Office of Water
Environmental Protection Agency
August 2010
E PA 800-R-10-001
Climate Change Vulnerability Assessments:
A Review of Water Utility Practices
-------
Acknowledgements
This report was prepared for EPA by Stratus Consulting Inc., Boulder, CO, under subcontract to
Abt Associates, Inc., under EPA Contract EP-C-07-023, Work Assignment 2-13. Karen Metchis
served as Technical Project Officer and principal reviewer of this document.
EPA wishes to acknowledge the principle authors, Jason M. Vogel and Joel B. Smith for their
diligence and expertise in preparing this report. Further thanks go to EPA ORD project officer,
Jill Neal, and ORD scientists Jeffrey Yang, Ph.D., and Thomas Johnson, Ph.D., for their
comments and suggestions in helping shape this report.
This report represents one product resulting from the 'First National Expert Workshop on Water
Utilities and Climate Change," held January 20-21, 2009, in Washington, D.C.
Acknowledgement also goes to Elizabeth Corr, Robert Cantilli, EPA Office of Water, and Jeff
Yang, Ph.D., ORD, for their collaboration in formulating the workshop.
Notice
The U.S. Environmental Protection Agency, through its Office of Water and Office of Research
and Development, funded and managed the development of this report. Any opinions expressed
in this report are those of the author(s) and do not necessarily reflect the views of the Agency,
therefore, no official endorsement should be inferred. Any mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
-------
Contents
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 2
3. APPROACHES TO ASSESSING CLIMATE CHANGE VULNERABILITY 4
3.1 Top-down Modeling Assessments 5
3.1.1 Scenario analyses 6
3.1.2 Sensitivity analyses 6
3.1.3 Paleoclimate or historic analyses 7
3.2 Bottom-up Threshold Analyses 7
3.3 Policy Environment as a Constraint 8
4. SOURCES OF CLIMATE INFORMATION 9
4.1 The Instrumental Record 10
4.2 Paleoclimate Data 10
4.3 Literature Reviews 11
4.4 Climate Projections 12
4.4.1 Climate scenarios 12
4.4.2 Global climate models 14
4.4.3 Downscaling 16
4.4.4 Combining methods: Climate projections with paleoclimate data 18
5. MODELING CHANGES IN WATER RESOURCES 20
5.1 Water Quantity 20
5.2 Water Demand 21
6. SUMMARY 23
7. RECOMMENDATIONS FOR FURTHER STUDY 25
8. REFERENCES 26
-------
1. Executive Summary
Climate change poses a variety of challenges for water management, and there is a need to
develop methods for understanding and managing risk. While much has been written about the
projected impacts of climate change at the continental or regional scale, scientists are quick to
caution decision makers about using projections based on global circulation models (GCMs) for
local decision making. This 'uncertainty' about specific impacts on local systems has raised
concern about the ability of water resource managers to plan for climatic and hydrological
changes at the local scale, and has spurred recent activity to develop methods for understanding
vulnerabilities, including how to downscale climate models.
This study examines and documents the steps taken by some of the leading utilities in an attempt
to identify the emergent characteristics of water utility climate change vulnerability assessments.
By examining the approaches taken and articulating the steps, information, and judgments
needed for such decision making, we hope to contribute to the collaborative problem solving
among the user and research communities who are working to further refine and validate such
procedures.
The study describes the activities of eight water utilities who have conducted climate
vulnerability assessments: East Bay Municipal Utility District, City of Boulder Utilities
Division, Denver Water, Massachusetts Water Resources Authority, New York City Department
of Environmental Protection, Portland Water Bureau, Lower Colorado River Authority, and
Seattle Public Utilities.
The following general observations can be made. First, while the intention was to evaluate both
drinking water and wastewater utilities in the systems evaluated, most vulnerability assessment
work focused on water quantity and water demand and in only one case on water quality.
Second, while utility managers typically possess expertise about their systems, their hydrological
and management models, and the local hydrology, they have more limited access to climate
change information. Utilities are able to examine impacts on their operations due to changes in
climate-sensitive variables (e.g., flow, demand), however, in many cases, in order to understand
the changes in climate that affect those variables, utilities are engaging outside climate expertise.
Third, in each case study presented in this report, utilities are compensating for the 'uncertainty'
about climate change by evaluating a range of scenarios or models, and in most cases the
resulting information is used to test the robustness of existing decision making or planning
against different future, plausible scenarios.
Climate change is a complex issue, and more work is needed to establish reliable practices for
incorporating climate change into water utility decisions and planning. At the same time,
utilities appear to have benefited from their various efforts to understand their potential
vulnerabilities and to evaluate long term planning options. Despite the uncertainty, there are
reasonable and prudent steps utilities and other water managers can take to better understand and
manage climate risk.
1
-------
2. Introduction
For many decades, water utilities have proactively assessed the ability of their systems to provide
a reliable supply of drinking water or adequate wastewater services under assumptions of
population change, municipal expansion (e.g., annexation), technological innovation, and
changes in regulations. These were the foreseeable challenges utilities addressed to ensure that
ratepayers received high quality water services at minimal cost.
Historically, utilities assumed stationarity of climate in their water resource planning - the idea
that natural systems fluctuate within an unchanging envelope of variability. Recently, however,
this assumption has been challenged, forcing managers to rethink future water resources
planning and management with regard to climate change (Milly et al., 2008). In response, over
the last decade, water utilities have expanded their risk assessment efforts to address changing
climate conditions.
Generally speaking, vulnerability assessments done by water utilities are grounded in a thorough
understanding of their water system. On a practical level, this means that vulnerability studies
use a variety of tools and models that have been developed and refined to reflect the local
hydrology, climate, infrastructure, operations, and demands that are unique to a particular water
utility. Consequently, no single tool currently exists that can comprehensively address the
vulnerability of diverse water systems to climate change. For example, most utilities have
operational or management models that estimate the day-to-day operation of their water or
wastewater systems. By necessity, these operational models are either unique to a particular
utility (e.g., a proprietary system model) or in some cases, customized versions of a tool used by
many utilities (e.g., the Water Evaluation and Planning, or WEAP, system model).
The purpose of this report is to illustrate the approaches used by water utilities to assess their
vulnerability to climate change. This is a review of best practices in this emerging effort across
the industry for the purpose of informing utilities considering engaging in this issue about the
various methods used by their peers. The report does not judge or evaluate the efforts of utilities
or the merits of different vulnerability assessment methods, but describes the efforts of eight
utilities as broad approaches, tools, and methods worthy of consideration by other utilities
depending upon their needs, available resources, and other factors.
In this report we examined the ways in which several water utilities evaluated their potential
vulnerability to climate change. Utilities included in this study were identified from published
studies on water resources and climate change, including six Water Research Foundation reports
(Strange et al., 2009; WRF, 2010a, 201 Ob, 2010c, 2010d, 201 Of), two reports by the Water
Utility Climate Alliance (Barsugli, et al., 2009; Means et al., 2010), and participants in the
ongoing Joint Front Range Climate Change Vulnerability Study (WRF, 2010e). We also
included utilities selected from the list of participants in two recent workshops focused on water
resources and climate change: the National Drinking Water Advisory Council's Climate Ready
Water Utilities Working Group and U.S. Environmental Protection Agency's (EPA's) First
National Expert and Stakeholder Workshop on Water Infrastructure Sustainability and
2
-------
Adaptation to Climate Change (U.S. EPA, 2009).l In total, we identified 50 water utilities as
potential candidates for in-depth case studies. Initial review, however, revealed that many of the
selected utilities were interested in tracking the climate change issue and learning what others
had done, but had not yet engaged in their own climate change vulnerability assessments. If this
was the case, the utility was eliminated from the initial list for further study.
We identified five utilities that had published reports on their own climate change vulnerability
work, and another five utilities that appeared to have conducted vulnerability work but had not
published the modeling results. Because of time and resource limitations, the study was limited
to eight of these 10 utilities. The two utilities not included were Inland Empire Utilities Agency
and King County Wastewater Treatment Division. The eight utilities included in the assessment
included:
> East Bay Municipal Utility District (EBMUD)
n A water supply and wastewater utility serving 1.3 million customers.
> City of Boulder Utilities Division
° A water supply and wastewater utility serving 113,000 customers.
> Denver Water
n A water supply utility serving 1.3 million customers.
> Massachusetts Water Resources Authority (MWRA)
° A water supply and wastewater utility serving 2.2 million customers.
> New York City Department of Environmental Protection (NYCDEP)
n A water supply and wastewater utility serving 9.2 million customers.
> Portland Water Bureau (PWB)
n A water supply utility serving 860,000 customers.
> Lower Colorado River Authority (LCRA)
n A conservation and reclamation district that manages water supply along a 600-
mile stretch of the Colorado River in Texas, operates six dams, and helps
communities plan and coordinate their water and wastewater needs.
> Seattle Public Utilities (SPU)
n A water supply utility serving 1.35 million customers.
Finally, we studied these eight utilities' climate change vulnerability analyses in depth to identify
tools and approaches used by those water utilities in their assessments. The remainder of this
report synthesizes the insights from the analysis of these eight utilities in the following sections:
approaches to assessing climate change vulnerability, sources of climate information, modeling
changes in water resources, summary, and recommendations for further study.
1. A forthcoming report from the Water Research Foundation (WRF, 2010d) offers case study information
relevant to this report, but the final draft was not publicly available for inclusion in this report.
-------
3. Approaches to Assessing Climate
Change Vulnerability
Most water managers engaged in climate vulnerability analyses have a strong technical
understanding of their water systems, including local hydrology, historical operating conditions,
and standard operational practices.2 However, they typically are not climatologists and have
limited experience assessing risks from climate change. As a result, a water utility can either
engage outside expertise to do a sophisticated exploration of the implications of climate change
or they can make do with the information and capabilities readily available. Utilities that choose
not to engage outside experts are generally limited to analyses based on their internal
understanding of their water system and/or sensitivity analyses to assess what hypothetical
climate changes would mean for their water system. In contrast, utilities that engage outside
expertise can generate climate projections specific to their watershed. This more resource-
intensive approach enables computationally sophisticated vulnerability analyses.
Approaches to assessing climate change risks are generally classified as either "top-down"
modeling assessments or "bottom-up" threshold analyses (e.g., Miller and Yates, 2006; Freas
et al., 2008; Stratus Consulting and MWH Global, 2009). This categorization is useful, and,
generally speaking, utilities choose one of these two approaches to initiate their climate
vulnerability assessment. However, it should be noted that these two approaches are not mutually
exclusive and utilities often incorporate elements of both, either in series or in parallel. The
choice of one approach over another, however, has significant implications for study design and
resource needs, making them useful as categorical descriptions. The following utilities initially
used a top-down assessment:
> City of Boulder Utilities Division
> Denver Water
> MWRA
> NYCDEP
> PWB
> SPU
The following utilities initially used a bottom-up assessment:
> EBMUD
> LCRA
In general, top-down assessments are model and data driven. They often are more time and
resource intensive than bottom-up assessments, sometimes requiring expertise beyond the
2. Furthermore, most utilities take land use changes, population projections, and economic development into
account in infrastructure design and operational planning. Climate, however, often gets lumped in with
hydrological considerations and does not rise to the level of an independent planning consideration.
-------
capacity of many utilities. In general, bottom-up assessments are driven by knowledge of the
utility system itself. Such assessments often focus on defining critical system thresholds or
specific decisions that may be sensitive to climate change. This can often be done qualitatively,
and the necessary expertise often resides within the utility. As suggested above, these two
approaches are not mutually exclusive.
3.1 Top-down Modeling Assessments
Six utilities in our study initiated their vulnerability assessments with a top-down approach:
Boulder Utilities (Smith et al., 2009), Denver Water (Laurna Kaatz, personal communication,
April 8, 2009), MWRA (Stephen Estes-Smargiassi, personal communication, February 26,
2010), NYCDEP (2008; Major et al., 2007), PWB (Palmer and Hahn, 2002), and SPU (Palmer,
2007). This method projects future climate conditions specific to the watershed of concern and is
the most computationally and resource intensive of the analysis approaches used by utilities.
Because of this, the top-down modeling assessments done by the utilities examined for this
report were often completed by the water utility in conjunction with an academic institution or
consulting firm with expertise in climate modeling. For example, SPU and PWB partnered with
the University of Washington's Climate Impacts Group (UW-CIG), NYCDEP partnered with the
City University of New York Institute for Sustainable Cities (CISC) and Columbia University's
Center for Climate Systems Research (CCSR), LCRA used CH2M Hill as a consultant, and the
City of Boulder Utilities Division engaged Stratus Consulting and Hydrosphere (now AMEC).
Top-down assessments can involve relying on historic hydrology, using paleoclimate records,
engaging in literature reviews, or they can be done using climate projections. All these options
are discussed in more detail in Section 3 (Sources of Climate Information). By far, the most
common form of top-down assessment involves generating downscaled climate projections for a
utility's local watershed using global climate model (GCM) output. To do this, the analysis
generally includes assumptions about future emission levels (i.e., global socioeconomic
development scenarios), the effects of those emission levels on global climate (e.g., temperature
and precipitation), the translation of global climate effects to the region or watershed, and the
specification of climate changes on hydrology (e.g., streamflow), as well as operational,
management, and demand models.
All of the utilities we studied in depth assessed a number of climate variables; however, the most
common climate variables of interest were those that fed into their watershed-specific hydrology
models or system-specific management or operations models. Most often, this included
temperature and precipitation data or streamflow data at the spatial and temporal scales of each
individual utility's hydrology, management, operations, or other models.3 Therefore, to make
meaningful comparisons in this study, we focused our investigation of the different modeling
approaches to develop projections of temperature and precipitation. It is worth noting, however,
3. This conclusion was also reached in a study of the Water Utility Climate Alliance utilities (Barsugli et al.,
2009).
5
-------
that some utilities also investigated issues of special concern for their water system. For example,
NYCDEP was concerned about source water quality, especially due to heavy precipitation events
and resulting turbidity, and MWRA was concerned about likely impacts on the safe yield of its
sources.
Top-down modeling has three important subcategories: scenario analyses, sensitivity analyses,
and paleoclimate data or historic climate observations to define temperature and precipitation
patterns for water system planning purposes. Each of these subcategories is described in detail in
the following sections.
3.1.1 Scenario analyses
The scenario approach to a top-down assessment begins by defining at least one, but more
commonly two or three plausible greenhouse gas (GHG) emission futures that merit
consideration by a water utility. A scenario analysis then works step by step through GCMs,
downscaling those results,4 and finally using those results in water system specific hydrology,
demand, operational, and/or management models as appropriate to investigate the vulnerability
of a water system to each scenario. Section 3.4 (Climate Projections) discusses these individual
steps in greater detail. In many cases, scenarios are developed to capture a wide range of
plausible futures by incorporating an extensive range of models, emissions scenarios, and
projected demands. As an example, the SPU study applied the scenario analysis technique and
used GCMs and socioeconomic projections to assess a "middle of the road," a "warmer and
wetter," and a "warmer and drier" scenario. MWRA's approach used the outcome of all GCMs
to estimate the probability distributions of likely changes in temperature and precipitation and
then utilized those distributions in the statistical downscaling. This methodology, developed at
NCAR, avoids using a scenario tied to a single GCM.
3.1.2 Sensitivity analyses
Another form of the top-down approach uses a sensitivity analysis, incorporating the use of
incremental changes in climate such as 1°, 2°, and 3°C (1.8°, 3.6°, and 5.4°F) annual temperature
increases combined with +/- 0%, 10%, or 20% annual changes in precipitation.5 PWB used this
type of sensitivity analysis to bracket a plausible range of climate-altered future hydrology. But
utilities can also engage in a sensitivity analysis, as was done by EBMUD, with a bottom-up
4. The term "downscaling" has come to mean the use of higher resolution regional climate models (RCMs;
dynamic downscaling) or complex statistical approaches (statistical downscaling), but in this case can also
refer to simple downscaling, which typically involves adding the temperature increase in a GCM grid to
observed temperatures for stations in the grid box and multiplying the percentage change in precipitation by
the observed precipitation record.
5. Sometimes sensitivity analyses use arbitrary increments, but they can also be informed by the scientific
literature. Sensitivity analyses can use different changes in temperature and precipitation by season, but this
approach was not used in the vulnerability analyses investigated for this report.
-------
approach to avoid the computationally and resource intensive steps of a scenario analysis. A
sensitivity analysis does not require using GHG emissions projections, using climate model
projections, or downscaling climate model output.
3.1.3 Paleoclimate or historic analyses
A third top-down approach incorporates paleoclimate studies or historic climate observations to
define temperature and precipitation patterns for water system planning purposes. The method
often defines a worst case scenario (e.g., three consecutive years of the drought of record) or
examines the water system effects of droughts in the paleorecord outside of observed variability.
While the City of Boulder Utilities Division, Denver Water, and EBMUD used paleoclimate
studies to complement their climate change vulnerability studies, none of them relied exclusively
on paleoclimate or conservative extrapolations from historic data.6
3.2 Bottom-up Threshold Analyses
The bottom-up approach to climate change vulnerability assessments is grounded in knowledge
of the water system itself. Using this approach, a qualitative system assessment is done to
determine which system components are potentially vulnerable to change. The results of the
assessment can be used to focus further study on specific impacts of concern.
EBMUD, which used this approach, stated, "In a 'Bottom-Up' approach, the most critical
vulnerabilities of the District's water supply system are identified, the causes of those
vulnerabilities are articulated, and then steps are taken to better address and solve the
vulnerability in the face of climatic uncertainty" (EBMUD, 2009, p. 4-16). They noted their
concept of a bottom-up approach is adapted from the AwwaRF (now the Water Research
Foundation) publication, Climate Change and Water Resources: A Primer for Municipal Water
Providers (Miller and Yates, 2006).
More specifically, the EBMUD research approach involved a portfolio evaluation that first
identified potential portfolio components (e.g., new reservoirs, expanded reservoir storage,
increased conservation, conjunctive use, water reclamation, desalination, interbasin transfers),
screened those components for technical, environmental, and economic feasibility, and then
constructed alternate portfolios of multiple components that could meet projected demands
(e.g., increased conservation and conjunctive use, or water reclamation and interbasin transfers).
Then, a preliminary portfolio analysis was conducted using a combination of the WEAP system
model and the district's EBMUDsim model - known collectively as the "W-E model." Portfolios
that performed poorly under current hydrological conditions were eliminated and the remaining
portfolios were subjected to detailed analyses under anticipated climate change conditions using
6. The lack of evidence for sole reliance on these types of analyses in the cases investigated for this report is
largely a product of the case selection methodology, which focused on utilities with a published record of
sophisticated vulnerability work.
7
-------
the W-E model. By screening out portfolios that perform poorly under current conditions,
EBMUD implicitly limited options to those that could be implemented under current conditions
without significant reduction of reliability, but which presumably would improve the system
under climate change relative to the current system. By comparing the climate change scenarios
to a baseline scenario, the sensitivity of the EBMUD water system to each component was
assessed to identify critical vulnerabilities and identify portfolios that addressed those
vulnerabilities.
LCRA also used a bottom-up approach, and in CH2M Hill's report for the authority, the
threshold approach was defined in this way: "This threshold approach identifies system
components that are dependent on the status of climate variables (e.g., precipitation, temperature,
etc.) and the overall system risk to climate change, resulting in a preliminary risk assessment
based on the professional judgment of experts who know the system and the planning area"
(CH2M Hill, 2008, p. 3-2). This approach is a qualitative or semi-quantitative analysis that
consisted of (1) identifying the climate variables of importance and exploring the sensitivity of
LCRA to these variables; (2) determining water system responses to a range of potential climate
changes; (3) assessing the vulnerability of LCRA to climate change impacts; (4) assessing
system performance according to the uncertainty associated with climate change factors driving
LCRA vulnerability; and (5) evaluating overall system risk and identifying areas in need of
further analysis.
3.3 Policy Environment as a Constraint
A key constraining factor affecting any utility's ability to assess climate change vulnerability is
the level of support of the local community about climate change, in particular utility ratepayers,
boards of directors, and locally elected officials. In many circumstances, it is not feasible for a
utility to engage in a "climate change" vulnerability study because there is no political support
for the effort. Therefore, the policy environment directly impacts a utility's approach to climate
change vulnerability assessment. For example, New York City Mayor Michael Bloomberg has
taken climate change seriously during his tenure, providing political support for climate change
adaptation efforts. As a result, NYCDEP has been able to do sophisticated analyses and make
long-term investments to assess potential risks. Many water resource managers in less supportive
political environments, however, do not have the political support needed to take on climate
change directly or explicitly and must use historic or paleoclimate data rather than downscaled
climate model data to justify operational changes or investments that require a rate increase. If
support does not exist for climate studies, a water manager might elect to do a sensitivity
analysis as a lower-profile way to understand the potential vulnerabilities of their water system to
changes in climate conditions - climate change or climate variability.7
7. Water utilities might also select these methods for other reasons, such as a lack of funding or because
uncertainty in the projections does not yield enough value added to justify the effort of an intensive climate
change vulnerability assessment.
-------
Vulnerability to changing environmental conditions is often incorporated into contemporary
utility planning efforts even if those efforts are not billed as climate change vulnerability
assessments. Indeed, during research for this report, we discovered a utility that reported taking
sea level rise into account in their infrastructure design as far back as 1989, but without
advertising the analysis as "climate change." In summary, the policy environment in which a
utility operates may dictate what approaches to climate vulnerability assessment are acceptable,
how such studies are conducted, what data sources are used, and other important aspects of
vulnerability assessment design.
4. Sources of Climate Information
Regardless of how a climate change vulnerability assessment is framed, climate information that
operates at the spatial and temporal scale of a utility's hydrology, planning, and/or operational
models is almost always used in vulnerability assessments.8 Most commonly, temperature and
precipitation estimates at daily, weekly, or monthly timesteps are input into utility-specific
hydrology, demand, operations, and/or management models to assess the effects of different
climate conditions on water supply or wastewater services.9
Temperature and precipitation estimates can be derived from a number of sources, including the
instrumental record, paleoclimate studies of variability beyond the historical record, literature
reviews, and downscaling GCM projections to utility-specific watershed(s) to generate projected
future climate conditions. These sources of climate information have different associated
logistical and resource constraints and each is explored in more depth below. The key criterion
for selecting a source of climate information is that the approach and the data can be justified,
sometimes only to a technical audience, but increasingly to ratepayers, boards of directors, and
elected officials as well.
Utilities need credible climate information in order to consider climate change impacts on their
water systems. Historically, most utilities that address climate change issues have relied on
partnerships with academic and research institutions to help them select appropriate climate data,
however, many utilities are increasingly turning to private consulting companies to help them
with this task. Small- and medium-sized utilities, however, often lack the financial resources
needed to engage consultants to provide technical guidance on climate change vulnerability
studies and their ability to partner with academic institutions is not clear. The scalability of the
climate information approaches described below must be carefully considered for potential
8. Note that climate information includes more than GCM output, including historic weather station data,
stream gauge data, paleoclimate reconstructions, and more. However, when GCM output is used, it typically
has to be downscaled or assumptions made to reduce the temporal and spatial resolution to be usable in utility
hydrology or operational models.
9. Climate models often provide monthly average changes that must then be combined with the observed
record at a daily timestep to generate projected daily changes in climate.
-------
applicability beyond the relatively large and highly-resourced water utilities that have led the
movement to consider climate change explicitly in their infrastructure and operational planning.
4.1 The Instrumental Record
The instrumental record is often used to define utility-specific models for demand, hydrology,
operations, and management. These models can be simple statistical relationships or complex
physical models of local hydrology that incorporate vegetation cover, aspect, slope angle, and
soil type in addition to precipitation and temperature. Regardless, the instrumental record forms a
fundamental basis for most utility vulnerability assessments in one form or another.
Utility vulnerability assessments have historically used the drought of record or flood of record
to assess the vulnerability of a water system by applying the most extreme climate conditions on
record to contemporary circumstances. In many cases, this type of analysis can be useful for
identifying current vulnerabilities of a water system to current climate. ° Although none of the
utilities discussed in this report relied exclusively on this assessment methodology, many utilities
still routinely rely on this method due to its familiarity and ease of use.11 It is important to note
that using only the instrumental record implicitly retains the assumption of climate stationarity -
fluctuation within an unchanging envelope of variability - as observed typically within the last
century or less.12
4.2 Paleoclimate Data
Some researchers have begun developing proxy records for precipitation and streamflow based
on tree rings and geological evidence.13 These researchers have found evidence of dry periods
prior to the 20th century that were more persistent and perhaps more intense than what is
observed in the instrumental record (e.g., Woodhouse and Lukas, 2006a). Because natural
climate variability can often be quite large compared to changes projected from climate change,
some utilities have used paleoclimate data as a complement to their climate vulnerability
assessments.
10. With a changing climate, there should be more caution about interpreting the historic record. What may
have been a 100-year drought or flood calculated from the historic record may shift significantly with
projected climate change, or even by adding the paleorecord to historic data.
11. For example, both Contra Costa Water District and Santa Clara Valley Water District in California make
conservative assumptions about "worst-case scenarios" for drought planning instead of engaging in climate
change-specific studies as described in this report.
12. The exception to this would be to identify trends in the historic record and extrapolate them into the future.
None of the utilities surveyed carried out such an approach.
13. The scientific field of dendrohydrology involves using tree-ring qualities, such as the width of the annual
growth rings, to estimate pre-instrumental streamflow of a specific river.
10
-------
For example, EBMUD used paleoclimate reconstructions of overflow on the Sacramento River
to examine the resilience of their drought planning sequence (Meko and Woodhouse, 2005).
Their use of paleoclimate data was largely qualitative, but the analysis demonstrated EBMUD's
drought planning sequence would likely function adequately under the drought conditions
evidenced in the paleorecord. Denver Water has also used 375-year tree ring reconstructions of
the Colorado and Platte River basins in conjunction with their water supply simulation model -
PACSM - to estimate the frequency and severity of drought within Denver Water's collection
system (Woodhouse and Lukas, 2006a). Boulder Utilities used a 300-year tree ring
reconstruction of hydrology for similar purposes (Hydrosphere, 2003).
Tree-ring reconstructions hold much promise because of an excellent network of tree-ring data,
especially (but not limited to) the western United States (TreeFlow, 2010). The use of
paleoclimate reconstructions, however, is limited because tree-ring reconstructions are not
available for all areas, they can be difficult to translate into the modeling environment for water
management, good relationships with hydrology have only been established for the growing
season - giving an incomplete record of annual hydrology, and the instrumental records
necessary for high-quality calibration of tree-ring data can be difficult to obtain. For these
reasons, and because of the specialized skill set necessary to conduct such studies, paleoclimate
data may not be feasible for many water utilities (Garrick and Jacobs, 2005).
4.3 Literature Reviews
Literature reviews can generate information for projected future climate conditions for many
utilities at low cost. The most common sources for initiating climate literature reviews are the
Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC, 2007) and the
U.S. Global Change Research Program's report Global Climate Change Impacts in the United
States (Karl et al., 2009). The National Oceanic and Atmospheric Administration (NOAA) is
currently developing a climate information web portal (http://www.climate.gov), and NOAA's
Regional Integrated Sciences and Assessments (RISA) program offices often provide regional
literature and data resources (http://www.climate.noaa.gov/cpo_pa/risa/).
With the dramatic increase of climate research across the United States, regional specification of
GCM output can often be found in the scientific peer-reviewed literature or on web portals.
For example, EBMUD decided not to expend resources on developing its own downscaled
climate projections because sufficient data could be obtained in works of the IPCC (2007), the
U.S. National Research Council (NRC, 2002), and the U.S. Geological Survey (Dettinger, 2005).
Using these data, EBMUD ran a sensitivity analysis on their water system using the following
factors:
> Increased customer demand resulting from a 4°C (7.2°F) increase in air temperature
between 1980 and 2040;
> Changes in the timing of runoff in the Mokelumne River corresponding to 2°, 3°, and 4°C
(3.6°, 5.4°, and 7.2° F) increases in air temperature; and
11
-------
> Reductions in Mokelumne River runoff corresponding to a 10% and 20% decrease in
precipitation.
EBMUD combined these three factors to derive seven scenarios that represented an adequate
range of conditions to analyze the potential sensitivity of the EBMUD water system to climate
change.
4.4 Climate Projections
The development of watershed-specific projected future climate conditions is the most
computationally and resource intensive approach to gathering climate information. Data obtained
by this method were employed by many utilities, including the City of Boulder Utilities Division
(Smith et al, 2009), Denver Water (Laurna Kaatz, personal communication, April 8, 2009), SPU
(Palmer, 2007), MWRA (Stephen Estes-Smargiassi, personal communication, February 26,
2010), NYCDEP (2008), PWB (Palmer and Hahn, 2002), and LCRA (CH2M Hill, 2008).
Climate projections consist of three main elements: the use of socioeconomic scenarios of GHG
emissions, the selection of GCMs, and the region-specific downscaling of climate model output.
Each of these elements is described in greater detail below. Also described below is the
innovative combination of climate projections with other data sources for analytical purposes.
It is important to note that the complexity and resource intensity of making climate projections
for water resources is being rapidly reduced. For example, the U.S. Bureau of Reclamation now
provides a downscaled dataset of statistically downscaled monthly climate projections from 1950
to 2099 that consists of 112 climate projections based on 16 climate models and three GHG
emissions scenarios downscaled to a spatial resolution of 1/8 degree across the coterminous
United States (Maurer et al, 2007). The National Center for Atmospheric Research (NCAR) is
also developing a dataset of GCMs combined with different regional climate models (RCMs)
(UCAR, 2007). As these datasets and tools become more widespread, the time and resource
threshold to engage in climate projections likely will be lowered. CH2M Hill's assessment of the
LCRA system used the Bureau of Reclamation dataset as the foundation for its analysis.
4.4.1 Climate scenarios
Utility climate scenarios generally consist of two distinct elements: GHG emissions scenarios
and defined time periods. Probabilities have not been assigned to the emissions scenarios (they
are all assumed equally likely to occur) and, as a result, probabilities have not been assigned to
climate change scenarios.14
GHG emissions scenarios have been developed to identify plausible future global energy use,
economic growth, land-use change, population growth, technological innovation, and other
14. NCAR has developed probabilities of regional changes in temperature and precipitation based on analysis
of GCM output (Tebaldi et al., 2005, 2006), but these probabilities were not used by the utilities examined in
this study.
12
-------
factors that could affect future emissions of GHGs. Most utilities in our analysis used GHG
emissions scenarios defined in the 2000 IPCC Special Report on Emissions Scenarios (SRES;
see Box I).15
In general, water utilities have selected two or three GHG emissions scenarios to provide
substantially different future GHG emission levels. Using multiple GHG emissions scenarios
avoids putting too much emphasis on any single projection of the future. For example, the SPU
study used emissions scenarios A2 and Bl; NYCDEP used emissions scenarios A2, A1B, and
Bl; and LCRA used emissions scenarios A2 and Bl. While, by definition, these scenarios are
equally plausible, water utilities have generally selected both a low emissions/environmentally
friendly scenario (i.e., Bl) and a high emissions/development-oriented scenario (i.e., A2) to
ensure representation of a wide range of plausible futures.
Box 1. IPCC SRES scenarios.
> The Al storyline and scenario family describes a future world of very rapid economic growth, global
population that peaks in mid-century and declines thereafter, and the rapid introduction of new and
more efficient technologies. Major underlying themes are convergence among regions, capacity
building, and increased cultural and social interactions, with a substantial reduction in regional
differences in per capita income. The Al scenario family develops into three groups that describe
alternative directions of technological change in the energy system. The three Al groups are
distinguished by their technological emphasis: fossil intensive (A1FI), non-fossil energy sources
(AIT), or a balance across all sources (A1B).
> The A2 storyline and scenario family describes a very heterogeneous world. The underlying theme is
self-reliance and preservation of local identities. Fertility patterns across regions converge very slowly,
which results in continuously increasing global population. Economic development is primarily
regionally oriented and per capita economic growth and technological change are more fragmented and
slower than in other storylines.
> The Bl storyline and scenario family describes a convergent world with the same global population
that peaks in midcentury and declines thereafter, as in the Al storyline, but with rapid changes in
economic structures toward a service and information economy, with reductions in material intensity,
and the introduction of clean and resource-efficient technologies. The emphasis is on global solutions
to economic, social, and environmental sustainability, including improved equity, but without
additional climate initiatives.
> The B2 storyline and scenario family describes a world in which the emphasis is on local solutions to
economic, social, and environmental sustainability. It is a world with continuously increasing global
population at a rate lower than A2, intermediate levels of economic development, and less rapid and
more diverse technological change than in the Bl and Al storylines. While the scenario is also oriented
toward environmental protection and social equity, it focuses on local and regional levels.
Source: Nakicenovic et al., 2000 (p. 4-5).
15. IPCC is now moving toward a set of scenarios based on radiative forcing and not tied to specific
socioeconomic scenarios (Moss et al., 2010).
13
-------
Furthermore, nearly all utilities utilizing climate projections looked at two or three future time
periods, often defined by relevance to their planning processes. Typically, projections were made
for approximately 30 years for operations or strategic planning and approximately 50 years for
infrastructure and capital improvement plans. For example, the SPU study used three time
periods (2025, 2050, and 2075); Portland used two time periods (2025 and 2045); and LCRA
used two time periods (2050 and 2080). Each time period generally represented an average of 10
of more years of climate data to smooth out year-to-year variation in model output.
4.4.2 Global climate models
After selecting appropriate scenarios, the water utilities generally identified a subset of GCMs to
use in their study. GCMs are generally selected from the World Climate Research Programme's
Coupled Model Intercomparison Project (CMIP3).16 CMIP3 contains data contributed by 23 of
the world's leading climate models (see Table 1).
Table 1. GCMs in the World Climate Research Programme's CMIP Phase Three
Model IPCC First
designation published Sponsor
BCC-CM1
BCCR-BCM2.0
CCSM3
CGCM3.1(T47)
CGCM2.1(T63)
CNRM-CM3
CSIRO-MK3.0
ECHAM5/MPI-OM
ECHO-G
FGOALS-gl.O
GFDL-CM2.0
GFDL-CM2.1
GISS-AOM
2005
2005
2005
2005
2005
2004
2001
2005
1999
2004
2005
2005
2004
Beijing Climate Center
Bjerknes Centre for Climate Research
National Center for Atmospheric Research (NCAR)
Canadian Centre for Climate Modeling and Analysis
Canadian Centre for Climate Modeling and Analysis
Meteo-France/Centre National de Recherches Meteorologiques
(CNRM)
Commonwealth Scientific and Industrial Research Organisation
(CSIRO) Atmospheric Research
Max Plank Institute for Meteorology
Meteorological Institute of the University of Bonn,
Meteorological Research Institute of the Korea Meteorological
Administration (KMA), and Model and Data Group
National Key Laboratory of Numerical Modeling for
Atmospheric Sciences and Geophysical Fluid Dynamics
(LASG)/mstitute of Atmospheric Physics
U.S. Department of Commerce/NCAA Geophysical Fluid
Dynamics Laboratory (GFDL)
NOAA GFDL
National Aeronautics and Space Administration (NASA)
Goddard Institute for Space Studies (GISS)
Country
China
Norway
USA
Canada
Canada
France
Australia
Germany
Germany/
Korea
China
USA
USA
USA
16. Note that CMIP is in the process of updating model outputs in anticipation of the Fifth Assessment Report
of the IPCC. The next iteration of CMIP will be called CMIP5 - skipping "CMIP4" - in order to align the
numbering with the IPCC reports.
14
-------
Table 1. GCMs in the World Climate Research Programme's CMIP Phase Three (cont.)
Model IPCC First
designation published Sponsor
GISS-EH
GISS-ER
INM-CM3.0
IPSL-CM4
MIROC3.2 (high
resolution)
MIROC3.2 (medium
resolution)
MRI-CGCM2.3.2
PCM
UKMO-HadCM3
UKMO-HadGEMl
2004
2004
2004
2005
2004
2004
2003
1998
1997
2004
NASA/GISS
NASA/GISS
Institute for Numerical Mathematics
Institut Pierre Simon Laplace
Center for Climate System Research (University of Tokyo),
National Institute for Environmental Studies, and Frontier
Research Center for Global Change (JAMSTEC)
Center for Climate System Research (University of Tokyo),
National Institute for Environmental Studies, and Frontier
Research Center for Global Change (JAMSTEC)
Meteorological Research Institute
NCAR
Hadley Centre for Climate Prediction and Research/Met Office
Hadley Centre for Climate Prediction and Research/Met Office
Country
USA
USA
Russia
France
Japan
Japan
Japan
USA
UK
UK
This climate model output, however, can vary dramatically by region. Consequently, for this and
other practical considerations, most water utilities select a subset of climate models to use in
their study. Three common methods for selecting models are described below.
One method for selecting climate models is hindcasting, which refers to using climate models to
project climate over years for which instrumental data are available. The ability of various
models to replicate observed climate is often used as a means of selecting a subset of climate
models for use in the study under the justification that those models more accurately model the
climate of that watershed or region. For example, in the first phase of its work, the NYCDEP
Climate Change Task Force selected three GCMs namely the National Center for Atmospheric
Research (NCAR), Goddard Institute of Space Studies (GISS) and European Center Hamburg
Model (ECHAM) (NYCDEP, 2008). In the second phase of work, NYCDEP evaluated the -20
GCMs for various combinations of five meteorological variables (precipitation, average,
maximum and minimum temperatures and wind speed), over four seasons using probability
based skill scores. Since no single model performs best for all variables and seasons they arrived
at the most objective way to choose a subset of models to identify the models with the highest
mean skill scores (averaged across all variables).
A second method for selecting climate models is to purposefully select models that project a
range of climate conditions. This was the strategy employed in the SPU study (Polebitski et al.,
2007a). The three GCM-scenario combinations selected by the UW-CIG were the GISS Model
ER/B1 combination, the MPIECHAM5/A2 combination, and the Institut Pierre Simon Laplace
(IPSL) CM4/A2 combination. These models have performed well in other studies when
replicating the temperature and precipitation trends of the Pacific Northwest during the
20th century (Mote et al., 2005). The MPI ECHAM5/A2 model represents a middle-of-the-road
15
-------
scenario with moderate warming and precipitation increase. The IPSL-CM4/A2 model scenario
is significantly wetter and warmer, and the GISS-ER/B1 scenario is slightly drier and warmer.
The LCRA study (CH2M Hill, 2008) used a third method for selecting climate models that is
effectively a hybrid of the two methods described above. Initially, six models were selected for
further assessment based on expert judgment of the leading models for California selected by
researchers at the Scripps Institute of Oceanography: NCAR CCSM3.0, NCARPCM1, GFDL
CM2.1, CNRM CMS, CCSRMICRO3.2 (medres), and MPIECHAM5.17 Projected changes in
temperature and precipitation from all six models under two emissions scenarios (A2 and Bl)
and two time periods (2050 - averaged 2036-2065; and 2080 - averaged 2066-2099) were
plotted onto a grid representing the 10th and 90th percentile of all 112 downscaled model
simulations from the U.S. Bureau of Reclamation. Scenarios that plotted consistently outside of
this range were discarded, resulting in the selection of two climate models for the LCRA study
that were considered to provide a reasonable range of potential climate change effects in the
Colorado River Basin.
4.4.3 Downscaling
Finally, all climate projections must be downscaled to the spatial and temporal resolution to be
useful inputs into water system models. This is generally done using one of three downscaling
methodologies: simple, statistical, or dynamic downscaling (see Box 2).
For example, the PWB study used simple downscaling. It started with relatively coarse spatial
resolution model results - on the order of three degrees or approximately 300-km grids (Palmer
and Hahn, 2002). UW downscaled the data to one-degree or approximately 100-km grids using
the Symap algorithm. However, even at one degree, the model results were still coarse.
Consequently, the climate change signal for each model and time period was calculated as the
difference between average monthly temperature and precipitation from a control run simulating
current climate and future climate model predictions. This yielded delta values by month for
2025 and 2045 that indicated the modeled change in temperature and precipitation. These deltas
were then applied to an observed climate dataset used by UW as input into its Distributed
Hydrology, Soil-Vegetation Model (DHSVM) to characterize the hydrology of PWB's Bull Run
watershed. This represents a form of simple downscaling because it used coarse GCM data and
related them to the observed record to define a dataset at the spatial and temporal scale of the
utility's models.
17. NCAR CCSM3.0 and GFDL CM2.1 were ultimately chosen for use in the LCRA study.
16
-------
Box 2. Downscaling methodologies
Simple downscaling involves adding projected changes in temperature and precipitation from GCMs to
weather observations for a historical control period. This preserves observed spatial differences and
locally observed seasonality of weather, and allows for the selection of different historic control periods
for different purposes (e.g., a drought year or a year of heavy flooding). However, this method does not
account for differences in climate change within a grid box.
Statistical downscaling is a method by which climate change can be projected for a particular location or
small geographic area. Variables found in GCMs such as pressure are correlated with observations such as
observed temperature or precipitation. The correlation uses GCMs' simulation of current climate. This
correlation is used to project changes in the predicted variable(s) based on the estimated changes in
GCMs. Statistical downscaling is computationally easier than dynamic downscaling, but does not allow
for changes in the relationship between the predictor and predicted variables.
Dynamic downscaling uses RCMs to give higher resolution projections than GCMs. These models are
like GCMs, but simulate only a portion of the globe and can therefore have much smaller grid boxes than
GCMs and can incorporate sub-GCM-grid scale variables (e.g., mountain ranges, lakes). RCMs are
"nested" within GCMs using "boundary conditions" from GCMs to drive them. RCMs typically have
resolutions of 50 km or less. RCMs have the advantage of simulating dynamic relationships between
climate variables, but require extensive computing power.
A key limitation of all three downscaling methods is that they depend on GCMs. Errors in GCMs are
typically not corrected by downscaling techniques, although each technique will supply sub-GCM-grid
scale variations in climate.
NYCDEP used a similar change factor approach, using the difference in daily GCM control and
future scenarios to calculate change factors that were subsequently applied to local
meteorological records used to drive watershed and reservoir models. In the first phase of
NYCDEP climate change work, monthly change factors (additive or multiplicative) were
calculated from pooled daily data associated with each month in each scenario. In the second
phase of the work the change factor methodology was evaluated, and an improved method was
developed that calculated 25 additive change factors for 25 equally spaced bins of the frequency
distribution associated with each month's pooled data. City of Boulder Utilities Division also
used simple downscaling, but combined GCM output with paleoclimate reconstructions.
The LCRA study used a different technique for downscaling. The selection of GCMs for the
LCRA study ultimately led to eight datasets - two models (NCAR CCSM3.0 and GFDL
CM2.1), two scenarios (A2 and Bl), and two time periods (2050 - averaged 2036-2065; and
2080 - averaged 2066-2099). These eight datasets were statistically downscaled from the model
grid scale to regional or watershed scales by retrieving the appropriate temporal-spatial data in
the archived U.S. Bureau of Reclamation downscaled dataset (1/8 degree; 12-km grids).18 A suite
18. The Bureau of Reclamation developed an archive of statistically downscaled and bias-corrected climate
projections for the contiguous 48 states at 1/8 degree resolution, which is what was used in the LCRA study
(see http://gdo-dcp.ucllnl.org/downscaled cmip3_projections/dcplnterface.html). It should be noted that it is
possible to statistically downscale GCM projections using a variety of techniques independent of the U.S.
Bureau of Reclamation dataset. The SPU study provides an example of a different technique.
17
-------
of data processing tools and geographic information systems were applied to these data to
delineate the datasets for the State of Texas and the Colorado River Basin. Simulated future
climate was analyzed by comparing the 2050 and 2080 periods with a reference period of
1970-1999 (CH2M Hill, 2008). This use of sophisticated statistical techniques to downscale
GCM output to local watersheds is known as statistical downscaling.
The SPU study conducted its own method of statistical downscaling (Polebitski et al., 2007b).
First, the SPU study did a bias correction of GCM simulation of hindcast climate using historic
data. Those bias corrections were then applied to coarse (GCM grid scales) future climate data.
The bias-corrected coarse grid was then related to a regional grid (1/8 degree) using scaling
factors (additive factors for temperature and multiplicative factors for precipitation). The
regional, bias-corrected future climate data are then related to specific weather station locations
and again bias-corrected to yield a monthly time-series at the station level. These station level
changes can be combined with historic data to yield scenarios of daily data that preserve historic
climate variability.
For its statistical downscaling, the MWRA used a non-parametric method for generating the
requisite daily weather sequences, specifically the K-nearest neighbor (K-NN) bootstrapping
technique (Yates et al., 2003). This method samples the historical record in order to generate a
large set of individual weather sequences that replicates the statistical characteristics of local
weather but are consistent with the range of GCM-derived temperature and precipitation trends.
Using probability density functions derived from 21 GCMs following the procedure outlined in
Tebaldi et al. (2005), the K-NN algorithm was used to develop individual sequences of weather
variables for the key weather station locations that are used by the system simulation model
(WEAP) of MWRA's supply system.
The final downscaling methodology is dynamic downscaling. Dynamic downscaling uses RCMs
to translate GCM output into higher-resolution regional climate data. Dynamic downscaling
captures the effects of mesoscale features such as narrow mountain ranges, complex
land/waterbody interactions, or variations in land use and land cover in a way that GCM
currently cannot. Until recently, the computing power necessary for dynamic downscaling to
provide useful data was prohibitive and the results were not high-resolution enough to improve
upon statistical downscaling approaches. Current research into RCMs and dynamic downscaling
shows promise. For example, UW-CIG is engaged in research to develop a fully functional RCM
for the Pacific Northwest. NYCDEP also worked with Columbia University's Center for Climate
Systems Research (CCSR) and the City University of New York to establish a dynamic
downscaling strategy for future assessments.
4.4.4 Combining methods: Climate projections with paleoclimate data
The study conducted for the City of Boulder combined climate model output with paleoclimate
data (Smith et al., 2009). To our knowledge, this is the first study to do so in the United States.
The study matched paleoclimate reconstructions of streamflow in Boulder Creek with years
during the observed climate record that had similar streamflow. The climate in the similar years
18
-------
was used as a proxy for the climate of the paleorecord. In other words, the reconstruction of
streamflow in a paleo year (e.g., 1635) was then matched to a year in the observed record with a
similar annual streamflow. A best match between reconstructed streamflow and years in the
observed record did not always exist. Therefore, the technique used a random process to select
years from the observed record with similar characteristics. The model was run many times to
yield different combinations of the observed record and mimic the reconstruction. This
introduces many combinations of reconstructions, which allows for more variability in results.
The Boulder study used a simple downscaling of GCM data. The changes in temperature and
precipitation were added to or multiplied by temperature and precipitation, respectively, in the
proxy record. This yielded a set of climate change scenarios that mimicked the year-to-year
variability of the paleorecord, but with long-term average changes in climate from the GCMs
imposed on the set.
The reconstruction of streamflow for Boulder Creek utilized a 437-year record from 1566 to
2002 (Woodhouse and Lukas, 2006b). The reconstruction technique estimates 65% (i.e., has an
o
R of 0.65) of the observed streamflow variance.
The study used output from three GCMs under three emissions scenarios for the decades of the
2030s and 2070s. The selected emissions scenarios (A2, A1B, and Bl) represented a wide range
of future conditions. As with SPU, the GCMs were selected to capture a wide range of potential
changes in regional climate. While all climate models project that the central Rocky Mountains
and adjacent plains will become warmer, the models disagree as to whether precipitation will
increase or decrease. To overcome this, the study used a relatively wet model (the Canadian
Climate Model), a relatively dry model (GFDL Version 0), and a model roughly in the middle of
the range of model output (GFDL Version 1).
19
-------
5. Modeling Changes in Water Resources
Water utilities use a variety of models and analytical techniques for various purposes. As a
general organizational distinction, it simplifies things to examine how a utility approaches water
quantity and water demand separately.19
5.1 Water Quantity
Water utilities may use one or more water management model(s) for day-to-day operations and
long-term water planning. Many, but not all, also use a hydrologic model to translate temperature
and precipitation data into streamflow, reservoir storage, evaporative loss, and other variables
input directly into water system management models. The specific constellation of models used
by a utility often determines what climate information it needs to carry out a climate change
vulnerability assessment.
PWB, for example, input temperature and precipitation data into the DHSVM to generate
averaged annual hydrographs. DHSVM is a physically based hydrology model that characterized
the entire Bull Run watershed into 150-m grid cells with grid-specific data on soil and vegetation
type, soil depth, vegetation height, and surface elevation and slope. DHSVM represents the more
detailed end of utility hydrology models that have been empirically characterized for a number of
parameters in addition to temperature and precipitation.
PWB applied the UW-CIG developed climate altered streamflows through DHSVM and then
input the data into PWB's Supply and Transmission Model (STM), an operational model of
PWB's terminal water storage and groundwater resources. STM operates at a daily timestep and
simulates the flow of water throughout the water transmission system using seasonally varying
rule curves for the reservoirs, as well as modeling water releases for hydropower production and
instream flows. Groundwater operations are coordinated with reservoir operations to enable the
investigation of a variety of operating alternatives using variables such as length of drawdown
period, amount of groundwater pumped during drawdown, minimum storage during drawdown,
and water used during drawdown. As a part of a Water Research Foundation project, the PWB
used the STM data sets to look at how the WEAP platform worked in comparison with the
outputs of the STM, however, the WEAP model was not used for the initial climate change study
done by UW-CIG. For purposes of this analysis it is important to note the use of multiple
operational models for different purposes by a single utility.
19. Some utilities in this study, such as NYCDEP, were also concerned about climate change effects on water
quality. However, generally speaking, water quality involves a greater variety of models (e.g., reservoir
specific eutrophication models) and/or micro-climate processes that often occur at a higher resolution than can
be captured by current GCMs or RCMs (e.g., intense precipitation events). The scope of this report did not
allow us to fully explore water quality modeling.
20
-------
In order to use SPU's systems model, climate change parameters needed to be converted to
inflow. UW-CIG used the DHSVM hydrology model to produce climate-altered hydrologic
datasets based on climate model output for use in the Conjunctive Use Evaluation (CUE). SPU
then used the results from DHSVM in its CUE systems model - a weekly timestep simulation
model of the Cedar and Tolt River systems - for water supply planning. CUE is used for
90
calculating and evaluating the firm yield and reliability of Seattle's water supply system and
potential future water supply projects. CUE results indicated that yield decreased under all
climate change scenarios for all time periods. SPU also ran several planning scenarios through
CUE to determine whether available supply could be increased to compensate for anticipated
shortfalls.
The use of river gauge data as input into CUE was a policy decision for SPU, because it also has
the capability to use the Seattle Forecast Model (SEAFM), a proprietary hydrology model
calibrated to the SPU watersheds and used in operational forecasting and operations planning.
UW-CIG chose to use DHSVM instead of SEAFM (or its successor models) for its climate
change vulnerability analysis because UW-CIG developed DHSVM and it was available to all
watersheds in the broader UW-CIG led regional study in which SPU participated.
NYCDEP has a very complex modeling environment. It currently does analysis using the
Variable Source Loading Function (VSLF) or the SWAT watershed models. These can in turn
provide inputs to one of two reservoir models: a one-dimensional reservoir eutrophication model;
or a two-dimensional reservoir turbidity transport model (CEQual W2). The OASIS reservoir
system operation model requires measured or simulated water inputs to the reservoirs and
simulates the storage and transfer of water between the 19 reservoirs comprising the NYC water
supply system. The watershed models, take daily temperature and precipitation data to generate
streamflow and evapotranspiration, as well as a number of water quality parameters. Outputs of
the watershed models can be used to drive both the OASIS system model and reservoir water
quality models. A knowledge of reservoir operations is an important determinate of reservoir
water quality, and a needed input to the reservoir models. For future climate simulations, where
historical operations are not known, OASIS simulation are needed to specify reservoir operation
scenarios associated with a given climate scenario. All of the above models were in use at
NYCDEP to evaluate the impacts of changes in watershed management, land-use, and reservoir
operation policies. This integrated modeling system was easily adapted for evaluation of the
effects of climate change, once credible future scenarios of the meteorological variables needed
to drive the models were available.
5.2 Water Demand
Changes in climate can also have an important effect on water demand. Demand means different
things to different utilities depending upon their customer base and the timeframe of their
assessment. Several utilities evaluated in this report provided interesting examples of different
20. Firm yield is based on the 98% reliability standard.
21
-------
approaches or concerns to project changing water demand. EBMUD was focused primarily on
exterior water use by urban and suburban water customers, while SPU and PWB were focused
on the balance between population growth and climate altered demand over time.
According to EBMUD, customer demands are projected to vary predominantly based on
temperature change. While indoor water use is not expected to change significantly with global
warming, outdoor water use could change dramatically. To account for this, 2040 customer
demands were re-normalized using a projected temperature increase of 2.15°C (3.87°F) between
2005 and 2040 with no change in precipitation. EBMUD's analysis indicated that a 20%
reduction in precipitation had little influence on overall customer demands compared to the
projected temperature change. The demand model suggested a 3.6% increase in customer
demands by the year 2040 (EBMUD, 2009).
The SPU study examined the effect of climate change on water demand using a dual approach of
regression analysis and forecast modeling. First, SPU performed a regression analysis of peak
season consumption for 1982-2007 using monthly consumption data, maximum temperature,
and rainfall at SeaTac Airport for May through September. This relationship was assumed to
hold in the future. SPU had already developed a demand forecasting model for its 2007 Water
System Plan, which forecasted non-climate altered demand change over time. Under this model,
demand was forecasted to decrease below historic levels through 2050, but increase above
historic levels by 2075. Applying the results of the regression analysis to these forecasts adjusts
demand slightly upward due to climate change in 2025 and 2050. But in 2075, the climate-
induced increase becomes more significant on top of the additional forecasted upswing from the
non-climate altered demand model.
In 2002, PWB examined the effects of climate-altered streamflow on system operations from a
demand perspective using their STM. The evaluation investigated the climate and population
growth impact on demand and supply separately and then jointly to discern the discrete effects of
climate change. The process for calculating the climate impact on demand was derived from the
Joint Institute on the Study of Atmosphere and Oceans (JISAO) report, Impacts of Climate
Variability and Change in the Pacific Northwest (JISAO-CIG, 1999). This study used seven
featured years and the ECHAM4 climate scenario for detailed analysis. Growth in demand not
considering changes in climate decreased the average minimum storage dramatically. This was
exacerbated (approximately doubled) by the impacts of climate on hydrology and demand.
PWB also utilized their water demand econometric model to estimate near- and long-term water
demand that established a relationship between total water demand and selected economic,
demographic, seasonal, and day-to-day weather variables. This econometric model is used as a
forecasting tool by projecting changes in the economic and demographic variables. In order to
gauge the effects of climate change on demand, past weather years with patterns that
approximate climate projections are selected to forecast future demands. For example, in
projecting demands in 2050, PWB uses 1948 as the wettest, or best-case scenario (corresponding
to a 3.67°F/2.04°C increase in average peak season temperature), 1980 as a middle-of-the-road
scenario (corresponding to a 4.83°F/2.68°C increase in average peak season temperature), and
22
-------
1991 as the driest, or worst-case scenario (corresponding to a 6.72°F/3.73°C increase in average
peak season temperature).
6. Summary
We investigated assessments of climate change vulnerability conducted by eight utilities to
determine the emergent characteristics of water utility climate change vulnerability assessments
(see Table 2). We found that two standard approaches were used for the risk assessments: the
top-down modeling assessment and the bottom-up threshold analysis. While both methods
produce important risk information, the selection of an assessment model is often determined by
available fiscal and technical resources as well as the policy environment faced by the utility.
These two approaches are not mutually exclusive, and are often run in series or in parallel to
provide a robust vulnerability assessment. Notably, top-down assessments come in a variety of
types, including scenario analyses, sensitivity analyses, and paleoclimate or historic analyses.
Sources of climate data used by the utilities included the instrumental record, paleoclimate data,
literature reviews, and climate projections from GCMs. While information from each of these
sources played an important role in one or more utility vulnerability assessments, climate
projections offered the most computationally demanding approach. Climate projections required
the selection of GHG emissions scenarios and defined time periods, the selection of a subset of
GCMs (typically by hindcasting, selecting a broad range of outputs, or eliminating outliers), and
downscaling the GCM data to the spatial and temporal scale necessary for input into utility
hydrology, management, or operations models through simple, statistical, or dynamical
downscaling techniques.
Finally, the climate information was run through utility-specific models to determine the effects
of projected climate conditions on water supply and demand. We found that on the supply side,
the models typically used included both hydrology and operational or management models.
Demand modeling by utilities ranged from service-area-specific correlations between customer
demand and temperature/precipitation to utility-specific demand models developed for long-
range planning.
23
-------
Table 2. Aspects of utility vulnerability assessment efforts
Boulder
Denver
Water
MWRA
Assessment
approach
Top-down
(paleo)
Top-down
(sensitivity,
scenario in
process)
Top-down
NYCDEP Top-down
PWB
SPU
LCRA
(scenario)
Top-down
(sensitivity)
Top-down
(scenario)
Bottom-up
(scenario)
EBMUD Bottom-up
a.
b.
(sensitivity)
The 1% annual
Climate
model
selection
Purposeful
selection for
range
Current study:
Purposeful
selection for
range
All models
(Yates/NCAR
project)
Probability
based skill
score approach
Purposeful
selection for
range
Purposeful
selection for
range
Hindcast and
purposeful
selection for
range
Conducted
sensitivity
analysis
GHG
emissions
scenarios
A2,A1B,
Bl
Current
study: A2,
A1B,B1
A1B
A2,A1B,
Bl
1% annual
increase in
C02a
2007
study: A2,
Bl,2002
study: A2
A2,B1
N/A
increase in C02 scenario was
Because EBMUD chose to use a
applicable.
Downscaling
Simple
Current
study:
statistical
Statistical
Delta change
factor
methodology
by month
and
frequency
distribution,
statistical
Statistical
Statistical
Statistical
(Bureau of
Reclamation
dataset)
N/A
Time
periods
2030, 2070
In process
study: 2040
and 2070
Weekly
projections
through
2060
2046-2065
and 2081-
2100
2025, 2045
2007 study:
2000, 2025,
2050, 2075
2002 study:
2000, 2020,
2040
2050, 2080
N/A
System
models
Boulder Creek
Model
PACSM, ESP
WEAP
OASIS,
CEQual W2,
1-D
Eutrophication
STM, WEAP
Conjunctive
Use Evaluation
(CUE)
WAMWRAP,
EBMUDSim,
WEAP
often used prior to development of the SRES
sensitivity analysis, the climate modeling
Runoff
models
CLIRUN
Sacramento
Soil
Moisture
coupled
with
Anderson
Snow- 17,
WEAP
abed
watershed
model
VSLF,
SWAT
DHSVM
DHSVM
Variable
Infiltration
Capacity
N/A
scenarios.
; aspects of this table are not
24
-------
7. Recommendations for Further Study
This report describes a range of practices used by eight water utilities to evaluate the effects of
climate change on water quantity and water demand using hydrological, management and
operational models. Climate change is a complex issue and will require ongoing work to
establish reliable practices for incorporating climate change into water utility decisions and
planning. Additional questions to develop this line of inquiry further include:
• How are decision makers and rate payers responding to these types of analyses?
• Would the results vary if different methods were used by the same utility?
• How do the different methods compare in effectiveness, over time?
• What other decisions, utility models, etc. need climate vulnerability assessments?
• How are vulnerability assessments being conducted for understanding climate
impacts on managing:
o Water quality?
o Intense precipitation and storms (stormwater, floods, wind)?
o Sea level rise?
As other water resource managers gain more experience with climate vulnerability assessments,
it would benefit the water resource management field to continue to document lessons learned in
an effort to create, over time, a solid foundation of acceptable industry practices.
25
-------
8. References
Barsugli, J., C. Anderson, J.B. Smith, and J.M. Vogel. 2009. Options for Improving Climate Modeling
to Assist Water Utility Planning for Climate Change. Final Report. Prepared for Water Utility Climate
Alliance. December.
CH2M Hill. 2008. Climate Change Study: Report on Evaluation Methods and Climate Scenarios.
Final Report. Lower Colorado River Authority and San Antonio Water System. November.
Dettinger, M.D. 2005. From climate-change spaghetti to climate-change distributions for 21st Century
California. San Francisco Estuary and Watershed Science 3(1). Available:
http://repositories.cdlib.org/jmie/sfews/vol3/issl/art4.
EBMUD. 2009. WSMP 2040: Water Supply Management Program 2040 Plan. East Bay Municipal
Utility District. October.
Freas, K., B. Bailey, A. Munevar, and S. Butler. 2008. Incorporating climate change in water planning.
Journal of the American Water Works Association 100:6. June.
Garrick, D. and K. Jacobs. 2005. Tree-ring records inform water management decisions: Workshop
strives for better integration of past with present. Southwest Climate Outlook. May.
Hydrosphere. 2003. Simulation of Hypothetical Climate Change Scenarios Using the Boulder Creek
Water shed Model. Hydrosphere Resource Consultants (now AMEC), Boulder, CO.
IPCC. 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to
the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, S. Solomon, D.
Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.). Cambridge
University Press, Cambridge, UK and New York.
JISAO-CIG. 1999. Impacts of Climate Variability and Change: Pacific Northwest. A Report of the
Pacific Northwest Regional Assessment Group for the US Global Change Research Program. Joint
Institute for the Study of Atmosphere and Oceans - Climate Impacts Group. Office of Global
Programs at the National Oceanographic and Atmospheric Administration. Seattle, WA. November.
Karl, T.R., J.M. Melillo, and T.C. Peterson. 2009. Global Climate Change Impacts in the United
States. Cambridge University Press, New York.
Major, D.C., R. Horton, and C. Rosenzweig. 2007. Integrated Modeling Project for Water Quantity
and Quality: Guidelines for Project Final Report. Climate Impact Assessment of Environmental
Infrastructure Systems. New York City Department of Environmental Protection. June 29.
Maurer, E.P., L. Brekke, T. Pruitt, and P.B. Duffy. 2007. Fine-resolution climate projections enhance
regional climate change impact studies. Eos Trans. AGU88(4iy.504.
26
-------
Means III, E.G., M.C. Laugier, and J.A. Daw. 2010. Evaluating Decision Support Planning Methods
for Incorporating Climate Change Uncertainties into Water Planning. Final Report. Prepared for
Water Utility Climate Alliance. January.
Meko, D.M. and C.A. Woodhouse. 2005. Tree-ring footprint of joint hydrologic drought in
Sacramento and Upper Colorado River basins, western USA. Journal of Hydrology 308:196-213.
Miller, K. and D. Yates. 2006. Climate Change and Water Resources: A Primer for Municipal Water
Providers. Jointly sponsored by Awwa Research Foundation, Denver, CO and University Corporation
for Atmospheric Research, Boulder, CO. Awwa Research Foundation, American Water Works
Association, and IWA Publishing.
Milly, P.C.D., J. Betancourt, M. Falkenmark, R.M. Hirsch, Z.W. Kundzewicz, D.P. Lettenmaier, and
R.J. Stouffer. 2008. Stationarity is dead: Whither water management? Science 319:573-574.
Moss, R.H., J.A. Edmonds, K.A. Hibbard, M.R. Manning, S.K. Rose, D.P. van Vuuren, T.R. Carter, S.
Emori, M. Kainuma, T. Kram, G.A. Meehl, J.F.B. Mitchell, N. Nakicenovic, K. Riahi, S.J. Smith, R.J.
Stouffer, A.M. Thomson, J.P. Weyant, and T.J. Wilbanks. 2010. The next generation of scenarios for
climate change research and assessment. Nature 463:747-756.
Mote, P.W., E. Salathe, and C. Peacock. 2005. Scenarios of Future Climate for the Pacific Northwest.
University of Washington: Climate Impacts Group.
Nakicenovic, N., J. Alcamo, G. Davis, B. de Vries, J. Fenhann, S. Gaffm, K. Gregory, A. Grubler, T.Y.
Jung, T. Kram, E.L. La Rovere, L. Michaelis, S. Mori, T. Morita, W. Pepper, H. Pitcher, L. Price, K.
Raihi, A. Roehrl, H.-H. Rogner, A. Sankovski, M. Schlesinger, P. Shukla, S. Smith, R. Swart, S. van
Rooijen, N. Victor, and Z. Dadi. 2000. Emissions Scenarios. A Special Report of Working Group III of
the Intergovernmental Panel on Climate Change. Cambridge University Press, New York.
UCAR. 2007. North American Regional Climate Change Assessment Program. Available:
http://www.narccap.ucar.edu/. Accessed May 28, 2010.
NRC. 2002. Abrupt Climate Change: Inevitable Surprises. National Academies Press: Washington,
D.C. Available: http://www.nap.edu/openbook.php7isbn = 0309074347.
NYCDEP. 2008. The New York City Department of Environmental Protection Climate Change
Program Assessment and Action Plan: Report 1. May.
Palmer, R.N. 2007. Final Report of the Climate Change Technical Committee. A report prepared by
the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle,
WA.
Palmer, R.N. and M. Hahn. 2002. The Impacts of Climate Change on Portland's Water Supply: An
Investigation of Potential Hydrologic and Management Impacts on the Bull Run System. Portland
Water Bureau. January.
27
-------
Polebitski, A., L. Traynham, and R.N. Palmer. 2007a. Technical Memorandum #4: Approach for
Developing Climate Impacted Meteorological Data and its Quality Assurance/Quality Control. A
report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply
Planning Process, Seattle, WA.
Polebitski, A., M.W. Wiley, and R.N. Palmer. 2007b. Technical Memorandum #2: Methodology for
Downscaling Meteorological Data for Evaluating Climate Change. A report prepared by the Climate
Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA.
Smith, J.B., K. Strzepek, L. Rozaklis, C. Ellinghouse, and K.C. Hallett. 2009. The Potential
Consequences of Climate Change for Boulder Colorado's Water Supplies. Prepared for NOAA's
Climate Program Office. February.
Strange, E.M., D.R. Lane, and C.N. Herrick. 2009. Utility Guidance for Mitigating Catastrophic
Vegetation Change in Watersheds. Project #4009. Water Research Foundation, Denver, CO.
Stratus Consulting and MWH Global. 2009. Implications of Climate Change for Adaptation by
Wastewater andStormwater Agencies. Water Environment Research Foundation Report No. CC2R08.
Tebaldi, C., J.M. Arblaster, K. Hayhoe, and G.M. Meehl. 2006. Going to extremes: An
intercomparison of model-simulated historical and future change extremes. Climatic Change 79:185-
211.
Tebaldi, C., R.L. Smith, D. Nychka, and L.O. Mearns. 2005. Quantifying uncertainty in projection of
regional climate change: ABayesian approach to analysis of multi-model ensembles. Journal of
Climate 18:1524-1540.
TreeFlow. 2010. Streamflow reconstructions from treerings website. Available: http://treeflow.info/.
Accessed March 30, 2010.
U.S. EPA. 2009. Proceedings of the First National Expert and Stakeholder Workshop on Water
Infrastructure Sustainability and Adaptation to Climate Change. Publication No. EPA 600/R-09/010.
Available: http://www.epa.gov/nrmrl/wswrd/wqm/wrap/workshop.html.
Woodhouse, C.A. and JJ. Lukas. 2006a. Multi-century tree-ring reconstructions of Colorado
Streamflow for water resource planning. Climatic Change. DOT: 10.1007/sl0584-006-9055-0.
Woodhouse, C. and J. Lukas. 2006b. Streamflow Reconstructions for Boulder Creek, South Boulder
Creek, and the Colorado River. Report to City of Boulder and Hydrosphere Resource Consultants.
AMEC Consulting, Boulder, CO.
WRF. 2010a. Analysis of Changes in Water Use Under Regional Climate Change Scenarios. Project
#4263. Project in progress; completion year 2013. Water Research Foundation, Denver, CO.
WRF. 2010b. Estimating Salinity Effects Due to Climate Change on the Georgia and South Carolina
Coasts. Project #4285. Project Description; completion year 2012. Water Research Foundation,
Denver, CO.
28
-------
WRF. 2010c. Evaluating Effects of Climate Change on Water Utility Planning Criteria and Design
Standards. Project #4154. Project Description and Abstract; completion year 2010. Water Research
Foundation, Denver, CO.
WRF. 2010d. Incorporating Climate Change Information in Water Utility Planning: A Collaborative,
Decision Analytic Approach. Project #3132. Project Description and Abstract; completion year 2010.
Water Research Foundation, Denver, CO.
WRF. 2010e. Project Snapshot: Joint Front Range Climate Change Vulnerability Study #4205.
Available:
http ://www. waterresearchfoundation. org/research/TopicsAndProj ects/proj ectSnapshot. aspx?pn = 4205.
Accessed March 30, 2010.
WRF. 201 Of. Water Utilities and Climate Change: A Research Workshop on Effective System
Adaptation. Project #4228. Project Description; completion year 2010. Water Research Foundation,
Denver, CO.
Yates, D., S. Gangopadhyay, B. Rajagopalan, and K. Strzepek (2003). A technique for generating
regional climate scenarios using a nearest neighbor algorithm, Water Resources Research, 39(7),
1199.
29
------- |