Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Simulating High-Elevation Hydropower with Regional

Climate Warming in the West Slope, Sierra Nevada


David E. Rheinheimer 1; Joshua H. Viers 2; Jack Sieber 3; Michael Kiparsky 4; Vishal K. Mehta 5;
and Scott T. Ligare 6
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Water systems in snowmelt-dominated hydroregions such as California’s Sierra Nevada mountains are sensitive to regional
climate change, hydropower systems in particular. In this study, a water resources management model was developed for the upper west
slope Sierra Nevada to understand the potential effects of regional climate warming on hydropower at the watershed scale, a scale that has
been largely neglected but is important for hydroregional planning. The model is developed with the Water Evaluation and Planning system
(WEAP) and includes most water management infrastructure in the study region. Hydropower is simulated assuming historical long-term
electricity demand and a spill minimization rule. The method is suitable for simulating generation for most of the main watersheds in the
region. To assess the potential effect of climate warming, uniform air temperature increases of 0°C, 2°C, 4°C, and 6°C were considered, with
no change in precipitation, to approximate regional warming through 2100. The highly productive northern Sierra Nevada sees large re-
ductions in hydropower generation with decreases in annual runoff. The central watersheds see less reduction in annual runoff and can adapt
better to changes in runoff timing. Generation in southern watersheds, which are less productive, decreases. Results from this study can help
identify which watersheds might easily adapt to climate change, where hydropower is likely to conflict with other uses, and where more
detailed operational studies are needed. DOI: 10.1061/(ASCE)WR.1943-5452.0000373. © 2014 American Society of Civil Engineers.
Author keywords: Hydro power; Simulation models; Climate change; Reservoirs.

Introduction The high-elevation, upper west slope of the Sierra Nevada pro-
vides most of the water for California’s vast water supply system,
Hydropower systems are at the management nexus of energy sup- supporting hydropower generation, ecosystems, recreation, and
ply, water supply, freshwater ecosystems and ecosystem services, municipal and agricultural water supply for much of the state.
recreation, and flood control. High-elevation hydropower systems Hydrologic patterns in the Sierra Nevada are typical of a Mediter-
are particularly sensitive to global climate warming in snowmelt- ranean-montane climate, with a cool, wet period from about
dominated regions, where major shifts in runoff magnitude and tim- November to April and warm, dry period from about May to
ing are expected (Barnett et al. 2005; Bates et al. 2008; IPCC 2007), October (Null et al. 2010). Of California’s in-state hydropower
exacerbating existing trade-offs between competing water uses generation capacity of 14,000 MW, about 134 high-elevation
(Kundzewicz et al. 2007). The Sierra Nevada in California, USA, hydropower plants in the Sierra Nevada contribute approximately
is one such snowmelt-dominated hydroregion with extensively 8,800 MW, representing 63% of in-state hydropower capacity and
developed hydropower systems vulnerable to climate change. This 16% of in-state generation capacity from all sources (California
paper describes a water resources management model developed Energy Commission 2009; Franco and Fagundes 2005). The value
for the upper west slope of the Sierra Nevada, applied to assess of this hydropower is disproportionate to its total output because
spatial trends of potential effects of climate warming on hydro- high-elevation hydropower systems are typically operated when
power generation across the entire region. electricity prices are high and to respond to short-term fluctuations
in energy demand (Madani and Lund 2009; Olivares 2008; Vicuña
1
Postdoctoral Researcher, Center for Watershed Sciences, Univ. of et al. 2011).
California, One Shields Ave., Davis, CA 95616 (corresponding author). Many studies over the last three decades have focused on
E-mail: drheinheimer@ucdavis.edu climate change effects on California’s water resources (Vicuna and
2
Associate Director, Center for Watershed Sciences, Univ. of California,
Dracup 2007; Franco et al. 2011), including effects on hydropower.
One Shields Ave., Davis, CA 95616. E-mail: jhviers@ucdavis.edu
3
Senior Scientist, Stockholm Environment Institute, 11 Curtis Ave.,
Studies consistently show substantial air temperature increases and
Somerville, MA 02144. modest precipitation decreases in California by the end of the cen-
4
Associate Director, Wheeler Institute for Water Law & Policy, Univ. of tury, though California will retain its Mediterranean climate (Cayan
California at Berkeley, Berkeley, CA 94720. et al. 2008; Dettinger 2005; Franco et al. 2011). Specific statewide
5
Staff Scientist, Stockholm Environment Institute, 400 F St., Davis, CA estimates of warming range from about 1.5°C to 6°C with business-
95616. as-usual greenhouse gas emissions (Franco et al. 2011) with greater
6
Water Resources Control Engineer, State Water Resources Control warming expected in the northern Sierra Nevada (Hayhoe et al.
Board, 1001 I St., Sacramento, CA 95814. 2004). Warming alone has and will decrease the fraction of precipi-
Note. This manuscript was submitted on November 4, 2011; approved
tation falling as snow, increasing winter precipitation runoff and
on April 18, 2013; published online on April 20, 2013. Discussion period
open until October 1, 2014; separate discussions must be submitted for in- decreasing spring snowmelt runoff, shifting runoff to earlier in the
dividual papers. This paper is part of the Journal of Water Resources Plan- year (Cayan et al. 2008; Dettinger et al. 2004; Null et al. 2010;
ning and Management, Vol. 140, No. 5, May 1, 2014. © ASCE, ISSN Young et al. 2009). Warming will also reduce annual runoff, even
0733-9496/2014/5-714-723/$25.00. with no change in precipitation (Null et al. 2010).

714 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014

J. Water Resour. Plann. Manage. 2014.140:714-723.


Investigations of the effects of regional climate change on Cal-
ifornia’s hydropower generation have historically been quite gen-
eralized over the entire state or detailed and limited to one or two
specific hydropower projects. California-wide, Madani and Lund
(2010) estimated mean hydropower generation decreases of
1.3% assuming a change in runoff timing only and 19.7% assuming
reduced runoff (drier conditions), providing an approximate range
for expected changes. Vicuña et al. (2011) focused on two specific
systems in the Sierra Nevada, the Big Creek system in the San Joa-
quin River (R.) watershed and the Upper American River Project in
the American R. watershed, and estimated end-of-century reduc-
tions in generation of 10.4% in the former and 12.2% in the latter,
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

averaged over a range of downscaled general circulation models.


Mehta et al. (2011), meanwhile, estimated a reduction of approx-
imately 22% in the Middle Fork Project in the American R. water-
shed, assuming 6°C warming and historical precipitation levels. No
previous study in California has assessed spatial trends of climate
change effects on hydropower across a larger region, primarily be-
cause of the greater modeling challenges associated with achieving
higher spatiotemporal accuracy.
The watershed as a spatial unit of analysis, if sufficiently large,
can allow for coarser scale water system modeling while still pro-
viding useful insights about spatial trends within a larger hydrore-
gion (Muñoz and Sailor 1998). Here, the “watershed” is considered
to be large enough to including at least one major river, with pos-
sibly several hydropower systems. The watershed scale, however
defined, can be useful for hydroregional planning. For example,
management of regional freshwater ecosystems is necessarily at
the watershed-to-catchment scale (Viers 2011). Recognizing this,
others have begun to use the watershed as a spatial unit for analysis
in the western Sierra Nevada. Young et al. (2009) developed a
hydrologic model of the western Sierra Nevada, calibrated using Fig. 1. Features included in SIERRA
rivers from the 15 largest watersheds in the region. Null et al.
(2010) assessed the results of Young et al. (2009) to further under-
stand the vulnerability of the same 15 watersheds to changes in assessment and results. SIERRA also builds on the water manage-
hydrologic patterns. Similarly, Ficklin et al. (2012) used mostly ment simulation model of Mehta et al. (2011), which focuses on the
the same rivers for calibration of a different hydrologic model Cosumnes, American, Bear, and Yuba River (CABY) watersheds
of the Sierra Nevada. within the study region.
The main goal of this study was to develop and apply a water As with the study by Mehta et al. (2011), SIERRA uses the
resources simulation model that builds on the work of Young et al. water management module of the Water Evaluation and Planning
(2009) and the related hydropower study of Mehta et al. (2011). system (WEAP), a modeling platform commonly used in water
Objectives included developing a network model that simulates planning studies [e.g., Thompson et al. (2012)] that integrates a
dominant water management systems of the upper west slope Sierra physically based, lumped hydrology model with a priority-based
Nevada in sufficient detail to assess spatial trends of climate change water resources management simulation model (Yates et al. 2005).
effects on high-elevation hydropower without the operational detail To simulate water management, WEAP requires infrastructure rep-
required for specific systems. The intent was thus to help bridge the resentation, infrastructure parameters, initial and boundary condi-
gap between single-system studies and general, statewide studies to tions, and operating rules. Rules include assignment of priorities,
further the scientific basis for regional watershed planning. To do which range from 1 (highest) to 99 (lowest). WEAP uses a tradi-
this, a coarse water management simulation model was developed tional linear programming solution method to apportion water as
for 15 watersheds in the upper west slope of the Sierra Nevada equitably as possible among water demands with the same priority
(Fig. 1), called the Sierra Integrated Environmental and Regulated (Yates et al. 2005). System infrastructure and operations rules are
Rivers Assessment (SIERRA) model. The rest of this paper de- summarized here, with greater detail found in Rheinheimer (2011).
scribes the model methods and limitations, model performance,
and application of the model to assess regional climate warming
effects on hydropower generation. Study Area and Infrastructure
The study area includes 13 of the 15 largest watersheds in the high-
elevation, upper west slope of the Sierra Nevada, with most water-
Methods sheds defined by the large, low-elevation, multipurpose terminal
“rim” dams just above California’s Central Valley floor (Fig. 1).
The SIERRA model builds on the unregulated hydrologic model of The watersheds range in size from 730 km2 to 9,400 km2 , totalling
the west slope Sierra Nevada developed by Young et al. (2009), 47,700 km2 , and collectively drain approximately 26.2 × 109 m3
including the same watershed and subwatershed definitions and per year (Null et al. 2010). Watersheds include the Feather
the same weekly time step. Though the weekly time step was (FEA); American, Bear, and Yuba (ABY); Mokelumne (MOK);
used, this paper focuses on the seasonal time step in performance Calaveras (CAL); Stanislaus (STN); Tuolumne (TUO); San

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014 / 715

J. Water Resour. Plann. Manage. 2014.140:714-723.


Joaquin (SJN); Kings (KNG); Kaweah (KAW); Tule (TUL); and because they are better able to represent more intricate relationships
Kern (KRN) Rivers. Two watersheds in the region—Cosumnes among system components, multiple nonmarket water uses such as
(COS) and Merced (MER)—are omitted for lack of major infra- found in the Sierra Nevada, and contractual operations policies
structure. The Yuba, Bear, and Kern River watersheds are defined (Sigvaldason 1976). Within a simulation framework, hydropower
by low-elevation, low-volume diversion dams. Null et al. (2010) operations decisions can be from an optimization subroutine or
provides more detailed characteristics of these watersheds. from explicit rules (Rani and Moreira 2010), such as derived ana-
The SIERRA model simulates management of reservoirs, hy- lytical rules (Lund 2000; Lund and Guzman 1999).
dropower, water supply, and in-stream flow requirements above Despite decades of research in modeling multireservoir systems,
the rim dams (exclusive) in the study area. Fifty-eight (58) reser- developing and applying either an embedded optimization method
voirs listed by the California Data Exchange Center (CDEC; or system-specific analytical rules within a short time frame for a
cdec.water.ca.gov) are included, out of about 300 with dams rou- region as extensively developed as the Sierra Nevada remains chal-
tinely inspected (http://www.water.ca.gov/damsafety), with most lenging. Rather than focus on modeling every hydropower system
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

small (< ∼2 × 106 m3 ) reservoirs excluded. Rim reservoirs are accurately, a first approximation of hydropower operations was
omitted, with major exceptions of New Bullards Bar Reservoir sought. Here, two methods are used to define water demand for
(ABY) and Lake Isabella (KRN). Most high-elevation systems hydropower. The first method, called the water year index method
are operated independently of the rim dams, except for Hetch (WYIM), uses historical observations to approximate hydropower
Hetchy, which provides some flood-control functions for the release requirements. The second method, called the spill demand
low-elevation New Don Pedro Reservoir. Other infrastructure method (SDM), is based on water demand and is used to
and managed features include 86 fixed-head hydropower plants, minimize spill.
16 variable-head hydropower plants, 126 conveyances (canals,
flumes, tunnels, and pipelines), 25 water supply demand nodes,
and 109 in-stream flow requirements. All high-elevation hydro- Water Year Index Method
power plants are included, except for a few small, private, isolated The modeling objective was to represent average historical diver-
hydropower plants. Demands for water supply are included where sions for hydropower at the seasonal scale rather than simulate ac-
data were available or where a diversion clearly exists from public tual operations in any given week. To do this, the model followed
documents and maps. No important water management features the approach of Mehta et al. (2011), who demonstrated that mean
were excluded, such that exclusions would have very minor effects weekly hydropower operations can be coarsely represented by es-
on model results at the spatial resolution considered. tablishing a historical correlation between a discrete water year type
(WYT) and diversions for hydropower during any given week.
Such an approach accounts for two important features of historical
Reservoirs operations but neglects others, described in further detail in a fol-
Reservoir storage capacities and initial storage were obtained from lowing section. First, hydropower operational goals are to meet en-
CDEC and the U.S. Geologic Survey (USGS; http://www.usgs ergy demand patterns (Jacobs et al. 1995). Second, over the long
.gov) or estimated from available data. Volume-elevation relation- term, hydropower operations depend on total amount of water
ships are used to determine surface area, needed for estimating available, as determined by runoff patterns and storage capability
evaporation, and for variable-head hydropower. The USGS (Madani and Lund 2009). It is assumed that historical hydropower
volume-elevation data are used where available or estimated using releases over the long term reflect historical energy demand. Thus,
a second-order polynomial fit using historical volume and elevation if energy requirements are assumed constant, as in this study, it is
data from CDEC. assumed that hydropower operators will attempt to continue to
Reservoir operations for recreation, water supply, and flood con- maintain long-term historical operations patterns, even with a
trol were defined by setting requisite volumes for the inactive zone, changed hydrology. This is the basis for the WYIM, which tries
buffer zone, and conservation zone of reservoirs (Yates et al. 2005). to maintain historical operations. A major limitation of the WYIM
A maximum conservation zone level, or rule curve, is used to create is that it does not account for the long-term changes in runoff
flood space. Rule curves are from public documents describing timing expected with warming; this is accounted for by the SDM.
operations objectives (e.g., for flood control operations) and from The WYIM modifies the approach of Mehta et al. (2011) for
observed reservoir data and knowledge about broader operational quicker application to many hydropower plants across a region us-
objectives within specific hydropower systems. ing a water year index. Whereas Mehta et al. (2011) assumes a step-
Lake evaporation is estimated using the Penman equation wise, discrete relationship between regional water availability, the
(Penman 1948), modified by Van Bavel (1966) and Kohler and WYIM assumed a continuous, linear relationship. Annual regional
Parmele (1967), as described by Dingman (2002). Meteorological water availability is approximated using a supraregional water year
data is from Daymet (daymet.ornl.gov), as described and used by index (WYI).
Young et al. (2009). Surface area is derived from reservoir volume- Energy demand E for a powerhouse can be represented with an
elevation curves. expression that includes variable percent α of constant energy
generation capacity Emax :

Hydropower Et ¼ αt Emax ð1Þ


Hydropower systems are difficult to model correctly. Multireser- where t is the time step. The percent αt is calculated for each time
voir hydropower systems are best modeled with multireservoir op- step and each powerhouse using historical observations. For each
erations optimization methods (Labadie 2004; Rani and Moreira time step, a linear fit is calculated from historical flow and WYI
2010), reflecting operational goals of maximizing profit (Jacobs values for the time step. The slope mt and intercept bt of the linear
et al. 1995) and, therefore, minimizing spill (Madani and Lund fit are used to calculate αt :
2009). However, simulation models, including an optimization-
based simulation such as WEAP, are still widely used, especially mt · WYI þ bt
at ¼ ð2Þ
in planning and management studies (Rani and Moreira 2010) Qmax

716 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014

J. Water Resour. Plann. Manage. 2014.140:714-723.


where Qmax = maximum turbine capacity. In implementation, that historically occurred during the wet season for later generation
Eq. (2) is modified to ensure that 0 ≤ at ≤ 1, except for pumped during the dry period.
storage plants, where αt can be negative to allow reverse flows.
The California Department of Water Resources (DWR) Sacra- Spill Demand Method
mento Valley WYI (WYISac ) and San Joaquin Valley WYI (WYISJ )
were used to approximate water availability for the northern water- One important limitation of the WYIM is a lack of adaptation to
sheds (Feather through American) and southern watersheds changes in runoff timing due to warming. The WYIM implicitly
(Mokelumne through Kern), respectively. These WYIs, estimated hedges during the wet season, by operating hydropower plants
by DWR from inflows to several regional rivers, are continuous and at less than capacity, based on historical inflow patterns, electricity
have units of million acre-feet (MAF) per year (1 MAF ¼ 1.23× prices, and so on. With greater precipitation-driven winter runoff,
109 m3 ) (CDEC 2010). The time series for WYISac and WYISJ the WYIM would hedge too much, generating at less than capacity
were calculated for each atmospheric warming scenario using even when there is spill, which is undesirable (Jacobs et al. 1995).
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

simulated runoff from Young et al. (2009) and estimates of The SDM explicitly accounts for this.
Sacramento River runoff. Historical powerhouse flows included To ensure that hydropower generation is always prioritized over
all available data from 1970 to 2008, obtained from USGS gauges spill, the SDM requires that reservoir inflows in excess of existing
and water utilities. Available period-of-record data ranged from 7 to downstream demands (storage, hydropower, water supply, and
39 years, with a median of 28 years. environmental releases) be released to generate hydropower. The
Energy generation capacity Emax is assumed constant in all SDM is implemented by setting Qsd equal to 100% of turbine flow
high-head hydropower plants, such that capacity, though with a priority lower than other local demands.
This method is used for all hydropower plants but for three dis-
Emax ¼ ρghηQmax ð3Þ tinct purposes. First, the SDM ensures that the WYIM does not
result in overhedging for peaking plants when inflows are high.
Second, the SDM ensures that hydropower plants that lack up-
where ρ = density (assumed 1,000 kg=m3 ); g = gravitational con- stream storage always divert as much as possible, constrained only
stant (9.81 m=s2 ); h = fixed hydropower head; η = plant efficiency by water availability, in-stream flow requirements (IFRs), and
(assumed 90%); and Qmax = hydropower turbine flow capacity. facility capacities. Finally, the SDM ensures de facto coordination
Plant head and capacity are from publicly available documents, among plants operated in a series.
such as Federal Energy Regulatory Commission (FERC) licenses,
and observed data, primarily USGS gauge data.
Fig. 2 demonstrates the WYIM, including its strengths and Water Demand
inherent limitations, by showing the relationship between WYISJ Diversions for municipal and agricultural water demand is limited
and hydropower turbine flow for Big Creek No. 1 powerhouse above the large rim dams, though they can be important because
(San Joaquin watershed) for two weeks of the year: July 25–31 they have a higher priority than hydropower and play a central role
and November 7–13. As with most hydropower plants below large in some systems (e.g., the Hetch Hetchy system in the Tuolumne R.
reservoirs, the linear relationship between WYI and turbine flow is watershed). Demands are modeled based on historical mean weekly
stronger when reservoirs are full and water is more plentiful flows, using data provided by water agencies, publicly available
(e.g., July) and weaker during drier weeks, when reservoirs are gauge data, or descriptions of operations from agency documents.
empty or refilling (e.g., November). During wet periods, hydro- All demands are independent of water year type, except for deliv-
power and other users can take as much as is available, even in eries to San Francisco from the Hetch Hetchy system (Tuolumne R.
drier years. During dry periods, however, hydropower operators watershed). Because historical diversions to San Francisco are neg-
are more selective, basing release decisions on many factors other atively correlated with WYISJ , they were determined by regressing
than just water availability, such as wholesale electricity prices, pri- historical annual diversions against WYISJ , with fixed intra-annual
ces of other energy sources, air temperature at load centers, hydro- delivery patterns.
power operation type, and agreements with other users. Results
from this method reflect the relative consistency in storage hedging
In-Stream Flow Requirements
In-stream flow requirements (IFRs) represent minimum flow
25.0 requirements below dams and diversions. Most IFRs identified
Hydropower turbine flow (m 3/s)

July 25-31 Qt = 1.32WYISJ + 9.38 in regulatory documents such as FERC licenses are included, with
R² = 0.57
20.0 more complex, short-term IFRs, such as occasional pulse flows and
Nov. 7-13
recreation flows, omitted. Many IFRs depend on facility-specific
15.0 or regional WYTs. Most WYT definitions use a combination of
year-to-date flows and flow forecasts for the remainder of the water
10.0
Qt = 0.99WYISJ + 2.15 year. For expediency, WYTs were computed assuming perfect
R² = 0.12 foresight of the water year type using the unimpaired hydrology
data, without forecasting.
5.0

0.0 Priority Setting


1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
WEAP allocates water in part based on assigned priorities, with
San Joaquin Valley WYI (billion m 3) allocations first to higher priority uses within each time step. To
ensure that upstream, nonconsumptive demands are met before
Fig. 2. Historical San Joaquin Valley water year index (WYI) and Big
downstream demands, upstream demands are assigned higher
Creek No. 1 powerhouse turbine flow for the weeks July 25–31 and
priorities than downstream demands. Demand types are generally
November 7–13
prioritized in order from highest to lowest priority as follows:

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014 / 717

J. Water Resour. Plann. Manage. 2014.140:714-723.


(1) IFRs, (2) water supply demand, (3) hydropower, and hydrologic model of the Sierra Nevada calibrated with the spatial
(4) reservoirs. and temporal resolution needed for weekly scale accuracy of all
Some priorities were adjusted as needed to reflect real system facilities.
operations, such as assigning a higher priority during the refill
period for some reservoirs. Hydropower plants immediately below Climate Change Representation and Sensitivity
a reservoir, which used the WYIM, have priorities equal to the res- Climate change was represented as uniform, homogeneous in-
ervoir. Per the SDM, lower elevation hydropower plants in a series creases in air temperature over a historical climate with precipita-
have priorities lower than the upstream-most plant. tion unchanged. Historical variability also remained unchanged.
Actual climatic changes are expected to be heterogeneous and
nonuniform and will include changes in magnitude, timing, and
Inflow Hydrology variability of temperature and precipitation (Franco et al. 2011).
SIERRA represents inflows as artificial tributaries to real river This inhibits the ability to make more advanced analyses related
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

locations. For this study, the water years (WY) 1981–2000 from to changes in the climate variability expected with warming.
the weekly inflow data described by Young et al. (2009) were used, Additionally, SIERRA is sensitive to climate change only though
modified for the CABY region by Mehta et al. (2011). This period hydrologic processes. However, climate warming will also affect
includes representative recent historical runoff conditions, includ- other management drivers, such as land use disturbance (Miller
ing an extended drought (1987–1992), the wettest year on record et al. 2009).
(1983), and the flood year of record (1997). Young et al. (2009)
used WEAP to estimate runoff in the western Sierra Nevada by Rim Dams
simulating hydrologic processes within “catchments” consisting Operation of rim dams can affect upstream operations in some
of homogeneous physical characteristics and meteorological con- cases but are not currently present in SIERRA. Inclusion of rim
ditions (Yates et al. 2005). Catchments were defined by Young et al. dams and upstream flood-control operations would enable a better
(2009) by intersecting “subwatersheds” with 250-m elevation understanding of flood risk and control, including the possibility of
bands. Subwatersheds were defined by locations of management utilizing higher reservoirs for some flood control, as is already done
interest (“pour points”), such as dam or flow gauge locations. in some instances (Hickey et al. 2003).
Flows from each catchment within a subwatershed were aggregated
to determine the subwatershed contribution to flow at each pour
point. Young et al. (2009) used spatially interpolated historical Calibration and Model Assessment
(WY 1980–2000) Daymet (daymet.ornl.gov) climate data. They Parameters and operating rules used were derived from historical
calibrated their model at the watershed scale using monthly flows data, as described previously, so calibration was limited to adjust-
at the outlets of 13 of the 15 watersheds, omitting BAR and ing priorities, particularly for reservoirs in series or parallel in com-
CAL for lack of observed data, and for snow water equivalent at plex systems. Priorities were adjusted based on a visual comparison
15 high-elevation locations. of observed and modeled differences in storage among reservoirs.
No adjustments were made to the inflow hydrology data set.
Climate Change Scenarios The performance of turbine flow for individual power plants and
hydropower generation for each watershed was assessed following
To represent climate change, this study used the runoff data from
the guidance of Moriasi et al. (2007), who reviewed a range of met-
Young et al. (2009), who applied a sensitivity analysis approach
rics for simulating watershed processes. For each of mean seasonal
(Wilby et al. 2009) to calculate effects of regional climate change
and annual turbine flow for power plants and total hydropower gen-
on runoff. Specifically, Young et al. (2009) considered uniform in-
eration for watersheds, the authors calculated the Nash-Sutcliffe
creases in air temperature of 0°C, 2°C, 4°C, and 6°C, approximating
model efficiency index (NSE) (Nash and Sutcliffe 1970) and
regional warming through 2100 (Franco et al. 2011) and no change
root-mean-square error standardized by the standard deviation of
in regional precipitation. Potential changes to other, nonmeteoro-
the observed series (RSR) (Moriasi et al. 2007). The NSE indicates
logical factors, such as land use disturbance (Miller et al. 2009)
how well the model performs compared to the historical mean and
and electricity demand patterns (Miller et al. 2008), were not
can range from −∞ to 1, with values exceeding zero indicating
incorporated.
better performance. The NSE is not a perfect predictor, necessitat-
ing additional metrics and visual assessments (Moriasi et al. 2007;
Limitations Jain and Sudheer 2008). The following criteria were used to
determine whether the model results can be considered “satisfac-
The key limitations to the methods and inputs are summarized to
tory” or better: seasonal NSE > 0.60, annual NSE > 0.70, seasonal
provide a better interpretation of model results and to suggest
RSR ≤ 0.60, annual RSR ≤ 0.50, and mBias within 15%. These
potential future improvements.
are more stringent than suggested by Moriasi et al. (2007), who
Hydropower Logic considered the monthly time step.
Hydropower operations rules represent the record of long-term Across the study region, SIERRA simulates mean seasonal
diversions rather than explicitly responding to fluctuations in hydropower turbine flow as shown in Fig. 3, where each point
resource availability and energy demand. Further, integration of in Fig. 3 represents a single powerhouse (eight powerhouses are
energy prices would allow a more appropriate sensitivity to warm- omitted for lack of observed data). In aggregate, the model tends
ing because warming affects energy prices through changing to under-represent diversions for hydropower in the summer (JAS)
demand (Miller et al. 2008). and fall (OND) and slightly over-represent diversions in winter
(JFM) and spring (AMJ). Table 1 summarizes turbine flow model
Inflow Hydrology performance metrics at multiple temporal scales, indicating that
The inflow hydrology data set calibration was limited to the larger most individual plants are less well simulated than in aggregate.
watersheds used here, limiting accurate representation of all facili- The most well-modeled hydropower plants also have the greatest
ties, as discussed in a subsequent section. There is currently no historical diversions and energy generation, which is partly

718 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014

J. Water Resour. Plann. Manage. 2014.140:714-723.


Table 1. Summary of Hydropower Turbine Flow Performance Metrics Table 2. Hydropower Generation Performance Metrics by Basin
NSE RSR mBias (%) NSE RSR mBias (%)
Non-exceedance (%) Seasonal Annual Seasonal Annual Seasonal Basin Seasonal Annual Seasonal Annual Seasonal
100 0.90 0.83 4.18 2.91 203 FEA 0.67 0.84 0.58 0.40 5
75 0.63 0.47 1.05 1.15 24 ABY 0.73 0.95 0.52 0.22 2
50 0.20 0.21 0.89 0.89 1 MOK −0.61 −0.39 1.27 1.18 −37
25 −0.11 −0.33 0.61 0.73 −14 STN 0.70 0.76 0.54 0.49 −11
0 −16.46 −7.49 0.32 0.41 −82 TUO 0.12 0.04 0.94 0.98 −29
SJN 0.74 0.86 0.51 0.37 −8
Note: Higher NSE is better; lower RSR is better; mBias closer to zero is
KNG 0.68 0.88 0.56 0.35 5
better.
KAW 0.68 0.73 0.57 0.52 −4
TUL 0.78 0.84 0.46 0.40 −9
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

reflected in Fig. 3. Conversely, the least well-modeled plants tend KRN 0.72 0.67 0.53 0.58 14
to be small, though small powerhouses were not necessarily poorly Note: Bold values indicate satisfactory or better.
represented. In general, the model better represented operations of
power plants with low storage capacities compared with total up-
stream runoff. Thus, small run-of-river power plants were generally of operations for one major reservoir (Salt Springs) and poor rep-
well simulated as were larger power plants. In contrast, small- to resentation of hydrology above an important power plant (Tiger
medium-size power plants below larger reservoirs were less well Creek). However, TUO and MOK are retained in subsequent analy-
modeled with the method used. ses because MOK contributes relatively little hydropower and TUO
In contrast to individual powerhouses, the model is generally, changes little with warming, resulting in little overall effect on ag-
though not entirely, satisfactory or better for representing hydro- gregate results.
power generation by watershed, indicating the watershed a suitable These assessments indicate that SIERRA is appropriate for use
unit for spatial analysis with this model. Table 2 lists watershed for watershed-scale hydropower assessments (exclusive of MOK
performance metrics for hydropower generation, whereas Fig. 4 and TUO) rather than for individual facilities, consistent with
compares mean seasonal observed and modeled generation by coarse inflow hydrology calibration (Young et al. 2009) and the
watershed. Based on the mentioned thresholds, simulation of all approximation method used for modeling hydropower. Further im-
watersheds except for TUO and MOK can be considered satisfac- provements are needed in inflow hydrology and operational logic to
tory or better. The KRN is less than satisfactory at the annual time simulate hydropower operations with greater spatial detail.
step. The poor representation of TUO and MOK is due to inherent
limitations in the generalized method used for representing hydro-
power operations in more operationally complex systems and the Results with Warming and Discussion
sensitivity of the model to local hydrology. The TUO is operated
primarily for water supply for San Francisco, rather than hydro- Hydropower generation sensitivity to warming were assessed in ag-
power, rendering the WYIM for hydropower less useful for gregate and by basin. Results described are best used as relative
TUO. Poor MOK performance is due to inadequate representation measures, due to the model limitations discussed. Systemwide, hy-
dropower generation increases in JFM and decreases in all other
seasons (data not shown), resulting in a total decrease of about
1,700 GWh, approximately 8%, with 6°C warming compared to
historical climate conditions. This represents approximately 1%
of California’s mean annual in-state electricity production and
5% of annual in-state hydroelectricity production for years
1990–2001 (Murtishaw et al. 2005).
Across most watersheds, hydropower generation decreases
compared with historical generation with long-term warming
(Table 3); changes by watershed vary substantially with near-
and medium-term warming (Fig. 5). However, there is a discernible
trend in changes latitudinally. Mean annual hydropower generation
decreases in the northern watersheds (FEA and ABY) under all
warming scenarios, by about 11% (FEA) to 15% (ABY) with
long-term warming. Exclusive of MOK and TUO, generation in
central watersheds changes relatively little compared to the
northern watersheds and even increases somewhat with lesser
warming (Fig. 5). For example, SJN sees only increases in gener-
ation, even with long-term warming. Generation trends with warm-
ing in the southern watersheds, which produce relatively little
energy compared to northern and central watersheds, are some-
where in between—generation decreases, but magnitude decreases
are small. These trends are generally both quantitatively and quali-
tatively consistent with results reported by others (Mehta et al.
2011; Rheinheimer et al. 2013; Vicuña et al. 2011), as discussed
previously.
Fig. 3. Mean observed and simulated mean hydropower turbine flow
Seasonal changes in hydropower generation within watersheds
by season
with warming (Fig. 6) combined with an understanding of specific

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014 / 719

J. Water Resour. Plann. Manage. 2014.140:714-723.


Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Mean observed and modeled hydropower generation by watershed

Table 3. Mean Annual Hydropower Generation Change with þ6°C


compensated for in other seasons, JFM in particular. Though
Warming by Watershed MOK and TUO are poorly modeled, the better modeled watersheds
also show this trend.
þ0°C þ6°C Change Change Throughout the Sierra Nevada warming causes less snowmelt-
Watershed (GWh=year) (GWh=year) (GWh=year) (%)
driven runoff in the late spring and early summer and greater pre-
FEA 3,827 3,392 −435 −11 cipitation-driven events in the winter (Null et al. 2010; Young et al.
ABY 5,736 4,889 −847 −15 2009). As this change in timing occurs, runoff initially becomes
MOK 693 625 −68 −10 more uniform throughout the year. Some hydropower systems, par-
STN 918 899 −19 −2
ticularly in the central watersheds, benefit from this increased uni-
TUO 1,256 1,261 6 0.5
SJN 4,267 4,311 44 1 formity with some warming (þ2°C and þ4°C) by being able to use
KNG 1,476 1,396 −79 −5 a greater percentage of inflows and spilling less. Rheinheimer et al.
KAW 45 40 −6 −12 (2013) also report a similar phenomenon for hydropower in the
TUL 38 35 −2 −6 Upper Yuba River watershed. Hydropower generation with greater
KRN 605 543 −62 −10 warming (þ6°C) is also influenced by changes in runoff timing, but
Total 18,861 17,391 −1,468 −8 in most cases it is dominated by magnitude reductions. The com-
bination of increased winter runoff, which increases winter spill,
and lower total annual runoff reduces annual generation with 6°
infrastructure explain the annual scale responses to regional climate C warming. The degree to which a hydropower system can benefit
change. In the north, generation decreases in JAS with little to from changes in runoff timing, such as in the central watersheds,
no compensatory increase in JFM generation. By contrast, in cen- depends on both changes in total magnitude and the current
tral watersheds reductions in JAS generation are consistently physical and operational configuration of the system. In SJR,

(a)

(b)

(c)

Fig. 5. (a) Absolute mean annual hydropower generation with warming by watershed; (b) absolute mean annual hydropower generation change with
warming by watershed; (c) relative mean annual hydropower generation change with warming by watershed; an asterisk indicates unsatisfactory
calibration metrics

720 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014

J. Water Resour. Plann. Manage. 2014.140:714-723.


(a)

(b)

(c)
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. (a) Absolute mean seasonal hydropower generation with warming by watershed; (b) absolute mean seasonal hydropower generation change
with warming by watershed; (c) relative mean seasonal hydropower generation change with warming by watershed; an asterisk indicates unsatis-
factory calibration metrics

for example, annual runoff magnitude changes are relatively small efforts should be made to reduce uncertainty to better understand
(Null et al. 2010), whereas the large Mammoth Pool Reservoir has the magnitude and nature of climate effects and adaptation poten-
existing extra storage capacity to be able to accommodate shifting tial. For example, there are indications that there may be greater
runoff patterns, resulting in minimal reductions in runoff with potential to adapt hydropower systems to climate warming in
þ6°C warming. the central watersheds of the Sierra Nevada, such that improve-
Limitations of the model, including inaccuracies in inflow hy- ments in hydrologic and operations models would be useful in
drology and operational logic, also contribute to the observed identifying that potential. Meanwhile, it appears that there is much
trends in changes with warming. The rainfall-runoff model used less adaptation potential in the northern watersheds. In this case,
was calibrated to watershed outlets and a few snow gauge locations reductions in uncertainty with improved modeling efforts can help
(Young et al. 2009), resulting in poorly simulated runoff in specific identify the most efficient ways to manage increased water scarcity.
locations within watersheds, which affects operations of specific
powerplants, as discussed previously. Operational logic, such as the
WYIM and SDM, also affects system response to changing inflows Conclusions
but was found to be less of an issue than other influences.
Though modeled long-term changes are greater than mean bias, In this study, the authors developed a water system operations
which translates to uncertainty in the magnitudes of changes pre- model (SIERRA), performed a basic performance assessment,
sented here, based on the combination of representation of seasonal and applied SIERRA to consider the sensitivity of seasonal hydro-
generation patterns, anticipated hydrologic changes, and knowl- power generation to regional climate warming. The hydropower
methods used are satisfactory for simulation of seasonal and annual
edge of historical operational patterns in specific locations, the au-
hydropower generation across larger watersheds, consistent with
thors anticipate these trends to be representative of potential future
the inflow hydrology model used, but require further refinement
conditions for most watersheds. Thus, while specific quantitative
for greater spatial and temporal specificity.
changes reported here would change with improved input and
These results also further the scientific basis for better under-
methods, these results provide a preliminary qualitative indication
standing the potential vulnerabilities of different water-use sectors
of how spatial and temporal patterns of hydropower in the Sierra
to anticipated regional climate change. However, further scrutiny of
Nevada will change with regional warming.
specific locations across the Sierra Nevada are needed to better
There are important water resources management implications
understand how regional climate change could affect local trade-
that stem from this study, particularly given the uncertainty present. offs between water uses. Improvements to SIERRA, with better
Modeling specific water systems over large areas in sufficient detail hydrologic representation and project-specific modeling, are neces-
to make actionable decisions remains elusive, due to the human sary to achieve the finer spatial and temporal resolution needed to
resources required to develop such detailed models. Thus, there re- make these kinds of assessments. Such improvements are also
mains a trade-off between modeling accuracy and spatial and tem- needed to answer key questions related to changes in climate vari-
poral resolution. There is uncertainty in all components of the ability and water year types expected with warming. Results from
climate change effects modeling spectrum, from world develop- this study indicate that improvements should be prioritized in the
ment assumptions, to general circulation models and regional northern watersheds to better understand and help resolve potential
downscaling, to hydrologic models, to water resources manage- water management conflicts likely to occur with the major de-
ment models, partly due to numerous approximations made at each creases in runoff magnitude and other hydrologic changes expected.
level. In the face of such approximations and both propagated and
accumulated uncertainty, there is still useful information from each
modeling level that can inform long-term water resources planning Acknowledgments
(Kiparsky et al. 2012).
In this study in particular, results indicate both where there may This work was funded in part by the California Energy Commis-
be greater adaptation potential within specific systems and where sion’s Public Interest Energy Research (PIER) program. We thank

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014 / 721

J. Water Resour. Plann. Manage. 2014.140:714-723.


the Pacific Gas & Electric Company and the San Francisco Public of the Intergovernmental Panel on Climate Change, M. L. Parry,
Utilities Commission for providing historical data, the Stockholm O. F. Canziani, J. P. Palutikof, P. J. van der Linden, and C. E. Hanson,
Environment Institute for providing and supporting WEAP and hy- eds., Cambridge University Press, Cambridge, U.K., 211–272.
drology data, and students Jason Emmons and Shannon Brown for Labadie, J. W. (2004). “Optimal operation of multireservoir systems: State-
their data-gathering and modeling assistance. Finally, we also thank of-the-art review.” J. Water Resour. Plann. Manage., 10.1061/(ASCE)
0733-9496(2004)130:2(93), 93–111.
the anonymous reviewers of this article for their thoughtful critique.
Lund, J. R. (2000). “Derived power production and energy drawdown rules
for reservoirs.” J. Water Resour. Plann. Manage., 10.1061/(ASCE)
0733-9496(2000)126:2(108), 108–111.
References Lund, J. R., and Guzman, J. (1999). “Some derived operating rules for
reservoirs in series or in parallel.” J. Water Resour. Plann. Manage.,
Barnett, T. P., Adam, J. C., and Lettenmaier, D. P. (2005). “Potential
10.1061/(ASCE)0733-9496(1999)125:3(143), 143–153.
impacts of a warming climate on water availability in snow-dominated
Madani, K., and Lund, J. R. (2009). “Modeling California’s high-elevation
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

regions.” Nature, 438(7066), 303–309.


hydropower systems in energy units.” Water Resour. Res., 45(9),
Bates, B. C., Kundzewicz, Z. W., Wu, S., and Palutikof, J. P. (2008).
W09413.
“Climate change and water. Technical paper of the Intergovernmental
Panel on Climate Change.” IPCC Secretariat, Geneva. Madani, K., and Lund, J. R. (2010). “Estimated impacts of climate warming
California Data Exchange Center (CDEC). (2010). “Chronological recon- on California’s high-elevation hydropower.” Clim. Change, 102(3–4),
structed Sacramento and San Joaquin valley water year hydrologic clas- 521–538.
sification indices.” 〈http://cdec.water.ca.gov/cgi-progs/iodir/wsihist〉 Mehta, V. K., et al. (2011). “Potential impacts on hydrology and hydro-
(June 1, 2010). power under climate warming of the Sierra Nevada.” J. Water Clim.
California Energy Commission. (2009). “2009 Integrated energy policy Change, 2(1), 29–43.
report.” Sacramento, CA. Miller, J. D., Safford, H. D., Crimmins, M., and Thode, A. E. (2009).
Cayan, D. R., Maurer, E. P., Dettinger, M. D., Tyree, M., and Hayhoe, K. “Quantitative evidence for increasing forest fire severity in the Sierra
(2008). “Climate change scenarios for the California region.” Clim. Nevada and southern Cascade Mountains, California and Nevada,
Change, 87(S1), 21–42. USA.” Ecosystems, 12(1), 16–32.
Dettinger, M. D. (2005). “From climate-change spaghetti to climate-change Miller, N. L., Hayhoe, K., Jin, J., and Auffhammer, M. (2008). “Climate,
distributions for 21st century California.” San Francisco Estuary & extreme heat, and electricity demand in California.” J. Appl. Meteorol.
Watershed Sci., 3(1). Climatol., 47(6), 1834–1844.
Dettinger, M. D., Cayan, D. R., Meyer, M., and Jeton, A. E. (2004). Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner, R. L.,
“Simulated hydrologic responses to climate variations and change in Harmel, R. D., and Veith, T. L. (2007). “Model evaluation guidelines
the Merced, Carson, and American River basins, Sierra Nevada, for systematic quantification of accuracy in watershed simulations.”
California, 1900–2099.” Clim. Change, 62(1–3), 283–317. Trans. Am. Soc. Agric. Biol. Eng., 50(3), 885–900.
Dingman, S. L. (2002). Physical hydrology, Prentice Hall, Upper Saddle Muñoz, J. R., and Sailor, D. J. (1998). “A modelling methodology for
River, New Jersey. assessing the impact of climate variability and climatic change on
Ficklin, D. L., Stewart, I. T., and Maurer, E. P. (2012). “Projections of 21st hydroelectric generation.” Energy Convers. Manage., 39(14),
century Sierra Nevada local hydrologic flow components using an 1459–1469.
ensemble of general circulation models.” J. Am. Water Resour. Assoc., Murtishaw, S., Price, L., de la Rue du Can, S., Masanet, E.,
48(6), 1104–1125. Worrell, E., Sathaye, J. (2005). “Development of energy balances for
Franco, G., Cayan, D. R., Moser, S., Hanemann, M., and Jones, M.-A. the state of California.” California Energy Commission, Sacramento,
(2011). “Second California assessment: Integrated climate change CA.
impacts assessment of natural and managed systems. Guest editorial.” Nash, J. E., and Sutcliffe, J. V. (1970). “River flow forecasting through
Clim. Change, 109(1), 1–19. conceptual models: Part I, A discussion of principles.” J. Hydrol.,
Franco, G., and Fagundes, M. (2005). “Potential changes in hydropower 10(3), 282–290.
production from global climate change in California and the western Null, S. E., Viers, J. H., and Mount, J. F. (2010). “Hydrologic response and
United States.” California Energy Commission, Sacramento, CA. watershed sensitivity to climate warming in California’s Sierra Nevada.”
Hayhoe, K., et al. (2004). “Emissions pathways, climate change, and PLoS ONE, 5(4), e9932.
impacts on California.” Proc. Natl. Acad. Sci. U.S.A., 101(34),
Olivares, M. A. (2008). “Optimal hydropower reservoir operation with
12422–12427.
environmental requirements.” Ph.D. dissertation, Univ. of California,
Hickey, J. T., Bond, M. V., Patton, T. K., Richardson, K. A., and Pugner,
Davis, CA.
P. E. (2003). “Reservoir simulations of synthetic rain floods for
Penman, H. L. (1948). “Natural evaporation from open water, bare soil, and
the Sacramento and San Joaquin River basins.” J. Water Resour.
grass.” Royal Soc. London Proc., 193(1032), 120–145.
Plann. Manage., 10.1061/(ASCE)0733-9496(2003)129:6(443),
443–457. Rani, D., and Moreira, M. M. (2010). “Simulation-optimization modeling:
Intergovernmental Panel on Climate Change (IPCC). (2007). “Climate A survey and potential application in reservoir systems operation.”
change 2007: Synthesis report. Contribution of working groups I, II Water Resour. Manage., 24(6), 1107–1138.
and III to the fourth assessment report of the Intergovernmental Panel Rheinheimer, D. E. (2011). “Modeling multi-reservoir hydropower systems
on Climate Change.” Geneva. in the Sierra Nevada with environmental requirements and climate
Jacobs, J., et al. (1995). “SOCRATES: A system for scheduling hydroelec- warming.” Ph.D. dissertation, Univ. of California, Davis, CA.
tric generation under uncertainty.” Ann. Oper. Res., 59(1), 99–133. Rheinheimer, D. E., Yarnell, S. M., and Viers, J. H. (2013). “Hydro-
Jain, S. K., and Sudheer, K. P. (2008). “Fitting of hydrologic models: A power costs of environmental flows and climate warming in
close look at the Nash-Sutcliffe index.” J. Hydrol. Eng., 13(10), California’s Upper Yuba River watershed.” River Res. Appl., 29(10),
981–986. 1291–1305.
Kiparsky, M., Milman, A., and Vicuña, S. (2012). “Climate and water: Sigvaldason, O. T. (1976). “A simulation model for operating a
Knowledge of impacts to action on adaptation.” Annu. Rev. Environ. multipurpose multireservoir system.” Water Resour. Res., 12(2),
Resour., 37(1), 163–194. 263–278.
Kohler, M. A., and Parmele, L. H. (1967). “Generalized estimates of Thompson, L. C., Escobar, M. I., Mosser, C. M., Purkey, D. R., Yates, D.,
free-water evaporation.” Water Resour. Res., 3(4), 997–1005. and Moyle, P. B. (2012). “Water management adaptations to prevent
Kundzewicz, Z. W., et al. (2007). “Freshwater resources and their manage- loss of spring-run Chinook salmon in California under climate change.”
ment.” Climate change 2007: Impacts, adaptation and vulnerability. J. Water Resour. Plann. Manage., 10.1061/(ASCE)WR.1943-5452
Contribution of working group II to the fourth assessment rep. .0000194, 465–478.

722 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014

J. Water Resour. Plann. Manage. 2014.140:714-723.


Van Bavel, C. H. M. (1966). “Potential evaporation: The combination Water Evaluation and Planning [computer software]. Stockholm Environ-
concept and its experimental verification.” Water Resour. Res., 2(3), ment Institute, U.S. Center, Somerville, MA.
455–467. Viers, J. H. (2011). “Hydropower relicensing and climate change.” J. Am.
Vicuna, S., and Dracup, J. (2007). “The evolution of climate change impact Water Resour. Assoc., 47(4), 655–661.
studies on hydrology and water resources in California.” Clim. Change, Wilby, R. L., et al. (2009). “A review of climate risk information for
82(3), 327–350. adaptation and development planning.” Int. J. Climatol., 29(9),
Vicuna, S., Dracup, J. A., Lund, J. R., Dale, L. L., and Maurer, E. P. (2010). 1193–1215.
“Basin-scale water system operations with uncertain future climate Yates, D., Sieber, J., Purkey, D., and Huber-Lee, A. (2005). “WEAP21: A
conditions: Methodology and case studies.” Water Resour. Res., demand-, priority-, and preference-driven water planning model: Part 1,
46(4), W04505. Model characteristics.” Water Int., 30(4), 487–500.
Vicuña, S., Dracup, J. A., and Dale, L. (2011). “Climate change impacts on Young, C. A., et al. (2009). “Modeling the hydrology of climate change in
two high-elevation hydropower systems in California.” Clim. Change, California’s Sierra Nevada for subwatershed scale adaptation.” J. Am.
109(1), 151–169. Water Resour. Assoc., 45(6), 1409–1423.
Downloaded from ascelibrary.org by University of California, Berkeley on 04/16/14. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT © ASCE / MAY 2014 / 723

J. Water Resour. Plann. Manage. 2014.140:714-723.

You might also like