettinger, M. D., and Smith, L. J., 1996. Potential Effects of Climate Change on

Streamflow, Eastern and Western Slopes of the Sierra Nevada, California and Nevada.

U.S. Geological Survey Water Resources Investigations Report 95-4260.

Larson, K.L, White, D.D., Gober, P. and Wutich, A., 2015. Decision-Making Under Uncertainty

for Water Sustainability and Urban Climate Change Adaptation. Sustainability 7(11),

14761-14784.

Legates, D.R., and McCabe Jr., G.J., 1999. Evaluating the Use of 'Goodness-of-Fit' Measures in

Hydrologic and Hydroclimatic Model Validation. Water Resour. Res. 35 (1), 233-241.

Los Angeles Department of Water and Power (LADWP), 2010. Urban Water Management Plan

2010.

Los Angeles Department of Water and Power (LADWP), 2013. Water Resources Department

Los Angeles Department of Water and Power (LADWP), 2015. Los Angeles Department of

Water and Power 2015 Briefing Book.

28
 http://d3n8a8pro7vhmx.cloudfront.net/themes/5595df952213930462000001/attachments/

original/1430386934/LADWP_Water_System_Fact_Sheet.pdf?1430386934

Luo, L., Apps, D., Arcand, S., Xu, H., Pan, M., and Hoerling, M, 2017. Contribution of

Temperature and Precipitation Anomalies to the California Drought During 2012-2015.

Geophysical Research Letters 44, 3184-3192.

Maggioni E., 2015. Water Demand Management in Times of Drought: What Matters for Water

Conservation. Water Resources Research 51, WR016301. doi: 10.1002/2014WR016301.

McFarlane, D., Stone, R., Martens, S., Thomas, J., Silberstein, R., Ali, R., and Hodgson, G.,

2012. Climate Change Impacts on Water Yields and Demands in South-Western

Australia. Journal of Hydrology 475, 488-498.

Metropolitan Water District of Southern California (MWD), 2010. The Regional Urban Water

Management Plan. Water Resource Management Group. Available at

http://www.mwdh2o.com/mwdh2o/pages/yourwater/RUWMP/RUWMP_2010.pdf

Milano, M., Ruelland, D., Dezetter, A., Fabre, J., Ardoin-Bardin, S., and Servat, E., 2013.

Modeling the Current and Future Capacity of Water Resources to Meet Water Demands

in the Ebro Basin. Journal of Hydrology 500, 114-126.

Mishra, A.K. and Singh, V.P., 2010. A Review of Drought Concepts. Journal of

Hydrology, 391(1), 202-216.

Nakicenovic N. et al, 2000. Special Report on Emissions Scenarios: A Special Report of

Working Group III of the Intergovernmental Panel on Climate Change. Cambridge

University Press, Cambridge, U.K. 599.

http://www.grida.no/climate/ipcc/emission/index.htm

29
National Oceanic and Atmospheric Administration, 2015. U.S. Seasonal Drought Outlook.

http://www.cpc.ncep.noaa.gov/products/expert_assessment/sdo_summary.html

Panagopoulos, G.P., 2014. Assessing the Impacts of Socio-Economic and Hydrological Factors

on Urban Water Demand: A Multivariate Statistical Approach. Journal of Hydrology

518, 42-48.

Parandvash, G.H., and Chang, H., 2016. Analysis of Long-term Climate Change on Per Capita

Water Demand in Urban Versus Suburban Areas in the Portland Metropolitan Area,

USA. Journal of Hydrology 538, 574-586.

Polebitski A.S., Palmer R.N., Waddell P., 2011. Evaluating Water Demands Under Climate

Change and Transitions in the Urban Environment. Journal of Water Resources Planning

and Management. doi: 10.1061/(ASCE)WR.1943-5452.0000112

Pulido-Calvo, I., Gutirrez-Estrada, J.C., Savic, D., 2012. Heuristic Modelling of the Water

Resources Management in the Guadalquivir River Basin, Southern Spain. Water

Resources Management 26, 185-209 (doi: 10.1007/s11269-011-9912-0).

Renwick, M. E., and Green, R. D., 2000. Do Residential Water Demand Side Management

Policies Measure Up? An Analysis of Eight Water Agencies. Journal of Environmental

Economics and Management 40, 37-55.

Rinaudo, J. D., Neverre, N., Montiginoul, M., 2012. Simulating the Impact of Pricing Policies on

Residential Water Demand: A Southern France Case Study. Water Resources

Management 26, 2057-2068. doi: 0.1007/s11269-012-9998-z

Rodrguez-Daz, J.A., Weatherhead, E.K., Knox, J.W., Camacho, E., 2007. Climate Change

Impacts on Irrigation Water Requirements in the Guadalquivir River Basin in Spain. Reg

Environ Change 7(3), 149-159.

30
Roy S.B., Chen L., Girvetz E.H., Maurer E.P., Mills W.B., Grieb T.M., 2012. Projecting Water

Withdrawal and Supply for Future Decades in the U.S. Under Climate Change Scenarios.

Environmental Science & Technology 46, 2545-2556. doi: 10.1021/es2030774

Ruth M., Bernier C., Jollands N., Golubiewski N., 2007. Adaptation of Urban Water Supply

Infrastructure to Impacts from Climate and Socioeconomic Changes: The Case of

Hamilton, New Zealand. Water Resources Management 21, 1031-1045. doi:

10.1007/s11269-006-9071-x

Sampson, D.A., Quay, R. and White, D.D., 2016. Anticipatory Modeling for Water Supply

Sustainability in Phoenix, Arizona. Environmental Science & Policy 55, 36-46.

Sullivan, A., White, D. D., Larson, K. L. and Wutich, A., 2017. Towards Water Sensitive Cities

in the Colorado River Basin: A Comparative Historical Analysis to Inform Future Urban

Water Sustainability Transitions. Sustainability 9(5), 761.

Sustainable City pLAn, 2015. Office of the Los Angeles Mayor Eric Garcetti.

http://plan.lamayor.org/wp-content/uploads/2017/03/the-plan.pdf

U.S. Census Bureau, 2010. Metropolitan and Micropolitan Statistical Areas.

http://www.census.gov/popest/data/historical/2010s/vintage_2011/metro.html

Wada, Y., Van Beek, L.P.H. and Bierkens, M.F., 2011. Modelling Global Water Stress of the

Recent Past: On the Relative Importance of Trends in Water Demand and Climate

Variability. Hydrology and Earth System Sciences 15(12), 3785-3805.

Wilhite, D.A. ed., 2016. Droughts: A Global Assesment. Routledge.

Wilson, T. S., Sleeter, B. M., and Cameron, D. R., 2015. Future Land-Use Related Water

Demand in California. Environmental Research Letters 11, 054018.

31
Zhou, S.L., McMahon, T.A., Walton, A., and Lewis, J., 2000. Forecasting Daily Urban Water

Demand: A Case Study of Melbourne. Journal of Hydrology 236, 153-164.

Zhou, S.L., McMahon, T.A., Walton, A., and Lewis, J., 2002. Forecasting Operational Demand

for an Urban Water Supply Zone. Journal of Hydrology 259, 189-202.

32
 Figure Captions

Fig. 1. Scenario values for the five variables in the MLR model used to project water demand.

See Table 2 for sources and scenario descriptions

Fig. 2. Comparing actual versus fitted MLR model of monthly water demand in Los Angeles, 1970-2014

Fig. 3. Projections of water demand from 2015-2050 highlighting high, medium, and no population
growth scenarios (Horizontal line indicate 50th percentile of projections with dashed lines depicting 25th
and 75th percentiles).

Fig. 4. Projections of water demand from 2015-2050 highlighting various Tier 1 price increase scenarios
(Horizontal line indicate 50th percentile with dashed lines depicting 25th and 75th percentiles).

Fig. 5. Projections of water demand from 2015-2050 under the baseline scenario with different climate
models*
(*The baseline scenario assumes 20% increase in population, 50% increase in price, and 20% increase in
conservation.)

Fig. 6. Graph showing probability of meeting LADWP pLAn target of reducing residential water demand
by 25% as function of Population and Price in 2035. Dark blue indicates probability of meeting water
demand target; Light blue indicates not meeting water demand target.

33
34
 Monthly Water Demand in Million Cubic
Meters

0
20
30
40
50
60

10

Actual
Month/Year
MLR Model

35
 1.2
Water Demand (Billion Cubic Meters) Low Population Growth

1 Medium Population Growth

High Population Growth

0.8

0.6

0.4

0.2

0
2015 2020 2025 2030 2035 2040 2045 2050
Year

36
 1.2
Water Demand (Billion Cubic Meters) No Increase in Tier 1 Price

1 Medium Increase in Tier 1 Price

High Increase in Tier 1 Price

0.8

0.6

0.4

0.2

0
2015 2020 2025 2030 2035 2040 2045 2050
Year

37
A2 climate scenario

B1 climate scenario

No change in climate

38
39
 Highlights

Residential water demand was projected from 2015-2050 under various scenarios

Water demand in Los Angeles is projected to increase by 36% from 2014 to 2050

Climate change will increase water demand in Los Angeles by 5% from 2014 to 2050

Population growth significantly decreases chances of meeting 2035 water demand goal

Stormwater capture and recycled water for groundwater storage can help reach goal

40