Actionable Science in Practice: Co-producing Climate Change Information for Water Utility Vulnerability Assessments

Technical Report (PDF Available) · May 2015with 328 Reads 
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
·
Affiliation: Water Utility Climate Alliance
Cite this publication
Abstract
This report documents a collaborative Water Utility Climate Alliance effort, the Piloting Utility Modeling Applications (PUMA) project. The PUMA project was an effort to produce actionable science through close collaboration between climate experts and utility personnel to meet the needs of four water utilities. Instead of asking climate experts what they thought utilities should do regarding climate change, four WUCA utilities agreed to forge partnerships with scientific institutions to explore how to integrate climate considerations into their specific management context. This report documents those four utilities’ experience between the start of the PUMA project in 2010 and the writing of this report in late 2014. A fundamental goal of this report is to draw lessons from these four distinct projects regarding best practices in the co-production of actionable science. We attempt to display how each team went about tailoring climate information to specific decision-maker needs, show what worked and did not work, and inform future research and investment along the boundary between climate science and adaptation decision-making. Below is a big-picture summary of our conclusions, based on the experience of all four PUMA utilities: } Assessment was local, and one size did not fit all. Although each PUMA project sought to illuminate a similar question – the impact of climate change on drinking water supplies – the four utilities pursued widely different approaches in service of that goal. } The scientist and utility-manager learning process was a two-way street. In practice, the climate modelers themselves often had as much to learn about how water utilities model their systems as the water utility personnel had to learn about how climate modelers project future climate. } Water utilities sometimes needed to customize approaches to using climate model output. General circulation model (GCM) output, downscaling techniques, and even baseline observational datasets used to validate climate projection tools frequently needed to be customized for use in local assessments; this included correcting these climate model outputs to accurately reflect local conditions. } Utilities required flexibility in exploring different methods to use climate model output. Each of the PUMA utilities followed a different path, using different data, models, and techniques to get started and learn about climate change in general and enhance applicability to local circumstances in particular. } Utilities found that they needed to consider using a bottom-up as well as a top-down approach to climate modeling. A bottom-up approach begins by asking what is important in the context of a specific utility and a top-down approach begins by exploring what the science can tell us about how climate may change. PUMA utilities found value in both approaches. } Information on changes in extreme event impacts was a major need for water utilities. Although climate models do not easily capture extreme events, such events were the most sought-after projections for many of the utilities’ PUMA projects. } Understanding local hydrology was critical. A good understanding of local hydro-meteorology was important in understanding the impacts of changes in temperature, precipitation, solar radiation, winds, and other key variables on water supply sources. } Utilities and scientists learned to adopt a “don’t hesitate to innovate” strategy. Some of the most successful aspects of the PUMA project occurred when water utilities and their scientific partners decided to create something new to meet their needs. For lessons learned on a case-by-case basis, please refer to each case in Sections 3.1 through 3.4; for detailed conclusions across the project, please see Section 4: Conclusions for an Applied Research Agenda for Climate Services.
Advertisement
Prepared for:
Water Utility Climate Alliance
The Central Arizona Project, Denver Water, the Metropolitan Water District of Southern
California, the New York City Department of Environmental Protection, the Portland Water
Bureau, the San Diego County Water Authority, the San Francisco Public Utilities Commission,
Seattle Public Utilities, the Southern Nevada Water Authority, and Tampa Bay Water
Project Manager: David Behar, San Francisco Public Utilities Commission
Contract Manager: Keely Brooks, Southern Nevada Water Authority
Lead Author
Jason M. Vogel
Contributing Authors
Joel B. Smith, Stratus Consulting
Megan O’Grady, Stratus Consulting
Paul Fleming, Seattle Public Utilities
Kavita Heyn, Portland Water Bureau
Alison Adams, Tampa Bay Water
Don Pierson, New York City Department of Environmental Protection
Keely Brooks, Southern Nevada Water Authority
David Behar, San Francisco Public Utilities Commission
May 2015
Actionable Science in Practice: Co-producing
Climate Change Information for Water Utility
Vulnerability Assessments
Final Report of the Piloting Utility Modeling
Applications (PUMA) Project
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Table of Contents
Acknowledgments ......................................................................................................................... ii
List of Acronyms .......................................................................................................................... iii
Executive Summary ...................................................................................................................... v
Section 1 Introduction ........................................................................................................... 1
Section 2 PUMA Project Process ......................................................................................... 5
Section 3 PUMA Project Outcomes: Four Utility Project Profiles ................................... 7
3.1 New York City Department of Environmental Protection ..................................... 8
3.1.1 NYCDEP’s PUMA project summary ......................................................... 8
3.1.2 NYCDEP issues of interest ......................................................................... 9
3.1.3 How the PUMA project will affect utility decisions ................................ 12
3.2 Portland Water Bureau .......................................................................................... 14
3.2.1 PWB’s PUMA project summary .............................................................. 14
3.2.2 PWB issues of interest .............................................................................. 15
3.2.3 How the PUMA project will affect utility decisions ................................ 19
3.3 Seattle Public Utilities........................................................................................... 20
3.3.1 SPU’s PUMA project summary ................................................................ 20
3.3.2 SPU issues of interest................................................................................ 21
3.3.3 How the PUMA project will affect utility decisions ................................ 24
3.4 Tampa Bay Water ................................................................................................. 25
3.4.1 TBW’s PUMA project summary .............................................................. 25
3.4.2 TBW issues of interest .............................................................................. 26
3.4.3 How the PUMA project will affect utility decisions ................................ 29
Section 4 Conclusions for an Applied Research Agenda for Climate Services ............. 30
References ............................................................................................................ 33
Suggested Resources ........................................................................................... 35
Appendix A Applying Climate Model Outputs 101 for Water Utilities ............................ A-1
Appendix B PUMA Project Points of Contact..................................................................... B-1
Page i
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Acknowledgments
The authors of this report would like to thank the Water Utility Climate Alliance (WUCA) in
general and Piloting Utility Modeling Applications (PUMA) project members in particular for
funding and supporting this project. We highly appreciate the guidance provided during the
course of the project by the WUCA members that participated: Alan Cohn, Elliot Schneiderman,
and Lorraine L. Janus, New York City Department of Environmental Protection; Lorna Stickel
and Edward Campbell, Portland Water Bureau; and Tirusew Asefa, Tampa Bay Water.
In addition, we would like to express our gratitude to the following PUMA climate science
counterparts for their valuable input throughout the project and for this report:
The Pacific Northwest Climate Impacts Research Consortium, including:
Dr. Philip Mote, Ms. Megan Dalton, and Ms. Kathie Dello, Oregon State
University
Dr. Bart Nijssen and Ms. Tzu-Hsin Cindy Chiao, University of Washington
Dr. John Abatzoglou and Dr. Katherine Hegewisch, University of Idaho
Ms. Wendy Graham, University of Florida
Ms. Aavudai Anandhi Swamy, Kansas State University.
Page ii
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
List of Acronyms
AR IPCC Assessment Report
AR4 IPCC Fourth Assessment Report
AR5 IPCC Fifth Assessment Report
BCCA Bias-correction and constructed analog
BCSA Bias-correction and stochastic analog
BCSD Bias-correction and spatial disaggregation
CCAWWG Climate Change and Water Working Group
CCRUN Consortium for Climate Risk in the Urban Northeast
CDF Cumulative distribution function
CIRC Climate Impacts Research Consortium
CMIP Coupled Model Intercomparison Project
CMIP3 Coupled Model Intercomparison Project Phase 3
CMIP4 Coupled Model Intercomparison Project Phase 4
CMIP5 Coupled Model Intercomparison Project Phase 5
COAPS Center for Ocean-Atmospheric Prediction Studies
CUNY City University of New York
DHSVM Distributed Hydrology-Soil-Vegetation Model
FSU Florida State University
GCM General circulation model or global climate model
GHG Greenhouse gas
GIS Geographic information system
IPCC Intergovernmental Panel on Climate Change
MACA Multivariate Adaptive Constructed Analog
NASA-GISS NASA Goddard Institute for Space Studies
NCAR National Center for Atmospheric Research
NOAA National Oceanic and Atmospheric Administration
NYCDEP New York City Department of Environmental Protection
OSU Oregon State University
Page iii
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
PRISM Parameter-elevation Relationships on Independent Slopes Model
PRMS Precipitation-runoff Modeling System
PUMA Piloting Utility Modeling Applications
PWB Portland Water Bureau
RCM Regional climate models
RCP Representative Concentration Pathway
RISA Regional Integrated Sciences and Assessments
SD Statistically distributed
SDBC Spatial disaggregation and bias-correction
SDSM Statistical Downscaling Model
SECC Southeast Climate Consortium
SPU Seattle Public Utilities
SRES Special Report on Emissions Scenarios
TAR IPCC Third Assessment Report
TBW Tampa Bay Water
UF University of Florida
UID University of Idaho
USGCRP U.S. Global Change Research Program
USGS U.S. Geological Survey
UW University of Washington
VIC Variable Infiltration Capacity
WRF Weather Research and Forecasting
WUCA Water Utility Climate Alliance
Page iv
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Executive Summary
This report documents a collaborative Water Utility Climate Alliance effort, the Piloting Utility
Modeling Applications (PUMA) project. The PUMA project was an effort to produce actionable
science through close collaboration between climate experts and utility personnel to meet the
needs of four water utilities. Instead of asking climate experts what they thought utilities should
do regarding climate change, four WUCA utilities agreed to forge partnerships with scientific
institutions to explore how to integrate climate considerations into their specific management
context.
This report documents those four utilities’ experience between the start of the PUMA project in
2010 and the writing of this report in late 2014. A fundamental goal of this report is to draw
lessons from these four distinct projects regarding best practices in the co-production of
actionable science. We attempt to display how each team went about tailoring climate
information to specific decision-maker needs, show what worked and did not work, and inform
future research and investment along the boundary between climate science and adaptation
decision-making.
Below is a big-picture summary of our conclusions, based on the experience of all four PUMA
utilities:
Assessment was local, and one size did not fit all. Although each PUMA project sought
to illuminate a similar question – the impact of climate change on drinking water supplies
– the four utilities pursued widely different approaches in service of that goal.
The scientist and utility-manager learning process was a two-way street. In practice, the
climate modelers themselves often had as much to learn about how water utilities model
their systems as the water utility personnel had to learn about how climate modelers
project future climate.
Water utilities sometimes needed to customize approaches to using climate model
output. General circulation model (GCM) output, downscaling techniques, and even
baseline observational datasets used to validate climate projection tools frequently needed
to be customized for use in local assessments; this included correcting these climate
model outputs to accurately reflect local conditions.
Utilities required flexibility in exploring different methods to use climate model output.
Each of the PUMA utilities followed a different path, using different data, models, and
techniques to get started and learn about climate change in general and enhance
applicability to local circumstances in particular.
Page v
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Utilities found that they needed to consider using a bottom-up as well as a top-down
approach to climate modeling. A bottom-up approach begins by asking what is
important in the context of a specific utility and a top-down approach begins by exploring
what the science can tell us about how climate may change. PUMA utilities found value
in both approaches.
Information on changes in extreme event impacts was a major need for water utilities.
Although climate models do not easily capture extreme events, such events were the most
sought-after projections for many of the utilitiesPUMA projects.
Understanding local hydrology was critical. A good understanding of local hydro-
meteorology was important in understanding the impacts of changes in temperature,
precipitation, solar radiation, winds, and other key variables on water supply sources.
Utilities and scientists learned to adopt a “don’t hesitate to innovate” strategy. Some of
the most successful aspects of the PUMA project occurred when water utilities and their
scientific partners decided to create something new to meet their needs.
For lessons learned on a case-by-case basis, please refer to each case in Sections 3.1 through 3.4;
for detailed conclusions across the project, please see Section 4: Conclusions for an Applied
Research Agenda for Climate Services.
Page vi
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
1. Introduction
The Water Utility Climate Alliance (WUCA) is a coalition of 10 of the nations largest water
providers (see Figure 1).
1
Together, they supply drinking water for more than 43 million people
in the United States. WUCA was formed in 2007 to better understand the effects of climate
change on water-related infrastructure and water resource supplies. In addition to sharing the
coalitions experiences with independently assessing climate vulnerability and identifying
adaptation actions, WUCA has engaged in a number of collaborative efforts to advance the
understanding of available climate science and how it can support water resource decision-
making. The coalition has also published two white papers: one on improving climate modeling
in support of water utilities, and a second on decision support planning methods. Both are
available online (http://www.wucaonline.org).
This report documents a new collaborative WUCA effort, the Piloting Utility Modeling
Applications (PUMA) project. This report describes the lessons learned from four WUCA case
studies focused on understanding and assessing science products for use in vulnerability
assessments (see the text box, WUCA’s four steps to adaptation).
Along with the release of this report, WUCA is releasing another white paper featuring case
studies of utilities that are actively engaged in step three of the framework, planning. The case
studies highlighted are incorporating climate vulnerability assessments into decision-making
processes as diverse as long-term water supply planning and day-to-day capital investment
decisions. In many cases utilities have added to or completely changed their planning models and
methods to properly address climate change and other future challenges. This companion white
paper, Embracing Uncertainty: A Case Study Examination of How Climate Change is Shifting
Water Utility Planning, is also available online (http://www.wucaonline.org).
Formally, the goal of the PUMA project is …to identify state-of-the-art modeling tools and
techniques that can be used by water utilities to assess potential climate change impacts on their
systems and watersheds.” However, the motivation for the PUMA project also includes
collaborating with climate scientists to generate an applied research agenda developed through
the experience of four member utilities.
1. WUCA member utilities include the Central Arizona Project, Denver Water, the Metropolitan Water District
of Southern California, the New York City Department of Environmental Protection, the Portland Water
Bureau, the San Diego County Water Authority, the San Francisco Public Utilities Commission, Seattle Public
Utilities, the Southern Nevada Water Authority, and Tampa Bay Water.
Page 1
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
The WUCA mission
WUCA provides leadership in assessing and adapting to the potential effects of climate change through
collaborative action. The coalition seeks to enhance the usefulness of climate science for the adaptation
community and to improve water management decision-making in the face of climate uncertainty.
See www.wucaonline.org
.
Figure 1. The 10 utilities in WUCA, formed in 2007, provide drinking water to over 43
million people in the United States.
WUCA’s four steps to adaptation
1. Understanding: Utilities develop an understanding of climate science, climate change projections,
techniques for downscaling projections to regional scales, and the capabilities and limitation of the
science for applied uses. Understanding is also a fundamental outcome for each step in the adaptation
framework, as it continuously evolves and expands as utilities progress through or revisit these steps.
2. Assessing: Utilities use the understanding gained in the first step to perform analyses aimed at identifying
potential impacts on their water systems from climate change and to better appreciate vulnerabilities to
future climate changes.
3. Planning: In light of the looming challenges of climate change, utilities begin incorporating climate
science and assessments into water utility planning and identifying adaptation strategies. Often this step
leads utilities to examine the robustness of their planning methods, models, data, and fundamental system
assumptions.
4. Implementing: Utilities make decisions and implement actions aimed at adapting to climate change and
reducing system vulnerabilities.
Page 2
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Actionable science and co-production of knowledge
Early in their work, WUCA members invested significant time interacting with the climate
science community, including university researchers, federal science agencies, and climate
modeling centers. These interactions made obvious the disconnect between the critically
important but highly complex science in the peer-reviewed literature, and the need for climate
information on the part of decision-makers such as water utilities. Utility decision-makers are
engineers, planners, appointed board members, and elected officials, and none of these audiences
spend significant time reading the peer-reviewed literature. Climate science leaders are tenured
or tenure-track academics, Intergovernmental Panel on Climate Change (IPCC) authors, and
federal science agency officials, often from academia themselves, who historically spend little
time interacting with decision-makers. Cultural, communication, and even linguistic differences
between these communities can be profound – and the reward systems of each group do not
generally incentivize working with the other.
In this environment, WUCA sought to define its members’ science, data, and climate service
needs and identified a term to convey these needs: “actionable science.” Introduced at a
U.S. Environmental Protection Agency adaptation conference in January 2009, actionable
science was defined as follows:
Data, analysis, and forecasts that are sufficiently predictive, accepted, and
understandable to support decision-making, including capital investment
decision-making.
The definition, including italicized emphases, was carefully-crafted to make it clear that
decision-makers were seeking high-quality and understandable information, but not perfect
information (“sufficiently); that science was needed to inform decision-making, but not
dictate action (“support”); and that the stakes for actionable science for utilities involve
potentially expensive infrastructure investments using taxpayer and ratepayer dollars
(“capital investment”). In subsequent months and years, the term actionable science in one
form or another has been embraced by a range of entities endeavoring to respond to the
needs of society for usable climate information. These entities include the U.S. Army Corps
of Engineers, a federal agency consortium called the Climate Change and Water Working
Group (CCAWWG),
2
the U.S. Global Change Research Program (USGCRP), the Global
Framework for Climate Services, and, most recently, the Presidents Climate Action Plan
and the Executive Order 13653 announced in November 2013 (WMO, 2011; USACE, 2012;
2. CCAWWG members include the U.S. Army Corps of Engineers, the U.S. Bureau of Reclamation, the
U.S. Geological Survey, the Federal Emergency Management Agency, the National Oceanic and Atmospheric
Administration, and the U.S. Environmental Protection Agency.
Page 3
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
USGCRP, 2012; EO 13653, 2013; Executive Office of the President, 2013; Raff et al.,
2013).
In 2014, a federal advisory committee appointed to advise the Secretary of the Interior on
department programs providing adaptation science drafted a definition of actionable science that
builds upon the WUCA definition. Although not final at the time of this report’s publication, this
definition has been circulating and has appeared in literature, including recent USGCRP
planning documents. This new definition states:
Actionable science provides data, analyses, projections, or tools that can support
decisions regarding management of the risks and impacts of climate change. It is
ideally co-produced by scientists and decision-makers and creates rigorous and
accessible products to meet the needs of stakeholders (ACCCNRS, 2014).
The new definition adds the term “co-production” which also appears in the title of this paper,
and is intended to convey the idea that science in service of adaptation is not a one-way street,
but a collaborative venture between scientists and decision-makers in which the needs and skills
of each come into play throughout that collaboration. Co-production best practice precludes the
“loading dock” model, wherein climate information is generated without the input of the
decision-maker and then provided to that decision-maker, ostensibly ending the responsibility of
the “provider.” Similarly, co-production means that a decision-maker cannot simply describe his
or her needs and expect the scientist to simply fulfill them. Co-production requires an iterative,
collaborative process across the borders between science and policy that draws upon the unique
needs, experience, and even the limitations of each party, providing the strongest possible
underpinning for societal action in response to the consequences of climate change.
The PUMA project arose from these conversations and was envisioned essentially as an effort to
produce actionable science in a co-production environment to meet the needs of four water
utilities. Instead of asking climate experts what they thought utilities should do regarding climate
change, four WUCA utilities agreed to forge partnerships with scientific institutions to explore
how to integrate climate considerations into their specific management context. This report
documents those four utilities’ experience between the start of the PUMA project in 2010 and the
writing of this report in late 2014. A fundamental goal of this report is to draw lessons from these
four distinct projects regarding best practices in the co-production of actionable science. We
attempt to display how each team went about tailoring climate information to specific decision-
maker needs, show what worked and did not work, and inform future research and investment in
the boundaries between climate science and adaptation decision-making.
Our intent is to cover the most important features and lessons learned, project-by-project, at the
time of writing this report. We invite interested parties seeking more detail or information on the
evolution of the projects after this report was written to contact individual project leaders;
contact information appears in Appendix B.
Page 4
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
2. PUMA Project Process
For the PUMA project, four utilities engaged in a modeling
process to better understand how climate changes might affect
their water systems through a chain-of-models” exercise. In
addition, one utility, Seattle Public Utilities (SPU), deployed a
“bottom up,” metrics/threshold approach to querying downscaled
climate data to understand how frequently existing thresholds of
interest to SPU are going to be exceeded in the future to create
climate storylines.” The chain-of-models concept refers to the
sequence of models that experts use to apply climate information
to a water utility context. For example, the outputs from general
circulation models (GCMs), which are also called global climate
models, become inputs into techniques for increasing the
resolution of GCM data (commonly called downscaling); the
resulting outputs become inputs into a hydrology model; the
outputs of the hydrology model become inputs into a system
management or operations model; and the outputs of the utility
models can help define the climate change impacts on water
supply, water quality, and other parameters that water utilities
commonly evaluate to facilitate planning (see Figure 2). In this
way, GCM data can help identify potential climate change
impacts on water system performance. For three of the utilities,
this involved determining how to increase the resolution of GCM
outputs and integrate them into their existing utility models; for
one utility, Portland Water Bureau (PWB), the work focused on
considering climate change in selecting and developing a
hydrologic model and obtaining downscaled climate data to
enable future work on climate change.
In this report, we do not detail specific potential water supply
impacts at each pilot utility. Instead, the main purpose of the
PUMA project is broader: to understand how climate modeling
projections, when used in conjunction with existing utility
management tools, can help address utility planning needs, to
explain how utilities can choose to use that information to
support adaptation, and to explore the nature of productive
collaboration between climate scientists and decision-makers.
This report addresses the following topics:
Figure 2. Illustration of the
chain
-of-models concept.
Page 5
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
The climate modeling tools that the four utilities selected to obtain climate data and why,
the climate modeling tools they did not select and why, and how the utilities obtained
climate data to use in their assessments
Experiences in incorporating the data into hydrologic modeling tools to project impacts
of climate change on water resources and resource management
The value of the results in decision-making and potential next steps
Lessons learned in the effort to bridge the gap between climate science providers and
climate science users.
The utilities wanted to engage in this process in parallel with each other so they could share their
experiences and learn from each other, so they could provide a roadmap for peer utilities that
might consider engaging in a similar effort, and so interested scientists and climate service
organizations could learn more about how to effectively translate climate science for adaptation
planning purposes. The PUMA project has accomplished this to date by holding regular
conference calls throughout the course of the project, including both utility and science leads for
all four PUMA projects. To track their experience with this process, WUCA hired Stratus
Consulting to act as an independent historian and observer of the process that unfolded at each
utility over the subsequent 37 months. The Stratus Consulting team sent out a series of three or
four surveys to each utility and engaged in a series of follow-up interviews with the PUMA
teams to gather information about how each project evolved over time.
The remainder of this report profiles the experience of each utility’s PUMA project. This
includes a brief project summary, several key issues of interest that arose in the project, and how
the PUMA project is expected to affect utility decisions. These profiles are not intended to
provide an exhaustive list of every issue that each utility addressed, but rather to call out some of
the more interesting and insightful experiences of each utility. The report ends with Conclusions
for an Applied Research Agenda for Climate Services, which draws on the lessons these four
utilities learned and what those lessons mean for the coproduction of knowledge between climate
modelers and water utilities.
For readers who may not yet be familiar with climate models and downscaling techniques, we
recommend that you first turn to Appendix A, Applying Climate Model Outputs 101 for Water
Utilities. This appendix provides a technical overview of the global climate modeling process,
including discussions of model selection; reconciling large-scale versus local-scale climatic
processes, commonly referred to as bias correction and downscaling; issues concerning time
steps and time periods; and other topics. This basic discussion of the application of climate
model outputs provides an adequate basis for understanding the context in which the four utilities
made decisions during the course of the PUMA project.
Page 6
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3. PUMA Project Outcomes: Four Utility
Project Profiles
The PUMA project leadership team selected four utilities to participate in the PUMA project, all
with varying characteristics in areas such as previous experience assessing climate change,
service area size, and primary climate change impact of concern. The four utilities selected were:
New York City Department of Environmental Protection (NYCDEP)
Portland Water Bureau (PWB)
Seattle Public Utilities (SPU)
Tampa Bay Water (TBW).
Each of these utilities partnered with local scientific climate change experts, many of whom are
part of the National Oceanic and Atmospheric Administrations (NOAAs) Regional Integrated
Sciences and Assessments (RISA) program. Each science partner played a different role, but in
general the science partners helped select, obtain data from, and better resolve the GCMs and
GCM data, based on extensive discussion of each utility’s precise needs. The PUMA utilities and
their scientific partners were:
NYCDEP the City University of New York (CUNY) Institute for Sustainable Cities,
Columbia University, the NASA Goddard Institute for Space Studies (NASA-GISS),
Cornell University, and the Consortium for Climate Risk in the Urban Northeast
(CCRUN)
PWB the Pacific Northwest Climate Impacts Research Consortium (CIRC), which
includes the University of Idaho (UID), the University of Washington (UW), and Oregon
State University (OSU)
SPU – CIRC, which includes UID and OSU
TBW – the Southeast Climate Consortium (SECC), which includes the University of
Florida (UF) and the Florida State University (FSU) Center for Ocean-Atmospheric
Prediction Studies (COAPS).
Page 7
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3.1 New York City Department of Environmental Protection
NYCDEP overview
Number of customers: 9.2 million
Gallons of water produced per day: 1.1 billion
Service types: Drinking water supply, wastewater and storm water management
Supply sources: Mainly surface, with access to groundwater
Water treatment: Not filtered (will be partially filtered once the Croton filtration facility is brought back
online); treated with chlorine, ultraviolet light, phosphoric acid, sodium hydroxide, and fluoride; Alum is
applied during high-turbidity events
Primary concern for climate change: Water quality
Project highlight: New York City created a new delta-change technique to increase the resolution of GCM
data called the statistically distributed (SD)-delta method, which is simple to use yet provides insight on
extreme events
3.1.1 NYCDEPs PUMA project summary
NYCDEPs PUMA project exemplifies a well-resourced and sophisticated utility that has
explored complex scientific methods without choosing the most expensive or most complicated
technique. Ultimately, NYCDEP used a relatively simple technique to increase the resolution of
climate model data, using already developed hydrologic, reservoir water quality, and
management models as a pragmatic way to start looking at climate change impacts. NYCDEP
saw its PUMA project as part of an internally developed Climate Change Integrated Modeling
Project that was already underway before the start of PUMA.
Through its PUMA project and preceding climate change work, NYCDEP has focused on
identifying possible impacts and then developing operational policies to minimize the effects of
those impacts. They were particularly concerned about impacts that would affect their supply,
quality, and operations, such as changes in the timing of winter run-off, reduction in winter
snowpack, changes in the thermal structure of reservoirs, and an increase in the severity of
extreme events.
Although NYCDEPs climate change work has generated useful insights on changes in winter
precipitation and peak runoff, the utility believes these water supply issues are manageable.
NYCDEPs more challenging concern lies in water-quality issues, such as high turbidity driven
by extreme precipitation. NYCDEP has found in its PUMA-related work that examining the
effects of extreme precipitation events on water quality under climate change is particularly
challenging because current climate models provide only limited information on this topic.
Page 8
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3.1.2 NYCDEP issues of interest
NYCDEP issue 1: Selecting GCMs
The process by which NYCDEP decided on which GCM data to use is instructive. The New
York City PUMA team members initially selected four GCMs
3
for which they could easily
access model output online. The team members evaluated each models fit for their region by
using a probability-based skill score comparing baseline GCM outputs with historical data.
However, after analyzing the data from the four models, they discovered that one model had the
best fit for temperature while a different model had the best fit for precipitation. Thus, they could
not find one best model.
This led NYCDEP to shift their strategy and pull data from a larger suite of GCMs. This time,
team members first considered which models provided data for the variables needed to run their
existing management models. Variables included air temperature (average, maximum, and
minimum), precipitation, solar radiation, and wind speed, all at daily time steps. Based on these
needs, NYCDEP was able to select a subset of 15 to 20 Coupled Model Intercomparison Project
Phase 3 (CMIP3) GCMs from which to pull data for its analysis. At this point, NYCDEP needed
to convert GCM output to higher resolution to compare it against historical station data.
NYCDEP issue 2: Developing future climate scenarios
NYCDEP selected the delta-change method
4
to develop future climate scenarios from GCM
output. These GCM scenarios provided a daily time series of derived meteorological variables
that served as the driving data behind the hydrologic and water quality models. The major
advantage of the delta-change method was the ease and speed of application, and the direct
scaling of local historical observations to form a scenario based on changes suggested by the
GCM simulations. One disadvantage to this approach was that the temporal sequencing of storm
events in the derived time series remained unchanged from the historical record; this method
may not be helpful in circumstances where changes in event frequency and antecedent conditions
are important to the impact assessment.
However, NYCDEP felt more confident using historical events, altering the events’ magnitude as
a foundation for hypothetical future events. This approach empowered utility staff to apply their
own institutional knowledge of past events when considering climate change, and enabled
NYCDEP to develop hypothetical droughts or storms. The utility could then ask broad questions
3. CCGCM, GISS, CCSM3, and ECHAM5/MPI-OM.
4. In the delta-change method, an additive (or in the case of precipitation, multiplicative) change factor is
calculated as the difference (or in the case of precipitation, the ratio) between a GCM variable derived from a
current climate simulation (GCM baseline) and derived from a future climate scenario (GCM future”) taken
at the same GCM grid location.
Page 9
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
such as, “What if the drought of the 1960s happened again, but was more intense because of
climate change? If New York experienced a hurricane followed by a tropical storm, as it did in
2011, how would this type of event be different under an altered future climate?
NYCDEP issue 3: Addressing water quality issues caused by extreme events
NYCDEP was motivated in large part by the potential impacts of climate change on water
quality, especially by causing high turbidity in NYCDEP’s unfiltered water supply. Turbidity
must remain below a certain level to maintain the utility’s Filtration Avoidance Determination.
Other water quality issues, such as climate impacts on disinfection byproducts and dissolved
organic carbon, also motivated NYCDEP. All of these water-quality impacts are generally driven
by extreme events, such as intense rain events or extended periods of very high temperatures.
Notably, such extreme events can also affect operations and damage infrastructure, and,
according to NYCDEP, were worth understanding for those reasons as well.
Currently, climate scientists provide many extreme event projections in qualitative terms because
GCMs cannot sufficiently capture quantitative changes in extreme events. For example, the
intensity of hurricanes is expected to increase because more energy will be available from
warmer sea surface temperatures. However, GCMs cannot currently quantify such increases in
intensity precisely.
NYCDEP considered these qualitative statements regarding changes in extremes useful from a
planning perspective, but not useful in a chain-of-models analysis, which requires numeric
inputs. This led NYCDEP to develop a method that could be used in a chain-of-models analysis
that also represented extreme eventsa SD-delta method:
In the SD-delta method, modelers derive multiple change factors for each month across
the cumulative distribution functions (CDFs) of daily meteorology values from hindcasts
and projections from GCMs. Examples of meteorology values are total daily precipitation
and maximum air temperature.
To derive the series of change factors, the CDF is divided into a series of bins of daily
meteorology values, with a change factor calculated specifically for each bin. The method
can be adapted for bins of any size.
To develop a climate-altered time series for a meteorological variable, these monthly
change factors are then applied to observed data for the location of interest. When the
change factors for the bins at the tails of the CDF are used, this provides information
about extreme events.
Page 10
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
During the initial application of the SD-delta method, NYCDEP resolved the CDFs of GCM
variables into 25 bins, each including 4% of the data. NYCDEP applied the SD-delta method to
assess the effects of projected climate change on turbidity. For example, the wettest 4% of days
may lead to significant increases in precipitation and thus increased streamflow and turbidity. As
shown in Figure 3, the result was a significant increase in projected winter turbidity.
Figure 3. Comparison of NYCDEP’s mean monthly observed turbidity (black line) in
Ashokan Reservoir – East Basin with range of simulated 2080–2100 turbidity from five
climate change scenarios (maroon bars).
Although the SD-delta method is useful in extracting projected changes in the intensity of
extreme events from climate model output, it retains one well-documented drawback in common
with the conventional delta-change method. In both the conventional delta-change and SD-delta
methods, the sequencing of events in future scenarios is determined by the sequencing of events
in the historical record. Such replication of weather sequencing in the future is not realistic.
To address this shortcoming, NYCDEP has begun looking at stochastic weather generators
5
to
investigate the effects of extreme events. Stochastic models enable the evaluation of future
scenarios with event sequencing that is not driven by the historical record.
5. A stochastic weather generator is a tool that produces a synthetic time series of weather data for a location,
based on the statistical characteristics of observed weather at that location.
0
1
2
3
4
5
Jan Feb Mar
Apr May Jun
Jul Aug Sep Oct Nov Dec
Page 11
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
NYCDEP issue 4: Bringing scientific expertise in house
NYCDEP has a rich history of diverse partnerships with external climate scientists and experts.
Starting in 2003, NYCDEP’s Bureau of Environmental Planning and Analysis created a Climate
Change Task Force that consisted of members within NYCDEP as well as outside participants,
such as climate and impact scientists from Columbia University; an environmental engineering
consulting firm, HydroQual; and the New York City Office of Environmental Coordination, the
Office of Long-term Planning and Sustainability, and the Law Department. The task force
conducted an adaptation assessment that was published in 2008.
More recently, NYCDEP has partnered with scientists at Cornell University and it also continues
to work collaboratively with the local RISA,
6
the CCRUN, on a number of collaborative
projects. Most notably, however, NYCDEP has developed an extensive internal network of
scientific capacity. In a unique partnership, the City University of New York (CUNY) Institute
for Sustainable Cities hires full-time post-doctoral scientists to work with NYCDEP staff at
NYCDEP offices. NYCDEP scientists, along with faculty from various support universities who
serve as part-time project advisors, then closely oversee the work of the post-doctoral scientists.
The program has allowed NYCDEP to maintain a broader and more continuous scope in its
climate change work and provides a mechanism for technology and knowledge transfer from
scientists with state-of-the-art expertise to NYCDEP staff scientists. Although post-doctoral
scientist turnover can be problematic, NYCDEP staff judge this program as a major success and
as value-added to its climate impacts work.
3.1.3 How the PUMA project will affect utility decisions
The climate change team members at NYCDEP are optimistic that their findings from the
PUMA project will influence decision-making both within and outside their utility. Within
NYCDEP, where possible, the team proactively encouraged climate change considerations. For
instance, they started voluntarily and proactively including the results of their research in a report
capturing progress on source water protection to maintain the U.S. Environmental Protection
Agency’s Filtration Avoidance Determination. In recent years, the NYCDEP climate change
team noticed an increase in questions about climate change impacts from within the utility.
Superstorm Sandy, Tropical Storms Irene and Lee, and other extreme events in the area, such as
the 2007 floods in New York City, have contributed to the increased discussion around climate
change impacts. These extreme events have heightened interest about climate change concerns at
the highest levels within NYCDEP and increased support of climate change research within the
utility. However, because many of NYCDEP’s ongoing modeling results still involve a fair
6. NOAA’s RISA program supports research teams that help expand and build the nation’s capacity to prepare
for and adapt to climate variability and change. RISA is explicitly charged with collaborating and partnering
with public and private climate data user communities.
Page 12
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
amount of uncertainty, it is not yet undertaking any large, climate-specific operational or
management decisions for water supply based on these results, but is instead engaging in a no-
regrets strategy as described in the box below.
7
Decision-makers at NYCDEP have also started to get questions from outside the organization
regarding climate change. They have participated in region-wide post-Sandy meetings about how
climate change could affect regulatory decisions to increase preparedness for similar types of
extreme events in the future. These meetings have created opportunities to leverage the results
from their PUMA project and other climate change research directly into the decision-making
process both within and outside the utility.
NYCDEPs climate change modeling work allowed the team to have answers ready for internal
and external inquires, and reinforced the climate change team’s emphasis on no-regrets
adaptation planning when possible (see the text box, Example of a no-regrets strategy in
NYCDEP).
Example of a no-regrets strategy in NYCDEP
No-regrets strategies are those that provide benefits under current climate conditions and potential future
climate conditions. When a utility spends money on a no-regrets strategy, it will reduce the utility’s risk to
existing utility stressors while making it more resilient to future climate change, ensuring the investment is
worthwhile regardless of which climate future unfolds. For example, a main focus within NYCDEP over the
past five to six years has been to invest in increasing its ability to blend water. Blending means combining
water from various sources to meet water quality requirements. To expand its ability to blend water,
NYCDEP has built a better operations modeling system and embarked on infrastructure projects such as
connecting aqueducts that expand their capacity to draw and mix water from multiple sources. These
approaches help provide NYCDEP with a greater flexibility to address a multitude of current and future
challenges to its system, including the impacts of climate change. When NYCDEP climate change team
members present information to their decision-makers, they encourage no-regrets strategies when possible.
7. Note that NYCDEP has already undertaken major projects to protect wastewater facilities from storm surge
and sea level rise, where the direction of change is clearer and the need more immediate.
Page 13
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3.2 Portland Water Bureau
PWB overview
Number of customers: 935,000
Gallons of water produced per day: 101 million
Service type: Drinking water supply
Supply sources: Surface and groundwater
Water treatment: Not filtered; treated with chloramine
Primary concern for climate change: Shifts in the hydrograph for the primary surface water supply that
could affect the timing and length of reservoir drawdown and changes in extreme precipitation events that
could affect turbidity; both of these could influence use of the secondary groundwater supply
Project highlight: PWB evaluated a range of hydrology models against its needs and selected one within
budget that can integrate climate modeling data for long-term planning purposes
3.2.1 PWBs PUMA project summary
PWB first examined potential climate change impacts to its primary surface water supply, the
Bull Run watershed, through a 2001 study. In the PUMA project, the utility wanted to evaluate
results from the newer generation of climate models to see what, if anything, had changed.
Although its overall goal was similar to other utilities, the PWB team was able to approach the
work from a unique perspective. Because the utility did not yet have an existing in-house
hydrologic model, PWB was in a better position to begin its discussions with a focus on
outcomes, without worrying about process or sunken costs in an existing utility hydrologic
model.
PWB’s team members decided they wanted to focus their PUMA research on a long-term
evaluation of the impacts of climate change on the Bull Run watershed. With this objective in
mind, the members set out to select a hydrologic model that would best fit their needs, and which
could be used to translate GCM outputs, such as temperature and precipitation, into future stream
flow projections that could inform long-term planning.
PWB has never used a hydrologic model to conduct day-to-day operations, and has instead relied
on internal modeling tools for operational purposes. However, PWB needed a hydrologic model
for its climate impacts assessment. At first, PWB considered implementing the same hydrologic
model as a peer utility, but that hydrologic model was expensive and complicated to use because
it was designed for operational purposes. PWB team members decided that they did not need to
invest in a complex hydrologic model designed to support operations to investigate the impacts
of climate change on their watershed. However, they decided the ability to investigate climate
Page 14
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
change impacts was worth the investment in a relatively inexpensive and easy-to-use hydrologic
model customized for the Bull Run watershed. Their model review revealed that they would be
able to obtain answers to their research questions without investing in an expensive hydrologic
model.
3.2.2 PWB issues of interest
PWB issue 1: Selecting a hydrologic model
To select a hydrologic model that could help PWB investigate the impacts of climate change to
its system, the PWB team worked with a group of hydrologic modelers from UW. The PWB and
UW team held a series of workshops to select a hydrology model that included participation by
several bureau staff members and modelers from both UW and UID.
The first workshop evaluated eight possible hydrologic models according to the following
criteria:
Non-proprietary software
Ability to process multiple runs through scripting
Ability to simulate hydrologic processes at appropriate spatial and temporal scales
General ease of setup and use
Reputation of the model and use in other Northwest climate change studies
Cost to the utility, including initial, setup, and operating costs.
The initial workshop resulted in the selection of three models for further evaluation: (1) the
Distributed Hydrology-Soil-Vegetation Model (DHSVM), (2) the Precipitation-runoff Modeling
System (PRMS), and (3) the Variable Infiltration Capacity (VIC) model.
Following this workshop, UW calibrated and validated the three hydrologic models for the Bull
Run watershed to compare observed streamflows to simulated streamflows from the three
models. UW assessed the performance of these models (called model skill) against several
statistical measures, including percent bias and Nash-Sutcliffe Efficiency, to determine model
goodness-of-fit over a 30-year period (1976–2005).
At the second workshop, UW presented the results of its statistical analysis, and UID shared
information on the appropriate use of climate data. Participants evaluated the three hydrology
model finalists primarily according to three criteria in addition to model skill: the cost of the
model, the ease of use by PWB personnel, and the amount of time required to run the model (see
Figure 4 for the hydrograph comparison of the three hydrologic models assessed by PWB; note
the variation in spring runoff across models). For model skill, the PWB team focused on the
models’ abilities to replicate the historical annual and monthly hydrographs, particularly peak
Page 15
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
flows. The PWB team determined that peak flows were of greatest concern in the hydrograph
because they can lead to turbidity and water-quality concerns. At the end of the workshop, the
PWB team selected the PRMS model, which was developed by the U.S. Geological Survey
(USGS) and is used in hydrologic studies, many related to climate change.
Figure 4. Comparison of PWB’s mean monthly observed streamflows with simulated
streamflows from three hydrologic models.
PRMS was chosen in part because of its relative affordability and ability to provide the necessary
outputs for PWB’s needs. PRMS has a good fit with daily and annual flows, and the best fit with
monthly flows (Figure 4). It also has the benefit of ongoing technical support from USGS. The
PWB team can also run the PRMS model relatively quickly, which facilitates a fast learning
curve and allows the utility to quickly process a large amount of GCM data for future in-house
climate impacts analyses.
At the third and final workshop, UW scientists trained PWB staff on the PRMS hydrologic
model and on integrating the model with the climate modeling data provided by the UID team.
By the end of the workshop, several staff members could run the model.
Page 16
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
PWB issue 2: Difficulties developing a hydrologic dataset
To set up and run the PRMS model, the PWB team used regional topography, soil, and
vegetation data from geographic information system (GIS) datasets. However, because PWB has
only one meteorological station in the watershed, the UW team recommended the use of a
historical gridded meteorological dataset (Livneh et al., 2013) to calibrate the model against
historical stream flows for the watershed. One meteorological station would not capture the
topographic climate gradients within the entire watershed, especially as they relate to air
temperature and the amount and phase of precipitation (i.e., snow or rain). The Livneh et al.
dataset is at a 6-kilometer resolution and is the most current dataset available for the coterminous
United States. It is an update to the widely used Maurer et al. (2007) dataset, and applies the
same methodology. It uses 1915–2011 historical climatology data from NOAAs Cooperative
Observer Program, as well as data from the Parameter-elevation Relationships on Independent
Slopes Model (PRISM).
The PRMS model calibration initially showed persistent differences in average monthly flows
between PRMS and the historical data. These differences were because of bias in the Livneh
et al. dataset, not because of problems with the PRMS model. While the Livneh et al. dataset is a
robust dataset, its resolution did not capture the micro-climate and orographic effects
8
present in
the Bull Run watershed, which is located on the flank of Mount Hood and near the Columbia
River Gorge. Specifically, the dataset did not accurately represent the water balance of the
watershed, in terms of amount of precipitation, evapotranspiration, and runoff. Consequently, the
UW team needed to bias-correct
9
the Livneh et al. dataset to fit historical conditions in PWBs
watershed, mainly by increasing average monthly precipitation.
The initial calibration of PRMS thus required more effort than expected, in large part because
good historical data, at the scale required for the watershed, were not readily available. These
historical data had to be developed and bias corrected, as described above, to enable further
climate-impact studies. With the bias correction to the Livneh et al. dataset accomplished, the
PWB team could compare historical stream flows from the hydrologic model with observed
streamflows and fine-tune the model to ensure the best fit. With this task completed, the PWB
team is now able to run downscaled temperature and precipitation data from the GCMs through
the PRMS model.
8. An orographic effect is a change in atmospheric conditions (e.g., humidity, precipitation, wind speed)
caused by a change in elevation, typically related to mountain ranges.
9. Note that this type of bias correction is not referring to the climate models, but to correcting historical data
for the hydrologic model.
Page 17
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
PWB issue 3: Selecting GCMs and increasing data resolution
While PWB staff did not spend time experimenting with GCM data during the PUMA project,
they did work with climate modelers at UID to develop downscaled GCM data for their use. At
first, UID provided PWB with a subset of five GCMs that were representative of the multi-model
mean and extremes, for use if PWB did not have the resources to run more GCMs.
However, given the characteristics of the selected PRMS hydrologic model, the PWB team
decided to use all available downscaled GCMs provided by UID as inputs into PRMS to develop
climate-altered hydrologies for future time periods. The main reason for this is that several
climate scientists involved in this project, including Dr. Phil Mote with the CIRC (also the
PUMA science and climate service lead since the outset of the project), advocated for the use of
at least 10 GCMs to capture the range of uncertainty in future climate offered by current climate
models.
The UID team chose a 20-model subset of the 41 Coupled Model Intercomparison Project
Phase 5 (CMIP5) models because they provide daily climate data for the following variables:
minimum temperature, maximum temperature, precipitation, downwelling shortwave radiation,
specific humidity, maximum relative humidity, minimum relative humidity, and surface wind
components. These variables were necessary for the UID team to downscale the GCMs using the
Multivariate Adaptive Constructed Analog (MACA) method. Not all GCMs provide these
variables at a daily time step. The MACA datasets include historical (19502005) and future
(2006–2100) time periods. Furthermore, an important component of the MACA process was the
use of the bias-corrected Livneh et al. dataset in training the GCMs during downscaling.
Normally, MACA uses a 4-kilometer resolution historical training dataset for the GCMs.
However, in this case, because UW had essentially customized the Livneh et al. dataset to
6-kilometers for the Bull Run watershed, the UID team felt that using this dataset to train the
GCMs would lead to more location-specific downscaled data.
10
UID also provided PWB with a ranking of the original, non-downscaled, 20 GCMs based on
how well they represented the historical climate record for the Northwest. This information will
be useful to PWB as it conducts its in-house analysis; it will be a way to quality check the
downscaled data. If some downscaled datasets do not result in a good historical fit with observed
streamflows, it may be because the relevant GCM has a low ranking in terms of fit to historical
climate in the Northwest.
10. Note that PWB and SPU used the same statically downscaled datasets (MACA and Livneh et al.), but
customized each differently based on their different watershed context and hydrologic modeling systems. Due
to overlap in the researchers working on both the PWB and SPU projects, both utilities received consistent
datasets, which allows them to share results with each other with some degree of comparability.
Page 18
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
PWB issue 4: Developing in-house capacity
From the beginning, PWB was motivated to build institutional capacity through their PUMA
project. Partly because PWB was unable to update the results of their 2001 climate impacts study
internally, utility personnel decided that they should use the PUMA project as an opportunity to
develop their technical in-house expertise and capability to do future climate impacts analyses,
instead of relying on consultants and climate scientists alone. This required not only selecting a
hydrologic model that was easy to use, but also training personnel in the use of that hydrologic
model and in climate impacts analysis.
Through this experience, PWB came to several realizations about the interplay between climate
science and water utility experts. First, the PWB team found it necessary to simplify some of the
statistical and technical information that the climate scientists provided to solicit input from other
utility staff who were not as familiar with climate or hydrologic models. PWB also learned that
educating climate scientists about the qualities and operation of the PWB water system was
necessary to help the scientists generate applicable hydrologic and climate inputs. A key lesson
for PWB was not to assume that climate scientists understand water utility operations and
modeling environments.
3.2.3 How the PUMA project will affect utility decisions
The main objective of PWB’s PUMA project was to select a hydrologic model that could help
the utility with future climate change impacts research. With a hydrologic model selected and
GCMs identified, downscaled, and refined, PWB plans to continue its climate assessment work
by running the outputs from the full suite of 20 GCMs through the PRMS model.
Going forward, PWB team members will focus on understanding how they might portray and
use the model results in the future, including how to integrate the results into PWB’s long-range
planning processes. Additionally, the City of Portland has completed a Climate Change
Preparation Strategy, which includes actions that PWB will take to help prepare for the impacts
of climate change. Results from this project could inform the update to this strategy in future
years.
Page 19
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3.3 Seattle Public Utilities
SPU overview
Number of customers: 1.3 million
Gallons of water produced per day: 120 million
Service types: Drinking water supply; wastewater, solid waste, and storm water management
Supply sources: Surface and groundwater for peak season and emergency use
Water treatment: SPUs largest water source, the Cedar watershed, is unfiltered and all water is treated with
chlorine, ozone, and ultraviolet light; SPUs second surface water source, the Tolt watershed, is filtered and
treated with ozone and chlorine. Both sources provide for corrosion control and fluoridation.
Primary concern for climate change: Impacts of climate change on water availability/reliability, impacts of
climate change on conditions of operational interest to SPU (e.g., number of days with precipitation greater
than 2 inches), timing of the onset of fall rains and the effects on reservoir refill strategies, atmospheric
rivers (corridors of moisture in the atmosphere) and their effects on flooding, and possible increases in
forest fire incidence or intensity on protected watersheds.
Project highlight: SPU is using the chain-of-models approach for assessing impacts on supply through a co-
production model, and is generating the climate-altered hydrology and water supply impacts assessment in-
house. It is complementing this with a “bottom-up” approach using SPU-specific metrics to query the
downscaled climate projections and generate climate storylinesthat have operational implications for
SPU and that help to describe projected changes in the climate. SPU was still completing its analysis as this
report went to press; this summary is preliminary and subject to change.
3.3.1 SPUs PUMA project summary
SPU has conducted three climate impact assessments of its water supply, with the most recent
one completed in 2008. The utility decided to use its PUMA project as an opportunity to update
this 2008 study
11
by using CMIP5 climate model outputs, increasing the sophistication of its
analysis by increasing the number of climate models considered and by “rounding out” the
traditional focus on water supply to examine other, complementary research questions. The
utility also viewed the project as an opportunity to build institutional capacity by assuming the
responsibility of generating climate-altered hydrologies in-house using downscaled
meteorological data, as well as the water supply impacts analysis. SPU was particularly
interested in four questions:
11. The previous 2008 SPU study used three GCMs and two emissions scenarios. The spatial resolution of that
data was increased by using a delta-change technique with quantile mapping applied to the historical
meteorological record. Furthermore, the mapping (i.e., transformation functions) were not uniform across a
grid cell and were different for each meteorological station.
Page 20
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
1. How might climate change affect long-term water supply availability or reliability?
2. How will climate change affect some baseline conditions of operational interest to SPU?
3. What effect will climate change have on the timing of the onset of fall rains and on
atmospheric rivers?
4. How might climate change affect forest fire frequency and intensity?
SPU’s PUMA project amounted to a sophisticated chain-of-models approach complemented by a
bottom-up analysis of the downscaled meteorological data, GCM outputs, and literature review.
In addition, two issues SPU wanted to explore through the PUMA project were the effect, if any,
that climate change would have on the timing of the return of fall rains and on the advent of
atmospheric rivers. These two phenomena are of operational and long-term planning interest to
SPU:
SPU has a relatively small storage capacity in its two reservoirs compared to other water
systems, which may have multi-year storage. As a result, SPU largely depends upon an
annual refill cycle that is initiated by the return of fall rains. A delay in the fall rains
could pose some challenges for SPU.
Atmospheric rivers have a dual nature from SPU’s perspective. The sustained, large
volumes of precipitation they generate can create challenges for SPU’s drainage,
wastewater, and drinking water systems. However, atmospheric rivers can also play a
“drought buster” role, providing much-needed precipitation in the midst of a dry period.
In the winter of 2014, SPU witnessed this second quality: a series of four minor
atmospheric rivers arrived in Washington State just as the state was contemplating
declaring a drought. The storms increased SPU’s supply/snowpack roughly twofold in
approximately four weeks.
3.3.2 SPU issues of interest
SPU issue 1: Using an operational metric/threshold approach
With their PUMA project, SPU decided to complement the top-down chain-of-models approach
with a bottom-up approach in which the utility identified operational metrics/thresholds of
relevance and those to query the downscaled projections. For example, SPU identified a 24-hour,
2-inch precipitation event as a metric for long-duration events that generally cause problems for
its drainage and wastewater system. The utility also identified multiple consecutive days over
80°F with no precipitation as a metric of conditions that raise concerns about possible forest
Page 21
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
fires. Nearly 30 metrics were developed, some of which were developed in consultation with
other city departments, such as Seattle City Light and the Department of Transportation.
The metric/threshold approach helped make the climate projections utility-specific, defining
variables of concern to monitor over time. This approach may make climate projections more
relevant by placing them in the context of how climate change may exacerbate current
vulnerabilities. Finally, SPU thought that the use of operational metrics/thresholds could help
develop “climate storylines” (i.e., descriptions of how current conditions of interest to SPU are
expected to change in the future). The utility felt that the storylines would be a nice complement
to the chain-of-models analysis and facilitate communication about the potential effects of
climate change on its systems and services.
SPU issue 2: Challenges in disaggregating daily data
SPU’s hydrology model HFAM requires hourly data to run, but the MACA-downscaled GCM
data the utility was using were recorded as daily values. As a consequence, the SPU team used
the HFAM guidelines on temporal disaggregation of daily data. When hourly data were missing,
these guidelines call for allocating the daily precipitation amount equally across the 5 hours
between 4:00 a.m. and 9:00 a.m., with the remaining 19 hourly precipitation values set to zero.
Unfortunately, following these guidelines led to significant discrepancies between observed and
simulated monthly average stream flows. The total annual stream flow was similar. However,
compared to observations the simulation showed less runoff during the early winter months and
more runoff during the spring. The entire hydrograph was incorrect in both magnitude and
timing.
The SPU-CIRC team traced this problem to the method of temporal disaggregation. Because the
4:00 a.m. to 9:00 a.m. period is a particularly cold time of day, choosing that five-hour period to
evenly spread the daily precipitation amount resulted in a bias of too much snow accumulation in
the winter. SPU’s watersheds are in a transition zone, where temperatures regularly cross the
freezing point. The arbitrary precipitation allocation over-predicted precipitation falling as snow
and under-predicted precipitation falling as rain in the winter; this led to predictions of lower
winter flows, more snow available to melt during the spring, and higher spring flows.
SPU plans to resolve this problem by selecting a different temporal disaggregation of
precipitation (e.g., allocating daily precipitation during a warmer time of the day and choosing
the five-hour period that minimizes the error in the hydrograph). CIRC’s sensitivity analysis
identified the 7:00 a.m.–12:00 p.m. window as the one that minimizes the error and yields the
most realistic hydrograph. SPU may explore other methods of temporal disaggregation in future
assessments, after the PUMA project. Although the discrepancy initially caused concern, joint
investigations into the discrepancy illustrated the benefits of co-production and co-learning, as
well as the importance of testing the projections and assumptions.
Page 22
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
SPU issue 3: Developing institutional capacity
In addition to viewing the PUMA project as a mechanism to obtain the best available climate
change projections and enhance its knowledge, SPU also viewed PUMA as an opportunity to
continue to build its institutional capacity. Building adaptive capacity within SPU has been a
primary objective of its climate program since the program launch. Through PUMA, SPU
continued an evolution in SPU’s capacity building that started with its first assessment. In that
assessment, an analysis of the impacts on SPU’s supply was done for SPU by outside experts. In
its second assessment, SPU played a more active role and used climate-altered hydrologies
generated by its collaborating researchers to conduct the impacts analysis in-house. Through the
PUMA project, SPU decided to use the downscaled meteorological projections that CIRC
generated via MACA to develop climate-altered hydrologies in-house as well as the subsequent
water supply impacts analysis. As such, SPU embedded itself in the chain-of-models approach,
which involved more staff time than in past studies, and co-produced knowledge with CIRC (see
Figure 5 for a summary of the evolution of SPU’s climate change assessment).
Figure 5. Representation of the increasing complexity of SPU climate studies and the
increasing capacity of SPU staff to manage the chain of models required to do an
assessment.
Page 23
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3.3.3 How the PUMA project will affect utility decisions
SPU is a leader in integrating climate change science into utility planning. The utility has one
full-time staff person in the Directors Office in charge of the Climate Resiliency Group within
the organization. However, despite the strong support for the issue within SPU, incorporating
climate into its planning and decision-making is still an operational challenge. Big hurdles exist
in integrating climate change into SPU’s line of businesses and bridging the gap between
information generated through assessments and information needed to inform decisions.
Fundamentally, this effort to mainstream climate change into current practices raises important
issues about appropriate uses of climate projections. It also brings to light the need to adjust
existing decision-making frameworks to better address the uncertainties and projected impacts of
climate change.
Some of the concrete ways in which SPU integrates and shares climate change information
within its organization include:
Stage Gates: Stage Gates is SPUs asset-management governance process in which SPU
managers make decisions at specified points (e.g., Stage Gates) in the project conception
and delivery cycle before the project can advance to the next phase or stage. The process
now includes high-level climate questions to ensure that project managers consider
changes to climate in their project evaluations. The PUMA products may be useful for
considering climate impacts with Stage Gates decisions.
Science Talks: SPU holds Science Talks that are open to all SPU staff. These talks
constitute a forum to present the latest climate change information and discuss how it can
be integrated into employees’ work. These talks represent a way to share information and
gain buy-in from other staff.
Strategic Business Plan/Water System Plans: SPU just adopted a six-year strategic
business plan that prominently features climate change. SPU has also used its state-
mandated Water System Plan as a venue to include climate change impact assessments
on long-range water supply plans.
Climate College: SPU is considering starting a series of trainings for utility personnel
concerning different climate-related topics. If this training comes to fruition, it would be
another venue at which the PUMA team could present their findings.
Results of the research on these two phenomena was not available at the time this report was
completed, but SPU intends to share the results as appropriate when they become available and
to integrate that information into utility decisions.
Page 24
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
3.4 Tampa Bay Water
TBW overview
Number of customers: 2.3 million
Gallons produced per day: 220260 million
Types of service: Drinking water supply
Supply sources: Surface, groundwater, desalination
Water treatment: Monochloramines disinfection, pH control for groundwater; filtration/ozone/
monochloramines disinfection for surface water; surface water pretreatment and membrane reverse osmosis
for seawater desalination; all treated waters are blended and disinfected with monochloramines
Primary concern for climate change: Changes in seasonal rainfall patterns both temporally and spatially at
the daily time scale
Project highlight: After conducting a thorough comparative analysis of several techniques for increasing the
resolution of GCM data, TBW developed a new method that captured the spatial-temporal relationships of
rainfall that drive west-central Florida’s hydrology
3.4.1 TBWs PUMA project summary
TBW started considering climate change impacts to its utility before the PUMA project. They
saw the PUMA project as an opportunity to continue its strong partnership with the UF Water
Institute and the SECC. TBW’s goals for the PUMA project were twofold: (1) to increase the
relevance of climate, climate change data, and tools to planning and operating Florida public
water supply utilities in general, and TBW in particular; and (2) to develop collaborative
relationships with climate scientists and hydrologists in academia and NOAAs climate services
to promote development and availability of usable climate data for hydrologic modeling.
TBW did not have an immediate need for climate information developed through its PUMA
project, which allowed the utility to take a more exploratory and deliberative approach to its
project. The TBW team decided to use the project to explore several modeling options and
decide on the best methodology, so that when a major decision needs to be considered using
future projections, the TBW team will have the modeling process ready.
Results of the TBW teams climate modeling efforts showed that every GCM projected an
increase in future temperature, but the models showed differences among future precipitation
estimates. When run through TBW’s integrated surface-water/groundwater hydrologic model,
the differences in precipitation estimates translated into significant differences in future stream
flow projections. The uncertain precipitation signal overwhelmed the more certain temperature
signal in estimating hydrologic implications of projected future changes.
Page 25
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
The TBW team concluded that the spatial-temporal relationships of rainfall were critical to
understanding west-central Florida’s hydrology. Because current methods of increasing the
resolution of GCM data did not incorporate this important relationship, TBW’s PUMA team
developed a new technique to increase the spatial resolution of climate data that would account
for this locally important aspect of rainfall.
3.4.2 TBW issues of interest
TBW issue 1: Increasing the resolution of GCM data
As part of their PUMA project, TBW team members explored several options to convert GCM
data to higher-resolution data. They first considered dynamical downscaling using the MM5
regional climate model (RCM). MM5 was originally evaluated because it was a fully operational
model, the modeling community accepted MM5, and the UF Water Institute’s climate modelers
were familiar with it. However, the TBW team members ultimately decided not to use MM5 for
the following reasons:
They found that it could not reproduce the climatology of the Tampa Bay region without
significant bias-correction
TBW and the UF Water Institute did not have the expertise to improve the raw MM5
predictions for the region
MM5 was a black box model, over which the scientists at UF and TBW had limited
understanding and no control
Running 20 years of simulations took the UF Water Institute a long time; the modeling
team needed consistent access to a high-performance computer.
In the end, the TBW members decided to work more closely with a climate modeler who
understood the physics and parameterization of the RCMs and downscaling, so they sought out
further partnerships with FSU. The team then partnered with the FSU COAPS to conduct two
types of downscaling to see which would produce the most accurate regional climate simulation
for Florida. They looked at both dynamical and statistical approaches.
The team examined the use of dynamical downscaling techniques already in use at COAPS.
These techniques were the FSU COAPS Land-atmosphere Regional Reanalysis downscaling for
the Southeast United States at a 10-kilometer resolution (CLARReS10) and the FSU COAPS
Land-atmosphere Regional Ensemble Climate Change Experiment for the Southeast United
States at a 10-kilometer resolution (CLAREnCE10).
Page 26
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Although the team members discovered that dynamical downscaling methods produced the best
fit for their regions historical climate, the extreme time and computational expense associated
with dynamical downscaling of GCMs made these tools impractical and unwieldy for sustained
use. In the end, the TBW PUMA team decided to use statistical downscaling techniques as the
basis for future analyses, supplemented by three dynamically downscaled future projections
TBW had already developed using CLAREnCE10.
Knowing that the spatial and temporal distributions of precipitation were the key factors in
reproducing west-central Florida’s hydrology, the modeling team compared the following three
statistical downscaling techniques in a retrospective analysis:
The bias-correction and spatial disaggregation (BCSD) method
A modified version of BCSD that reverses the order of spatial disaggregation and bias-
correction (SDBC)
The bias-correction and constructed analog (BCCA) method.
In addition, the modeling team developed a new statistical downscaling technique in an attempt
to capture more information on the spatial and temporal distributions of precipitation:
The bias-correction and stochastic analog (BCSA) method, which Dr. Syewoon Hwang
and Dr. Wendy Graham, both of the UF Water Institute, developed for the project.
The team compared each method’s reproduction of precipitation using spatial and temporal
statistics, transition probabilities, wet/dry spell lengths, spatial correlation indices, and
variograms for wet (June through September) and dry (October through May) seasons (see
Figure 6). After comparing results from each of these downscaling methodologies, the team
found that the BCCA method underestimated mean climatology of daily precipitation; the BCSD
and BCCA methods underestimated temporal variability; and the BCSD, SDBC, and BCCA
methods all underestimated spatial variability in precipitation. The technique developed for
TBW’s PUMA project, BCSA, was a better fit than the BCSD, BCCA, or SDBC methods for the
west-central Florida region, where reproducing small-scale spatiotemporal precipitation
variability is important. In other words, the TBW team developed a new technique to increase
GCM data resolution based on its needs and particular regional characteristics.
Page 27
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Figure 6. Comparison of spatial variability of TBW’s different downscaling methods. The
variograms show the ability of TBW’s different downscaling methods, dynamical downscaling
(R2 + RCM, CCSM + RCM) and statistical downscaling (CCSM + BCSD, CCSM + SDBC,
CCSM + BCSA), to replicate the spatial variability that exists in the observed record (red) for
the wet season (June–September, on left) and the dry season (October–May, on right). The
BCSA statistical downscaling method from the CCSM GCM best overlies the observed record
in both the wet and dry seasons.
TBW issue 2: Needing to understand local hydrology
TBW had considerable experience developing and using hydrologic models for water resource
issues in west-central Florida. This experience led the utility to understand the importance of
small-scale spatiotemporal precipitation variability in understanding west-central Floridas
hydrology. Having this fundamental knowledge of local hydrology and access to hydrologic
models were key for developing an understanding of how climate change would affect TBW’s
water supplies. Had TBW not had this knowledge before engaging in a climate impacts study, it
might have generated output that was unusable. Consequently, the TBW team believes it is
critical to understand local climate and hydrologic relationships before simply using output from
GCMs.
Page 28
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
TBW issue 3: Developing a strong scientist/utility relationship
Utility and scientific personnel involved in the TBW project have cultivated a strong
collaborative working relationship through the PUMA project, with both sides eager to discover
what the science reveals. They have also engaged other scientific partners as needed for the
research, such as the climate science team at the FSU COAPS for their downscaling work. Much
of their progress in the project was because of the attitude the primary researchers have taken
toward setting goals and evaluating results. Above all, they have taken an incremental approach,
going where the research takes them rather than strictly adhering to a prescribed statement of
work.
The TBW team identified how the partnership works to create a successful co-production
dynamic: regular contact and organized interaction. In addition to the PUMA project, the utility
and science leads for this PUMA project have also been working together on other projects, such
as projects through the Florida Water and Climate Alliance and the SECC, a group that meets
quarterly. Between these various projects, the two researchers have been in touch at least every
two months. They have also planned agendas in advance of their meetings, even if just a
telephone call, to maximize the value of each meeting and to stay on track.
3.4.3 How the PUMA project will affect utility decisions
TBW does not have an immediate need for the information provided through the PUMA project
in decision-making. This means the PUMA team members have had the luxury of investing the
time for this research. They have been able to explore all options and decide on the best
methodology, instead of being pressured by a hard deadline. They have explored the best route
for mainstreaming climate data into their decision-making, so when a major decision that needs
to consider future projections arises, the team will have the modeling process ready. However,
the TBW team’s work has generated growing interest both within and outside TBW. The PUMA
team has had the opportunity to present its work to the senior leadership at TBW, and to other
utilities and scientists in various regional forums. TBW has been able to have an effect on other
water utilities in the region through the Florida Water and Climate Alliance.
Page 29
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
4. Conclusions for an Applied Research Agenda for
Climate Services
The formal goal of the PUMA project was “…to identify state-of-the-art modeling tools and
techniques that can be used by water utilities to assess potential climate change impacts on their
systems and watersheds.” The PUMA project was also motivated in part by the frustrations of
decision-makers who were unable to penetrate the highly technical world of climate science.
Other features of this environment that PUMA utilities had observed and sought to address
included the paucity of scientists skilled and interested in translating complex science for a lay
audience; a lack of research comparing and contrasting the various sources of climate projections
and information; and relatively underdeveloped climate services, especially as compared to the
burgeoning need at a local and regional level. In sum, the PUMA project was essentially an
effort to learn by doing, to generate experience by creating collaboration between water
managers and climate scientists, and to inform best co-production practice while generating an
applied research agenda and set of outcomes specific to the needs of those utilities. The
following are some conclusions based on the PUMA experience:
Assessment was local, and one size did not fit all. Assessment approaches can vary
widely dependent on local needs. Although each PUMA project sought to illuminate a
similar question the impact of climate change on drinking water supplies – the four
utilities pursued widely different approaches in service of that goal.
The scientist and utility-manager learning process was a two-way street. None of the
PUMA utilities was a passive recipient of expert information from a climate modeler. All
four utilities partnered with climate science experts to co-produce information that was
useful to the specific water utility. In practice, the climate modelers themselves often had
as much to learn about how water utilities model their systems as the water utility
personnel had to learn about how climate modelers project future climate. For example,
TBW developed a work plan for its PUMA project, but when those expectations turned
out to be unrealistic, utility personnel worked with their climate modeler counterparts to
redefine the investigation to produce useful information.
Water utilities sometimes needed to customize approaches to using climate model
output. GCM output, downscaling techniques, and even baseline observational datasets
used to validate climate projection tools frequently need to be customized for use in local
assessments. Several utilities’ PUMA projects spent significant resources on comparing
historical climate data for their specific systems/watersheds to ensure that future
projections will be as accurate as possible (e.g., PWB, SPU, TBW).
Page 30
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Utilities required flexibility in exploring different methods to use climate model output.
Each of the PUMA utilities followed a different path, using different data, models, and
techniques to increase the resolution of GCM data. The point should not be to find the
perfect method, but to get started and learn about climate change in general, and what
your utility needs from climate models in particular. For example, NYCDEP did a
tremendous amount of work using the CMIP3 models and a simple delta method for
increasing the resolution of GCM data. In that process, the utility learned that it needed to
know more about extreme events, and developed the SD-delta method to serve that need.
Utilities found that they needed to consider using a bottom-up as well as a top-down
approach to climate modeling. A bottom-up approach begins by asking what is
important in the context of a specific utility. A top-down approach begins by exploring
what the science can tell us about how climate may change. The utilities profiled here
used both approaches to find success in generating useful information. For example,
TBW did not feel any immediate need to answer specific questions and was consequently
more top-down and exploratory in its climate modeling work. SPU, on the other hand,
took a bottom-up approach and drove its work based on pre-existing questions
concerning likely impacts, including the onset of fall rains and atmospheric rivers. SPU
also developed a suite of metric/thresholds with which to query the downscaled
meteorological data in order to understand how conditions of interest may change in the
future.
Information on changes in extreme event impacts was a major need for water utilities
Although climate models do not easily capture extreme events, such events were the most
sought-after projections for many of the utilities’ PUMA projects. For example,
NYCDEP was most concerned about water-quality issues, including intense precipitation
leading to turbidity events. SPU was concerned about intense precipitation leading to
combined sewer overflow events or urban flooding, as well as extreme heat leading to
catastrophic wildfires in its protected watershed. Because of this, several PUMA utilities
developed their own approaches/methods to be able to more accurately model extreme
events that matter to their systems (e.g., NYCDEP).
Understanding local hydrology was critical. Climate information, in the form of
temperature, precipitation, solar radiation, winds, and other key variables, is only useful
if a utility can translate that information into realistic water supply sources (e.g., stream
flows, groundwater). This means that a good understanding of local hydro-meteorology is
important to understand the impacts of climate change. TBW’s conclusion that the
spatiotemporal variability in precipitation drove west-central Florida’s hydrology and
PWB’s need to bias-correct historical climate datasets to match observed stream flow
both indicated the critical nature of understanding local hydrology.
Page 31
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Utilities and scientists learned to adopt a “don’t hesitate to innovatestrategy. Some of
the most successful aspects of the PUMA project occurred when water utilities and their
scientific partners decided to create something new to meet their needs. Examples
included the 25-bin approach that NYCDEP developed to capture a sense of extreme
events using a delta method, and the BCSA downscaling technique that the TBW team
developed to capture spatiotemporal variability in rainfall.
Page 32
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
References
ACCCNRS. 2014. Advisory Committee on Climate Change and Natural Resource Science.
Available: https://nccwsc.usgs.gov/acccnrst
EO 13653. 2013. Executive Order 13653 of November 1, 2013. Preparing the United States for
the Impacts of Climate Change. Available: http://www.whitehouse.gov/the-press-
office/2013/11/01/executive-order-preparing-united-states-impacts-climate-change.
Executive Office of the President. 2013. The Presidents Climate Action Plan. June. Available:
http://www.whitehouse.gov/sites/default/files/image/president27sclimateactionplan.pdf.
IPCC. 1996. Climate Change 1995: The Science of Climate Change: Contribution of Working
Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, Cambridge, UK and New York.
IPCC. 2013. Climate Change 2013: The Physical Science Basis, T.F. Stocker, D. Qin, G-K
Plattner, M.M.B. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P.M. Midgley
(eds.).Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change. Chapter 11 report graphics. Cambridge University Press, Cambridge,
UK and New York.
Kanamitsu, M., K. Yoshimura, Y. Yhang, and S. Hong. 2010. Errors of interannual variability
and multi-decadal trend in dynamical regional climate downscaling and its corrections. J
Geophys Res 115:D17115.
Livneh, B., E.A. Rosenberg, C. Lin, B. Nijssen, V. Mishra, K.M. Andreadis, E.P. Maurer, and
D.P. Lettenmaier. 2013. A long-term hydrologically based dataset of land surface fluxes and
states for the conterminous United States: Update and extensions. J Climate 26:9384–9392.
Available: http://dx.doi.org/10.1175/JCLI-D-12-00508.1.
Maurer, E.P., L. Brekke, T. Pruitt, and P.B. Duffy. 2007. Fine-resolution climate projections
enhance regional climate change impact studies. Eos, Transactions American Geophysical Union
88(47):504.
Mearns, L.O., S.C. Pryor, and V. Magaña. 2014. Downscaling of climate information. In Climate
Change in North America, Regional Climate Studies, G. Ohring (ed.). Springer International
Publishing, Switzerland.
Raff, D., L. Brekke, K. Werner, A. Wood, and K. White. 2013. Short-term Water Management
Decisions: User Needs for Improved Climate, Weather, and Hydrologic Information. Prepared
Page 33
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
by and for the U.S. Army Corps of Engineers, the Bureau of Reclamation, and the National
Oceanic and Atmospheric Administration. January. Available:
http://www.ccawwg.us/docs/Short-
Term_Water_Management_Decisions_Final_3_Jan_2013.pdf.
USACE. 2012. USACE 2012 Climate Change Adaptation Plan and Report. U.S. Army Corps of
Engineers. June. Available:
http://www.corpsclimate.us/docs/2012_USACE_Adaptation_Plan_and_Report_23_June_2012%
20final.pdf.
USGCRP. 2012. The National Global Change Research Plan 2012–2021: A Strategic Plan for
the U.S. Global Change Research Program. April. Available:
http://downloads.globalchange.gov/strategic-plan/2012/usgcrp-strategic-plan-2012.pdf.
Wilby, R.L. and C.W. Dawson. 2012. The Statistical DownScaling Model: Insights from one
decade of application. International Journal of Climatology 33(7):1707–1719.
WMO. 2011. Climate Knowledge for Action: The Global Framework for Climate Services –
Empowering the Most Vulnerable. WMO-No. 1065. World Meteorological Organization,
Switzerland. Available: ftp://ftp.wmo.int/Documents/SESSIONS/Cg-XVI/English/DOCs/d11-
1%281%29_1065_HLT_report_en.pdf.
WUCA. 2009. Options for Improving Climate Modeling to Assist Water Utility Planning for
Climate Change. Water Utility Climate Alliance. December. Available:
http://www.wucaonline.org/assets/pdf/pubs_whitepaper_120909.pdf.
Page 34
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Suggested Resources
Other WUCA reports
WUCA. 2009. Options for Improving Climate Modeling to Assist Water Utility Planning for
Climate Change. Water Utility Climate Alliance. December. Available:
http://www.wucaonline.org/assets/pdf/pubs_whitepaper_120909.pdf.
WUCA. 2010. Decision Support Planning Methods: Incorporating Climate Change Uncertainties
Into Water Planning. January. Available:
http://www.wucaonline.org/assets/pdf/pubs_whitepaper_012110.pdf.
WUCA. 2015. Embracing Uncertainty: A Case Study Examination of How Climate Change Is
Shifting Water Utility Planning. Water Utility Climate Alliance. May. Available:
http://www.wucaonline.org
NYCDEP PUMA technical papers
Anandhi, A., A. Frei, S.M. Pradhanang, M.S. Zion, D.C. Pierson, and E.M. Schneiderman. 2011.
AR4 climate model performance in simulating snow water equivalent over Catskill Mountain
watersheds, New York, USA. Hydrological Processes 25(21):3302–3311.
Anandhi, A., A. Frei, D.C. Pierson, E.M. Schneiderman, M.S. Zion, D. Lounsbury, and A.H.
Matonse. 2011. Examination for change factor methodologies for climate change impact
assessment. Water Resources Research 47(W03501). doi: 10.1029/2010WR009104.
PWB PUMA technical papers
Abatzoglou, J. and K. Hegewisch. 2013a. Climate Downscaling. University of Idaho. December
20.
Abatzoglou, J. and K. Hegewisch. 2013b. Climate Model Evaluation and Projections Synopsis.
University of Idaho. December 20.
Chiao, T-H C., B. Nijssen, and D.P. Lettenmaier. 2013a. PUMA Project Phase I – Hydrologic
Model for the Bull Run Watershed for Climate Change Analysis (Final Report). University of
Washington, Department of Civil and Environmental Engineering. December 21.
Chiao, T-H C., B. Nijssen, and D.P. Lettenmaier. 2013b. Technical Memorandum #1 for the
Portland Water Bureau PUMA Project Phase 1 Hydrologic Model for the Bull Run Watershed
Page 35
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
for Climate Change Analysis. University of Washington, Department of Civil and Environmental
Engineering. April 29.
Livneh, B., E.A. Rosenberg, C. Lin, B. Nijssen, V. Mishra, K.M. Andreadis, E.P. Maurer, and
D.P. Lettenmaier. 2013. A long-term hydrologically based dataset of land surface fluxes and
states for the conterminous United States: Update and extensions. J. Climate 26:9384–9392.
Available: http://dx.doi.org/10.1175/JCLI-D-12-00508.1.
SPU PUMA technical papers
Dalton, M.M. 2014. Technical Memorandum #4 for Seattle Public Utilities PUMA Project:
“Return of Fall Rains Analysis.” Climate Impacts Research Consortium. August 22.
Dalton, M.M. and K. Hegewisch. 2014. Technical Memorandum #3 for Seattle Public Utilities
PUMA Project: “Evaluation of Historic MACA-Downscaled Station Data. Climate Impacts
Research Consortium. June 2.
Hegewisch, K. and J. Abatzoglou. 2013. Technical Memorandum #2 for Seattle Public Utilities
PUMA Project: Climate Downscaling.Climate Impacts Research Consortium.
Rupp, D.E., J.T. Abatzoglou, K.C. Hegewisch, and P.W. Mote. 2013. Evaluation of CMIP5 20th
century climate simulations for the Pacific Northwest USA. J. Geophys. Res. Atmos. 118. doi:
10.1002/jgrd.50843.
TBW PUMA technical papers
Hwang, S. and W. Graham 2013a. Tampa Bay Water Climate Variability and Projections on
Regional Water Supplies Project: Technical Report. September.
Hwang, S. and W.D. Graham. 2013b. Development and comparative evaluation of a stochastic
analog method to downscale daily GCM precipitation. Hydrol Earth Syst Sci Discuss 10:2141–
2181. doi: 10.5194/hessd-10-2141-2013.
Hwang, S., W. Graham, A. Adams, and J. Guerink. 2013c. Assessment of the utility of
dynamically-downscaled regional reanalysis data to predict streamflow in west central Florida
using an integrated hydrologic model. Regional Environmental Change. doi: 10.1007/s10113-
013-0406-x.
Hwang, S., W. Graham, J.L. Hernández, C. Martinez, J.W. Jones, and A. Adams. 2011.
Quantitative spatiotemporal evaluation of dynamically downscaled MM5 precipitation
predictions over the Tampa Bay Region, Florida. J. Hydrometeor 12:1447–1464.
Page 36
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
A. Applying Climate Model Outputs 101 for
Water Utilities
In this section, we provide a general overview of general circulation models (GCMs), which are
also called global climate models. We also discuss how a utility or their climate science partners
might integrate GCMs into water utility models
12
to support utility decision-making. This section
discusses GCMs; emissions scenarios; GCM selection; increasing the resolution of GCM data,
often known as downscaling; time steps; time periods; and integration of GCM outputs into
utility models. We developed this appendix for decision-makers who are not familiar with GCMs
or how their outputs might be used in utility models.
A.1 GCMs
In general, GCMs simulate the world’s climate. They are often used to project what the Earths
climate will be like in the future under assumptions about future emissions of greenhouse gases
(GHGs). The outputs of these models have been used to inform long-term planning and
investment decisions by water and other natural resource managers.
GCMs divide the world into a grid, with each grid box having the same degrees of latitude and
longitude. All GCMs do not have uniformly sized grid boxes; they vary in size from
approximately 60 miles to more than 200 miles (roughly 1 to 3°). Because of the large size of
these grid boxes, small-scale climate features, such as mountain ranges, lakes, and irregular
coastlines, which are often locally important, are not represented by GCMs. To begin to focus
the global outputs on a particular region, researchers will isolate the grid box (or a set of
contiguous grid boxes) of interest for each GCM and capture the model results for that grid box.
This process allows the outputs from a global model to focus on a regional area of interest.
In this appendix, we discuss two generations of GCMs assembled as part of the Coupled Model
Intercomparison Project (CMIP): the Phase 3 (CMIP3) models and the Phase 5 (CMIP5)
models.
13
The CMIP3 model runs were completed in the early 2000s for the IPCC Third
Assessment Report (TAR) and Fourth Assessment Report (AR4). CMIP5 model runs were
completed in the late 2000s and early 2010s for the IPCC Fifth Assessment Report (AR5). When
the PUMA project began, the most widely used set of GCMs was from the CMIP3 generation;
12. For a more in-depth discussion of the science of climate modeling, see Options for Improving Climate
Modeling to Assist Water Utility Planning for Climate Change (WUCA, 2009).
13. Phase 4 (CMIP4) was skipped in order to align CMIP numbering with the IPCC Assessment Reports
(ARs).
Page A-1
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
however, over the course of the project, the CMIP5 models were released. Some of the PUMA
utilities stayed with CMIP3 for their analysis, while others used CMIP5.
A.2 Emissions Scenarios
A key component of the climate modeling process is the selection of emissions scenario(s).
Emissions scenarios are plausible narratives of future global social and economic development
with associated GHG concentration levels and radiative forcings.
14
They imply various future
levels of GHG emissions, taking into consideration different development futures that factor in
social changes such as economic conditions and population growth. Higher emissions scenarios
will accelerate climate changes. Lower emissions scenarios, some of which assume reductions in
future GHG emissions, would result in less severe climate changes.
In its last two assessments, the IPCC produced two types or “familiesof emissions scenarios:
the Special Report on Emissions Scenarios (SRES) and Representative Concentration Pathways
(RCPs). The SRES scenarios were developed in 2000 and used for the TAR and AR4. They start
with explicit assumptions about population and development and use these assumptions to define
the emissions scenario trajectories. The RCP scenarios were developed in response to a request
in 2007 to update, streamline, and modify the SRES scenarios, and were used in projections
provided in the AR5 published in 2013. In contrast to SRES, RCPs assume specific changes in
radiative forcing (i.e., roughly how much additional energy is trapped in the atmosphere by
GHGs). Socioeconomic scenarios that present pathways on how these levels of radiative forcing
can be reached are being developed.
A.3 GCM Selection
Researchers will sometimes select a subset of the GCMs that best simulate observed climate for
a specific area. They compare modelsre-creation of historical climate with the observational
record. This is a way to eliminate models that perform relatively poorly in simulating current
climate. However, there is no guarantee that models that best simulate current and past climate
are the most reliable models for simulating future changes in climate. Nonetheless, the ability of
models to simulate current and past climate patterns (e.g., which regions get more precipitation,
season cycles of climate) is a widely used measure of a models relative skill.
14. Radiative forcing is defined by the IPCC as, “the perturbation to the energy balance of the Earth-
atmosphere system (in watts per square meter) following, for example, a change in the concentration of carbon
dioxide or a change in the output of the Sun; the climate system responds to the radiative forcing so as to re-
establish the energy balance. A positive radiative forcing tends to warm the surface and a negative radiative
forcing tends to cool the surface” (IPCC, 1996, p. 49).
Page A-2
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
Uncertainties in Climate Modeling
Some major uncertainties concerning future climate change include (IPCC, 2013):
1. Future emissions of GHGs and other emissions, such as aerosols that affect climate.
2. The amount the Earth’s climate will warm in response to rising GHGs and other emissions, referred to
as climate sensitivityand typically expressed as the increase in average global temperature associated
with a doubling of atmospheric carbon dioxide levels.
3. The patterns of change in regional climate as distinguished from global or continental change. This
includes which regions will warm more than others, as well as which regions will get wetter or drier.
4. Natural climate variability. Variability ranges from short-term changes in climate (e.g., an increase or
decrease in number of days with precipitation, change in precipitation intensity, change in heat waves)
to seasonal variability (e.g., if winters get wetter and summers driers) to inter-annual variability
(e.g., changes in the El Niño Southern Oscillation) to changes in drivers of decadal variability such as
the Pacific Decadal Oscillation. Distinguishing between natural variability and climate change driven
by increased GHG concentrations is challenging, especially in the near term, which is defined as several
decades into the future.
Good practice captures a reasonably wide range of model projections of change in key variables, such as
temperature and precipitation, to reflect uncertainty across model projections; this requires the use of many
models. Furthermore, when presenting modeling results, a discussion of these uncertainties should always
be included.
When researchers select a subset of GCMs and identify appropriate emissions scenarios (see
discussion below), they can then obtain the simulations of current and future climates from the
GCMs. They can download this information from various websites that publish GCM results (or
use tools that provide GCM data). They can then isolate the grid box or grid boxes for their
region to obtain projected changes to variables such as temperature, precipitation, and solar
radiation.
The PUMA teams identified a subset of the GCMs to use in their analyses. Teams generally
selected a set of models based on how well they simulate observed climate. However, the output
from some climate models is not easy to obtain or the time scale of the output (e.g., daily or
hourly versus monthly) is not at the scale needed for local analysis. This also drives the decision
on which models to use. Since not all model outputs are available from one single location and
not all variables are available at the same scale, this affects the decision on how many models are
used in the local evaluation. For instance, PWB selected 20 CMIP5 models because these models
have all the available variables needed for MACA downscaling. The teams sometimes further
refined that selection to identify a number of models that encompass a range of outcomes (e.g., a
range of temperature and precipitation projections). They also sometimes further reduced the
number of models used to those that provide output that can be directly used in their operations
and management models, such as temperature and precipitation projections at geographic and
time scales consistent with their operations and management models.
Page A-3
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
A range of models are generally used because no single model can represent a wide range of
potential changes in future climate. Typically, utilities will try to capture a wide range of model
projections, particularly on changes in precipitation. Change in this variable is very important for
water utilities and the climate models tend to project a wide array of changes in precipitation.
This often includes some models projecting increases while others project decreases.
It is important to avoid selecting a range of outputs that is too narrow (e.g., selecting only models
that project reduced precipitation in a region when some models project increased precipitation).
Thus, often a combination of models is selected (e.g., relatively “wetand relatively “dry”
models that encompass a substantial range of projections of changes in precipitation).
Often a middle outcome, such as a model with results in the middle of the range or reflecting an
average of models, is also presented to not only include models that are toward the ends of the
distribution range. Another option is to use average projections from a number of GCMs. It has
been shown that when model simulations of current climate are averaged, the average is
generally closer to observations than individual model simulations of current climate. But the
average of models can smooth out spatial or temporal variability and change in extreme climate.
It is generally considered more prudent to rely on a range of model output, particularly with
changes in precipitation, rather than just the average of the model output. An alternative to using
an average of the model output is to select a climate model whose projection of climate change is
approximately in the middle of the range of all the models.
Many of the PUMA utilities decided to use a full complement of GCMs so that the full range of
climate model output could be analyzed. But this increase in the number of GCMs and the
number of model runs per GCM increased significantly from CMIP3 to CMIP5 and running all
models may run into resource limitations at a utility.
A.4 Increasing the Resolution of GCM Data (Downscaling)
GCM outputs alone often do not capture the level of spatial detail desired by water utilities and
other decision-makers. The large size of the grid boxes does not account for topographic and
other key differences within a grid box that may be critical (e.g., to accurately simulate
precipitation in a particular watershed). The term “downscaling” generally has been used to refer
to techniques to translate relatively coarse GCM outputs into more spatially disaggregated
outputs that much better reflect the variance of climate conditions within a grid box.
To alleviate this problem, further refinement of GCM output is done to present regionally varied
climate, which much more closely corresponds to actual climate conditions. The simplest
approach, called the delta methodor delta-change technique involves adjusting or
combining the GCM estimated climate with observations to create a more spatially and
temporally plausible dataset for a climate-altered future. The delta method adds (typically for
Page A-4
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
temperature) or multiplies (typically for precipitation) the estimated change in a variable in a grid
box to all observations within the grid box. In this approach, the change is assumed to be
uniform across a grid box,
15
but the estimated projections of temperature and precipitation will
vary reflecting spatial differences in observations (e.g., higher altitudes will still be cooler than
lower altitudes). Although it is not formally downscaling, this method provides a relatively
simple approach to disaggregation of coarse GCM output.
A second and likely most commonly used approach is a family of statistical downscaling
methods for bias correction and data disaggregation. Bias correction adjusts the GCM’s
estimates of current climate to align with observed climate. The bias correction is applied across
a distribution of observations (i.e., from high to low temperature and precipitation). Spatial
disaggregation techniques are then used to develop higher spatial resolution observational
datasets (e.g., Mauer et al., 2007) applies BCSD at 1/8th degree, or about 12 km)
16
and in turn
higher resolution climate change projections. Some PUMA partners that used these methods
include TBW with the BCSA method (a customized technique) and PWB and SPU with the
MACA method.
17
It is important to note that these methods do not estimate how GCM
projections of change at a coarse scale may vary within a grid box based on physical conditions
such as local topography. This is because the adjustment to GCM output to have it better align
with observations is also applied to model projections of future climate using those same
observations.
The last two approaches discussed in this section attempt to simulate how change in climate may
vary at much higher resolution than the GCMs simulate. Statistical Downscaling Models
(SDSMs)
18
statistically relate larger-scale meteorological conditions to local-level conditions.
Typically, site-specific variables such as temperature or precipitation from individual weather
stations are statistically correlated with larger-scale climate variables such as pressure patterns.
19
This correlation is then combined with a GCM’s simulation of change in the larger-scale variable
to estimate potential future site-specific temperature and precipitation. Wilby and Dawson (2012)
SDSM is an example of this technique, which was not employed by any of the PUMA teams.
15. In some cases, GCM projections can be interpolated across grid boxes to smooth out changes and avoid
sudden changes at the boundaries of grid boxes. In other cases, such as used by NYCDEP, the GCM changes
can be divided into bins.
16. See http://gdo-dcp.ucllnl.org/downscaled_cmip_projections/dcpInterface.html for more information on the
BCSD technique.
17. See http://maca.northwestknowledge.net for more information on the MACA technique.
18. Note that BCSD techniques are also widely referred to as statistical downscaling. This section is not
discussing those methods, but so-called statistical downscaling models.
19. For example, site-specific temperature or precipitation may be correlated with a 500-mb geopotential
height (Mearns et al., 2014).
Page A-5
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
An important limitation of both statistical downscaling methods and statistical downscaling
models is that they assume the statistical relationships that exist today will remain the same
under climate change. This assumption, known as “stationarity” (e.g., that past recorded climate
behavior will remain the same in the future), is likely incorrect, though in what way and to what
degree is not yet known.
Dynamical downscaling uses RCMs that are similar to GCMs but only simulate a section of the
Earth (e.g., North America), and do so at a much higher resolution than GCMs (e.g., 50 km and
less rather than approximately 200 km). By doing so, RCMs attempt to capture important
features such as mountains, coastlines, and large water bodies that influence regional climate
conditions. Unlike statistical techniques, RCMs can account for physically based changes in
climate. RCMs are nested” within GCMs (i.e., a high-resolution RCM is run with conditions at
the boundary of the model taken from a GCM). Examples of RCMs used by the PUMA utilities
include the Weather Research and Forecasting (WRF)
20
model used by NYCDEP, a mesoscale
model known as MM5
21
used by NYCDEP and TBW, and TBWs use of output from FSU
COAPS Regional Spectral Model (adapted from Kanamitsu et al., 2010). Even though they are
run at a higher resolution than GCMs, at their current level of development in the United States,
RCM output has generally been at lower resolution than most users’ desire and may not entirely
resolve local-scale atmospheric processes. They are also significantly resource intensive, which
limits the ability of water utilities to use them (they are frequently developed and run by large-
scale research institutions).
It is important to note that while all of these techniques provide much greater spatial detail than
what can be obtained from a GCM, it is not the case that any technique to increase GCM data
resolution will correct errors from GCMs or reduce the range of projections from multiple
GCMs. In fact, scientists often warn that each and every step in the assessment process,
including downscaling of GCM projections, may introduce additional uncertainties to the
process.
A.5 Time Steps
One factor the PUMA teams considered in selecting their GCMs was the temporal resolution of
the outputs. Some model outputs are saved as a single value per day, while other outputs are
saved with finer temporal refinement, such as at a three hourly scale. In the climate modeling
field, these timescales are often referred to as time steps. The available output time steps were
20. WRF is a collaboration among the National Center for Atmospheric Research (NCAR), NOAA, the Air
Force Weather Agency, the Naval Research Laboratory, the University of Oklahoma, and the Federal Aviation
Administration.
21. MM5 is the fifth-generation Pennsylvania State University and NCAR Mesoscale Model.
Page A-6
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
a critical decision factor for the water utilities because their utility models (e.g., hydrologic
models, operations or management models, flooding models) are designed to receive inputs at
specific time steps. Note that average monthly change in climate is a common time step for
climate model output. This can be satisfactory for estimating change in water supply through
operations models in many cases.
22
Flooding, on the other hand, tends to be very sensitive to
smaller time steps such as daily or sub-daily (e.g., hourly) precipitation amounts. For modeling
combined sewer overflow events, time steps as short as five minutes were reported as being used
by PUMA utilities in their own modeling. In the end, it was most common for PUMA utilities to
select GCM data at a daily time step and then disaggregate that data into subdaily time steps
when necessary.
A.6 Time Periods
Utilities and climate modelers must consider another temporal dimension in the climate
modeling process. GCM projections are typically calculated by averaging 20–30 future yearly
projections to a single simulated year. If only one, or even several, projected years are used, that
year or years could be a simulated bad drought year or a particularly wet period, and represent
climate variability more than climate change.
23
Thirty years of observed weather is typically
used to define a climate (e.g., the period 1981–2010 is used to define average climate) and a 30-
year (or even a 20-year) period of simulated future climate can be used to define a climate in the
future. So GCM yearly projections might be considered in a near-term grouping (i.e., 2035 being
the average of 2020–2050), mid-term (i.e., 2065 being the average of 2050–2080), and long-term
(i.e., 2085 being the average of 2070–2100).
An additional decision is on how far into the future climate change is simulated. This is also
important because the magnitude of climate change is greater further into the future, but also
more uncertain. Impacts many years from now may not be as relevant to a water utility. Each
team needs to make the decision about what time periods it will examine depending on what the
utility’s needs or interests are. For example, if a utility is making a long-term investment in a
large piece of infrastructure, it might consider long-term projections (e.g., out to 100 years from
the present). A simple rule-of-thumb is that projected changes within three decades of current
climate (e.g., projecting from 2015 out to 2045) will likely be dominated by natural variability.
22. This is the case when a water system uses reservoirs, which can attenuate the daily variability of stream
flow. But for many utilities who rely on run-of-river flows, monthly averaged climate output is not sufficient
to analyze impacts to water supply.
23. On the other hand, when people select a GCM based on long-term averages, they lose the details of month-
to-month or year-to-year variability. This means that certain interesting aspects of future GCM projections
could be lost through averaging.
Page A-7
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
The signal from climate change typically takes several decades to clearly emerge from the
“noise” of natural variability.
A.7 Climate Models into Utility Models
The GCMs provide climate model projections for specific variables such as temperature or
precipitation which can then be used in utility models to examine how changes in climate might
propagate through water systems and affect utilitiesabilities to meet demands. This type of
exercise is often referred to as a chain-of-models” because it involves inputting data from one
model into another model.
The concept of chain-of-models refers to the sequence of models that must be used to apply
climate information in a water utility context. For example, the outputs from a GCM can be used
as inputs into a hydrology model, and that hydrology model translates the GCM output into
streamflow or groundwater recharge to be used as inputs into a system management or
operations model. The outputs of those utility models can then help define the impacts of climate
change on water supply, water quality, and other parameters commonly modeled by water
utilities. In this way GCM data can help to identify potential impacts on water system
performance.
Page A-8
Piloting Utility Modeling Applications Project Final Report Stratus Consulting
B. PUMA Project Points of Contact
PUMA project utility leads
Utility Project lead Email Phone
SPU Paul Fleming paul.fleming@seattle.gov 206-684-7626
PWB Kavita Heyn kavita.heyn@portlandoregon.gov 503-823-4724
TBW Alison Adams
aadams@tampabaywater.org 727-791-2314
NYCDEP Alan Cohn alanc@dep.nyc.gov
718-595-4536
PUMA project contacts
Project role Email Phone
Project Manager David Behar dbehar@sfwater.org 415-554-3221
Contract Manager Keely Brooks keely.brooks@snwa.com 702-822-3349
RISA/Science partner
Entity
Climate science
representative Email Phone
OSU Phil Mote pmote@coas.oregonstate.edu 541-737-5694
Page B-1
  • ... These efforts have culminated most recently in a research effort by the Water Utility Climate Alliance (WUCA) 1 explicitly aimed at exploring the co-production of scientific information that is useful for water utilities planning for climate change (Vogel et al., 2015). According to the final report of this research effort, '''co-productio n'. . ...
    ... According to the final report of this research effort, '''co-productio n'. . .is intended to convey the idea that science in service of adapta-tion is not a one-way street, but a collaborative venture between scientists and decision-makers in which the needs and skills of each come into play throughout that collaboration" (Vogel et al., 2015). The authors continue, ''Co-production requires an iterative, collaborative process across the borders between science and policy that draws upon the unique needs, experience, and even the limitations of each party, providing the strongest possible underpinning for societal action in response to the consequences of climate change" (Vogel et al., 2015). ...
    ... .is intended to convey the idea that science in service of adapta-tion is not a one-way street, but a collaborative venture between scientists and decision-makers in which the needs and skills of each come into play throughout that collaboration" (Vogel et al., 2015). The authors continue, ''Co-production requires an iterative, collaborative process across the borders between science and policy that draws upon the unique needs, experience, and even the limitations of each party, providing the strongest possible underpinning for societal action in response to the consequences of climate change" (Vogel et al., 2015). Effective co-production of information can lead to the development of new forecast products and models to address ''real world problems" (Feldman and Ingram, 2010). ...
    Article
    Full-text available
    This article explores the efforts of four water utilities to co-produce actionable science by forging partnerships with scientific institutions to explore integrating climate considerations into their specific management context. The experiences of these four utilities and their scientific partners, as part of the Piloting Utility Modeling Applications project of the Water Utility Climate Alliance, provide a wealth of empirical evidence to illustrate some of the core concepts formulated to explain how to produce usable information and how to link research to decision making. Through these four case studies of co-production, we identify three findings that bridge principles and practice: each utility engaged in contextualizing research; in building and leveraging knowledge networks; and in embracing an entrepreneurial approach to their research agenda. In several instances, unanticipated but innovative assessment techniques were developed by science partners in collaboration with water utilities to fit the utility’s specific needs. The paper concludes by discussing some of the hard realities of co-production illustrated by these cases that should be kept in mind by people contemplating similar projects.
  • ... The report, "Decision Support Planning Methods: Incorporating Climate Change Uncertainties into Water Planning," was produced to help water utilities consider and evaluate traditional and emerging planning techniques for use in their own climate adaptation efforts. WUCA and its member cities have continued their interest in the use of projections, including a set of case studies in how climate change is shifting water utility planning (Stratus Consulting and Denver Water 2015) and about producing actionable climate information for utility modeling applications (Vogel et al 2015). ...
    Technical Report
    Full-text available
    The Commission-approved Study Plan for the Project does not order evaluation of the Project's effects in the light of future climate change. NMFS proposes this new information collection or study according to the regulations implementing the ILP, 18 C.F.R. § 5.15 (e), for good cause. Significant new information, material to the study objectives has become available, in the form of climate change study methods and application developed since NMFS' initial study requests were submitted on July 22, 2014. Additionally, California has experienced record drought conditions from 2012 to 2015, culminating in snowpack levels in 2015 that are estimated to the be lowest in 500 years (Belmecheri et al., 2016). This information suggests that climate change is currently affecting the Project area and will continue to do so over the term of any new license issued to the Project. NMFS provides further explanation of good cause below, as required under the regulations. § 5.15 (e) (1): Any material changes in the law or regulations applicable to the information request; NMFS does not find this criterion applicable here. § 5.15 (e) (2): Why the goals and objectives of any approved study could not be met with the approved study methodology; The existing Commission-approved Study Plan does not order evaluation of the Project's effects in the light of future climate change. The existing Commission-approved Study Plan uses historical and static flow and water temperature conditions to evaluate the Project's effects. This approach limits the usefulness of assessments of Project effects on anadromous fishes and their
  • ... The report, "Decision Support Planning Methods: Incorporating Climate Change Uncertainties into Water Planning," was produced to help water utilities consider and evaluate traditional and emerging planning techniques for use in their own climate adaptation efforts. WUCA and its member cities have continued their interest in the use of projections, including a set of case studies in how climate change is shifting water utility planning (Stratus Consulting and Denver Water 2015) and about producing actionable climate information for utility modeling applications (Vogel et al 2015). ...
    Technical Report
    Full-text available
    The Glacier and Runoff Changes (GRC) Study determination from the Federal Energy Regulatory Commission (FERC) Dispute Resolution (April 26, 2013) requires for the literature review as "described in Revised Study Plan (RSP) section 7.7.4.1." The RSP describes the literature review method: to summarize the current understanding of the rate and trend of glacier retreat and the contribution of glacial mass wasting to the overall flow of the Upper Susitna watershed, include trend analyses of glacier retreat, temperature, and precipitation." However, the implied objective, to understand potential future changes in runoff associated with glacier wastage and retreat, cannot be met through a literature review alone because no such literature exists for the region of the Susitna basin. While the Glacier and Runoff Changes Literature Review Study (7.7) provides a reasonable review of some of the ways temperature and precipitation variability may impact glaciers, the climate literature review within is brief (one page), inadequate, and does not refer to key literature relevant to Alaska. However, it does point to a range of potential temperature and precipitation changes, an unambiguous reduction in ice volume, and implications for water chemistry. A literature review is inadequate as a method to understand the future changes in glaciers and runoff with changing climate for infrastructure planning and determining project impacts from the combined and in some instances, synergetic effects of both the project construction and operations and changing climate on biota in the river. Climate change has become a key lens through which resource management decisions must be evaluated and addressed. The existing FERC-approved Study Plan does not order evaluation of the combined effects of the Project and climate change. Given that this large project will greatly alter natural flows which wild anadromous fish are adapted to in the Susitna River, and will alter habitats that anadromous fish depend upon for various stages in their life histories, and climate change will also continue to affect these same flows and habitats, the project's effects are likely to exacerbate the effects of the project. The existing FERC-approved Study Plan uses historical and static flows (high, low and average water years) and water temperature conditions to evaluate the proposed Susitna-Watana hydropower project's (Project) effects. The approved glacial and runoff changes study is limited to review of existing literature relevant to glacial retreat, and summarizing the understanding of potential future changes in runoff associated with glacier wastage and retreat (hereafter referred to as the Glacial and Runoff Changes (GRC) literature review, Wolken et al (2014)). This literature review approach is not adequate to assess the combined risks of climate change and 7.7 Glacier and Runoff Changes
  • Poster
    Full-text available
    In 2010, four water utilities: Tampa Bay Water (TBW), Portland Water Bureau (PWB), Seattle Public Utilities (SPU), and NYC Department of Environmental Protection (NYDEP) embarked on the Piloting Utility Modeling Applications (PUMA) project to track in real time the process and challenges of co-producing actionable climate change data for decision making. These utilities pursued customized approaches based on specific utility needs and concerns. By working collaboratively with climate scientists, utilities were able to assess the ability and limitations of today's climate science to respond to those concerns. This poster summarizes the processes and lessons learned from the PUMA project.
  • Chapter
    The previous chapter summarized the application of global models to simulate climate. The horizontal resolution of these models—generally 1–3 degrees—while adequate for resolving sub-continental North American climate features, is insufficient for simulating the more detailed properties of regional climate. To achieve higher resolution, a variety of so-called downscaling methods have been developed. This chapter reviews these methods and the results of applying them to the regional climate change problem.
  • Article
    Full-text available
    This paper describes a publicly available, long-term (1915-2011), hydrologically consistent dataset for the conterminous United States, intended to aid in studies of water and energy exchanges at the land surface. These data are gridded at a spatial resolution of 1/16° latitude/longitude and are derived from daily temperature and precipitation observations from approximately 20 000 NOAA Cooperative Observer (COOP) stations. The available meteorological data include temperature, precipitation, and wind, as well as derived humidity and downwelling solar and infrared radiation estimated via algorithms that index these quantities to the daily mean temperature, temperature range, and precipitation, and disaggregate them to 3-hourly time steps. Furthermore, the authors employ the variable infiltration capacity (VIC) model to produce 3-hourly estimates of soil moisture, snow water equivalent, discharge, and surface heat fluxes. Relative to an earlier similar dataset by Maurer and others, the improved dataset has 1) extended the period of analysis (1915-2011 versus 1950-2000), 2) increased the spatial resolution from 1/8° to 1/16°, and 3) used an updated version of VIC. The previous dataset has been widely used in water and energy budget studies, climate change assessments, drought reconstructions, and for many other purposes. It is anticipated that the spatial refinement and temporal extension will be of interest to a wide cross section of the scientific community.
  • Article
    The Statistical DownScaling Model (SDSM) is a freely available tool that produces high resolution climate change scenarios. The first public version of the software was released in 2001 and since then there have been over 170 documented studies worldwide. This article recounts the underlining conceptual and technical evolution of SDSM, drawing upon independent assessments of model capabilities. These studies show that SDSM yields reliable estimates of extreme temperatures, seasonal precipitation totals, areal and inter-site precipitation behaviour. Frequency estimation of extreme precipitation amounts in dry seasons is less reliable. A meta-analysis of SDSM outputs shows a preponderance of research in Canada, China and the UK, whereas the United States and Australasia are under-represented. In line with the wider downscaling community, the most favoured sector of analysis is water and flood risk management which accounts for nearly half of all output; research in other sectors such as agriculture, built environment and human health is less prominent but growing. Over 50% of the studies are concerned with production of climate scenarios, comparison or technical refinement of downscaling methodologies. In contrast, there is relatively little evidence of application to adaptation planning and climate risk management. We assert that further attention to physically meaningful quantities such as wind speeds, wave heights, phenological and hazard metrics could improve uptake of downscaled products. Chronic uncertainty in boundary forcing continues to undermine confidence in downscaled scenarios so these tools are best used for sensitivity testing and adaptation options appraisal. Copyright © 2012 Royal Meteorological Society
  • Article
    Full-text available
    There are a number of statistical techniques that downscale coarse climate information from global circulation models (GCM). However, many of them do not reproduce the small-scale spatial variability of precipitation exhibited by the observed meteorological data which can be an important factor for predicting hydrologic response to climatic forcing. In this study a new downscaling technique (bias-correction and stochastic analog method, BCSA) was developed to produce stochastic realizations of bias-corrected daily GCM precipitation fields that preserve the spatial autocorrelation structure of observed daily precipitation sequences. This approach was designed to reproduce observed spatial and temporal variability as well as mean climatology. We used the BCSA method to downscale 4 GCM precipitation predictions from 1961 to 1999 over the state of Florida and compared the skill of the method to the results obtained with the commonly used bias-correction and spatial disaggregation (BCSD) approach, bias-correction and constructed analog (BCCA) method, and a modified version of BCSD which reverses the order of spatial disaggregation and bias-correction (SDBC). Spatial and temporal statistics, transition probabilities, wet/dry spell lengths, spatial correlation indices, and variograms for wet (June through September) and dry (October through May) seasons were calculated for each method. Results showed that (1) BCCA underestimated mean climatology of daily precipitation while the BCSD, SDBC and BCSA methods accurately reproduced it, (2) the BCSD and BCCA methods underestimated temporal variability because of the interpolation and regression schemes used for downscaling and thus, did not reproduce daily precipitation standard deviations, transition probabilities or wet/dry spell lengths as well as the SDBC and BCSA methods, and (3) the BCSD, BCCA and SDBC methods underestimated spatial variability in precipitation resulting in under-prediction of spatial variance and over-prediction of spatial correlation, whereas the new stochastic technique (BCSA) accurately reproduces observed spatial statistics for both the wet and dry seasons. This study underscores the need to carefully select a downscaling method that reproduces all precipitation characteristics important for the hydrologic system under consideration if local hydrologic impacts of climate variability and change are going to be accurately predicted. For low-relief, rainfall-dominated watersheds where reproducing small-scale spatiotemporal precipitation variability is important, the BCSA method is recommended for use over the BCSD, BCCA, or SDBC methods.
  • Article
    Monthly temperature and precipitation data from 41 global climate models (GCMs) of the Coupled Model Intercomparison Project Phase 5 (CMIP5) were compared to observations for the 20th century, with a focus on the United States Pacific Northwest (PNW) and surrounding region. A suite of statistics, or metrics, was calculated, that included correlation and variance of mean seasonal spatial patterns, amplitude of seasonal cycle, diurnal temperature range, annual- to decadal-scale variance, long-term persistence, and regional teleconnections to El Niño Southern Oscillation (ENSO). Performance, or credibility, was assessed based on the GCMs’ abilities to reproduce the observed metrics. GCMs were ranked in their credibility using two methods. The first simply treated all metrics equally. The second method considered two properties of the metrics: (1) redundancy of information (dependence) among metrics, and (2) confidence in the reliability of an individual metric for accurately ranking models. Confidence was related to how robust the estimate of the metric was to ensemble size, given that for most of the models only a small number of ensemble members (i.e., realizations of the 20th century) were available. A cursory comparison with 24 CMIP3 models revealed few differences between the two generations of models with respect to the statistics analyzed.
  • Article
    Full-text available
    The goal of this study was to evaluate the ability of dynamically downscaled reanalysis data to reproduce local-scale spatiotemporal precipitation and temperature data needed to accurately predict streamflow in the Tampa Bay region of west central Florida. In particular, the Florida State University Center for Ocean-Atmospheric Prediction Studies CLARReS10 data (NCEP DOE 2 reanalysis data (R2) downscaled to 10-km over the Southeast USA using the Regional Spectral Model (RSM) were evaluated against locally available observed precipitation and temperature data and then used to drive an integrated hydrologic model that was previously calibrated for the Tampa Bay region. Resulting streamflow simulations were evaluated against observed data and previously calibrated model results. Results showed that the raw downscaled reanalysis predictions accurately reproduced the seasonal trends of mean daily minimum temperature, maximum temperature and precipitation, but generally overestimated the monthly mean and standard deviation of daily precipitation. Biases in the temporal mean and standard deviation of daily precipitation and temperature predictions were effectively removed using a CDF-mapping approach; however, errors in monthly precipitation totals remained after bias correction. Monthly streamflow simulation error statistics indicated that the accuracy of the streamflow produced by the bias-corrected downscaled reanalysis data was satisfactory (i.e., sufficient for seasonal to decadal planning), but that the accuracy of the streamflow produced by the raw downscaled reanalysis data was unsatisfactory for water resource planning purposes. The findings of this study thus indicate that further improvement in large-scale reanalysis data and regional climate models is needed before dynamically downscaled reanalysis data can be used directly (i.e. without bias correction with local data) to drive hydrologic models. However, bias-corrected dynamically downscaled data show promise for extending local historic climate observation records for hydrologic simulations. Furthermore, results of this study indicate that similarly bias-corrected dynamically downscaled retrospective and future GCM projections should be suitable for assessing potential hydrologic impacts of future climate change in the Tampa Bay region.
  • Article
    Full-text available
    A variety of methods are available to estimate values of meteorological variables at future times and at spatial scales that are appropriate for local climate change impact assessment. One commonly used method is Change Factor Methodology (CFM), sometimes referred to as delta change factor methodology. Although more sophisticated methods exist, CFM is still widely applicable and used in impact analysis studies. While there are a number of different ways by which change factors (CFs) can be calculated and used to estimate future climate scenarios, there are no clear guidelines available in the literature to decide which methodologies are most suitable for different applications. In this study several categories of CFM (additive versus multiplicative and single versus multiple) for a number of climate variables are compared and contrasted. The study employs several theoretical case studies, as well as a real example from Cannonsville watershed, which supplies water to New York City, USA. Results show that in cases when the frequency distribution of Global Climate Model (GCM) baseline climate is close to the frequency distribution of observed climate, or when the frequency distribution of GCM future climate is close to the frequency distribution of GCM baseline climate, additive and multiplicative single CFMs provide comparable results. Two options to guide the choice of CFM are suggested. The first option is a detailed methodological analysis for choosing the most appropriate CFM. The second option is a default method for use under circumstances in which a detailed methodological analysis is too cumbersome.
  • Article
    Full-text available
    A new data set enhances the abilities of researchers and decision-makers to assess possible future climates, explore societal impacts, and approach policy responses from a risk-based perspective. The data set, which consists of a library of 112 fine-resolution climate projections, based on 16 climate models and three greenhouse gas emissions scenarios, is now publicly available. Monthly climate projections from 1950 to 2099 were downscaled to a spatial resolution of 1/8° (about 140 square kilometers per grid cell) covering the conterminous United States and portions of Canada and Mexico.