[Senate Hearing 110-1158]
[From the U.S. Government Publishing Office]



                                                       S. Hrg. 110-1158
 
                       IMPROVING THE CAPACITY OF
                       U.S. CLIMATE MODELING FOR
                     DECISION-MAKERS AND END-USERS

=======================================================================

                                HEARING

                               before the

                         COMMITTEE ON COMMERCE,
                      SCIENCE, AND TRANSPORTATION
                          UNITED STATES SENATE

                       ONE HUNDRED TENTH CONGRESS

                             SECOND SESSION

                               __________

                              MAY 8, 2008

                               __________

    Printed for the use of the Committee on Commerce, Science, and 
                             Transportation



                  U.S. GOVERNMENT PRINTING OFFICE
75-346                    WASHINGTON : 2012
-----------------------------------------------------------------------
For sale by the Superintendent of Documents, U.S. Government Printing Office, 
http://bookstore.gpo.gov. For more information, contact the GPO Customer Contact Center, U.S. Government Printing Office. Phone 202ï¿½09512ï¿½091800, or 866ï¿½09512ï¿½091800 (toll-free). E-mail, [email protected].  


       SENATE COMMITTEE ON COMMERCE, SCIENCE, AND TRANSPORTATION

                       ONE HUNDRED TENTH CONGRESS

                             SECOND SESSION

                   DANIEL K. INOUYE, Hawaii, Chairman
JOHN D. ROCKEFELLER IV, West         TED STEVENS, Alaska, Vice Chairman
    Virginia                         JOHN McCAIN, Arizona
JOHN F. KERRY, Massachusetts         KAY BAILEY HUTCHISON, Texas
BYRON L. DORGAN, North Dakota        OLYMPIA J. SNOWE, Maine
BARBARA BOXER, California            GORDON H. SMITH, Oregon
BILL NELSON, Florida                 JOHN ENSIGN, Nevada
MARIA CANTWELL, Washington           JOHN E. SUNUNU, New Hampshire
FRANK R. LAUTENBERG, New Jersey      JIM DeMINT, South Carolina
MARK PRYOR, Arkansas                 DAVID VITTER, Louisiana
THOMAS R. CARPER, Delaware           JOHN THUNE, South Dakota
CLAIRE McCASKILL, Missouri           ROGER F. WICKER, Mississippi
AMY KLOBUCHAR, Minnesota
   Margaret L. Cummisky, Democratic Staff Director and Chief Counsel
Lila Harper Helms, Democratic Deputy Staff Director and Policy Director
   Christine D. Kurth, Republican Staff Director and General Counsel
                  Paul Nagle, Republican Chief Counsel


                            C O N T E N T S

                              ----------                              
                                                                   Page
Hearing held on May 8, 2008......................................     1
Statement of Senator Kerry.......................................     1
Statement of Senator Stevens.....................................     3

                               Witnesses

Carlisle, Bruce K., Assistant Director, Office of Coastal Zone 
  Management, Executive Office of Energy, and Environmental 
  Affairs, Commonwealth of Massachusetts.........................    30
    Prepared statement...........................................    32
Hack, Ph. D., James J., Director, National Center for 
  Computational Sciences, Oak Ridge National Laboratory..........    11
    Prepared statement...........................................    13
MacDonald, Dr. Alexander (Sandy), Deputy Assistant Administrator 
  for Laboratories and Cooperative Institutes, Office of Oceanic 
  and Atmospheric Research, NOAA, U.S. Department of Commerce....     3
    Prepared statement...........................................     5
Reed, Daniel A., Chair, Computing Research Association (CRA).....    19
    Prepared statement...........................................    21
Sarachik, Edward, Emeritus Professor of Atmospheric Science, 
  Adjunct Professor of Oceanography, and Adjunct Professor of 
  Applied Mathematics at the University of Washington and Co-
  Director, Center for Science in the Earth System...............    26
    Prepared statement...........................................    28
Walsh, John E., Director, Cooperative Institute for Arctic 
  Research, International Arctic Research Center, University of 
  Alaska.........................................................    50
    Prepared statement...........................................    52

                                Appendix

Response to written questions submitted by Hon. Daniel K. Inouye 
  to 
  Dr. Sandy MacDonald............................................    68
Response to written questions submitted by Hon. John F. Kerry to:
    Bruce K. Carlisle............................................    67
    Dr. Alexander (Sandy) MacDonald..............................    68
    Dr. Daniel A. Reed...........................................    71
    Edward Sarachik..............................................    71


                       IMPROVING THE CAPACITY OF
                       U.S. CLIMATE MODELING FOR
                     DECISION-MAKERS AND END-USERS

                              ----------                              


                         THURSDAY, MAY 8, 2008

                                       U.S. Senate,
        Committee on Commerce, Science, and Transportation,
                                                    Washington, DC.
    The Committee met, pursuant to notice, at 2:41 p.m., in 
room SR-253, Russell Senate Office Building, Hon. John F. 
Kerry, presiding.

           OPENING STATEMENT OF HON. JOHN F. KERRY, 
                U.S. SENATOR FROM MASSACHUSETTS

    Senator Kerry. We will officially begin. I will not call 
you to order because I have never seen such a quiet, almost 
somnambulant crew. Are you all right? This whole audience is 
quiet.
    Welcome. We are really, really happy to have you here today 
and I apologize for the delay. We just have a little bit too 
much going on right now.
    This is really an essential component of our Nation's 
effort to try to understand climate change, and as you know, we 
are going to be having a very important debate here in a matter 
of weeks on this subject in the form of our cap and trade 
legislation. A lot of issues are being raised and are 
continually being raised with respect to our ability to be able 
to understand future climate impacts and what our response 
ought to be to those impacts. A lot of that will enter into the 
debate. No question.
    We are in good company, in a sense, today. There may not be 
as many of us as there are in London, but in England today 150 
of the world's top climate modelers are meeting, focusing on 
exactly the same set of issues that we are talking about here 
today.
    As our panel well knows, but just for the public's 
understanding and the record, we want to emphasize that climate 
modeling has been a subject of this Committee's inquiry for 
over 20 years now. Al Gore and I held the first hearings back 
in 1987, and then in 1988 Jim Hanson made the first comments 
with respect to climate change being upon us. Subsequently we 
have had many different hearings and meetings to try to better 
understand how we can define to people what we are looking at 
and what to expect. For the public, it is obviously very 
important in terms of policy.
    These models are the basis for the predictions of future 
temperature increases, sea level rise, storm surge, and the 
other impacts of global climate change. To date, the U.S. 
Government has played a key role in developing several of the 
world's best and most accurate models which serve as the basis 
for much of the information in the Fourth Assessment Report of 
the Intergovernmental Panel on Climate Change.
    However, we are now beginning to understand the limitations 
of these models. They currently cannot provide us with 
predictions on a regional or local level, and they cannot 
provide us information that is essential for states, 
communities, and resource managers as they adapt to the 
localized impacts of climate change. The models also tend to 
provide information over a long horizon, a long period of time, 
rather than on a decadal scale that would be more useful to 
some of these end-users.
    In addition, the models are currently not capable of 
identifying potential thresholds or tipping points which could 
also result in abrupt climate change impacts.
    One of the key issues that we are going to explore today is 
the issue of computing capacity. There are a number of 
different limitations. To run the models at the desired 
resolution, we need supercomputers a thousand times more 
powerful than we have today. While the United States has some 
of the most powerful supercomputing facilities in the world, we 
do not have a structure or strategy in place to coordinate the 
hardware, software, networking, and data storage functions 
required to produce the type of information that we need.
    As we consider this multitude of overlapping functions, our 
efforts have to be driven by the ultimate needs of the end-
users of this information, and I hope that Bruce Carlisle is 
going to keep us focused on that today. Bruce is the Assistant 
Director of the Massachusetts Office of Coastal Zone 
Management. He has been one of the leaders in creating a new 
state program called StormSmart Coasts, and this excellent 
program provides Massachusetts' cities and towns with the 
information and tools that they need to protect themselves from 
coastal storm damage and prepare for the impacts of climate 
change and rising sea levels. This first-of-a-kind program will 
serve as a model for the country, and I am proud of the work 
that has taken place in my home State.
    I would like to briefly also introduce the other witnesses 
and then recognize our Ranking Member of the Committee, Senator 
Stevens, and then welcome your testimony.
    Dr. Sandy MacDonald, Director of the Earth System Research 
Laboratory at the National Oceanic and Atmospheric 
Administration; Dr. James Hack, Director of the National Center 
for Computational Sciences at the Oak Ridge National 
Laboratory; Dr. Daniel A. Reed, Scalable and Multicore 
Computing Strategist at Microsoft; Dr. Edward Sarachik, Co-
Director of the Center for Science in the Earth System at the 
University of Washington; and Dr. John Walsh, Chief Scientist 
at the International Arctic Research Center, University of 
Alaska Fairbanks.
    Gentlemen, we are deeply appreciative for your taking time, 
some of you to travel considerable distance, and all of you to 
bring your expertise to the Committee today. We appreciate it 
very, very much.
    Senator Stevens?

                STATEMENT OF HON. TED STEVENS, 
                    U.S. SENATOR FROM ALASKA

    Senator Stevens. Thank you very much, Mr. Chairman. Sorry 
to be a little bit late. I thought we were going to have a 
vote, but it was canceled.
    Senator Kerry. So did I actually. I went over there and 
that is why I was late. I thought I was going to vote early and 
get back, but I wound up being late.
    Senator Stevens. I think everyone knows that Alaskans 
depend upon timely, accurate climate information for 
decisionmaking on a range of issues. The same with the rest of 
the country, home construction, transportation. But we 
particularly need it for fisheries and resource management. And 
I do support research and development of climate models that 
provide this information to Federal and State agencies, as well 
as local people who depend on it every day.
    I remain concerned, however, that the recent climate models 
with site-specific data may not be accurate enough for the 
planning on the State and community levels. It is essential 
that we have not only the best model capabilities, but also 
comprehensive climate data to use in those model simulations. 
Our Nation should make it a priority to improve both climate 
modeling and access to the necessary supercomputing 
infrastructure.
    I welcome the witnesses here today, as the Chairman has, 
including Dr. John Walsh who has traveled all the way from the 
International Arctic Research Center in Fairbanks to be here. I 
look forward to your testimony, John. Nice to have you here.
    I thank you very much for holding the hearing, Mr. 
Chairman.
    Senator Kerry. Thank you very much, Senator Stevens.
    So if we could begin with you, Dr. MacDonald, and we will 
just run right down the line. And if I could ask everybody--
each of your testimonies will be placed in writing in the 
record in full, and if you could summarize in approximately 5 
minutes, that way we can have a little more time to engage in a 
discussion between the panelists and ourselves. Thank you. Dr. 
MacDonald?

         STATEMENT OF DR. ALEXANDER (SANDY) MacDONALD, 
DEPUTY ASSISTANT ADMINISTRATOR FOR LABORATORIES AND COOPERATIVE 
 INSTITUTES, OFFICE OF OCEANIC AND ATMOSPHERIC RESEARCH, NOAA, 
                  U.S. DEPARTMENT OF COMMERCE

    Dr. MacDonald. Good afternoon, Senator Kerry and Vice 
Chairman Stevens. I am Dr. Alexander MacDonald. My friends call 
me Sandy. My job is Deputy Assistant Administrator for NOAA 
Research. And I thank you for inviting me to discuss the key 
role that NOAA has played in improving understanding and 
prediction of climate through the use of models.
    Advancement of the scientific community's knowledge and 
understanding of the way our planet's climate system works 
comes from three steps. One of them is that we improve our 
observations. Second, we improve our understanding, and third, 
computer modeling. It is like a three-legged stool--
observations, theory, and modeling--that together provide the 
foundation for our understanding of the way the climate system 
has changed in the past and how it may change in the future.
    NOAA proudly notes that the world's first global climate 
model was created by our scientists at the Geophysical Fluid 
Dynamics Laboratory, or GFDL. This climate model has been 
identified in the popular literature as one of the milestones 
of scientific computing, along with advances like the invention 
of the hand-held calculator and the Internet.
    A climate model really allows us to create a virtual earth 
in the computer. They divide the Earth into three-dimensional 
boxes, millions of boxes, that cover the entire Earth--called 
grid cells. At each of these grid cells, many calculations are 
performed over and over in order to simulate the processes that 
are important to climate. The size of the grid cells determines 
the resolution of the model. The smaller boxes give scientists 
information that is more refined.
    On this poster, what we see is the current resolution and 
generation of model, and we are looking at precipitation. So it 
just kind of shows a big general area of precipitation. In 
reality, we know that in the western United States the 
mountains all get a lot of precipitation and the valleys do not 
get so much. And of course, there is a great deal along the 
coast. In our new model resolution that we would like to run--
we see that detail in the precipitation. So this is what we are 
after, is getting the regional detail correct.
    Computer models of the Earth's climate have been central to 
NOAA's pursuit of its goal to understand climate variability 
and change and to enhance society's ability to plan and 
respond. These models have done so well that they have become 
central to our integrated assessments, such as the 2007 
Intergovernmental Panel on Climate Change that is used to 
inform industrial and government climate and energy analysis. 
We know that these models are important. They helped us 
understand that the Dust Bowl of the 1930s was due to ocean 
temperatures. They helped us to understand the El Nino/La Nina 
cycle and we are currently learning how the Atlantic Ocean 
circulation works. It is crucial to our science.
    There is an increasing need for the types of information 
that climate models provide. We have land managers in the 
western states that are dealing with prolonged periods of 
drought and requesting long-term regional temperature and 
precipitation data. The thing that we are after is getting that 
local scale so that we can help our decisionmakers in 
transportation, energy availability and for emergency 
preparedness. This is not just government. This is the public 
and industry also.
    However, today's models are limited in providing the level 
of climate information by two things. First, there are 
significant gaps in our understanding of how the climate system 
works, and second, we are limited by computing. We do not have 
the computing that we need to do some of these very regional 
kinds of things. The best of today's climate models really give 
us information on large geographic scale such as continental 
scale. These limitations can be addressed to a significant 
extent by increased access to large supercomputers.
    Our vision of a greatly improved climate prediction during 
the next 5 to 10 years would require approximately 100 times as 
much computing power over what is currently available.
    Climate models are crucial to providing reliable 
information on climate variability and change. More accurate 
projections of future climate will contribute to improved 
preparation at the Federal, State, and local levels and by the 
public and by industry.
    I look forward to working with the Committee.
    [The prepared statement of Dr. MacDonald follows:]

Prepared Statement of Dr. Alexander (Sandy) MacDonald, Deputy Assistant 
 Administrator for Laboratories and Cooperative Institutes, Office of 
  Oceanic and Atmospheric Research, NOAA, U.S. Department of Commerce

Introduction
    Good afternoon, Mr. Chairman and Members of the Committee. I am 
Alexander MacDonald, Deputy Assistant Administrator for Laboratories 
and Cooperative Institutes in the Office of Oceanic and Atmospheric 
Research at the National Oceanic and Atmospheric Administration (NOAA) 
in the Department of Commerce. Thank you for inviting me to discuss 
climate modeling and NOAA's key role in improving the understanding and 
prediction of global climate and how it is changing.
    NOAA's mission is to understand and predict changes in the Earth's 
environment and conserve and manage coastal and marine resources to 
meet our Nation's economic, social, and environmental needs. In support 
of the mission, NOAA researchers develop and use mathematical models 
and computer simulations to both improve our understanding and 
prediction of natural climate variability, as well as to identify and 
predict climate change. Climate models help create an informed society 
that uses a comprehensive understanding of the role of the oceans, 
coasts, and atmosphere in the global ecosystem to make the best social 
and economic decisions. The ongoing pursuit of these objectives--of 
increasing our knowledge of the complex global climate system and 
communicating the relevant information to stakeholders--is summarized 
in NOAA's climate goal ``to understand and describe climate variability 
and change so as to enhance society's ability to plan and respond.''
    Today, I will be discussing the societal demands for climate change 
information, how climate models are used to meet these demands, and how 
the Nation benefits by improving climate models.
Societal Demands for Climate Change Information
    Climate variability and change can have a profound influence on 
society. Recent evidence of global climate change includes multi-year 
droughts, warmer global surface temperatures, accelerating sea level 
rise, decreasing Arctic sea ice, retreating glaciers, the acidification 
of our oceans, and shifts in ecosystems.
    Federal, regional, state, and local decision-makers need credible 
climate information at increasingly finer geographic scales to adapt to 
and mitigate climate variability and change on time scales from seasons 
to centuries. Land managers in western states dealing with drought have 
requested long-term regional temperature and precipitation data, along 
with easily accessible and understandable tools for decision-support. 
Resource managers from numerous Federal agencies have requested site-
specific information to help plan for and manage the effects of climate 
change. Regions and municipalities have requested local information 
about climate change to improve long-term decision-making on 
transportation, energy availability, and for emergency preparedness.
    A broad scope of industries face operational challenges due to 
climate variability and change, including: utilities; integrated oil 
and gas; mining and metals; insurance; pharmaceuticals; building and 
construction; and real estate. Our understanding of how climate change 
impacts U.S. fisheries and the health of the world's ocean ecosystems 
will aid in effective long-term fleet planning and enhance the security 
of the Nation's food supply.
    More accurate predictions of future climate will contribute to 
improved preparation for and response to phenomena such as drought, 
hurricane activity, coastal inundation associated with storms and sea 
level rise, heat waves, poor air quality, and forest fires. The 
Nation's scientific community can provide this key information with 
comprehensive, state-of-the-art climate models (with related 
computational and data storage capabilities), that continue to advance 
the understanding of climate change and its potential consequences at 
local to global scales.

Using Climate Models to Meet Societal Demands
Climate Modeling to Inform Society
    Many advancements in the scientific community's knowledge and 
understanding of the way our planet's climate system works come about 
via a synthesis of improved observations, advancements in theory, and 
computer modeling. Like a sturdy three-legged stool, observations, 
theory, and modeling together provide the foundation for our 
understanding of the way the climate system has changed in the past, 
and how it may change in the future.
    Why are climate models so important for providing reliable 
information on climate change? Science generally proceeds from 
observations to theory, then to experiments to verify the theory's 
predictions against the observations, and finally further refinement or 
even refutation of the theory. In order to perform experiments, we need 
to replicate the system being studied. This poses a problem for the 
study of the Earth's climate, for there is only one Earth! The use of a 
computer model of the Earth--a ``virtual Earth''--allows us to perform 
``climate experiments.'' Other fields in which it is expensive or 
dangerous to perform real experiments make similar use of computer 
simulation. Car design is a good example--most designs are tested for 
aerodynamic efficiency and crash testing on a computer, before a design 
ever makes it to the shop floor. The design of nuclear weapons is 
another excellent example; given the ban on tests of these weapons, the 
United States is devoting significant resources to develop the ability 
to model nuclear detonations.
    Climate science and computer modeling of the Earth's climate have 
advanced greatly since the world's first coupled atmosphere-ocean 
global climate model was created in the late 1960s. At NOAA we proudly 
note that the world's first such climate model was created by 
scientists at NOAA's Geophysical Fluid Dynamics Laboratory (GFDL). The 
esteemed journal Nature identified this first climate model as one of 
the ``Milestones of Scientific Computing''--along with advances like 
the invention of the handheld calculator, the Internet, and CT 
scanners.
    Over the last four decades, climate models have improved as both 
scientific brainpower and high performance computing have been devoted 
to this work. During that time, climate modeling has gone from being of 
interest primarily to a fairly small segment of the scientific and 
academic community to being of great interest to a broad section of 
society--here in the United States and around the world. More than 
fifteen climate modeling centers now exist, including those run by NOAA 
partners at the National Science Foundation's National Center for 
Atmospheric Research (with additional support from the Department of 
Energy), and the National Aeronautics and Space Administration. NOAA 
has remained at the forefront of climate modeling through this 
transition. This is evident in NOAA/GFDL having produced not one, but 
two of the premier global climate models that played an integral role 
in last year's influential report issued by the Intergovernmental Panel 
on Climate Change (IPCC), for which the IPCC shared the 2007 Nobel 
Peace Prize.
    The best of today's climate models are most reliable on relatively 
large geographic scales (i.e., for regions comparable in size to a 
third of the contiguous 48 states, or larger), with increasing 
uncertainty associated with climate projections on smaller scales. 
Those climate model results are being used now for an increasing range 
of applications. Projected changes in surface temperature and 
precipitation patterns, storm tracks, ocean currents, and Arctic sea 
ice are only a few of the aspects of climate being examined intensively 
by experts in the academic, government, and private sector communities. 
The customer base for high-quality climate model results is rapidly 
increasing. At NOAA we actively support these efforts by making large 
amounts of our climate model output freely available. Consistent with 
the U.S. Climate Change Science Program's (CCSP) strategic plan, anyone 
can go to NOAA websites and download data files that document many of 
our climate model results. In this way, the output of NOAA's climate 
models becomes input into climate impact studies and assessments.
    However, the demand for scientifically credible projections (based 
on variable greenhouse gas scenarios) of future climate change goes 
beyond what currently can be offered. Today's models are limited in two 
primary aspects: (a) there remain significant gaps in our understanding 
of how the climate system works, and (b) models are constrained by 
available computing. This latter limitation means that while these 
models are at their best in simulating climate features at scales of 
several hundred miles and larger, there is increasing uncertainty in 
their simulation of smaller scale climatic features. In addition, some 
of the processes operating in the climate system on small geographic 
scales are missing, and yet these processes may be important for large-
scale climate. Both of these limitations can be addressed to a 
significant extent through the use of very large supercomputers. As an 
additional benefit, access to advanced supercomputers makes it easier 
for NOAA to attract and retain the world's best climate scientists, to 
run models that resolve phenomena at the scale of a single state or 
even city.

What is a Climate Model?
    Climate models divide the three-dimensional global ocean and 
atmosphere into millions of boxes referred to as grid cells. At each of 
those grid cells many calculations are performed over and over again in 
order to simulate the time evolution of processes important to the 
climate. The number and size of a climate model's grid cells are 
largely determined by the amount of computer resources available; more, 
smaller boxes results in more calculations which require more computing 
power. Higher geographic resolution (more, smaller boxes) are desirable 
for climate models for much the same reason people prefer the picture 
quality of a high definition TV as compared to a grainy YouTube video: 
higher resolution provides a more detailed representation of the 
features in which we are interested, which benefits both scientific 
researchers and stakeholders. As a point of reference, in NOAA's recent 
climate models atmospheric grid points were of a size such that one 
box's surface area covers about twice the land surface area of the 
Commonwealth of Massachusetts. That means Maine is covered by two 
boxes, North Dakota and Washington State by about 4 each, and Texas by 
13. Since it takes several grid boxes to properly define or resolve a 
pattern, we can say that today's global climate models are limited in 
their ability to fully resolve features on spatial scales much less 
than the size of the 48 contiguous states.
    We test our understanding of climate, as expressed in a computer 
model, by comparing how well that model does against observations of 
past climate. For instance, we might initialize our model of the 
Earth's climate with its known state in, say, 1750--the 
``preindustrial'' climate, then apply the history of all the known 
external forces on the climate--solar variability, volcanoes, 
industrial emissions, land use changes--and see how well we do in 
predicting the known history of the 20th century climate. Our successes 
and failures help us refine our theories and our understanding. It is 
possible to ``tune'' a model to perform well against a given metric of 
skill--say the global mean surface temperature but we use a wide range 
of metrics (e.g., temperature, rainfall patterns, number of storms, 
wintertime snow cover, etc.)--and the only way to do well against a 
diverse and comprehensive set of metrics is to represent the physical 
climate system with fidelity.
    Models of the Earth system have many components and feedback loops. 
Today's models include interactions among many components, including 
the ocean, atmosphere, sea ice, vegetation, ecosystems, and reactions 
between natural and industrial chemicals in the atmosphere. With 
increasing complexity, new challenges appear. For example, a key 
research area in the current generation of climate models is to capture 
the effect of aerosols, which include industrial pollutants, soot, 
dust, and sea spray on climate. Aerosols block sunlight directly, but 
they also impact the formation of clouds, a key player in the climate 
system. Progress in this key topic is delayed because our ability to 
represent such computationally expensive climate processes in our 
models has outpaced the available computing resources on which to run 
them.

Current Modeling Capabilities and Achievements
    Computer models of the Earth's climate have been central to NOAA's 
pursuit of its goal to ``understand climate variability and change and 
enhance society's ability to plan and respond.'' These models have done 
so well over time that they have now become central to the integrated 
assessments that are used to inform industrial and governmental climate 
and energy policy. The leading international assessments such the IPCC, 
and focused products from the CCSP, both synthesize results from 
computer models to answer key questions asked by policymakers.
    At the time of the first IPCC report in 1991, NOAA/GFDL's model was 
one of the few models capable of producing reasonably realistic 
simulations of the Earth's climate. Since then, several centers around 
the world have developed climate models, and the assessment reports are 
now based on ``model intercomparison projects,'' where coordinated 
computations are independently run by different centers around the 
world. It is a testimony to NOAA/GFDL models' pre-eminence in the field 
that in 2007, at the time of the IPCC Fourth Assessment Report, they 
are still seen as being at the very apex of climate modeling, on the 
basis of independent evaluations of their performance against a wide 
range of metrics of skill.
    Specific achievements of NOAA's current climate models are 
manifold. NOAA climate modeling has helped demonstrate that the U.S. 
Dust Bowl in the 1930s and the drought in the African Sahel of the 
1980s were both caused in part by changes in the temperatures of the 
oceans. Our current understanding of El Nino and of how El Nino affects 
the U.S. climate is based in large part on NOAA research with climate 
models. NOAA climate modeling first pointed to the importance of the 
circulation of the Atlantic Ocean as a potential source for abrupt 
climate change. Further, NOAA models have clarified the competition 
between warming due to increasing concentrations of long-lived 
greenhouse gases and cooling due to short-lived atmospheric particles 
generated by human activity.
    NOAA models have also been major contributors to the most recent 
Ozone Assessments conducted by the World Meteorological Organization 
(WMO), evaluating the response of the Antarctic ozone hole to the 
reductions in the emissions of chlorofluorocarbons that followed the 
Montreal Protocol and projecting the future evolution of the ozone 
shield. NOAA has also developed climate models with higher geographic 
resolution that are currently being used to develop climate change 
projections over North America, as part of the North American Regional 
Climate Change and Assessment Program.
    The computer models themselves represent an important NOAA product. 
NOAA/GFDL's Modular Ocean Model (MOM) is the world's most widely used 
numerical model for simulating ocean circulation at the global scale 
and for understanding and predicting ocean climate phenomena. 
Significant recent advances include the ability to directly predict 
sea-level changes as well as improved representations of the complex 
features of the ocean's heat and chemical distributions. Over 400 
scientists around the world are now using MOM to perform oceanographic, 
weather, and climate studies. It is used for operational weather 
forecasting at NOAA's National Weather Service.

Benefits from Improving Climate Models
    NOAA's state-of-the-art climate models were used extensively in the 
latest IPCC assessment, the most recent WMO ozone assessment, and the 
ongoing North American Regional Climate Change Assessment Program. But 
despite recognition from independent experts as being among the highest 
quality climate models in the world, the models are not able to meet 
the growing suite of societal demands for climate change predictions. 
Current models are limited by some remaining gaps in our understanding 
of how the climate system works, and in computer resources. The lack of 
adequate computer power prevents us from making optimal use of existing 
knowledge by extending our simulations to smaller geographic scales and 
including a more complete set of climate processes.
    An example of a gap in understanding that is holding back progress 
in climate modeling is our lack of understanding of the Greenland and 
Antarctic ice sheets, a major source of uncertainty in predicting the 
future sea level. Recent observations have highlighted the potential 
for rapid changes in the ice sheets and the inadequacies of current 
theories of ice sheet dynamics. Coordinated progress will be needed in 
ice sheet observations, a buildup of the human capacity in this 
research field, and experimentation incorporating new models of the ice 
sheets into our climate models. Another key gap is our inadequate 
understanding of the factors that control the Earth's cloud cover and 
how it might change as the Earth warms. This gap is a key source of 
uncertainty in predicting the magnitude of the warming resulting from a 
given change in atmospheric carbon dioxide.
    Improving understanding on such central questions is fundamental to 
progress, and we are confident that our climate models will improve as 
they begin to explicitly resolve smaller geographic scales. The scales 
that our models resolve are determined by the available computer 
resources. With currently available computer resources, our models are 
most reliable at simulating climatic features with geographic scales of 
several hundred miles and larger, with increasing uncertainty in the 
simulation of smaller scale phenomena. The following are some of the 
benefits related to the inclusion of smaller scale processes in models:

    1. Projections of temperature and precipitation on smaller scales 
        than those currently resolved adequately by climate models to 
        aid decision-makers and planners at the regional and local 
        levels. For example, trends in many local water resources are 
        affected by small-scale topographic features and land-use 
        patterns that are not represented in current climate models.

    2. Many of the greatest effects of climate change may come about 
        through changes in extreme events, such as hurricanes, heat 
        waves, droughts, and floods. The climate models used in the 
        recent IPCC assessment do not provide adequate simulations of 
        hurricanes, for example. Other extreme events, such as droughts 
        and floods, are strongly influenced by small-scale processes 
        that are not well resolved in these models.

    3. It is likely that small-scale ocean currents and other ocean 
        processes may play a crucial role in the future behavior and 
        stability of the Antarctic and Greenland ice sheet, with large 
        potential influences on sea level rise.

    4. The response of ecosystems to climate change, including the 
        cycling of carbon through the system, is highly uncertain in 
        current models. This is strongly influenced by limited 
        computational resources, preventing the inclusion of important 
        small-scale processes, such as intense ocean upwelling near the 
        coasts, which are crucial to the global cycling of carbon.

    5. Improved predictive capability to support integrated national 
        air quality policy and regional emission management strategies 
        for air quality and climate. The prediction of climate change 
        impacts on air quality could be better assessed by including 
        smaller scale processes into models.

Pathways to Climate Model Improvements
    The next generation of climate models that explicitly include 
smaller-scale processes has been developed in NOAA. Prototypes of these 
models have been tested, but computer resources in NOAA are inadequate 
to use these models for the comprehensive simulations of climate change 
that are necessary to provide stakeholders with robust predictions of 
climate change. We cite here two examples of next generation models 
that have been developed but are too computationally expensive to run 
extensively given current resources:

    1. A new climate model has been developed that resolves important 
        ocean features on scales as small as 20 miles (Figure 1). For 
        comparison, models used in the most recent IPCC assessment 
        resolve ocean features on scales of 200-300 miles. The 
        inclusion of the small-scale ocean features may produce large 
        improvements in understanding how ocean circulation responds to 
        global warming, with major climatic impacts. This includes how 
        much carbon dioxide the ocean will absorb (or outgas) in the 
        future, the response of marine ecosystems to global warming, 
        how El Nino will respond to global warming, and the potential 
        for abrupt climate change due to changes in the circulation of 
        the Atlantic Ocean.

        
        
    Figure 1. A new climate model has been developed that resolves 
crucially important ocean features on scales as small as 20 miles. 
Application of this model for comprehensive climate change predictions 
would deliver much more credible predictions of the ocean's response to 
global warming, including the effect on marine ecosystems, carbon 
uptake, and ocean acidification.

      Application of this model for comprehensive climate change 
        predictions would deliver much more credible predictions of the 
        ocean's response to global warming, including the effect on 
        marine ecosystems, carbon uptake, and ocean acidification. This 
        would also greatly improve the prediction of decadal scale 
        changes in the ocean that may strongly influence hurricanes and 
        droughts, as well as predictions of Arctic climate change and 
        sea ice. However, NOAA does not currently possess the 
        computational capability to use this model. Applying this model 
        for the next IPCC assessment report would require approximately 
        10 times NOAA's current computing resource, which is comparable 
        to the largest machines in the United States. NOAA does not now 
        have access to these systems.

    2. A global atmospheric model is being developed that resolves 
        processes on a geographic scale of about 10 miles. A regional 
        version of this model faithfully simulates Atlantic hurricane 
        activity (Figure 2). The global version will simulate important 
        high impact climate phenomena and small-scale variations of 
        rainfall around the world. Use of such a model for 
        comprehensive predictions of climate change would increase our 
        confidence in the prediction of how hurricanes will change in 
        the future. This model would also be a great improvement in our 
        ability to predict regional climate change over the United 
        States, including such features as future changes in western 
        U.S. snowpack with associated water resource implications 
        (Figure 3). The output from this model would be of substantial 
        value across a wide suite of applications, from water resource 
        and infrastructure development to agricultural planning.

        
        
    Figure 2. A regional version of a global model with 10 mile 
resolution can faithfully simulate Atlantic hurricane activity. The 
global version will simulate important high impact climate phenomena 
and small-scale variations of rainfall around the world.



    Figure 3. A prototype model with a resolution of 30 miles was used 
to support the North American Regional Climate Change Assessment 
Program (NARCCAP) and simulates substantially more of the features in 
the precipitation field in the western U.S. than do current models. A 
global model with 10 mile resolution is expected to improve the capture 
of the amount, timing, location, and type of precipitation in order to 
better predict water resource issues arising in the western U.S., a key 
concern that has been identified by NOAA customers.

      The use of this model in comprehensive climate change predictions 
        would provide climate change predictions on geographic scales 
        of ten to twenty miles. However, it would require approximately 
        50 times NOAA's current computing capability to apply this 
        model to the next IPCC assessment report. Although this level 
        of computing corresponds to roughly half of the Nation's entire 
        research high performance computing capacity, a limited set of 
        climate integrations with this model could be used to advance 
        our understanding of how climate change affects high-impact 
        phenomena.

    These fine resolution oceanic and atmospheric climate model 
components will advance our understanding of and ability to predict 
climate. But our ambition is to combine them into a fine resolution 
coupled climate prediction system that is commensurate with the 
requests of policymakers and stakeholders at the regional and local 
levels. In the next 5-10 years, NOAA will work toward advancing the 
fidelity and utility of our climate models and combining the advantages 
of finer resolution in both the oceans and the atmosphere while fully 
capturing their complex interactions. Fulfilling such a vision would 
require approximately 100 times as much computing power as is currently 
available.

Conclusion
    We now have a deeper understanding of the climate system and the 
delivery of climate information to the Nation as a direct result of 
NOAA scientists and their collaborators using high performance 
computing for numerical simulation. Climate models have demonstrably 
improved our ability to simulate the Earth's climate. However, the 
demand for scientifically credible projections of future climate change 
goes beyond what currently can be offered. Scientific advancements and 
the generation of new climate information products that arise from 
better climate models are intimately tied to the state-of-the-art 
computers that are devoted to running them. NOAA is poised to run 
advanced climate models that resolve regional scale features in the 
atmosphere and ocean, incorporate the effects of chemistry and aerosols 
on climate, and provide long lead-time predictions of high-impact 
climate phenomena such as drought and hurricane activity.
    Thank you again for inviting me to discuss climate modeling and 
NOAA's key role in improving the understanding and prediction of global 
climate. Robust climate models help NOAA to provide reliable 
information on climate change. Many advancements in the scientific 
community's knowledge and understanding of the way our planet's climate 
system works have come about via a synthesis of improved observations, 
advancements in theory, and computer modeling. I look forward to 
working with the Committee on any further information you may require 
for your deliberations on this topic.

    Senator Kerry. Thank you, Doctor.
    Dr. Hack?

 STATEMENT OF JAMES J. HACK, Ph.D., DIRECTOR, NATIONAL CENTER 
             FOR COMPUTATIONAL SCIENCES, OAK RIDGE 
                      NATIONAL LABORATORY

    Dr. Hack. Thank you, Senator Kerry, and Vice Chairman 
Stevens, for the opportunity to speak with you today on ways to 
improve the capacity of U.S. climate modeling. My name is James 
J. Hack and I serve as Director of the National Center for 
Computational Sciences, which is located at Oak Ridge National 
Laboratory, and provides the most powerful computing resources 
in the world for open scientific research. One of the prominent 
NCCS research focus areas is the exploration of the Earth's 
climate system and climate change.
    There are many scientific and technical challenges related 
to monitoring, understanding, predicting, and adapting to 
climate change. Observations of the climate system are the 
foundation for our improved understanding of climate change and 
for building the computer models that are used to project the 
evolution of the climate system. Computational research 
associated with the modeling and prediction of Earth's climate 
includes developing methods for simulating complex multiphase 
flow over a wide range of scales with high fidelity, with high 
efficiency on the most powerful supercomputer systems 
available, and in a software environment that needs to 
continually incorporate new knowledge and new theoretical 
concepts into the models.
    State-of-the-art climate models embody our best 
understanding of the many complex processes that are central to 
the climate system. The goal of such modeling efforts is to 
accurately represent the collective behavior of these processes 
as an interactive system. The models are continually developed, 
tested, and evaluated against observations. They are the best 
available tools for exploring how the climate system works and 
how it is likely to evolve.
    Despite their imperfections, climate models are remarkably 
capable of reproducing the climate of the past, which builds 
confidence in their projections of future climate. They are 
also remarkably consistent in their projections of continued 
warming of the climate system for the remainder of this 
century, which was more completely discussed in the Fourth 
Assessment Report of the U.N. Intergovernmental Panel on 
Climate Change. The release of the IPCC report signaled that 
the detection and attribution of climate change at global 
scales has essentially been resolved.
    So the community is now faced with a new set of urgent 
problems relating climate change to human health, water 
resources, food supplies, and changing risks to manage the 
natural ecosystems. Central to these problems is the demand for 
much more regional detail about climate change on time scales 
of resource and infrastructure planning. In order to address 
these issues, along with important questions on mitigation and 
adaptation strategies, the climate community needs to develop 
and undertake a new coordinated research program that is 
balanced and integrated among observation, theory, and 
computation.
    Meeting these future challenges will require advances in 
every aspect of the models' theoretical, observational, and 
computational foundation. Many of society's questions will 
require the development of a new generation of more 
comprehensive climate models, frequently referred to as Earth 
System Models, that predict the coupled chemical, 
biogeochemical, and physical evolution of the climate system. 
Addressing the science issues will require new observations and 
new methods of analysis, new theoretical understanding, and new 
features and models of the earth system that include the 
interactions between human and natural systems. These models 
will play an important role in synthesizing a broad range of 
observations and projecting the future responses of human 
society and the natural world to the evolving climate regimes.
    The models will also need to be exercised at unprecedented 
high resolution. The needed increases in complexity and 
resolution will require transformational changes in 
computational capability. A flexible leadership-class 
computational science infrastructure will continue to be 
essential to making these advances possible.
    So in conclusion, there is no single pacing item to the 
advancement of climate change science, but a collection of 
interrelated science and technology challenges. Many of the 
issues discussed in this testimony speak to the need for a 
balanced investment in computational infrastructure, climate 
science and modeling, climate observations, computer science, 
and applied mathematics. In the short and long term, 
computational capability remains a significant bottleneck and 
should remain a high priority investment.
    But as the science and complexity of climate simulation 
continues to grow, so will new technical and scientific 
challenges. Proactive investments in climate modeling science, 
software, algorithms, data management, and other pacing items 
will ensure that scientific progress can keep pace with the 
rapidly evolving computational environment. Strategic 
programmatic management of such a broad multidisciplinary 
activity will also likely prove to be the most effective way to 
ensure that any new investments have the desired impact on 
accelerating progress.
    This is an exciting opportunity for the Nation to lead the 
world in developing a better understanding of the consequences 
of climate change. Thank you again for the opportunity to 
address the Committee, and I look forward to answering any 
questions.
    [The prepared statement of Dr. Hack follows:]

 Prepared Statement of James J. Hack, Ph.D., Director, National Center 
       for Computational Sciences, Oak Ridge National Laboratory

    I thank Chairman Inouye, Vice Chairman Stevens, and the other 
Members of the Committee for the opportunity to speak with you today on 
ways to improve the capacity of U.S. climate modeling for decision-
makers and other end-users. My testimony draws on over two decades of 
developing global models of the climate system at the National Center 
for Atmospheric Research. My name is James J. Hack and I currently 
serve as director of the National Center for Computational Sciences 
(NCCS) at the Oak Ridge National Laboratory (ORNL). The ORNL NCCS 
provides the most powerful computing resources in the world for open 
scientific research. It is one of the world's premier computational 
science research environments supporting advances in our understanding 
of the physical world and using that knowledge to address our most 
pressing national and international concerns. My role as NCCS Director 
provides a unique perspective on how the application of leadership-
class computing technology in a computational science partnership with 
scientific investigators can radically accelerate basic progress for a 
variety of extremely demanding scientific domains. Examples of NCCS 
research focus areas are the simulation of complex biomolecular systems 
with applications to pharmaceuticals as well as more efficient biofuel 
generation, simulations that investigate the fundamental properties of 
materials, such as high temperature superconductors, and simulations 
exploring the processes that maintain and regulate Earth's global 
climate system.
    There are many scientific and technical challenges related to 
monitoring, understanding, predicting and adapting to climate change, 
especially on local and regional scales. Observations of the entire 
Earth, for instance, are the foundation for improved understanding of 
climate change and for computer models that accurately predict weather 
and climate. A newly emerging issue is the development of optimal 
methods for assimilating this broad range of physical, chemical, and 
biogeochemical observations into models of the Earth system in order to 
more completely describe the current state of the system. This is but 
one example of how the synthesis of models and observations is critical 
both for understanding the present climate and for simulating its 
evolution over the next several decades. Computational research 
associated with the modeling and prediction of Earth's climate system 
includes developing methods for simulating complex multiphase fluid 
motions over a wide range of scales with high fidelity and with high 
computational efficiency, as well as by the need to continually 
incorporate new theoretical and observational knowledge into global 
models. The rapid evolution of computer architectures creates its own 
challenge to fielding stable computational environments that support 
Earth system science.
    State-of-the-art climate models, such as those developed by NSF, 
NOAA, DOE Office of Science, and NASA programs embody our best 
understanding of the physical and biogeochemical processes that are 
central to the climate system. The goal of such modeling efforts is to 
accurately represent the collective behavior of these climate processes 
as an interactive system. These models are continually developed, 
tested, and evaluated against observations. Although they are the best 
available tools for exploring how the climate system works, they are 
not perfect. Uncertainties arise from shortcomings in our scientific 
understanding of the climate system, and in identifying the best 
mathematical approaches for representing those processes we do 
understand in numerical models.
    Despite these imperfections, climate models are still able to 
reproduce the climate of the past, which gives considerable confidence 
in their ability to simulate changes in future climate. For instance, 
climate modelers are able to test the role of various forcings in 
producing observed changes in climate over the past century. Such 
simulations have now reliably shown that global surface warming of 
recent decades is a response to the increased concentrations of 
greenhouse gases in the atmosphere. They are also remarkably consistent 
in their projections of continued warming of the climate system for the 
remainder of this century, as discussed in the Fourth Assessment Report 
(AR4) of the United Nations Intergovernmental Panel on Climate Change 
(IPCC). The release of the IPCC AR4 report, along with release of a 
series of Climate Change Science Program (CCSP) reports, signal that 
the detection and attribution of climate change at global scales has 
essentially been resolved. The global community is now faced with a new 
set of urgent problems relating climate change to human health, water 
resources, food supplies, changing risks of forests to fires and insect 
disease, and threats to managed and natural ecosystems. Central to 
these problems is the demand for much more regional detail on climate 
change on the time scales of resource and infrastructure planning. In 
order to address these issues, the community needs to develop and 
undertake a coordinated research program balanced and integrated among 
observation, theory and computation. Meeting future challenges in 
climate change science will require qualitatively different levels of 
scientific understanding, modeling capabilities, and computational 
infrastructure than are currently available to the climate science 
community. Many of society's questions will require the development of 
a new generation of more comprehensive climate models, frequently 
referred to as Earth System Models (ESMs) that predict the coupled 
chemical, biogeochemical, and physical evolution of the climate system. 
These models will also need to be exercised at unprecedented high 
resolution. The needed increases in complexity and resolution will 
require transformational changes in computational capability.
    Over the last 30 years, modeling capabilities have advanced 
considerably in their treatment of complexity, and the ability to treat 
ever finer scales of motion. Modern atmospheric models represent the 
observed equator to pole energy transport much more realistically than 
did earlier model generations. They also do a much better job of 
representing many detailed features of the observed mean climate state. 
These improvements have meant that global climate models are now 
routinely run with fully-interacting atmosphere, ocean, land surface, 
and sea ice components. These more realistic and complex models can now 
not only simulate observed changes over the past century in global mean 
climate, but also climate variability and change on continental scales. 
This includes the attribution of many of the observed large-scale 
changes in indicators of climate extremes consistent with a warming 
climate, such as the annual number of frost days, warm and cold days, 
and warm and cold nights. Models that contributed simulation results to 
the IPCC AR4 also generally agree that regions like the subtropics will 
dry, including the U.S. Southwest, while polar latitudes will receive 
more precipitation related to large changes and shifts in the 
extratropical storm tracks.
    On finer spatial scales, however, state-of-the-art climate models 
don't always agree on projected climate change impacts, either on 
decadal or longer time scales. It is also not clear that they can 
accurately project changes in extreme events, or can reliably simulate 
changes in low-frequency climate variability or the likelihood of 
abrupt change. Near-term investments in the climate science enterprise 
could lead to a significant quantitative improvement in the scientific 
community's ability to address these difficult but societally relevant 
questions, leading to improved guidance to policymakers and 
stakeholders charged with developing strategies for adapting to climate 
change.
    One immediate scientific challenge and opportunity is the 
incorporation of chemical and biogeochemical processes in climate 
models. The science surrounding the chemical and biogeochemical 
coupling of climate has become central to answering climate change 
questions, particularly those associated with the global carbon cycle. 
Addressing the science issues will require new observations and methods 
of analysis, new theoretical understanding, and new models of the Earth 
system that include the interactions between human and natural systems. 
These models will play pivotal roles in interpreting the paleoclimate 
records, in synthesizing and integrating observational measurements to 
study the current carbon cycle, and in projecting the future responses 
of human society and the natural world to evolving climate regimes.
    Another example of a pressing scientific challenge is the rate of 
sea level rise and the impact of that rise on coastal communities. 
Recent observations indicate ice sheets can dissipate on much more 
rapid timescales than from melting alone due to dynamical processes in 
large outlet glaciers and ice streams within the ice sheet. Faster 
disintegration of the ice sheets will contribute to faster sea level 
rise and will pose a greater risk of abrupt changes in the climate 
system. Abrupt climate change can also result from thresholds and 
nonlinearities in the response of climate to slower time scale forcing 
of the climate system. Examples include rapid changes in ocean 
circulation, large scale vegetation mortality and succession, release 
of methane frozen in ocean and permafrost, and megadroughts. The 
climate community will need to use models to identify thresholds of 
forcing in the climate system and explore the likelihood and impacts of 
such scenarios. The community's efforts to advance climate modeling and 
its application to science and technology options for mitigation and 
adaptation will require advances in essentially every aspect of the 
models' theoretical, observational, and computational foundation. 
Quantifying uncertainties in predictions will require a new level of 
integration between modeling and observational science. New 
mathematical methods and algorithmic techniques will also be required 
to address the fundamental challenges of multi-scale coupling of 
physical, dynamical, chemical and biogeochemical processes. A flexible 
leadership-class computing infrastructure has been and will continue to 
be a key factor in making these advances possible.
    As mentioned earlier, today's climate models are in strong 
agreement that global and continental-scale temperatures will continue 
to rise as a result of human activities. However, it is also important 
to improve our understanding of the likely changes in regional climate 
over the next few decades. Climate forecasts on decadal time scales are 
governed primarily by the history of the ocean circulation and the 
current atmospheric forcing. Therefore, climate forecasts on these time 
scales will require retrospective analyses of the global oceans to be 
able to accurately initialize the forecasts. The ocean is responsible 
for much of the inertia or near-term ``memory'' in the climate system. 
The development of ocean data assimilation techniques, largely an 
applied mathematics and algorithmic challenge, will be necessary to 
provide an initial ocean state for decadal prediction and represents a 
pacing item for seasonal, inter-annual, and decadal prediction. While 
assimilation has been extensively developed and used in the weather 
community, the climate community will need to evaluate which 
assimilation methodology is best suited for climate simulation and the 
creation of realistic initial states for climate change scenarios. 
Optimal interpolation and simple methods have so far been adequate for 
the ocean due to sparseness of data, particularly for salinity and for 
ocean properties at depths below 1000m. With the influx of new ocean 
data sets, advanced techniques will need to be examined. Recent 
progress in deploying large numbers of floats and the launch of new 
satellites that together will measure salinity profiles will greatly 
improve our ability to effectively constrain ocean models with 
assimilation. For example, assimilation of data from ARGO floats with a 
fully coupled climate model has shown great promise in determining the 
state of the climate system, although the assimilation process is 
extremely computationally demanding.
    Accurate projections of changes in the frequency of climate 
extremes at relatively high geographic and temporal resolution will be 
essential for the development of robust adaptation strategies. However, 
current climate models have been designed primarily to predict patterns 
of change at a coarser level. Much more research is required to 
understand how increasing model resolution and employing increasingly 
sophisticated parameterized treatments of non-resolvable processes may 
affect the ability of models to more accurately simulate changes in 
local extremes. In particular, the relationships between extreme 
statistics and synoptic-scale low-frequency variability are not 
understood.
    A better understanding of low-frequency variability is critical for 
the detection of climate-change signals. For Earth system modeling, it 
is important to characterize the natural modes of coupled variability 
in the carbon cycle, terrestrial ecosystems, and dynamic vegetation. It 
is also important to develop a better understanding of external forcing 
mechanisms, such as the role of solar variability in the broader 
context of the Sun-Earth system. Current understanding of these complex 
systems is limited by the length of the observational record. The wide 
dynamic range in the relevant space and time scales further complicates 
the coupling issues. New mathematical methods designed for multiscale 
systems hold promise for this class of problems, and these methods 
should be explored for efficient implementation in next generation 
models.
    As suggested earlier, a large number of significant impacts could 
follow from abrupt changes in the climate system. These occur when the 
gradual increases in climate forcing trigger an abrupt transition of 
the coupled system to a new state. Potential examples of abrupt change 
include dynamic dissolution of the ice sheets and bifurcations of the 
ocean circulation system. Characterization of abrupt climate change 
requires a new paradigm for climate change modeling, one in which the 
models are integrated over the full range of uncertainties in forcing 
and parameterized physics. Exploration of this phase space will require 
implicit formulations of the coupled system designed for fast 
equilibration combined with new mathematical techniques and a sustained 
petascale computing capability.
    Multiscale interactions also complicate treatment of the climate 
system. As with the broader issues of climate variability, process-
level understanding of things like the water cycle is limited by the 
lack of basic observations. While the absence of these data still 
represents a barrier to progress, near-term enhancements in 
computational capacity would permit the resolution of fundamental 
phenomena at the process level. Targeted investments in observational 
programs can provide much of the necessary data to validate high-
resolution process modeling studies of critical topics like aerosol-
cloud interactions, central to the climate model sensitivities that 
lead to discrepancies in projections of future climate on century-long 
time scales.
    Finally, there are significant software and computational hardware 
infrastructure challenges pacing progress in climate science. Many 
scientists have found the growing requirements to support the software 
on high performance computers as a distraction from the central 
scientific goals of improving climate models and answering fundamental 
questions about climate feedbacks and variability. This drawback is 
offset by the new scientific opportunities provided by dramatic 
increases in computational power. This becomes an issue of scientific 
productivity. What is needed is a software framework that not only 
scales from desktop to petascale, but also that supports multi-scale 
model development and process integration. As a closer connection with 
observational data and process studies is required to advance the 
science of regional climate prediction, the software must also become 
more closely integrated and supported across scales. A flexible and 
powerful software development environment will increasingly be required 
to support data assimilation and other data intensive frameworks. The 
limitations of existing software environments have emerged as key 
bottlenecks to progress where near-term investment would have important 
scientific payoffs.
    There are real opportunities to invest in climate change science to 
improve the utility of global models for decisionmakers and the broader 
end-user community. High impact opportunities for investment include 
computational facilities, theoretical efforts associated with model 
development, targeted observational programs and the development of 
novel computational algorithms. Investments in modeling will accelerate 
progress on improving the predictive skill of global climate models. 
The climate community needs to develop a new generation of Earth System 
Models based upon new and expansive requirements including the ability 
to more accurately reproduce major modes of natural variability, 
incorporating functionality for decadal-scale ensemble forecasts at 
very high spatial resolution, the flexibility to incorporate new data 
on the physical, chemical, and ecological climate system in the form of 
process representation (thereby increasing the fidelity of climate 
simulations), stronger connectivity with user communities for exploring 
adaptation and mitigation strategies, and the capability for two-way 
interactions among emissions, impacts, adaptation, and mitigation.
    Modeling over a large range of time scales to fully evaluate the 
couplings between biogeochemical cycles, chemistry, and ecology will 
present a significant computational challenge. The growth requirement 
of characteristic applications of climate change prediction models 
already more than doubles every year. High-resolution ocean circulation 
studies and cloud system resolving atmospheric simulations are already 
pushing the limits of petaFLOP systems that utilize many tens of 
thousands of processors. As regional climate prediction on decadal to 
century time scales becomes more important, the required computational 
power will approach the exaFLOP scale (one quintillion floating point 
operations per second) that will utilize 100K-1M processors. This will 
require a continued focus on fielding state-of-the-art leadership class 
computing facilities so that computational capability does not become a 
more critical pacing factor. Ancillary investments in software, 
networking, data storage, collaborative tool, and visualization 
technologies are necessary for balance. For example, climate science is 
both distributed and collaborative. As interest in climate science 
continues to grow and its scope broadens to encompass issues of 
ecosystem and economic impacts, and the evaluation of mitigation and 
adaptation strategies, the number of participants will also increase. 
The overall productivity of researchers and the quality of the research 
output can likely be improved significantly by the use of advanced 
collaboration technologies that distribute applications and data across 
the network. It is easy to project that climate research demands on 
networks will grow yet further as data volumes increase. With a growing 
number of participants in the climate science enterprise, and a growing 
diversity and volume of climate data, the need for new data and network 
resource management strategies and technologies will emerge. Modern 
visualization capabilities can also play an important role in the 
discovery of new scientific results and in the communication of the 
science to a broader community of stakeholders. For an area like 
climate modeling this is particularly important because of the societal 
relevance of our results to policymakers and those concerned about the 
consequences of climate change.
    New observational programs and data assimilation systems represent 
opportunities to improve our understanding of a variety of physical, 
chemical, biogeochemical, and ecological processes, reducing key 
uncertainties in modeling assumptions. Meteorological and oceanic 
analyses have become important tools for studying the mean state and 
variability of the current physical climate. These analyses are 
constructed using a model that is adjusted by incorporating 
observations during its numerical integration. These analyses have 
proved particularly useful for understanding the relationship between 
observations and the underlying dynamics of the climate system. It 
would be especially valuable to have a comparable analysis of 
biogeochemical and chemical cycles that could relate local and global 
biogeochemical processes to more completely describe the state of the 
global system. However, there are no existing analyses that encompass 
the physical, chemical, and biogeochemical processes in the climate 
system. Development of these analyses will require significant 
investment in assimilation systems for chemical and biogeochemical 
observations from in situ and satellite platforms. Much more advanced 
models will be required to understand the fidelity of the analysis 
system, which will further push the sophistication of global modeling 
activities.
    Investments in computational algorithms will increase scientific 
productivity using leadership-class computers for climate change 
simulation studies and improved simulation accuracy. There is a broad 
class of mathematical and numerical algorithms that are ready to be 
explored for application to the climate problem. For example, there are 
strong arguments for exploiting higher-resolution variable gridding 
configurations for the atmospheric component of a climate model. The 
computational demands of uniform ultra-high resolution configuration of 
a global atmospheric model would outstrip existing computational 
capability. An intermediate practical approach to dealing with 
resolution issues is to use a multi-resolution approach, such as nested 
refinement. These approaches will allow scientists to improve 
understanding of the multi-scale interactions in the climate system, to 
identify those of greatest importance, and to document their effects on 
climate. Ultimately, such research will help determine the best methods 
of including these multi-scale interactions in climate models, and it 
will help differentiate between those processes that can be better or 
newly parameterized versus those that cannot. Such techniques are 
already being explored by several research groups including the 
National Center for Atmospheric Research. With a nested or adaptive 
resolution approach the computational capability required could be 
reduced by an order of magnitude or more, and could make the goal of 
computing with such ultra-high resolution models more feasible. A final 
example of algorithm opportunities is the need to better characterize 
the uncertainty in simulation results. Ensembles and basic statistics 
are currently used to assess uncertainty due to natural internal 
variability intrinsic to the climate system. More formal methods for 
verification, validation and uncertainty quantification are needed from 
the computer science, mathematics and statistical science communities. 
A particular challenge is the sparse nature of observational data 
necessary to validate models.
    The Nation's climate modeling enterprise is likely to be 
increasingly driven by the need to obtain scientific results for a 
large and diverse group of users, including government officials, in a 
timely fashion. In such an environment, the development of innovative 
models, algorithms, and software must be managed as a project, as 
opposed to an open-ended research program. Some aspects of such an 
approach are well-understood, such as the need for planning, schedule 
visibility, and milestones. A more difficult problem is the potential 
dependence of success on delivering high-risk products in models, 
algorithms, and software on a particular schedule. Many of these 
products, such as new approximation methods, or new programming models, 
represent non-incremental departures from the current methods used in 
production climate models, but may be necessary to achieve National 
goals. Risk management in such a setting requires careful planning and 
a close and continuing collaboration between the climate, facilities, 
applied mathematics, and computer science communities. In addition to 
the research management needs, there will also be a need to ensure that 
end-users are sufficiently involved in the prioritization of research 
efforts, and that the resources and institutions exist to transfer the 
large volumes of information into the decisionmaking processes of 
various private and governmental users.
    In conclusion, there is no single pacing item to the advancement of 
climate change science, but a collection of interrelated science and 
technology challenges. Many of the issues discussed in this testimony 
speak to the need for a balanced investment portfolio in computational 
infrastructure, climate science, computer science, and applied 
mathematics. In the short and long term, computational capability 
remains a significant bottleneck and should remain a high priority 
investment. But as the science and complexity of climate simulation 
grows, so will new technical and scientific challenges. Immediate 
proactive investments in climate science, software, algorithms, data 
management, and other pacing items are needed for accelerated progress 
that can keep pace with the rapidly evolving computational environment. 
The management of these investments is also critical to success. 
Strategic management of such a broad multidisciplinary activity will 
likely prove to be the most effective way to ensure that new 
investments have the desired impact on accelerating progress.
                                 ______
                                 
                     The Climate Prediction Project
Global Climate Information for Regional Adaptation and Decision-Making 
        in the 21st Century

Challenge
    The world recognizes that the threat of global climate change is 
one of the most important problems facing humanity. To cope with the 
consequences of climate change, the peoples, governments, and economies 
of the world must develop mitigation and adaptation strategies, which 
will require investments of trillions of dollars. The development of 
science-based adaptation and mitigation strategies will only be 
possible through a revolution in regional climate predictions.

The Summit
    The World Modeling Summit for Climate Prediction was organized to 
develop a strategy to revolutionize prediction of the climate through 
the 21st century to help address the threat of global climate change.

Summit Declaration
    1. Improved prediction of the changes in the statistics of regional 
climate, especially of extreme events and high-impact weather, are 
required to assess the impacts of climate change and variations, and to 
develop adaptive strategies to ameliorate their effects on water 
resources, food security, energy, transport, coastal integrity, 
environment and health.
    2. Our current inadequacy in the provision of robust estimates of 
the risk to society, particularly from possible catastrophic changes in 
regional climate, is strongly influenced by limitations in computer 
power and the size of the scientific workforce.
    3. Climate prediction is among the most computationally demanding 
problems in science. It is both necessary and possible to revolutionize 
climate prediction: necessary because of the grand challenge posed by 
the changing climate, and possible building on the past accomplishments 
of prediction of weather and climate. However, the scientific expertise 
and the computing capability is not available in any single nation, and 
a comprehensive international effort is essential. Investing today in 
climate science will lead to significantly reduced costs of coping with 
climate change tomorrow.
    4. A Climate Prediction Project coordinated by WCRP, in 
collaboration with WWRP and the IGBP and involving the national weather 
and climate centers should be initiated to provide global climate 
information for regional adaptation and decision-making in the 21st 
century.
    5. As a part of the Climate Prediction Project, and in addition to 
enhancing the capacity of the world's existing national climate 
research centers, a World Climate Research Facility (WCRF) for climate 
prediction should be established that will enable the national centers 
to accelerate progress in improving operational climate prediction at 
decadal to multi- decadal lead times, enhancing understanding of the 
climate system, building global capacity, developing a trained 
scientific workforce, and engaging the global user community. The WCRF 
will argue for sustained, long-term, global observations that are 
needed to initialize, constrain and verify the models. An important 
component of the WCRF will be an archive of observations and model data 
with appropriate user interface and knowledge-discovery tools for 
diagnostic tests.
    6. The central component of the WCRF will be one or more dedicated 
high-end computing facilities that will enable the revolution in 
climate prediction by supporting the model resolution and complexity 
required for the most advanced and reliable representations of the 
climate system that technology and our scientific understanding of the 
problem can deliver. This computing capability, with systems at least 
10,000 times more powerful than the currently available computers, is 
vital for regional climate predictions to underpin mitigation policies 
and local and regional adaptation needs with robust estimates of risk. 
The computing capability will help advance our understanding of the 
processes responsible for climate variability and predictability, and 
provide a quantum leap in the exploration of the limits in our ability 
to reliably predict climate with a level of detail and complexity that 
is not possible at the national centers. It will also make it possible 
to bring to bear the latest and most innovative computer technology on 
the climate change problem, and provide a common modeling framework 
through an international computing laboratory and make it possible to 
conduct specialized experiments to advise decision-making in 
adaptation, mitigation. This project will permit scientists to strive 
toward kilometer-scale modeling of the global climate system, which 
will particularly benefit the simulation and prediction of tropical 
climate, helping many of the world's developing countries that are 
especially vulnerable to climate change.
    7. The WCRF will make it possible for the first time to deliver 
climate predictions with a reliable estimate of their uncertainty. To 
estimate the quality of a climate prediction requires an assessment of 
how accurately we know the current phase of natural climate 
variability, on which anthropogenic climate change is superimposed. But 
also the WCRF will enable the climate research community to assess how 
model uncertainties limit the skill of climate predictions. All 
elements of estimating the uncertainty in climate predictions pose an 
extreme burden on computing resources but also on the availability of 
observational data.
    8. The methodology of initializing weather and short-term climate 
prediction models with observations must be seamlessly extended to 
predictions of decadal variations and climate change. The understanding 
and representation of physical and biogeochemical processes and 
feedbacks must be improved to make reliable centennial projections.
    9. It may be possible that the WCRF will be funded in different 
ways, e.g., through public-private partnerships with corporate and 
foundation resources and through governmental treaties and agreements.
    The Climate Prediction Project has the potential to help humanity 
cope with the consequences of climate change.
    The lasting legacy of the Project will be to help the citizens of 
the world in the 21st century.

    Senator Kerry. Thanks very much, Dr. Hack.
    Dr. Reed?

  STATEMENT OF DR. DANIEL A. REED, CHAIR, BOARD OF DIRECTORS, 
              COMPUTING RESEARCH ASSOCIATION (CRA)

    Dr. Reed. Good afternoon, Mr. Chairman, and Mr. Vice 
Chairman. I am Daniel Reed, Chair of the Board of Directors of 
the Computing Research Association and a high performance 
computing researcher.
    Today I would like to make four points regarding the status 
and future of high performance computing for climate change 
modeling.
    It is clear we now face life and death questions, the 
potential effects of human activities and natural processes on 
our climate and their regional impacts. I believe high 
performance computing and computational science are among our 
best options to gain that understanding. HPC systems now bring 
detailed computational climate models to life. However, a 
recent Department of Energy study estimated that climate 
modeling could effectively use an exascale HPC system 
effectively. That is a computer 1,000 times faster than today's 
most powerful systems, and one nearly a billion times faster 
than today's PC's.
    Why are these climate models so complex? First, one must 
simulate many years to validate the models against 
observational data. Second, to understand possible 
environmental changes, one must model sensitivity to many 
conditions, including carbon dioxide emissions. Third, to 
understand the interplay of biogeochemical processes with 
public policies, one must evaluate model ensembles. And 
finally, one must study detailed regional effects such as 
hurricanes and storm surge, not just global ones. And I would 
add parenthetically that when I was at North Carolina, I spent 
a great deal of time working on precisely those issues, looking 
at the regional effects of storm surge and hurricanes.
    This leads to my second point, HPC availability for climate 
studies. In the 1980s, the importance of computing to science 
and the dearth of HPC facilities for research stimulated 
creation of the National Science Foundation and the Department 
of Energy's Office of Science Supercomputing Centers. They now 
provide much of the U.S. scientific HPC resources, including 
for climate change. Without question, our HPC infrastructure is 
enormously greater than 20 years ago, but so too are our 
expectations and our needs. Equally tellingly, most HPC 
resources are shared across many scientific disciplines, and 
only a portion of them support climate change studies.
    This brings me to my third point, high performance 
computing technology trends. Until the mid-1980s, high 
performance computing was defined by custom designed vector 
processors, those designed by the legendary Seymour Cray. The 
ubiquitous PC changed that, creating a new high performance 
computing model, one based on large clusters of PCs. By 
analogy, this was a shift from a single bulldozer to 1,000 
shovels. However, our 20-year free ride of increasing 
microprocessor performance, which is to say bigger shovels, has 
ended, and a second transition, multiple processors per chip, 
lots of small shovels, is underway. This multicore revolution 
will be even more disruptive, profoundly affecting the 
computing industry and, more pointedly, climate researchers. 
Simply put, we are now suffering the delayed consequences of 
limited Federal research investment in this domain.
    Moreover, the scientific data deluge from new instruments 
threatens to overwhelm our research institutions and the 
ability of climate researchers to integrate data with 
multidisciplinary models.
    This leads to my last point, climate high performance 
computing research and development and procurement models. We 
must explore new HPC hardware designs that better support 
scientific and national defense applications, recognizing that 
the design cost for these systems are rarely repaid by 
commercial sales. Thus, we must rethink our models for high 
performance computing research and development and procurement. 
Simply put, a million rowboats is no substitute for an aircraft 
carrier.
    We also need new programming models that simplify 
application development for multicore processors. Today climate 
modeling teams spend inordinate amounts of time tailoring 
software to HPC systems, time better spent on climate research. 
Climate analysis also requires diverse investments, as Dr. Hack 
mentioned. HPC facilities must be balanced with investments in 
software, storage, algorithms, and tools.
    In 2005, I chaired the President's IT Advisory Committee 
Report on Computational Science and in 2007 co-chaired the 
President's Council of Advisors on Science and Technology, 
PCAST, review of computing research. Both of those reports 
recommended an interagency strategic road map for research 
computing and high performance computing infrastructure.
    In summary, our challenges are to sustain both the research 
and the deployment of HPC systems needed to ensure our planet's 
health.
    Thank you very much for your time and attention. I look 
forward to questions.
    [The prepared statement of Dr. Reed follows:]

   Prepared Statement of Daniel A. Reed, Chair, Board of Directors, 
                  Computing Research Association (CRA)

    Good afternoon, Mr. Chairman and Members of the Committee. Thank 
you for granting me this opportunity to comment on current U.S. 
computational capabilities and the research and infrastructure needs to 
support climate modeling. I am Daniel Reed, Chair of the Board of 
Directors for the Computing Research Association (CRA). I am a 
researcher in high-performance computing; a member of the President's 
Council of Advisors on Science and Technology (PCAST); the former 
Director of the National Center for Supercomputing Applications (NCSA), 
one of NSF's high-performance computing centers; and Director of 
Scalable and Multicore Computing Strategy at Microsoft.
    I would like to make five points today regarding the status and 
future of high-performance computing (HPC) for climate change modeling, 
beginning with the relationship between HPC and climate change models.

1. High-end Computational Science: Enabling Climate Change Studies
    We know the Earth's climate has changed during the planet's 
history, due to the complex interplay of the oceans, land masses and 
atmosphere, the solar flux and the biosphere. Recently, the U.S. 
Climate Change Science Program and the Intergovernmental Panel on 
Climate Change (IPCC) \1\ concluded that climate change will accelerate 
rapidly during the 21st century unless there are dramatic reductions in 
greenhouse emissions. We now face true life and death questions--the 
potential effects of human activities and natural processes on our 
planet's ecosystem. I believe HPC tools and technologies provide one of 
our best options for gaining that understanding.
---------------------------------------------------------------------------
    \1\ R. Alley et al, Climate Change 2007: The Physical Science 
Basis, IPCC, Working Group 1 for the Fourth Assessment, WMO.
---------------------------------------------------------------------------
    In 2005, I was privileged to chair the computational science 
subcommittee of the President's Information Technology Advisory 
Committee (PITAC), which examined the competitive position of the U.S. 
in computing-enabled science. In our report, Computational Science: 
Ensuring America's Competitiveness,\2\ we noted that
---------------------------------------------------------------------------
    \2\ Computational Science: Ensuring America's Competitiveness 
President's Information Technology Advisory Committee (PITAC), June 
2005, http://www.nitrd.gov/pitac/reports/2005
0609_computational/computational.pdf.

        Computational science is now indispensable to the solution of 
        complex problems in every sector, from traditional science and 
        engineering domains to such key areas as national security, 
        homeland security, and public health. Advances in computing and 
        connectivity make it possible to develop computational models 
        and capture and analyze unprecedented amounts of experimental 
        and observational data to address problems previously deemed 
---------------------------------------------------------------------------
        intractable or beyond imagination.

    Computational science now constitutes the third pillar of the 
scientific enterprise, a peer alongside theory and physical 
experimentation. This is especially important in a field such as 
climate change studies, where the models are complex--multidisciplinary 
and multivariate--and one cannot conduct parametric experiments at 
planetary scale.
    Why then, is HPC especially critical to climate change studies? 
First, one must simulate hundreds to thousands of Earth years to 
validate models and to assess long-term consequences. This is practical 
only if one can simulate a year of climate in at most a few hours of 
elapsed time. Each of these simulations must be of sufficient fidelity 
(i.e., temporal and spatial resolution) to capture salient features. 
Today, for example, most climate models that are run for several 
hundred to several thousand simulated years do not explicitly resolve 
important regional features like hurricanes. These are large-scale, 
capability computing problems (i.e., ones requiring the most powerful 
computing systems).
    Second, to understand the effects of environmental changes and to 
validate climate models, one must conduct parameter studies (e.g., to 
assess sensitivity to different conditions such as the rate of CO2 
emissions or changes in the planet's albedo). Each of these studies 
involves hundreds to thousands of individual simulations. This is only 
practical if each simulation in the ensemble takes a modest amount of 
time. These are large-scale, capacity computing problems (i.e., ones 
requiring ongoing access to multiple, large-scale computing systems).
    Third, understanding the sensitivity of physical and biogeochemical 
processes to social, behavioral and economic policies requires 
evaluation of statistical ensembles and many model variants. These are 
hypothesis-driven computational scenarios that are only possible after 
the physical and biogeochemical processes are understood, requiring 
additional capacity and capability computing.
    This is a daunting problem--developing, validating and evaluating 
multidisciplinary climate models in time to provide the necessary 
answers to critical questions:

   How many simulation scenarios are necessary (minimally and 
        optimally).

   What model elements are needed for each scenario?

   What temporal and spatial resolution, along with physical 
        models, is affordable?

   What are the errors and uncertainties in model predictions?

   When must research end and production simulation begin to 
        produce policy guidance?

    Underlying these questions is the need for powerful computers to 
model climate change at regional and fine scales, and to support the 
sophisticated and computationally expensive algorithms needed to 
represent the complexities of both natural and human effects. We must 
also manage the tsunami of observational data now being captured via a 
new generation of environmental sensors, integrating high-resolution 
Earth system models with assimilated satellite and other data, 
supported by large data archives and intelligent data mining and 
management systems.
    Finally, we must develop the multiphysics algorithms and models 
needed to represent the complex interactions of biological, 
geophysical, chemical and human activities. New scientific and 
mathematical advances will also be required to quantify model 
uncertainty for such complex systems. This fusion of sensor data with 
complex models is large-scale computational science in its clearest and 
most compelling form. Equally importantly, those HPC systems must be 
available for researcher use.

2. High-Performance Computing Resource Availability
    In the early 1980s, HPC facilities were accessible only by a 
handful of U.S. researchers. Most access required both a national 
security clearance and partnership with one of the U.S. weapons 
laboratories or international travel--for access to computing research 
facilities outside the U.S. The rising importance of computing to 
science and the dearth of HPC facilities for open scientific research 
stimulated creation of the National Science Foundation (NSF) 
supercomputing centers and similar facility investments by the 
Department of Energy's (DOE) Office of Science. Although other agencies 
also support HPC facilities, NSF and DOE now provide the overwhelming 
fraction of the unclassified resources for computational science, 
including climate change.
    This NSF program and its descendents, the Partnerships for Advanced 
Computational Infrastructure (PACI) and the TeraGrid, continue to 
support academic researchers via consulting, HPC systems and archival 
storage. All of the NSF-supported resources, with the exception of the 
majority at the National Center for Atmospheric Research (NCAR), are 
allocated by peer review across all disciplines. The computing 
facilities at NCAR include peer-reviewed resources allocated for 
weather and climate research and the Climate Simulation laboratory 
(CSL) resources dedicated to climate change research. Historically, all 
NSF computing resources have been substantially over-subscribed, with 
unmet demand from academic researchers. Recently, however, NSF has 
funded a series of competitive hardware acquisitions to help address 
this shortfall, with the largest slated to sustain one petaflop \3\ on 
selected applications. \4\
---------------------------------------------------------------------------
    \3\ One teraflop is 1012 floating point operations/
second; one petaflop is one thousand teraflops, or 1015 
floating point operations/second; one exaflop is one thousand 
petaflops, or 1018 floating point operations/second.
    \4\ See the NSF Office of Cyberinfrastructure, http://www.nsf.gov/
dir/index.jsp?org=OCI for details on the NSF acquisition program.
---------------------------------------------------------------------------
    The DOE Office of Science also maintains a set of unclassified 
computing facilities, anchored by the National Energy Research 
Scientific Computing Center (NERSC), two leadership-class computing 
systems at Oak Ridge and Argonne National Laboratories, and a smaller 
facility at the Pacific Northwest Research Laboratory. The majority of 
DOE's NERSC resources are also allocated by peer review, with the 
requirement that the proposed use be relevant to the DOE Office of 
Science mission. Finally, the DOE leadership-class facilities target 
focused projects that could benefit from access to the largest-scale 
facilities in the country, including the climate change modeling 
program. Most of these resources are allocated by the INCITE 
initiative.\5\
---------------------------------------------------------------------------
    \5\ Department of Energy Innovative and Novel Computational Impact 
on Theory and Experiment (INCITE) initiative, http://
hpc.science.doe.gov/
---------------------------------------------------------------------------
    Our computational science infrastructure is enormously greater than 
twenty years ago. However, so are our expectations and needs--science 
and computing are now synonymous. Equally tellingly, because almost all 
of our NSF and DOE HPC resources are shared across disciplines, only a 
modest fraction of these systems is dedicated to climate change 
studies. Rather, researchers rely on a combination of proposal peer 
review and programmatic resource allocation to conduct climate change 
studies on a diverse array of HPC systems.
    At present, there is no truly large scale U.S. climate change 
computing research facility, architected, configured and dedicated to 
multidisciplinary climate change studies that can deliver timely and 
accurate predictions. A recent DOE study estimated that climate and 
environmental modeling could use an exascale system effectively (i.e., 
one thousand times faster than any extant computer system). Simply put, 
change modeling is a deep and challenging scientific problem that 
requires computing infrastructure at the largest scale.

3. Computing Evolution: Lessons and Challenges
    In the late 1970s and the 1980s, HPC was defined by vector 
processors, as exemplified by the eponymously named systems designed by 
the legendary Seymour Cray. These systems combined high-speed, custom 
processor design with fast memories and innovative packaging. 
Researchers and software developers were able to tune selected portions 
of their codes to the vector hardware, achieving unprecedented 
performance with modest effort.
    With the birth of the PC, a new approach to HPC began to emerge in 
the 1980s. The increasing performance and low cost of commodity 
microprocessors--the ``Attack of the Killer Micros''--transformed HPC. 
This new model of massive parallelism partitions computations across 
large numbers of processors. Via this approach, one can increase peak 
hardware performance to levels limited only by economics and 
reliability. However, achieving high performance on complex 
applications is more problematic and challenging, particularly for 
multidisciplinary applications. The climate change community expressed 
great concern about this disruptive technology transition during the 
1990s, with concomitant political controversy.
    Recognizing this technological shift, the associated challenges and 
the opportunities, the Defense Advanced Research Projects Agency 
(DARPA) launched an aggressive research and development program that 
engaged academia, industry and national laboratories. Other Federal 
agencies, notably the National Science Foundation (NSF), the Department 
of Energy's (DOE) Office of Science and the National Aeronautics and 
Space Administration (NASA), joined in the High-Performance Computing 
and Communications (HPCC) program.\6\
---------------------------------------------------------------------------
    \6\ The High-Performance Computing and Communications (HPCC) 
program became the Networking and IT Research and Development (NITRD) 
program, http://www.nitrd.gov/about/about_NITRD.html.
---------------------------------------------------------------------------
    In the 1990s, research flourished in computer architecture, system 
software, programming models, algorithms and applications. Computer 
vendors launched new initiatives, and parallel computing startup 
companies were born. Planning began for petascale systems, based on 
integrated hardware, architecture, software and algorithms research. 
After a promising start, much of the initiative faded and attention 
shifted elsewhere. The most notable exception was DOE's National 
Nuclear Security Administration (NNSA). Needing to certify the weapons 
stockpile without testing, NNSA embraced HPC to verify and validate 
weapon safety and readiness. The complex physics drove new algorithm 
and software development and acquisition of some of the world's most 
power computing systems, all based on massive parallelism and commodity 
microprocessors.
    While the U.S. computing industry largely abandoned purpose-built 
supercomputers in favor of commodity designs, Japanese vendors, notably 
Hitachi and Fujitsu, continued to develop and evolve vector 
supercomputers. In 2002, Japan announced the Earth Simulator--then the 
world's fastest computer. The Earth Simulator was designed specifically 
for large-scale climate and weather studies and drew on many years of 
vector computing research and development.
    Although the Japanese plan had long been public, it precipitated 
considerable concern. The interagency High-End Computing Revitalization 
Task Force (HECRTF) was chartered to assess the competitive position of 
the United States. I was privileged to chair the 2003 HECRTF community 
workshop and edited the associated community report.\7\ The Federal 
agencies produced a complementary report and a proposed action plan. 
Several agencies launched new programs, of which the largest and most 
visible were the NSF OCI petascale initiative and the DOE Office of 
Science's Scientific Discovery through Scientific Computing (SciDAC) 
\8\ and INCITE programs.
---------------------------------------------------------------------------
    \7\ The documents for the High-End Computing Revitalization Task 
Force (HECRTF), including the community workshop report, can be found 
at http://www.nitrd.gov/subcommittee/hec/hecrtf-outreach.
    \8\ Department of Energy, Scientific Discovery through Scientific 
Computing (SciDAC), http://www.scidac.gov/.
---------------------------------------------------------------------------
    Today, the majority of the world's largest HPC systems, dominated 
by U.S. laboratory and academic holdings, remain based on commodity 
building blocks and community-developed software. In this high-
performance ``monoculture,'' vendor profit margins are small, and 
competition for sales is intense, with limited vendor opportunity to 
recover research and development investments in alternative 
architectures. Equally worrisome, the pool of academic researchers in 
HPC and computational science is small, and research funding is 
limited.
    Without doubt, the explosive growth of scientific computing based 
on clusters of commodity microprocessors has reshaped the HPC market. 
The U.S. remains the undisputed world leader in this space. Petascale 
systems are being deployed by NSF and DOE for academic and laboratory 
research, and feasibility assessments of exascale systems \9\ are 
underway. Although this democratization of HPC has had many salutatory 
effects, including broad access to commodity clusters across 
laboratories and universities, it is not without its negatives.
---------------------------------------------------------------------------
    \9\ Modeling and Simulation at the Exascale for Energy and the 
Environment, Summer 2007, http://www.sc.doe.gov/ascr/ProgramDocuments/
TownHall.pdf.
---------------------------------------------------------------------------
    Not all aspects of climate change models map efficiently to the 
cluster programming model of loosely coupled, message-based 
communication. It is also unclear if we have the resources needed to 
address the climate change problem at appropriate scale and in a timely 
manner, particularly given dramatic changes now underway in computing 
technology.

4. The Brave New World: Multicore and Massive Data
    Over the past twenty years, computational science and HPC have 
exploited the ever-increasing performance of commodity microprocessors. 
Each new processor generation combined greater transistor density, new 
architectural techniques and higher chip power to deliver greater 
single processor performance. This tripartite evolution is now over. 
Although transistor densities on chip will continue to rise, physics 
and power constraints make it impractical to increase clock frequencies 
further. Future chip performance increases will depend on explicit 
parallelism and architectural innovations. No longer will current 
software execute faster in the future without change. Parallelism is 
now required, even at the chip level, to deliver greater performance.
    This multicore revolution--the placement of multiple, slower 
processors on each chip--poses major new challenges for the computing 
industry. It is just as disruptive as the transition from vector to 
parallel computing was fifteen years ago. Today's quad-core chips will 
soon be replaced by chips containing tens, then hundreds and perhaps 
thousands of cores (processors). The technical challenges are daunting, 
and we have no straightforward technical solutions that will hide this 
complexity from software developers.\10\ This will profoundly affect 
the software industry and scientific researchers.
---------------------------------------------------------------------------
    \10\ This realization recently motivated Microsoft and Intel to 
invest $20M in academic multicore research at the University of 
Illinois at Urbana-Champaign and the University of California at 
Berkeley, http://www.microsoft.com/presspass/press/2008/mar08/03-18UPC
RCPR.mspx.
---------------------------------------------------------------------------
    For multicore chips, new programming models and tools are needed to 
develop parallel applications, and existing software must be 
retrofitted. New chip architectures are needed to exploit rising 
transistor densities, support parallel execution and enable 
heterogeneous processing. New memory technologies and interconnects are 
needed to support chips with tens to hundreds of cores. Equally 
importantly, new algorithms are needed that map efficiently to these 
new architectures. All of these changes will affect parallel climate 
models now being developed and executed on clustered commodity systems. 
Today, we are suffering some of the delayed consequences of limited 
research investment in parallel computing--architecture, system 
software, programming tools, data management and algorithms.
    In addition to dramatic changes in processors and computation, our 
models of data capture and management are in flux. We can now generate, 
transmit, and store data at rates and scales unprecedented in human 
history. Many of our new environmental instruments can routinely 
produce many tens to hundreds of petabytes of data annually. The 
scientific data deluge threatens to overwhelm the capacity of our 
Federal institutions to manage, preserve and process and of our climate 
modeling researchers to access and integrate the data with 
multidisciplinary models. This data integration is critical to climate 
model validation.
    Although industry is developing massive data centers to host 
Internet search, social networks and software as a service, our 
research data infrastructure has not kept pace. Climate researchers 
need better data management tools, including provenance tracking, 
translation, mining, fusion, visualization, and analysis. We must not 
focus exclusively on computing, but on the fusion of sensors and data 
management with computing hardware and rich climate models.

5. Actions: A Sustainable, Integrated Approach
    One can and must draw several important, salutary lessons from the 
changing nature of computing technology. The U.S. HPC industry is now 
driven by business and consumer technology economics, with concomitant 
advantages and disadvantages. Large product volumes and amortized 
research and development costs lead to rapid innovation and 
technological change. However, those same consumer economics mean that 
today's HPC systems are built from commodity hardware and software 
components, and they are often ill-suited to the numerically and 
communication intensive nature of climate change models. In 
consequence, they rarely deliver a large fraction of their advertized 
peak performance.
    Given their unique attributes, the highest capability computing 
systems have a very limited commercial market. The high non-recurring 
engineering costs to design HPC systems matched to scientific and 
government needs are not repaid by sales in the commercial marketplace. 
Hence, we must rethink our models for research, development, 
procurement and operation of high-end systems. We must target 
exploration of new systems that better support the needs of scientific 
and national defense applications and sustain the Federal investment 
needed to design, develop and procure those systems. Today's approach 
is unlikely to provide the necessary resources to address the climate 
change model problem fully.
    New programming models and tools are also needed that simplify 
application development and maintenance and that target emerging 
multicore processors. Today, almost all parallel scientific 
applications are developed using low-level message-passing libraries. 
Climate modeling teams must have deep knowledge of application software 
behavior and its interaction with the underlying computing hardware, 
and they often spend inordinate amounts of time tailoring algorithms 
and software to hardware and software idiosyncrasies, time more 
profitably spent on science and engineering research.
    Climate change analysis requires large-scale data archives, 
connections to scientific instruments and collaboration infrastructure 
to couple distributed scientific groups. Any investment in HPC 
facilities must be balanced with appropriate investments in hardware, 
software, storage, algorithms and collaboration environments. Simply 
put, climate change modeling, as with all scientific discovery, 
requires a judicious match of computer architecture, system software, 
algorithms and software development tools.
    These facts illustrate the importance of a long-term, integrated 
research and development program that considers the entire 
computational science ecosystem, something I advocated as chair and co-
chair of two recent PITAC and PCAST subcommittees, respectively. Both 
the 2005 President's IT Advisory Committee (PITAC) report on 
computational science and the 2007 President's Council of Advisors on 
Science and Technology (PCAST) review of the Networking and Information 
Technology Research and Development (NITRD) program recommended 
creation of an interagency strategic roadmap for computational science 
and computing research. In particular, the 2005 PITAC report found that

        The continued health of this dynamic computational science 
        ``ecosystem'' demands long-term planning, participation, and 
        collaboration by Federal R&D agencies and computational 
        scientists in academia and industry. Instead, today's Federal 
        investments remain short-term in scope, with limited strategic 
        planning and a paucity of cooperation across disciplines and 
        agencies.

    The report also recommended creation of a long-term, interagency 
roadmap to

        . . . address not only computing system hardware, networking, 
        software, data acquisition and storage, and visualization, but 
        also science, engineering, and humanities algorithms and 
        applications. The roadmap must identify and prioritize the 
        difficult technical problems and establish a timeline and 
        milestones for successfully addressing them.

    In that same spirit, the 2007 PCAST review of the NITRD program, 
Leadership Under Challenge: Information Technology R&D in a Competitive 
World,\11\ which I co-chaired, reiterated the need for a strategic plan 
and roadmap for high-performance computing and noted that
---------------------------------------------------------------------------
    \11\ Leadership Under Challenge: Information Technology R&D in a 
Competitive World, President's Council of Advisors on Science and 
Technology (PCAST), August 2007, http://www.ostp.gov/pdf/
nitrd_review.pdf.

        The Federal NIT R&D portfolio is currently imbalanced in favor 
        of low-risk projects; too many are small-scale and short-term 
        efforts. The number of large-scale, multidisciplinary 
        activities with long time horizons is limited and visionary 
---------------------------------------------------------------------------
        projects are few.

    Based on these studies, I believe we face both great opportunities 
and great challenges in high-end computing for climate change. 
Computational science truly is the ``third pillar'' of the scientific 
process. The challenges are for us to sustain the research, development 
and deployment of the high-end computing infrastructure needed to 
enable discoveries and to ensure the health of our planet.
    In conclusion, Mr. Chairman, let me thank you for this Committee's 
interest in this question and its continue support for scientific 
innovation. Thank you very much for your time and attention. I would be 
pleased to answer any questions you might have.

    Senator Kerry. Thank you very much, Dr. Reed.
    Dr. Sarachik?

STATEMENT OF EDWARD SARACHIK, EMERITUS PROFESSOR OF ATMOSPHERIC 
    SCIENCE, ADJUNCT PROFESSOR OF OCEANOGRAPHY, AND ADJUNCT 
     PROFESSOR OF APPLIED MATHEMATICS AT THE UNIVERSITY OF 
                          WASHINGTON 
          AND CO-DIRECTOR, CENTER FOR SCIENCE IN THE 
                          EARTH SYSTEM

    Dr. Sarachik. Thank you, Senator Kerry and Senator Stevens. 
My name is Ed Sarachik. I thought I retired 6 months ago, but I 
seem to have not. I hold various appointments on the faculty at 
the University of Washington and I am Co-Director, along with 
Ed Miles, of the Center for Science in the Earth System at the 
University of Washington. It is a very interesting center 
because it goes from climate information to climate impacts, to 
dealing with stakeholders, and to raising the consciousness of 
stakeholders, both in the public and private domain.
    Basically I can say that although each region of the 
Pacific Northwest--and there are many climates within the 
Pacific Northwest--has unique problems. All of them need a 
skillful prediction of next season's climate--that is 
precipitation and temperature--and a knowledge of the future 
variability of climate. It is not just how the mean temperature 
is going to change. We do not respond to the mean. We respond 
to variability. You can imagine building for 70 degree 
temperatures and it would matter if the day is 110 and the 
evening is 30 or if it is going to be 70 degrees all the time. 
We respond to variability not to the mean.
    So the question is can we do these two problems. Can we 
make skillful predictions a season in advance, and can we 
figure out what the future variability of climate is going to 
be as the climate changes?
    And here the answer is yes and no. We can make predictions 
a season in advance. The reason it has been so cold this winter 
is because it has been a La Nina year, and that was predicted 
about 6 months ago. But we cannot do the variability correctly. 
Despite the fact that we are spending a fair amount of money 
building these models for the IPCC, the IPCC cannot do regional 
climate. It can only do climate on continental scales. 
Continental scales are not the scale at which applications are 
made.
    So what do we have to do? The basic reason we cannot do the 
variability correctly is that the known modes of variability, 
El Nino, Pacific Decadal Oscillation, and the North Atlantic 
oscillation, are simply not done correctly and in the right 
place by these models. In order to get them to do the right 
thing in the right place by these models, there certainly are 
modeling issues. As has been mentioned so far, resolution is 
one of those modeling issues, and for that resolution we need 
bigger computers, to be sure. But we also need access to these 
computers. At this moment, none of these big climate models are 
being run at universities because universities simply do not 
have the wherewithal to do the running of it. So access to the 
various places that would have interest in improving the 
variability of these models is absolutely crucial.
    The second leg of the stool, as I believe has been 
mentioned already, is observations. We do not have a climate 
observing system. If we do not have a climate observing system, 
we cannot know what the climate is in all of its specificity 
around the globe. In particular, we make observations, but 
these observations are not necessarily connected dynamically.
    And there is a certain amount of research that absolutely 
needs to be done on El Nino southern oscillation, on the 
Pacific Decadal Oscillation, on the North Atlantic Oscillation, 
and the effects of CO2 and various other 
constituents on the atmosphere.
    A lot of these things--modeling observations and research 
needs to be done in an integrated manner. If we do not have the 
observations, we cannot really do the modeling. If we do not 
have the research, we cannot really do the modeling. If we do 
not have the modeling, we cannot really integrate the 
observations. These things really do need integration and some 
method of putting them all together and going ahead in a 
consistent manner.
    There has been a lot of talk about a national climate 
service. We have a National Weather Service. The National 
Weather Service takes weather observations, integrates them, 
and puts out maps twice or four times a day. We have nothing 
similar for climate, and having a national climate service 
would go a long way toward solving some of the problems 
necessary for doing good regional information.
    Thank you.
    [The prepared statement of Dr. Sarachik follows:]

     Prepared Statement of Edward Sarachik, Emeritus Professor of 
  Atmospheric Science, Adjunct Professor of Oceanography, and Adjunct 
 Professor of Applied Mathematics at the University of Washington and 
          Co-Director, Center for Science in the Earth System

    My name is Edward Sarachik and I am Emeritus Professor of 
Atmospheric Science, Adjunct Professor of Oceanography, and Adjunct 
Professor of Applied Mathematics at the University of Washington. I am 
also Co-Director of the Center for Science in the Earth System 
(supported by NOAA) which contains two groups: a Climate Dynamics Group 
and a Climate Impacts Group. The Climate Dynamics Group studies the 
physical climate system relevant to the Pacific Northwest and the 
Climate Impacts Group examines the impacts of climate variability and 
change on the Pacific Northwest, and produces climate information 
products and derived predictions (e.g., streamflow forecasts) for a set 
of local stakeholders. The combined Center studies the general problem 
of making climate information useful to stakeholders in the Pacific 
Northwest. The range of our activities and a list of our stakeholders 
can be seen on our website: http://cses.washington.edu/.
    I have also chaired two National Research Council committees: one 
that produced an National Academy Press report Learning to Predict the 
Climate Variations Characteristic of El Nino and the other, Improving 
the Effectiveness of U.S. Climate Modeling, both highly relevant to 
this hearing. I also chair the advisory group for the International 
Research Institute for Climate and Society at Columbia University which 
deals with the same problem as that of this hearing but in an 
international context.

What do stakeholders want?
    They ask questions they would have asked in the absence of climate 
change: basically, some knowledge about the variability in the near 
future. Some examples from the Pacific Northwest:

   All stakeholders want to know next season's temperature and 
        rainfall.

   Power companies, city water utilities, and ski area 
        operators want to know whether next winter's snowpack will be 
        thick and long lasting or thin and early melting.

   Fishers want to know if next season's coastal mixed layer 
        will be deep or shallow, warm or cold.

   The tourist industry wants to know if next summer will be 
        clear or cloudy.

   Insurance companies and state flood control agencies want to 
        know if there be an unusual number of storms next winter, and 
        the probability that there will be destructive windstorms.

    Then they ask questions about the very long term, say 50 years from 
now:

   Individuals and developers want to know if they should build 
        near the ocean in the presence of rising sea level. Do they 
        need a sea wall?

   Foresters want to know what species of tree should be 
        planted in what climate regime. In particular, what will be the 
        future range of temperature and precipitation?

   Wineries want to know if it will be too warm for specific 
        grape varieties and whether or not irrigation will be needed.

   Everybody wants to know if it will get too warm for salmon 
        survival.

    The progression of climate in a given small region is not what we 
are used to from global warming simulations. For temperature, the 
global average smoothes the record and the year to year variability is 
about half a degree F. Local temperature record has a year to year 
variability about 5 +F. Since the year to year variability in a limited 
region is of order of the 50 year warming trend, constantly dealing 
with next year's climate over a long period of time gives practice 
about dealing with long term climate change since many (but not all) of 
the climate manifestations are similar.
    The problem of producing climate information relevant to 
decisionmakers' needs then becomes

   Skillfully predicting next year's temperature and 
        precipitation in a limited region.

   Accurately simulating future variability of temperature and 
        precipitation in a limited region.

Can existing climate models do this?
    The answer is both yes and no.
    Yes. Next years climate can be predicted using current climate 
conditions, especially in the tropical oceans, as a starting point--
this can only be done two or three seasons in advance. There are a 
number of groups in the world that produce such predictions and there 
exists a ocean observing system in the tropical Pacific that produce 
the current climate conditions. Estimates of the predictable part of 
seasonal temperature variability is about 30 percent for the Pacific 
Northwest and about 40 percent for the extreme Southeast part of the 
U.S. so that even if the prediction systems were perfect, only these 
percentages of future variations can be predicted. This makes 
predictions of next year's climate intrinsically probabilistic.
    No. Existing climate models used for the Intergovernmental Panel on 
Climate Change (IPCC) process are comprehensive global models and are 
designed for mitigation, on large space and time scales. The 
variability known to be important regionally (El Nino, Pacific Decadal 
Oscillation, North Atlantic Oscillation) in the current crop of models 
used in the IPCC has been neglected and is done poorly. The IPCC 
concentrates on global averages and freely admits that the smallest 
region for which the models are useful is the continental scale, about 
3000 mile. On scales smaller than continental scale, the models are not 
useful and downscaling to smaller space scales by higher resolution 
models using the large global models as boundary conditions can not be 
expected to improve the situation. The output of existing models can be 
corrected to agree with past climate conditions and the correction used 
for future climates but there is no agreed upon methods for doing this.
What is the best path to producing useful regional climate information?
    Ideally we want a comprehensive climate model (similar to the ones 
currently used for the IPCC process) but which does the known patterns 
of climate variability (El Nino, Pacific Decadal Oscillation, North 
American Oscillation, etc.) correctly and which is run globally at high 
resolution (20 miles rather than the current 100 miles).
    This requires:

        1. A set of model building institutions well resourced and 
        interacting with the entire public and private research 
        sectors.

        2. Far more capable supercomputers. And, equally important, 
        making these supercomputers and advanced models available to 
        the entire research community.

    Supercomputing is necessary, but it is not, by itself, sufficient. 
Also required is:

        3. A research program to investigate the nature of climate 
        variability (especially decadal variability) and assure the 
        global climate models are capable of doing variability 
        correctly and in the correct locations.

    All research ultimately depends on having good observations--since 
we do not have a climate observing system, all future progress in 
climate research will depend on implementing one. So also required is:

        4. A climate observing system producing regular and systematic 
        climate observations.

    Since the output of the climate observing system will never cover 
every point in the atmosphere, ocean and ice over the entire earth, the 
models themselves can be used for interpolation, just as current 
weather models are used to assimilate weather observations into 
consistent global fields. Therefore the last component required is

        5. A monthly analysis of the climate system using the 
        observations produced by the climate observing systems in 4. 
        and the models developed in 1. and 2.

    Because this hearing assumes it, it is hardly necessary to add:

        6. A distribution network for regional climate and resource 
        information interacting directly with local stakeholders.

    At least a major portion of 4, 5, and 6 could be accomplished by 
the establishment of a National Climate Service.
    It may seem strange that starting with models for simulating local 
climate information we wound up with far more comprehensive 
requirements, but the ability to produce useful regional climate 
information to meet stakeholder needs depends on a healthy climate 
infrastructure. This is precisely the situation in that the ability to 
produce weather information for public and private use would be 
impossible without the weather infrastructure contained within the 
National Weather Service (NWS) and the research that is enabled by the 
observations and analyses emerging from the NWS. The ability to provide 
climate information to address end-user needs depends generally on the 
health of the climate infrastructure and the climate community.

    Senator Kerry. Thank you very much, Doctor. Appreciate it.
    Mr. Carlisle?

      STATEMENT OF BRUCE K. CARLISLE, ASSISTANT DIRECTOR,

          OFFICE OF COASTAL ZONE MANAGEMENT, EXECUTIVE

          OFFICE OF ENERGY, AND ENVIRONMENTAL AFFAIRS,

                 COMMONWEALTH OF MASSACHUSETTS

    Mr. Carlisle. Senator Kerry and Senator Stevens, my name is 
Bruce Carlisle and I am the Assistant Director for the 
Massachusetts Office of Coastal Zone Management. I want to 
thank you for the opportunity to offer testimony on the 
importance of predicting the effects of climate change through 
a national modeling strategy and ensuring that such a strategy 
meets the needs of state coastal managers and local officials.
    Our presence today is also on behalf of the Coastal States 
Organization which represents the interests of the Governors 
from the 35 coastal states, commonwealths, and territories on 
issues relating to sound management of our coasts, Great Lakes, 
and oceans.
    This testimony will cover climate change issues in the 
coastal zone, focusing on the priority modeling and information 
needs of coastal zone managers around the country and 
highlighting the work being done in Massachusetts to build 
effective coastal flood plain management strategies from the 
ground up. Your continuing support for climate change modeling, 
along with the necessary research, monitoring, and computing 
infrastructure, is of critical and growing importance to 
coastal states and communities. One of the points I will 
emphasize is that while a national strategy for climate change 
modeling and assessment is necessary, to be truly effective, it 
must be connected to and coordinated with state, regional, and 
local partners.
    Throughout the Nation, our coastlines and extensive coastal 
floodplains play a significant role in protecting our homes, 
personal safety, providing recreational opportunities for all 
incomes, preserving our natural resources and quality of life, 
and maintaining our viable economies. Coastal counties host 
more than half of the Nation's population, support nearly half 
of the Nation's jobs, and generate more than half of its gross 
domestic product. With accelerated sea level rise, more 
frequent intense storms and shifts in precipitation and 
temperatures, the coastal zone will also feel the brunt of 
global climate change, and these areas will be subject to 
increased flooding, shoreline erosion, saltwater intrusion into 
fresh water aquifers, harmful algal blooms, and the loss of 
coastal habitats.
    For more than 30 years, state coastal managers like those 
at the Massachusetts Office of Coastal Zone Management have 
been leaders in integrating coastal hazard response and 
proactive planning into coastal zone management. As a key 
sector and end-user, we have identified the following 
priorities and urge Congress to provide support in addressing 
these needs.
    The first is high resolution data models and diagnostics to 
generate regional and local sea level rise scenarios. In 
addition, modification of wind speed and storm surge height 
models to assimilate changing storm intensity and frequencies 
and incorporate the unique configurations and characteristics 
of local embayments. Additionally, more information to better 
understand the effects of changing sediment transport, erosion, 
and accretion regimes on habitats and the important ecosystem 
services they provide. Additional modeling on climate change 
impacts to local or regional hydrological processes and the 
rate of saltwater intrusion into coastal aquifers.
    In Massachusetts and many other coastal states, coastal 
land use decisions are being made at the town and municipal 
level by local officials who are working with shrinking budgets 
and resources and often lack technical and scientific 
expertise. Communities are in need of current information and 
predictions, packaged and delivered through specific tailored 
guidance on how to put that information to use.
    To start to address such needs, the Massachusetts Office of 
Coastal Zone Management just launched its new StormSmart Coasts 
program. StormSmart Coasts is designed to give local decision-
makers and ultimately businesses and homeowners the information 
and tools they need to protect themselves from coastal storm 
damage and flooding and to prepare for sea level rise and 
climate change. We deliver StormSmart Coasts tools via an 
extensive website which translates complex technical 
information into user-friendly guidance and planning frameworks 
with links to the best information and data from around the 
Nation. Complicated concepts are reinforced through a series of 
short fact sheets explaining the tools and providing success 
stories.
    One of the basic building blocks of StormSmart Coasts is 
hazard identification mapping. The StormSmart Coasts website 
explains the limitation of current flood maps, which for most 
communities in Massachusetts are more than 20 years old and do 
not include the effects of erosion or sea level rise. 
StormSmart Coasts strongly advises planners to seek and use 
additional sources of data such as storm surge, shoreline 
change, and inundation maps to assess their true vulnerability 
to coastal storm damage.
    There are two pending bills that would assist in developing 
key Federal-state partnerships to support these needs. 
Massachusetts and the Coastal States Organization appreciate 
the work of Senator Kerry and strongly support the climate 
change research and monitoring activities proposed in the 
Global Change Research Improvement Act of 2007. Under the bill, 
particular attention will be focused on regional and state 
vulnerabilities to climate change.
    Massachusetts and the Coastal States Organization also 
support the climate adaptation provisions in America's Climate 
Security Act of 2007, particularly the specific allocation of 5 
percent of the emission allowance account to states which could 
be used for affected coastal communities to adapt to climate 
change. These provisions recognize that coastal states and 
communities are on the front lines of climate change and will 
need Federal support that is proportionate to this risk.
    As you work on a results-oriented national modeling 
strategy, you must specifically answer the kind of questions 
asked by all coastal communities looking to implement effective 
coastal floodplain management. What are the current risks to my 
community, and how will those risks change in the future?
    Thank you again for the opportunity to testify. I would be 
happy to respond to any questions that you may have.
    [The prepared statement of Mr. Carlisle follows:]

Prepared Statement of Bruce K. Carlisle, Assistant Director, Office of 
       Coastal Zone Management, Executive Office of Energy, and 
          Environmental Affairs, Commonwealth of Massachusetts

    Mr. Chairman and Members of the Committee, my name is Bruce 
Carlisle and I am the Assistant Director for the Massachusetts Office 
of Coastal Zone Management. I want to thank you for the opportunity to 
offer testimony on the importance of predicting the effects of climate 
change through a national modeling strategy, and ensuring that such a 
strategy meets the needs of state coastal managers and local officials, 
who will be the ultimate decision-makers and end-users of this 
information. Through my fourteen years of working on coastal policy, 
planning, and management, I am keenly aware of the coastal climate 
change information needs in the Commonwealth.
    My presence today is also on behalf of the Coastal States 
Organization (CSO), which since 1970, has represented the interests of 
the Governors from the 35 coastal States, Commonwealths, and 
Territories on Federal legislative, administrative, and policy issues 
relating to sound coastal, Great Lakes, and ocean management. CSO and 
its members have been actively engaged in this issue, and in November 
of last year, Dr. Braxton Davis, Chair of the CSO Climate Change Work 
Group and Director of the Science and Policy Division at South 
Carolina's Office of Ocean and Coastal Resource Management, gave 
testimony to your committee on the importance of climate change 
research to state and local resource managers.
    This testimony will cover climate change issues in the coastal 
zone, focusing on the priority modeling and information needs as 
conveyed by coastal zone managers around the country and highlighting 
the work being done in Massachusetts to build effective coastal 
floodplain management strategies from the ground up. Your continuing 
support for climate change modeling, along with the necessary research, 
monitoring, and computing infrastructure, is of critical and growing 
importance to coastal states and communities. One of the points I will 
emphasize is that while a national strategy for climate change modeling 
and assessments is necessary, to be truly effective, it must be 
connected to and coordinated with state, regional, and local partners.

Background
    Throughout the Nation, our coastlines and extensive coastal 
floodplains play a significant role in protecting our homes and 
personal safety, providing recreation opportunities for all incomes, 
preserving our natural resources and quality of life, providing 
spawning grounds critical to our fishing industry, and maintaining our 
viable local, regional, and state economies. The coastal zone will also 
feel the brunt of global climate change. More than half of the Nation's 
population lives in coastal counties, and key economic sectors are 
directly linked to the coasts and oceans. Coastal counties host nearly 
half of the Nation's jobs and generate more than half its gross 
domestic product. Through the combined effects of climate change--
accelerated sea level rise, more frequent and intense storms, and 
shifts in precipitation and temperatures--these areas will see 
increased flooding and shoreline erosion, changes in sediment 
transport, saltwater intrusion into groundwater aquifers and coastal 
rivers, increased harmful algal blooms, the loss of coastal wetland and 
coral reef habitats, and changes in population dynamics among marine 
and coastal species. Unless coastal decision-makers and officials start 
to plan for and implement effective measures to ensure coastal 
community resiliency, current and future development and activities--
when poorly sited and/or designed--will aggravate these impacts over 
time.
    For more than 30 years, coastal managers--like those at the 
Massachusetts Office of Coastal Zone Management--have been leaders in 
integrating coastal hazard response and proactive planning into coastal 
zone management. We work in close coordination with both Federal 
agencies and local communities. Our efforts on coastal shoreline and 
floodplain management are extensive and include such actions as: 
developing critical information (e.g., high-resolution shoreline change 
data and coastal high-hazard zone delineation), coordinating the 
state's Rapid Response Storm Damage Survey Team to help spur recovery 
efforts, and providing hands-on technical assistance to communities as 
they review development projects or develop beach management plans.

Think Globally, Act Locally
    Large-scale research, observation, and modeling are critical to 
improving our understanding of, and predictive capabilities for, global 
climate change. The 2003 National Strategic Plan for the U.S. Climate 
Change Science Program explains that while research focused on key and 
emerging climate change science areas is a high priority, directly 
supporting regional resource management efforts is also a critical 
component of the national strategy. The plan points to the development 
of scenarios and comparisons, the implementation and application of 
models, and the advancement of information supporting adaptation 
strategies as means of supporting decision-making at all levels. 
Addressing the limitations of regional- and local-scale analyses of 
potential climate change impacts and improving the availability of such 
diagnostics will greatly enhance their effectiveness in regional and 
local decision-making contexts. As a key ``sector'' and ``end-user,'' 
the CSO has identified the following priority information and products 
to address future impacts of climate change in the coastal zone, and we 
urge Congress to provide support in addressing these needs:

   Localized Sea Level Rise Scenarios--High-resolution coastal 
        topographic and bathymetric elevation data should be coupled 
        with region-specific tide data, sea level rise projections, and 
        other key input parameters to develop basic inundation models 
        for the assessment of lands and resources most vulnerable to 
        accelerated sea level rise. These regional models are an 
        important first step, but coastal states will need more 
        detailed and complex models that incorporate local, embayment-
        scale changes in coastal geomorphology, hydrological 
        conditions, and human alterations and responses (e.g., seawalls 
        and beach nourishment) to more adequately assess social, 
        environmental, and economic vulnerabilities of climate change. 
        Coastal states and communities would benefit from the 
        development of uniform methods for modeling local-scale 
        shoreline changes associated with varying sea level rise 
        projections.

   Storm Surge Models--Existing models that estimate wind 
        speeds and storm surge heights resulting from predicted storm 
        events need to be broadened to incorporate changing storm 
        intensities and frequencies as the result of global climate 
        change. Again, models that incorporate the unique 
        configurations of local embayments or coastline morphologies, 
        water depths, and physical features such as bridges and roads 
        are required to develop accurate storm surge predictions and 
        serve as effective planning tools for decisions being made 
        today about the siting of new development and public 
        infrastructure.

   Impacts on Coastal Habitats and Ecosystem Services--The 
        integrity of many coastal habitats, such as estuarine marshes 
        and beaches, are dependent on adequate sources of sediment 
        supply and the accretion of sediments at certain rates. To 
        predict changes to the these habitats and the important 
        ecosystem services they provide--such as flood protection, 
        wildlife habitat, and recreation--more information is needed to 
        better understand erosion and deposition cycles and to improve 
        our ability to predict the effects of accelerated rates of sea 
        level rise on sediment transport, and accretion and erosion. 
        Without sufficient vertical accretion, estuarine marshes, in 
        particular, are extremely vulnerable to being drowned by 
        accelerated sea level rise.

   Ground Water and Salt Water Intrusion--Climate change will 
        have significant effects on local hydrologic cycles through 
        altered precipitation, evapotranspiration, and soil moisture 
        patterns. These changes will lead to altered groundwater 
        recharge in watershed areas, which will change the groundwater 
        flow to coastal regions and thus the rate of saltwater 
        intrusion in coastal aquifers. Additional modeling on the 
        climate change impacts to local or regional hydrological 
        processes and coastal water resources is also needed to manage 
        coastal water supplies and estuarine biodiversity.

    In Massachusetts and many other coastal states, coastal land use 
decisions are all too often being made at the town and municipal level 
by local officials who are working with shrinking budgets and 
resources, and often lack technical and scientific expertise. 
Communities are in critical need of current information and 
predictions, packaged and delivered through specific, tailored guidance 
on how to put that information to use to make storm resilient 
communities a reality. Because state coastal programs provide high-
quality products, services, and hands-on assistance to these 
constituents, they are uniquely positioned for the implementation of 
coastal climate change adaptation strategies.

StormSmart Coasts
    Created by the Massachusetts Office of Coastal Zone Management, 
StormSmart Coasts is designed to give local decision-makers, and 
ultimately businesses and homeowners, the information and tools they 
need to protect themselves from coastal storm damage and flooding, and 
to prepare for sea level rise and climate change. The strategy for 
initially delivering the StormSmart Coasts tools includes an extensive 
website (www.mass.gov/czm/stormsmart) and a series of regional 
workshops. The website translates complex technical information into 
user-friendly guidance and regulatory models with links to the best 
information and data from around the Nation. Complicated concepts are 
reinforced through a series of short fact sheets explaining the tools 
and providing success stories (see attached examples). The next phase 
of delivery will be to provide targeted technical assistance for 
StormSmart tool implementation to a select handful of coastal 
communities, and then take the lessons learned from these efforts and 
translate and package them for use by other coastal communities within 
Massachusetts and nationwide.

A Partnership at All Levels
    Led by a Coastal Management Fellow provided by the National Oceanic 
and Atmospheric Administration's (NOAA) Coastal Services Center, the 
StormSmart Coasts program is very much a team approach. StormSmart 
Coasts would not have been possible without support and contributions 
from individuals and groups at all levels. The StormSmart Coasts 
program was strongly influenced by guidance and advice from an attorney 
specializing in floodplain and wetlands law, representatives from the 
national Association of State Floodplain Managers, hazard mitigation 
staff from our state Department of Conservation and Recreation, Federal 
Emergency Management Agency (FEMA) personnel, and local officials. 
Recognizing the value of StormSmart Coasts as a national model, the 
Coastal Services Center has selected Massachusetts to receive another 
Coastal Management Fellow starting this summer to implement StormSmart 
Coast strategies in specific Massachusetts coastal communities.

StormSmart Coasts and the Local Connection
    Throughout its development, StormSmart Coasts has benefited from 
extensive input and review from local officials--the key target 
audience for the program. By involving local officials at the earliest 
stages of program development, we have created tools that directly meet 
their needs, and packaged them in a format that they can easily 
understand, access, and successfully implement. Empowering local action 
is critical, because in the end, it is the decisions that are made 
locally that will determine if we can successfully adapt to climate 
change and be resilient to natural hazards so as to avoid such 
tragedies as experienced in the aftermath of Hurricane Katrina.

No Adverse Impact
    The StormSmart Coasts program is based around the concept of No 
Adverse Impact. No Adverse Impact is a set of ``do no harm'' principles 
for local communities to follow when planning, designing, or evaluating 
public and private development activities and storm-damage prevention 
measures. This approach clarifies that community leaders not only have 
the legal right to consider the cumulative impacts of their permitting 
decisions, they have the legal responsibility. No Adverse Impact tools 
and techniques ensure that private development, public infrastructure, 
and planning activities do not have direct or indirect negative 
consequences on the surrounding natural resource areas, private 
property, or other communities.

Applying Model Outputs to Coastal Land Use Decisions
    One of the basic building blocks of StormSmart Coasts is hazard 
identification and mapping. The StormSmart Coasts website explains the 
limitation of the current FEMA Flood Insurance Rate Maps, which are 
engineering estimates of the extent of the floodplain at the time of 
the mapping. For most communities in Massachusetts, those maps are more 
than 20 years old and do not include the effects of erosion or sea 
level rise. StormSmart Coasts strongly advises hazard mitigation 
planners to seek and use additional sources of data, such as storm 
surge, shoreline change, and inundation maps, to assess their true 
vulnerability to coastal storm damage. They need current and specific 
information, synthesized and adapted to suit their requirements to best 
plan for and strategically address coastal floodplain management 
issues, adapt to climate change issues, and reduce impacts for future 
generations.
    The Massachusetts Office of Coastal Zone Management has extensive 
experience packaging technical information for use by local decision-
makers. One example is our shoreline change maps, which measure and 
estimate the changes in the Massachusetts coastline as a result of 
natural erosion and accretion, as well as relative sea level rise. 
These maps and all accompanying data are available on our website 
(www.mass.gov/czm/hazards/shoreline_change/shorelinechangeproject.htm) 
with a fact sheet explaining how to use the maps. These resources 
receive thousands of hits per year and are used locally to supplement 
information provided by outdated flood maps.

The Time to Act Is Now
    It is very important to emphasize that this is not a problem only 
for the future. In an increasing number of communities along the 
Massachusetts coast, erosion and flooding impacts are increasingly 
causing damage even during today's minor storms. And with climate 
change, these impacts will only grow as storms increase in frequency 
and intensity.

Successful Strategies through Federal-State Partnerships
    Through the Coastal Zone Management Act amendment process, 
provisions should be developed to allow states and territories to 
develop specific coastal climate change adaptation plans and 
strategies. States also support increased funding for climate change 
activities and support legislation that would encourage NOAA and other 
agencies to assist the states via technical assistance, mapping, 
modeling, data, and forecasting products, and intergovernmental 
coordination. Federal activities related to coastal adaptation should 
be coordinated closely with states by involving coastal zone management 
programs early in the planning process.
    There are several emerging areas where state, Federal, and other 
partners are actively working on improved coordination and cooperation 
for more effective coastal and ocean management. One of these is the 
new Integrated Ocean Observing System (IOOS) initiative. Led by NOAA, 
the IOOS program seeks to integrate coastal and ocean observing 
capabilities, in collaboration with Federal and non-Federal partners, 
to maximize access to data and generation of information products and 
inform decisionmaking. Massachusetts has been participating in both the 
Northeast and Mid-Atlantic Regional Coastal Ocean Observing Systems, 
which are comprised of diverse partners including state and Federal 
agencies, academic institutions, and coastal and maritime interests. In 
both of these regions, remote observation technologies (e.g., 
instruments on buoys and high frequency radar) and the development of 
prototype products have been prioritized to address the issue area of 
coastal inundation. When fully operational, real-time observations on 
meteorological and oceanographic measurements will be integrated into 
interactive products such as a Gulf of Maine Storm Simulation and 
Prediction System.
    Another example of emerging synchronization is the Northeast 
Regional Oceans Council (NROC). Consisting of delegates from the six 
New England states and ex-officio members from Federal agencies, NROC 
was established in 2005 by resolution of the New England Governor's 
Association. The primary function of the council is to engage in 
efforts that require or benefit from regional actions to address issue 
areas of ocean and coastal ecosystem health, coastal hazards 
resiliency, ocean energy planning and management, and maritime 
security. By increasing communication and cooperation among regional 
interests, the council provides new forums for information exchange and 
strategic state-Federal collaboration on such actions as regional 
climate change activities and initiatives.
    Finally, the Joint Subcommittee on Ocean Science and Technology 
created the Interagency Working Group on Ocean and Coastal Mapping in 
response to recommendations of the U.S. Ocean Action Plan and the 2004 
National Research Council report, A Geospatial Framework for the 
Coastal Zone: National Needs for Coastal Mapping and Charting. The 
Interagency Working Group on Ocean and Coastal Mapping brings together 
Federal, state, industrial, academic, and nongovernmental organizations 
to coordinate the best use of mapping resources and to avoid 
duplication of effort. One of the first tasks for this group is to 
develop an inventory of ocean and coastal mapping data and activities. 
At a recent strategic planning workshop in February 2008, highlights of 
Federal ocean and coastal mapping activities were presented, and 
representatives from Massachusetts, Florida, and California provided 
updates of their current data collection and mapping activities, best 
practices, and challenges. All participants identified coordination, 
collaboration, and partnerships as keys to successful past and future 
efforts.

Legislative Opportunities
    There are two pending bills that could assist in developing these 
key Federal-state partnerships. Massachusetts and CSO appreciate the 
work of Senator Kerry and strongly support the climate change research 
and monitoring activities proposed in the Global Change Research 
Improvement Act of 2007 (S. 2307). The proposed legislation would 
establish a national climate service through NOAA to address weather, 
climate change, and climate variability affecting public safety, 
advancing the national interest in understanding, forecasting, 
responding, adapting to, and mitigating the impacts of both natural and 
human-induced climate change and climate variability. National level 
research, infrastructure, and coordinated outreach and communication 
mechanisms would directly support state and local policymakers by 
providing comprehensive national research to assist with regional 
adaptation and mitigation planning. Under the bill, existing Federal 
climate change research would be coordinated and particular attention 
would be focused on regional and state vulnerabilities to climate 
change, allowing communities to utilize national data to help address 
adaptation and mitigation on a localized level.
    Massachusetts and CSO also support the climate adaptation 
provisions in America's Climate Security Act of 2007 (S. 2191), 
particularly the specific allocation of 5 percent of the Emission 
Allowance Account to states, which can be used for specific purposes, 
one of which is to collect, evaluate, disseminate, and use information 
necessary for affected coastal communities to adapt to climate change. 
We are in favor of the expansion of the Adaptation Fund, funded through 
the emissions cap and trade program, to include coastal adaptation. 
These provisions recognize that coastal states and communities are on 
the front lines of climate change and will need Federal support that is 
proportionate to this risk.

The Future of a Successful Climate Modeling Partnership
    As state-level coastal managers, we can develop new tools and 
package available tools through programs like StormSmart Coasts. While 
we will always do the best we can with the information we have 
available, the current scarcity of regional- and local-scale, high-
priority data and information is alarming. For example, to improve our 
understanding of current and future coastal floodplains and high-hazard 
zones, we need topographical information in finer resolution than the 
coarse 10- to 20-foot contour intervals available today. Similarly, 
while there are hydrodynamic models that encompass regional systems 
(e.g., Gulf of Maine, Massachusetts Bay), these have not been tailored 
to the region's complex coastline and bathymetry, which includes 
numerous islands and shoals, and they lack the necessary field 
measurements for model verification and refinement. Without adequate 
data or resources, state and local decision-makers cannot accurately 
map the existing extent of the coastal floodplain, let alone project 
what that floodplain will look like in the next 30 years. Given the 
scientific complexity and levels of funding involved, state and local 
governments cannot possibly hope to fill this data gap alone. We are 
very pleased to know that the Federal Government is looking to fulfill 
this role, and we guarantee that if you get us the information we need, 
we are prepared to use it wisely. Our personal safety, ecosystems, and 
local and regional economies depend on it.
    But data alone cannot solve the problem--this information must get 
into the hands of the people who can use it to make better choices 
about development, redevelopment, and storm-damage protection, 
including municipal officials, business owners, and current and future 
homeowners in coastal floodplain areas.
    Through StormSmart Coasts, we have built the framework and have 
begun to work with coastal communities to implement results-oriented 
strategies. But ultimately, the effectiveness of those strategies is 
limited by the data, models, and diagnostics available--and the 
information generated through a strategic climate modeling approach 
that provides such decision-support resources as reliable estimates of 
sea level rise in the next few decades will be the key to future 
success. With this critical gap filled, local and state officials will 
be able to successfully implement real-world strategies to address this 
very real problem--creating a true partnership that maximizes the best 
of what all levels of government have to offer.

Conclusion
    As you move forward, we strongly encourage you to look at how state 
programs like StormSmart Coasts serve as successful examples--
demonstrating not only how states can fine-tune and package the data 
and information developed through the Federal climate change programs 
for the local decision-makers to use in a real-world context--but also 
how all levels of government can work together successfully. To ensure 
that you continue to build a results-oriented national climate modeling 
strategy, we strongly encourage you to work with state coastal 
managers, as well as local officials, to understand our specific needs. 
To be effective, such a strategy must specifically answer the kind of 
questions asked by all coastal communities looking to implement 
effective coastal floodplain management--what are the current risks to 
my community and how will those risks change in the future. Please help 
us put all of the pieces together so we can respond quickly and 
effectively to future coastal hazards.
    Thank you again for the opportunity to testify on the importance of 
national efforts for climate change modeling. I would be happy to 
respond to any questions that you may have.
                                 ______
                                 
                              Fact Sheet 1
Introduction to No Adverse Impact (NAI) Land Management in the Coastal 
        Zone
A legally sound way for municipalities to protect people and property
What Is NAI?
    No Adverse Impact (NAI) is a forward-thinking, fair, and legally 
defensible approach to coastal land management. In its broadest sense, 
it is a set of ``do no harm'' principles to follow when your community 
is planning, designing, or evaluating public and private development 
activities and storm-damage prevention measures.



    While seawalls and other structures can sometimes provide storm 
protection, they generally require regular expensive upkeep and often 
lead to other problems (including beach erosion). Marshfield, 
Massachusetts.

    NAI protects the rights of residents, businesses, and visitors in 
your community by requiring that public and private projects be 
designed and completed in such a way that they do not: (1) pose a 
threat to public safety, (2) increase flood or storm damage to public 
or private property, and/or (3) strain municipal budgets by raising 
community expenditures for storm-damage mitigation, stormwater 
management, emergency services, and disaster recovery efforts.

NAI: Local and Comprehensive
    Careful management of coastal floodplains is critical to protect 
people and property, and to reduce the financial strain on businesses, 
private property owners, and municipal budgets. While the Commonwealth 
of Massachusetts has passed regulations to help prevent storm damage, 
ultimately most of the authority and tremendous responsibility to 
manage floodplains is entrusted to local governments.
    Accurately evaluating the potential effects of proposed activities 
can be challenging, and requires looking both on and offsite, since 
damage often isn't confined to the parcel(s) under review. For example, 
the construction of a home may change stormwater flow and increase 
erosion (removal of sediment by water or wind) to surrounding 
properties. Similarly, new parking lots, roads, and buildings may 
redirect stormwater onto other properties instead of allowing it to be 
reabsorbed into the ground.



    In addition to being costly to repair, roads damaged by storms can 
become hazards for rescue personnel and others. This road in Rockport, 
Massachusetts, was destroyed by a 2007 nor'easter.

    Since each permit might be considered to set a precedent, it is 
critical that communities consider the potential cumulative effects of 
their decisions--a number of seemingly insignificant projects can 
collectively cause substantial damage. The NAI approach clarifies that 
community leaders not only have the legal right to consider the 
cumulative impacts of their permitting decisions, they have the legal 
responsibility. Increasingly, communities that permit projects that 
result in flooding or storm damage to other properties end up in land 
court. (See the StormSmart Coasts Fact Sheet 2, No Adverse Impact and 
the Legal Framework of Coastal Management). Adopting the NAI approach 
also gives your community the chance to clearly articulate a ``do no 
harm'' goal for all future land use.

The NAI Approach
    The Association of State Floodplain Managers (ASFPM), a national 
organization of professional flood hazard specialists from all levels 
of government, the research community, the insurance industry, and 
technical fields, identifies three different levels of floodplain 
management strategies: Basic, Better, and NAI.

   Basic: Approaches typically used to meet minimum Federal or 
        state requirements for managing floodplains and coastal areas 
        to minimize flood losses.

   Better: Activities that are more effective than the basic 
        level because they: (1) are tailored to specific situations, 
        (2) provide protection from larger floods, (3) allow for 
        uncertainty in storm magnitude prediction, and (4) serve 
        multiple purposes.

   NAI: Tools and techniques that go further than the measures 
        defined as ``better'' by ensuring that private development, 
        public infrastructure, and planning activities do not have 
        direct or indirect negative consequences on the surrounding 
        natural resource areas, private property, or other communities.
        
        
    ASFPM has created seven NAI Building Blocks, which can help 
communities to maintain and enhance flood protection. These building 
blocks--hazard identification and mapping; planning; regulations and 
development standards; mitigation; infrastructure siting and design; 
emergency services; and public outreach and education--are briefly 
introduced in the table on the next page. For more information, see 
ASFPM's Coastal NAI Handbook at www.floods.org, or the StormSmart 
Coasts website at www.mass.gov/czm/stormsmart.

                           NAI Building Blocks
------------------------------------------------------------------------
 NAI Building
     Block            Basic            Better                NAI
------------------------------------------------------------------------
Hazard          Use FEMA Flood    Gather and use    Incorporate coastal
Identification   Insurance Rate    detailed          hazard data (e.g.,
and Mapping      Maps for land     coastal hazard    erosion rates,
                 use decisions.    data (e.g.,       vulnerability of
                                   historic          environmentally
                                   erosion rates,    sensitive areas,
                                   actual observed   and sea-level rise
                                   extents of        rates and impacts)
                                   floodwaters)      into community-wide
                                   for land use      planning maps and
                                   decisions.        regulations.
------------------------------------------------------------------------
Planning        Use land use      Develop           Design special area
                 planning and      floodplain        management plans
                 zoning through    management        to: protect storm
                 a community       plans that        damage and flood
                 master plan.      include           control functions
                                   stormwater        of natural
                                   management and    resources, promote
                                   hazard            reasonable coastal-
                                   mitigation        dependent economic
                                   measures.         growth, and improve
                                   Promulgate        protection of life
                                   detailed          and property in
                                   guidance          hazard-prone areas.
                                   focusing on
                                   reducing flood
                                   damage.
------------------------------------------------------------------------
Regulations     Follow Federal    Adopt conditions  Preserve sensitive
and              Emergency         for siting new    areas through
 Development     Management        development.      bylaws and
Standards        Agency National   Regulate          regulations that
                 Flood Insurance   cumulative,       may: establish
                 Program           substantial       maximum densities
                 regulations.      improvements.     for development,
                                   Revise            restrict structures
                                   regulatory        between the
                                   tools for         shoreline and the
                                   addressing        setback line,
                                   erosion along     mandate vegetative
                                   shorelines        coastal buffers
                                   including:        rather than manmade
                                   relocation of     structures
                                   threatened        (bulkheads,
                                   buildings,        seawalls, or
                                   building          groins), minimize
                                   setbacks, beach   impervious cover,
                                   nourishment and   and preserve stream
                                   bio-              corridor and
                                   engineering,      wetland buffers.
                                   and               Regulate placement
                                   stabilization     of fill.
                                   of eroded
                                   areas.
------------------------------------------------------------------------
Mitigation      Use common        Elevate or        Stabilize shorelines
                 practices, such   relocate          with vegetation.
                 as flood          buildings.        Prohibit
                 proofing          Acquire land.     construction in
                 existing          Encourage         especially damage-
                 structures.       nonstructural     prone areas.
                                   methods for       Prevent filling of
                                   shoreline         wetlands and other
                                   protection.       lowlands. Nourish
                                                     beaches where
                                                     appropriate.
                                                     Protect watersheds.
                                                     Monitor corrective
                                                     efforts. Regulate
                                                     construction of
                                                     shore-protection
                                                     structures.
------------------------------------------------------------------------
Infrastructure  Respond to storm  Upgrade damaged   Prohibit major
Siting and       events as they    facilities to     public
 Design          occur. After a    more hazard-      infrastructure
                 storm, rebuild/   resistant         investments in
                 repair to         standards.        special flood
                 previous          Inventory         hazard areas.
                 condition.        hazard risks of   Ensure that roads,
                                   all public        sewer lines, and
                                   buildings.        utility upgrades
                                   Insure            don't encourage
                                   buildings for     development in
                                   all hazards (as   hazard-prone areas.
                                   appropriate).     Zone to prohibit
                                   Identify, and     construction in
                                   if possible,      high-hazard areas.
                                   relocate or       Locate new critical
                                   protect           facilities above
                                   ``critical        500-year flood-
                                   facilities.''     plain.
------------------------------------------------------------------------
Emergency       Create and use    Create and test   Create plans to
Services         generic hazard    community-wide    ensure that all
                 response plan.    hazard plans      people who want or
                                   that involve      need to be
                                   all local         evacuated can be
                                   boards and        moved to safe
                                   departments.      shelters, and post-
                                                     disaster plans that
                                                     improve community
                                                     flood resistance
                                                     through: willing
                                                     land acquisition,
                                                     determining which
                                                     structures are
                                                     ``substantially
                                                     damaged,'' and
                                                     ensuring that
                                                     appropriate
                                                     reconstruction
                                                     meets code
                                                     requirements.
                                                     Establish mutual
                                                     aid agreements with
                                                     neighboring
                                                     communities.
------------------------------------------------------------------------
Public          Answer questions  Periodically      Create comprehensive
 Outreach        and provide       inform            education and out-
and Education    information as    residents of      reach programs
                 requested by      coastal           using expertise of
                 public.           hazards,          state and Federal
                                   vulnerability,    agencies (when
                                   and mitigation    needed) to
                                   techniques        encourage community-
                                   through public    wide proactive
                                   workshops, and    storm preparation.
                                   in forums after   Establish coastal
                                   storm recovery.   hazard disclosure
                                                     requirements for
                                                     property sales.
------------------------------------------------------------------------

The Benefits of NAI
    While NAI strategies require investment in planning and 
implementation, they offer real benefits for your community. NAI can . 
. .

   Save money: Less damage means lower post-storm community 
        cleanup costs, fewer demands on public officials' limited time, 
        and reduced strain on public resources.

   Decrease litigation: NAI principles have been judicially 
        tested and courts have shown immense deference to regulations 
        that seek to prevent harm (for an example, see the StormSmart 
        Coasts Fact Sheet 3, A Cape Cod Community Prevents New 
        Residences in Floodplains). NAI can also help your community 
        avoid potential litigation over ineffectual flood management 
        practices that result in future damage or loss of life. (See 
        Fact Sheet 2, No Adverse Impact and the Legal Framework of 
        Coastal Management.)

   Reduce conflicts with property owners: NAI doesn't say 
        ``no.'' It says ``yes, if . . .'' It is a common-sense approach 
        that seeks to protect everyone's property by only allowing 
        projects that eliminate or mitigate their impacts.

   Reduce risk to people and public and private property: 
        Better planned and designed development and public 
        infrastructure is less likely to cause and suffer damage. An 
        NAI approach can help protect the beaches that are critical to 
        many communities' economies.

   Lower flood insurance rates: The Community Rating System 
        (CRS) is a Federal Emergency Management Agency (FEMA) program 
        that decreases flood insurance rates for communities with 
        effective hazard mitigation strategies. Many NAI strategies 
        qualify for CRS credits. For more information see the CRS 
        Resource Center at training.fema.gov/EMIWeb/CRS/.

   Increase your capacity to bounce back after a storm: Reduced 
        storm damage means less downtime and less costly clean up for 
        local businesses, which is especially important for small, 
        locally owned businesses that may otherwise struggle to stay 
        solvent during frequent or prolonged closures.

   Clarify your land use objectives: By adopting NAI 
        principles, your community can articulate the overarching goals 
        that help bring consistency and predictability to permitting.

   Preserve quality of life: With NAI you can help make your 
        community safer while preserving quality of life for your 
        citizens now and in the future. An NAI approach can help ensure 
        that your community resources, including beaches, public parks, 
        and other open spaces, are there to be enjoyed by future 
        generations.

For More Information . . .

   For more on the theory of NAI and its application in coastal 
        areas, see the Association of State Floodplain Managers website 
        (www.floods.org), especially their Coastal NAI Handbook. Also 
        see the StormSmart Coasts website at www.mass.gov/czm/
        stormsmart.

   For more on the legal issues surrounding coastal management, 
        see the StormSmart Coasts Fact Sheet 2, No Adverse Impact and 
        the Legal Framework of Coastal Management.

   For an example of NAI-type regulations at work, see the 
        StormSmart Coasts Fact Sheet 3, A Cape Cod Community Prevents 
        New Residences in Floodplains.

   For a more detailed look at the legal theory behind this and 
        similar cases involving land management in hazardous areas, see 
        the Association of State Floodplain Managers' No Adverse Impact 
        Floodplain Management and the Courts by attorneys Jon Kusler 
        and Ed Thomas, at www.floods.org.
                                 ______
                                 
                              Fact Sheet 2
No Adverse Impact and the Legal Framework of Coastal Management
How communities can protect people and property while minimizing 
        lawsuits

    Managing coastal floodplains is a challenging endeavor that 
sometimes is incorrectly thought to put local government's duty to 
protect people and property in direct conflict with property rights. 
Most local officials want to reduce the harm and costs associated with 
coastal storms, and recognize that unwise development can worsen the 
situation. Unfortunately, as our society has grown more litigious, it 
may seem harder for municipal governments to stay out of land court 
when preventing or conditioning development projects, even when there 
is good evidence that these projects may create problems for others. 
However, the No Adverse Impact (NAI) approach to land use management is 
an appropriate way to protect people, property, and property rights. 
(To learn more about NAI, see the StormSmart Coasts Fact Sheet 1, 
Introduction to No Adverse Impact (NAI) Land Management in the Coastal 
Zone.)
    While nothing can prevent all legal challenges, following the NAI 
approach can help to: (1) reduce the number of lawsuits filed against 
local governments, and (2) greatly increase the chances that local 
governments will win legal challenges to their floodplain management 
practices. The legal system has long recognized that when a community 
acts to prevent harm, it is fulfilling a critical duty. The rights of 
governments to protect people and property have been well recognized by 
the legal system since ancient times. Courts from the Commonwealth of 
Massachusetts to the U.S. Supreme Court have consistently shown great 
deference to governments acting to prevent loss of life or property, 
even when protective measures restrict the use of private property. 
This ``prevention of harm'' principle is the foundation of the NAI 
approach. The goal of this fact sheet is to provide local officials 
with information on how to use the NAI tools to confidently protect 
people and property in a fair and effective way, while avoiding 
lawsuits (even those alleging takings).
    Two key points:

        1. Communities have the legal power to manage coastal and 
        inland floodplains.

        2. Courts may (and often do) find that communities have the 
        legal responsibility to do so.

        
        
    These Sandwich homeowners proactively protected their property by 
planting beach grass. Vegetating dunes and banks can reduce erosion and 
slow floodwaters without adversely impacting other properties.

How NAI Can Help Your Community Avoid Lawsuits
    The best way to avoid losing in court is to stay out of court. One 
of the strengths of the NAI approach is that its clear goal (the 
prevention of harm) fosters and encourages cooperation between 
landowners and regulators as they work together to try to find 
solutions to the problems associated with proposed projects. Such 
collaboration is a great way to stay out of land court.
    When avoiding court isn't possible, following the NAI approach can 
greatly increase the chances that local governments will win in 
lawsuits arising from their floodplain management practices. The most 
common and historically problematical challenges that local officials 
face while trying to regulate use of private property are allegations 
of ``constitutional takings.''

Not all the uses an owner may make of his property are legitimate. When 
        regulation prohibits wrongful uses, no compensation is 
        required.''--The Cato Institute
    Takings background: This fact sheet summarizes a complex body of 
law under the so-called ``Takings Clause'' of the Fifth Amendment to 
the U.S. Constitution. This summary is not intended to be legal advice 
for any particular situation, and may not be relied upon as such. To 
determine whether a particular regulation would cause a taking, 
communities should consult with an attorney. Property owners file 
takings cases when they believe regulations violate their 
constitutional property rights. The legal basis for these arguments can 
be found in the Fifth Amendment of the U.S. Constitution, which 
prohibits the Government from taking private property for public use 
without compensation. The interpretation of the courts through the 
years has clarified that the Fifth Amendment encompasses more than an 
outright physical appropriation of land. In certain situations, the 
courts have found that regulations may be so onerous that they 
effectively make the land useless to the property owner, and that this 
total deprivation of all beneficial uses is equivalent to physically 
taking the land. In such a situation, courts may require the governing 
body that has imposed the regulation to either compensate the landowner 
or repeal the regulation.
    Needless to say, with local budgets strapped and coastal land 
values skyrocketing, it is rarely economically feasible for local 
governments to compensate landowners when, for example, prohibiting a 
house on a solid foundation in an area known to flood, or preventing 
the construction of a seawall to protect a home on an eroding bluff.
    NAI to the Rescue: It is critical that management decisions respect 
property rights and follow general legal guidelines (see the ``Legal 
Dos and Don'ts of Floodplain Management'' text box). The courts have 
made it very clear that property rights have limits. For example, both 
Commonwealth of Massachusetts and Federal laws acknowledge that 
property owners do not have the right to: be a nuisance, violate the 
property rights of others (for example, by increasing flooding or 
erosion on other properties), trespass, be negligent, violate 
reasonable surface water use and riparian laws, or violate the public 
trust.

The Four Types of Regulatory Takings
    The best way to understand how the NAI approach helps to prevent 
takings challenges is to look specifically at what the courts have 
decided may constitute a regulatory taking. In 2005, the U.S. Supreme 
Court ruled on a precedent-setting case (Lingle v. Chevron), which 
clearly established regulatory taking guidelines. In their unanimous 
decision, the Court determined that there are four ways for a 
regulation to be a taking. Each way is briefly discussed below, with a 
non-technical explanation of how they are relevant to an NAI approach. 
(For a more detailed legal explanation of these cases, see the latest 
edition of No Adverse Impact Floodplain Management and the Courts, 
published by the Association of State Floodplain Managers at 
www.floods.org.)
    1. A physical intrusion. Governments may not, without compensation, 
place anything on private property against the wishes of the owner. The 
case discussed (Loretto v. Teleprompter Manhattan) involved a New York 
City requirement that building owners allow the cable company to 
install a small cable box and cables on all residential buildings. 
Because the NAI approach doesn't generally promote structural 
solutions, this type of regulatory taking is unlikely to apply. 
However, if a community's NAI plan involves the placement of structures 
(culverts, for example) on private property, this ruling makes it clear 
that the community may be required to obtain the permission of the 
landowner or pay compensation.



    2. A total or near-total regulatory taking. If a regulation 
restricts property rights to such a degree that it eliminates all or 
essentially all economically viable uses of a piece of property, this 
may constitute a taking. The case reviewed (Lucas v. South Carolina 
Coastal Council) was filed by a landowner who was prohibited from 
building a home on a barrier beach. In their opinion, the Court clearly 
states that regulations aimed at preventing nuisance don't constitute 
takings. It warns, though, that governing bodies arguing that specific 
regulations are designed to prevent nuisances will need to demonstrate 
how they are addressing similarly situated nuisances (i.e., regulations 
may not be applied arbitrarily). The NAI approach can help your 
community to consistently articulate how potentially harmful projects 
are nuisances. When designing land use regulations, your community 
should always try to ensure that the owner retains at least some 
economically beneficial uses. This is both fair and helps establish the 
legal reasonableness of your regulations. Note that land uses that harm 
others are not legal or beneficial, and that beneficial uses don't 
necessarily include building residences or other structures, especially 
in hazardous areas. Where new regulations, even hazard-based 
regulations, could sharply decrease the market price of property, 
consider allowing the transfer of development rights to areas where 
your community would like growth to occur. To learn about transferable 
development rights, see www.mass.gov/envir/smart_growth_toolkit/pages/
mod-tdr.html.
    3. A significant, but not near-total regulatory taking. Courts 
hearing takings arguments should consider three factors that have 
``particular significance''--(a) the magnitude of the economic impact, 
(b) how severely the regulation affects ``investment-backed 
expectations,'' and (c) the character of the government in action. The 
central case discussed (Penn Central v. City of New York) concerned a 
denied expansion of Grand Central Station in New York City. The 
historic preservation regulation reviewed in this case seeks to protect 
neighborhood character--not to prevent physical harm. These are two 
very different things in the eyes of the law. The U.S. legal system 
sometimes requires governments to compensate landowners when property 
rights are compromised for community improvement, but less frequently 
when they prevent potential harm. There is no property right to use or 
develop land in a way that harms others, even if that use maximizes the 
particular site's economic potential. There is no constitutional or 
legal right to a good return on investments. Unfortunately, some people 
invest in land with erroneous ideas about what they are legally allowed 
do with it, and when forbidden to do as they wish, may argue that 
regulations have devalued their property. The courts have made it clear 
that while regulations designed to prevent harm may reduce the market 
value of a piece of property, they do not decrease its true value, and 
hence NAI-based regulations cannot trigger this aspect of a taking 
test. A 2005 Massachusetts Supreme Judicial Court decision upheld a 
coastal town's regulation prohibiting new residences in its coastal 
floodplain because the town successfully established that this 
regulation was designed to prevent harm and did not render the land 
valueless.



    For more information, see the StormSmart Coasts Fact Sheet 3, A 
Cape Cod Community Prevents New Residences in Floodplains.

    4. Insufficient relationship between the requirement and the 
articulated government interest. If a community conditions a permit, 
the requirements it exacts from the landowner must be related to the 
goals of the regulation and must be ``roughly proportional'' to the 
predicted impacts of the proposed development. In the two cases, Nollan 
v. the California Coastal Commission and Dolan v. City of Tigard, 
landowners were required to provide a public right of way as a permit 
condition, even though the proposed developments did not reduce public 
access. The NAI approach avoids this type of taking by tightly binding 
regulations to the specific goal of preventing harm.
    With these and other decisions, the courts have made it clear that 
governments may regulate land without compensation if they do so with 
the intent of preventing harm. Fairly applied No Adverse Impact 
regulations make the ``takings issue'' a non-issue.
    From the property rights perspective, it's worth noting that the 
Cato Institute, which advocates for limited government, individual 
liberty, and free markets, agrees that preventing landowners from 
causing harm to others does not constitute a taking:

        ``Owners may not use their property in ways that will injure 
        their neighbors. Here the Court has gotten it right when it has 
        carved out the so-called nuisance exception to the 
        Constitution's compensation requirement. Thus, even in those 
        cases in which regulation removes all value from the property, 
        the owner will not receive compensation if the regulation 
        prohibits an injurious use.''--Roger Pilon, Senior Fellow and 
        Director--Cato Institute (to the U.S. House of Representatives, 
        2/10/95)
``The takings clause was never intended to compensate property owners 
        for property rights they never had.''--Massachusetts Supreme 
        Judicial Court
Why You Should Manage Your Floodplains
    Protecting people and property is a fundamental duty of all levels 
of government. One of the most effective ways that local governments 
protect people and property is through the permitting process. Here, 
local officials can and should do what they can to reduce the 
likelihood that the development or use of property will cause harm.
    Communities should also be aware that in a growing number of 
states, courts are favoring plaintiffs that sue local governments for 
permitting projects that later cause damage to property (for example, 
permitting the construction of roads that back-up streams and increase 
flooding in the community). For more information on this trend, see No 
Adverse Impact Floodplain Management and the Courts (available at 
www.floods.org), where the authors found that a community is vastly 
more likely to be successfully sued for allowing improper development 
that causes harm than for prohibiting it.
    The take-home lesson: As a local official, you have been given the 
responsibility and the legal rights to manage coastal and inland 
floodplains. If you do so in a way that expressly seeks to prevent 
harm, the courts will support you.

For More Information . . .
    This is not and cannot be legal advice. To answer specific legal 
questions please see an attorney licensed in your jurisdiction. To 
learn more about the general legal framework of NAI-based floodplain 
management see:

   No Adverse Impact Floodplain Management and the Courts for 
        an excellent overview of the case history of NAI at 
        www.floods.org. While this document is designed for attorneys, 
        it is useful for anyone working in floodplain management.

   The StormSmart Coasts Fact Sheet 3, A Cape Cod Community 
        Prevents New Residences in Floodplains, which examines a 
        community's successfully defended NAI-type bylaw.

   The Coastal NAI Handbook at www.floods.org.

   The NAI section of the Association of State Floodplain 
        Managers website at www.floods.org.

   The Institute for Local Government's one-page publication, 
        10 Tips for Avoiding Takings Claims, at cacities.org/
        index.jsp?displaytype=11&zone=ilsg&section=
        land⊂_sec=land_property&tert=&story=20219.

   The American Planning Association's 1995 Policy Guide on 
        Takings at www.planning.org/policyguides/takings.html.

   The StormSmart Coasts website at www.mass.gov/czm/
        stormsmart.
                                 ______
                                 
                              Fact Sheet 3
Case Study--A Cape Cod Community Prevents New Residences in Floodplains
Lessons learned from Chatham's legally successful conservancy districts

    In a landmark 2005 ruling, the highest court in Massachusetts 
decisively affirmed the authority of municipalities to regulate or even 
prevent residential or other high-risk development in flood-prone areas 
without financial compensation to the property owners, so long as the 
regulation does not render the land entirely valueless.
    The case arose from the town of Chatham's refusal to permit the 
construction of a new home in a flood zone because the local zoning 
bylaw prohibited new residential units in the town's mapped 
floodplains. After multiple appeals by the landowner, the Massachusetts 
Supreme Judicial Court ruled on July 26, 2005, that the zoning bylaw 
was based on reasonable public interest, and did not render the lot 
economically worthless. Therefore, no compensation was due. The 
decision was not appealed.

The Zoning Bylaw
    Chatham's zoning bylaw designates ``conservancy districts'' 
encompassing all land in the town's 100-year floodplain as mapped in 
its most recent town-approved Flood Insurance Rate Maps. The goal of 
the bylaw is to protect people, property, and resources (see ``Chatham 
Conservancy District Purposes'' sidebar). The bylaw clearly delineates 
three types of activities in designated conservancy districts--
permitted uses, special permit uses, and prohibited uses--examples are 
shown in the table below.

                                      Examples from Chatham's Zoning Bylaw
----------------------------------------------------------------------------------------------------------------
           Permitted uses                      Special permit uses                     Prohibited uses
----------------------------------------------------------------------------------------------------------------
Fishing, cultivation, and harvesting
 of shellfish (including excavation
 of areas for cultivation and
 harvesting of marine foods);
 various horticulture activities

Outdoor recreation activities,
 provided that related structures do
 not destroy beneficial character of
 district

Floats

Maintenance of existing raised
 roadways Installation of utilities

Agriculture

Government dredging of navigation
 channels

Construction and maintenance of town
 landings and public boat launching
 ramps; nourishment of town beaches

Mosquito control by Cape Cod
 Mosquito Control Project
                                      Construction of certain structures,
Maintenance of existing channels and   including catwalks, piers, ramps,
 marine facilities                     stairs, boat shelters, tennis
                                       courts.



                                                                            Filling of land








----------------------------------------------------------------------------------------------------------------

``The takings clause was never intended to compensate property owners 
        for property rights they never had.''--Massachusetts Supreme 
        Judicial Court
The Case
    The lawsuit concerned a 1.8-acre parcel located in Chatham's mapped 
floodplain (and therefore, in a conservancy district). In 1998, the 
owner of the lot received an offer of $192,000 for the parcel, 
contingent upon the ability of the purchaser to obtain the permits 
necessary to build a home. The proposed home was to be elevated on open 
piles above the mapped 100-year flood elevation.
    Because the lot is located within a conservancy district, the 
town's Zoning Board (the district permitting authority) denied the 
building permit application. The owner of the lot responded by filing 
one suit against the Selectmen and Zoning Board and another against the 
town's Conservation Commission (the construction would have also 
violated a local wetlands bylaw), each suit alleging that the bylaws 
violated the owner's constitutional property rights, and that the town 
had thereby effectively ``taken'' her property (for more on 
constitutional takings, see StormSmart Coasts Fact Sheet 2, No Adverse 
Impact and the Legal Framework of Coastal Management). A Superior Court 
judge combined the two suits. After a two-day trial, which included 
testimony on the flood history of the property, the risks and impacts 
of its potential development, and the difficulty in safely evacuating 
the area, the Superior Court found insufficient evidence to support the 
plaintiff's claims that the bylaws had resulted in a regulatory land 
taking, and upheld the town's decision.
    When the plaintiff appealed the decision, the Massachusetts Appeals 
Court affirmed the Superior Court's decision. While acknowledging that 
the bylaw did severely constrict the possible uses of the lot, the 
Appeals Court noted that ``a land-use regulation may deprive an owner 
of a beneficial property use--even the most beneficial such use--
without rendering the regulation an unconstitutional taking.'' The 
Appeals Court further noted that:

        ``As a matter of Massachusetts law, restricting residential 
        development within the path of floodwater, the flood plain, is 
        a direct, logical, and reasonable means of safeguarding persons 
        and property from those hazards occasioned by a flood and 
        advances a substantial state interest, that is, the health, 
        safety, and welfare of the general public as well as that of 
        its individual members.''

        
        
    The arrow indicates the approximate location of the proposed home 
site. This satellite photograph also shows the breach in the barrier 
beach from 1987. The breach greatly increased the exposure of the lot 
and surrounding properties to wave and storm surge.

    The plaintiff then appealed to the Massachusetts Supreme Judicial 
Court, which, after reviewing the case, upheld the lower courts' 
rulings, citing a recent U.S. Supreme Court decision that had rendered 
zoning bylaws and ordinances valid under the U.S. Constitution so long 
as their application bears a ``reasonable relation to the State's 
legitimate purpose'' (such as protecting people and property).
    The decision also noted that while the regulation may have indeed 
reduced the market value of the property, the prevention of one 
potential use for a piece of property did not constitute a total 
taking. A witness for the plaintiff estimated that with the bylaw, the 
lot was worth at least $23,000--a substantial reduction but still more 
than a ``token'' interest, according to the decision which cited a 
(2001) case where the U.S. Supreme Court ruled that no compensation was 
due when a regulation reduced the appraised value of a parcel from 
$3,150,000 to $200,000.
    Finally, the decision noted that there was ample evidence showing 
that the construction of a home on the lot could have severe adverse 
impacts on the surrounding community. The plaintiff's expert testified 
that the proposed house could be picked up off its foundation and 
floated away by a severe storm, potentially damaging neighboring homes. 
The defendant offered testimony that efforts to evacuate the home 
during a flood would pose risks to rescue workers, as well as the 
home's occupants.



    A Nauset Beach home destroyed by a 2007 storm. As was noted in the 
Massachusetts Supreme Judicial Court's ruling, damaged structures like 
the one in this photo can create debris that may threaten other 
structures.

    The Massachusetts Supreme Judicial Court concluded that no 
compensation was due to the property owner, because: ``The taking 
clause was never intended to compensate property owners for property 
rights they never had.''
    The decision was not appealed.

Why Chatham Won the Case

        1.            The zoning bylaw had the clear goals of protecting
                       people and property.
        2.            While the bylaw prevents construction of new
                       homes, it leaves property owners with many
                       alternative uses. The land retains more than a
                       ``token'' value.
        3.            The law was fair, and applied to identifiable,
                       mapped areas (i.e., wasn't ``spot zoning,'' which
                       unfairly prevents one individual property owner
                       from using property in a certain way).
        4.            The town's emergency management experts testified
                       that evacuation of the areas would put rescue
                       workers at risk.
        5.            The town was willing to legally defend its
                       position.


                       
                       
    Top: The erosional beach near the proposed home site is prone to 
flooding and storm damage.
    Bottom: An area ofChatham in the floodplain where flooding can make 
evacuation difficult.

For More Information . . .

   For an overview of the legal framework of coastal management 
        in Massachusetts, see the StormSmart Coasts Fact Sheet 2, No 
        Adverse Impact and the Legal Framework of Coastal Management.

   For the text of the decision, see www.socialaw.com/
        slip.htm?cid=15382.

   For a copy of the bylaw see www.chatham-ma.gov/
        Public_documents/chat
        hamma_CommDev/Zbylaw2005.pdf.

   For a more detailed look at the legal theory behind this and 
        similar cases involving management of land in hazardous areas, 
        see the Association of State Floodplain Managers' No Adverse 
        Impact Floodplain Management and the Courts, by attorneys Jon 
        Kusler and Ed Thomas at www.floods.org.

   The Massachusetts StormSmart Coasts webpage: www.mass.gov/
        czm/stormsmart.

        
        
    As coastal areas of Massachusetts continue to change in response to 
erosion and storms, the relative risks to properties do too. While the 
risk to these homes near a new breach is obvious, homes on the mainland 
that were once protected by the shifting barrier island also face 
increased exposure. (Photo: Nauset Beach, Chatham.)




    Senator Kerry. Thank you very much, Mr. Carlisle.
    Dr. Walsh?

       STATEMENT OF JOHN E. WALSH, DIRECTOR, COOPERATIVE

          INSTITUTE FOR ARCTIC RESEARCH, INTERNATIONAL

          ARCTIC RESEARCH CENTER, UNIVERSITY OF ALASKA

    Dr. Walsh. Senator Kerry, Senator Stevens, thank you for 
the chance to speak today. I am John Walsh from the 
International Arctic Research Center. I am also Director of 
NOAA's Cooperative Institute for Arctic Research at the 
University of Alaska.
    In many respects, Alaska is ground zero for recent climate 
change. We are seeing a dramatic loss of ice, melting glaciers, 
warming permafrost. Senator Stevens is well aware of the 
changes that are ongoing in Alaska.
    I would like to focus on the gap between what the 
stakeholders need, what they are requesting of the climate 
community, and what climate models are actually delivering.
    Alaska has a diverse population. It ranges from small 
indigenous communities that are reliant on subsistence 
activities to growing urban areas and to an energy sector on 
which the rest of the U.S. depends. Much of the infrastructure 
is built on permafrost, and one of the main characteristics of 
Alaska's climate is its wide seasonal swings. In addition to 
that, there are tremendous spatial variations. The graphics in 
the written testimony provide an example of the extreme 
contrasts spatially in the variables such as temperature.
    I will cite one example for what that means for the weather 
and climate of Alaska relative to modeling. The interior 
valleys, the Yukon River Valley, the Tanana River Valley, are 
precipitation shadows in reality. The present climate models 
with their resolutions of 100 to 200 kilometers present these 
areas as maximum elevation regions with precipitation greater 
than the surrounding areas. So we are completely losing the 
precipitation signal over a large portion of the state because 
of inadequate resolution. The resolution that we need in order 
to capture fields such as precipitation is between one and two 
orders of magnitude finer than what we now have in the latest 
generation of global models.
    A high priority for Alaskans is the tailoring of model 
output to include the information that is most relevant to the 
needs of planners and the public, as well as other 
stakeholders. The variables that are carried in climate models 
are often not the ones that correspond most to the user needs. 
In Alaska, some examples of these user needs are information 
about the firmness of the ground for overland transportation, 
snow cover characteristics, vegetative dryness during the fire 
season, and wind chill temperatures in exposed areas.
    A recent illustration of the needs and the gap relative to 
the modeling capabilities is the attempt by Peter Larsen and 
his colleagues in Anchorage to estimate the economic risks to 
public infrastructure in Alaska as a result of climate change 
in the coming decades. The global models on which he based his 
scenarios have uncertainties in themselves, and they produced a 
range of a factor of two to three in the estimates of 
infrastructure costs to be expected from climate change over 
the next 50 years.
    But more importantly are the limitations on the 
availability of variables beyond temperature and precipitation 
which were used in the Larsen study. Infrastructure such as 
buildings and roads will clearly be affected by freeze-thaw 
cycles, by changes in snow loads, by the temperature extremes, 
the peak-wind events, and the occurrences of flooding. We are a 
long way from being able to obtain that type of information in 
a credible manner from today's climate models, and there has 
actually been little effort to translate model output into 
these quantities that the users and the stakeholders need. So 
the bridging of the models and the user needs is an emerging 
area of activity. This need is intertwined with this need for 
higher resolution and for more credible model simulations.
    So the people of Alaska are already calling for more 
detailed and robust information than they are receiving from 
climate models, and I would argue that the challenge of 
integrating user needs such as those in Alaska with advances in 
climate modeling gives us an opportunity to respond to our 
taxpayers at the regional scale and to serve as a prototype for 
some globally integrated climate delivery services.
    Thank you.
    [The prepared statement of Dr. Walsh follows:]

 Prepared Statement of John E. Walsh, Director, Cooperative Institute 
 for Arctic Research, International Arctic Research Center, University 
                               of Alaska

Climate Modeling for Decision-Makers and Stakeholders in Alaska
    Alaska's statewide annual average temperature has increased by 3.4 
+F since the mid-20th century, and the increase is much greater (6.3 
+F) in winter. The higher temperatures of the recent decades have been 
associated with an earlier snowmelt in spring, a reduction of summer 
sea ice coverage, a retreat of many glaciers, and a warming of 
permafrost. These surface changes, as well as their associated climate 
drivers, have two characteristics that require advances in modeling if 
projections of change are to meet the needs of decision-makers and 
planners. First, feedbacks between ice, snow and the atmosphere exert 
potentially strong leverage on high-latitude climate change, and these 
feedbacks introduce large uncertainties into simulations by existing 
climate models. For example, the recent retreat of summer sea ice is 
occurring at a faster rate than projected by any of the models in the 
recent Fourth Assessment (2007) of the Intergovernmental Panel on 
Climate Change (Stroeve et al., 2007). There are also indications that 
feedbacks may already be occurring between the earlier spring snowmelt 
and the surface energy budget, resulting in an increase of vegetative 
greenness (photosynthetic activity) in parts of Alaska (Euskirchen et 
al., 2007). Second, the surface changes are highly variable over small 
spatial scales, largely as a result of complex topography and coastal 
configurations around the region. The figure below illustrates the fine 
resolution required to capture the spatial variations in Alaskan 
climate.



    Figure 1. Average July daily high temperatures in Alaska for 1961-
1990. Color ranges are 40-45 +F (blue), 45-50 +F (green), 50-55 +F 
(yellow), 55-65 +F (orange), and 65-75 +F (darker red). Image is from 
the PRISM database (Daly et al., 2008).

    In contrast to the 2 km resolution in figure above, the grid cell 
dimensions (spatial resolution) of global climate models are typically 
100-200 km. Figure 2 below shows the smoothness of projected 
temperature changes obtained from the global models for Alaska. The 
mis-match of scales is even greater for precipitation, which is a 
variable that is of great interest to users of climate information 
pertaining to water supplies, inland transportation, forestry, and 
terrestrial ecology.



    Figure 2. Projected changes of annual mean temperature (+F) over 
Alaska for the late 21st century (2090), based on the B1 simulations by 
the models used in the IPCC's (2007) Fourth Assessment. Yellow denotes 
a warming of 3-5 +F, deep red a warming of 8-10 +F.

    How can the utility of climate projections be made more useful to 
decision-makers and stakeholders in Alaska? Based on the experience of 
the Alaska Center for Climate Assessment and Prediction (a NOAA 
Regional Integrated Sciences and Assessment Center), the greatest needs 
are: (1) downscaling of the coarse-resolution model output, (2) 
reduction of the uncertainty inherent in the model-derived projections, 
and (3) tailoring of model output to include variables and information 
more directly relevant to the needs of planners and stakeholders. In 
the remainder of this testimony, we address these needs and approaches 
to meeting these needs.
    The mis-match of scales between Figures 1 and 2 can be addressed by 
two types of downscaling: dynamical and statistical. Dynamical 
downscaling consists of the nesting of a high-resolution regional model 
inside a coarser-resolution global model. This approach has been tested 
in various regions of the world, and its effectiveness is highly 
dependent on the validity of the input supplied at the lateral 
boundaries by the global model. For Alaska, the approach is being 
applied to simulations of the mass balance of glaciers in southeastern 
Alaska. The nesting of finer grids inside coarse grids achieves 1 km 
resolution over the glaciers. Applications to other surface features 
(e.g., permafrost, ecosystem changes) are being developed. The second 
approach to downscaling is statistical in its nature. In this case, 
statistical algorithms (e.g., multiple regression equations) are 
developed to relate model-computed quantities and observational data 
for which sufficiently long records exist. The predictors can be either 
pre-selected or screened. This approach, which generally requires a 
priori knowledge of a system's behavior in order to select candidate 
predictors, has been used successfully in weather prediction, where the 
term ``Model Output Statistics (MOS)'' describes the products. The 
predictor fields can be model counterparts of the desired quantity 
(i.e., a model's grid-cell temperature can be used as a predictor of 
temperature at a specific location, e.g., a weather station), or the 
predictors can include other model variables such as wind, humidity and 
cloud cover from the target location's grid cell and/or from upstream 
grid cells. This approach has significant potential to meet user needs 
for site-specific scenario information, but it has not been applied 
extensively in Alaska.
    The reduction of the uncertainty in climate projections from global 
models is essential for the validity of applications such as 
downscaling, whether dynamical or statistical. While global models are 
improving over time (Reichler and Kim, 2008), a promising area for 
advancement is the selection of subsets of models that are most 
credible for the application at hand. In the case of Alaskan climate 
simulations, several global climate models used in the IPCC Fourth 
Assessment capture the present climate (including its seasonal cycle) 
more successfully than other models. Preliminary studies indicate that 
a composite over a subset of the best 5-7 models (out of the total of 
20-25 available models) provides the greatest skill in simulations of 
Alaska, the Arctic and the Northern Hemisphere. These models tend to 
project larger changes of temperature and precipitation over Alaska for 
the remainder of the 21st century. In this respect, selection of models 
based on quantitative metrics of performance can reduce the uncertainty 
of future climate projections. Such activity should be a high priority 
for user services provided by the climate modeling community.
    A high priority in climate research is the tailoring of model 
output to include variables and information most relevant to the needs 
of planners and stakeholders. The variables carried by climate models 
are not always the ones that correspond to user needs, which can 
include (for example) the firmness of the ground for overland 
transportation; snow cover characteristics; vegetative dryness during 
fire season, etc. A recent illustration of such needs is the attempt by 
P. Larsen (Nature Conservancy) to estimate the economic risks to public 
infrastructure in Alaska as a result of climate change in the coming 
decades. While global model uncertainties limit the robustness of such 
estimates, an even greater limitation is the availability of variables 
beyond temperature and precipitation. Infrastructure such as roads and 
buildings will clearly be affected by changes in the freeze-thaw 
cycles, snow loads, temperature extremes, peak-wind events and 
occurrences of flooding. There has been little effort to translate 
model output for Alaska into these quantities that are most relevant to 
infrastructure risks as well as to other concerns of users. The 
bridging of models and user needs is an emerging area of activity, and 
it is intertwined with the need for site-specific (downscaled) climate 
projections and for reduced uncertainty in climate model output.

References
    Daly, C., M. Halbleib, J.I. Smith, W.P. Gibson, M.K. Doggett, G.H. 
Taylor, J. Curtis, and P.A. Pasteris, 2008. Physiographically-sensitive 
mapping of temperature and precipitation across the conterminous United 
States. International Journal of Climatology, DOI: 10.1002/joc.1688.
    Euskirchen, E.S., A.D. McGuire, and F.S. Chapin III. 2007. Energy 
feedbacks of northern high-latitude ecosystems to the climate system 
due to reduced snow cover during 20th century warming. Global Change 
Biology 13, 2425-2438.
    IPCC, 2007. The Physical Basis of Climate Change. Working Group I, 
Fourth Assessment Report of the Intergovernmental Panel on Climate 
Change, Cambridge University Press, Cambridge, U.K., 996 pp.
    Reichler, T., and J. Kim, 2008. How well do coupled models simulate 
climate? Bulletin of the American Meteorological Society, 89, 303-311.
    Stroeve, J., M.M. Holland, W. Meier, T. Scambos and M. Serreze, 
2007. Arctic sea ice decline: Faster than forecast. Geophysical 
Research Letters, 34, L09501, doi:10.1029/2007GL029703.

    Senator Kerry. Thank you very much, Dr. Walsh.
    It was very helpful, all of you, and I might add here and 
there a little baffling and confusing as to how we make 
layman's sense out of what we need versus what your hopes are 
and requests are.
    I gather, in listening to you, that the principal obstacle 
that we have got is a human resource pressure. We have this 
whole issue of high-end computing, adequate access to it, which 
we just heard about. The appropriateness of software, software 
standards, protocols, et cetera, and what they are going to be. 
More observational data, which you have all said we have got to 
have, and obviously, adequate research funding.
    What I am trying to figure out, as I listen to this, and 
perhaps you can help us understand. Give it to us in relative 
terms. We have now had several rounds of the IPCC. We are 
gearing up globally to make certain decisions. Now, obviously, 
Dr. MacDonald, you would like resolution, and Dr. Walsh you are 
talking about that in terms of the ability to be able to really 
predict something for a community.
    To what degree are we able now to adequately predict in a 
way that allows public policymakers to make a smart decision? 
Is there an element of guesswork in this now? Is there a 
sufficient level of accuracy that you can make some 
predictions, and then we need to go further to make the others? 
Who wants to tackle that? Dr. Sarachik?
    Dr. Sarachik. I think right now we are able to make certain 
predictions. We are able to say certain things. For example, 
there was a recent paper by a group out of Scripps talking 
about, as time goes on, the snowpack will decrease. Those are 
firm results because they are based on science. As it gets 
warmer, you have less ice, and therefore you have less 
available water for irrigation and various other things because 
the snowpack serves as reservoirs.
    On the other hand, we cannot make very fine-scale decisions 
because we do not have very fine-scale information. And a lot 
of our people want to do things at the watershed level. Puget 
Sound, for example, an estuary, is a climate regime in itself 
that has a lot of very unique problems that simply cannot be 
dealt with until we can make predictions of variability and 
actual climate for the coming year on that scale, and we cannot 
do that right now.
    Senator Kerry. And to do that, what will it take?
    Dr. Sarachik. I think it takes a balanced program of--sorry 
to say this--observations, modeling, and research because I do 
not think one of them can be allowed to get ahead of the 
others. They are all necessary and you cannot make progress 
without doing them all.
    Senator Kerry. Who will define the balance?
    Dr. Sarachik. I think you can define the balance by asking 
the National Academy to define that balance. So far, nobody has 
really thought about what the progression is going to be, and 
at the moment everybody is arguing for his own little 
specialty, but nobody is arguing for everything going together. 
And that is what is really necessary.
    Senator Kerry. The answer to that means you have got to 
have resources in every sector. That is a resource-based 
request.
    Dr. Sarachik. I think it is a resource-based request and it 
is an organizational problem for Government because government 
is not presently organized to do this very well. There have 
been continuing complaints about the United States Global 
Change Research Program, for example, which is not able to 
focus money on problems because agencies have different needs 
and requirements. Therefore, you cannot really solve some major 
problems.
    One of the outstanding problems has been decadal 
variability. We do not have a program on decadal variability 
despite its being recommended for a very long time. It is 
absolutely crucial for both understanding ENSO and for 
understanding global warming.
    Senator Kerry. For understanding what first?
    Dr. Sarachik. Global warming and El Nino Southern 
Oscillation.
    Senator Kerry. Can you give us the kind of dollar-to-result 
connection, or is that completely speculative?
    Dr. Sarachik. We have actually done this in various 
committees of the National Academy that I have served on, and 
the answer is if we want the national climate service which 
will give us the largest scale of information correctly, the 
observations, modeling on the large scale that will interact 
with offices on a small scale, it would be on the order of $1 
billion a year.
    Senator Kerry. $1 billion a year. Where are we now?
    Dr. Sarachik. It is a little hard to say because each 
agency defines climate in different ways and, therefore, cannot 
bring resources to bear on problems.
    Senator Kerry. With respect to that, anybody else jump in 
when you want to here. I am not just targeting one person.
    Does that mean that we need to pull the effort of all these 
agencies under one roof? Would that serve the agency interests 
adequately if we did that?
    Dr. Sarachik. Or within a single agency. I do not know of 
any weather service which is not a single agency throughout the 
world. There are hundreds of weather services throughout the 
world, and they are all separate agencies I would think. A 
climate service should be either a separate agency or within a 
single agency in order to be able to accomplish a single goal.
    Senator Kerry. Dr. Reed, do you want to weigh in?
    Dr. Reed. Just to echo that comment. That challenge is 
mirrored on the side of computing research and infrastructure 
as well. It is scattered across many agencies. There is a loose 
confederation of programs, and as I alluded in my testimony, 
one of the perpetual recommendations of the community is 
tighter coordination of those activities to focus on the 
underlying R&D that enables the computational science, of which 
climate change is one example, but also the procurement process 
for open scientific research facilities that support problems 
like this.
    Senator Kerry. How far are we, in your judgment, on the 
current scale of what we are putting in and the current rate of 
progression in modeling? The modeling, you would all agree, is 
better today than it was 5 years ago, and that is better than 
it was 10 years ago. So we are making some progress. We are not 
making all the progress you would like to make. Is the progress 
we are making sufficient to responsively address the concerns 
of people like Mr. Carlisle, Dr. Walsh, and others who are 
trying to make decisions at a local level? Or is there a 
correlation here between the amount of money, energy, effort, 
leadership that ought to go in to accelerating the 
supercomputing capacity and the other things so that we are 
getting better real-time results?
    Does anybody want to take it? Yes, Dr. MacDonald. You can 
all have a shot at it.
    Dr. MacDonald. Senator Kerry, our scientists at GFDL feel 
that we do need a balanced program, but we are at a point where 
significantly higher resolution models really would give us 
much better regional information, and they made that point that 
it was not just that they thought that, that they had been able 
to test that concept, instead of running at 130 miles, running 
at like 30- and 40-mile resolution. So they tested that out, 
and that is where they got this figure that you see.
    Senator Kerry. What does it take us to get there?
    Dr. MacDonald. The increase by a factor of four in 
resolution with better physics takes something like 100 times 
as much processing as today.
    I want to add one more point.
    Senator Kerry. What is the limitation on that? Is it just 
the commitment to the funding or is there a technical 
limitation?
    Dr. MacDonald. I do not think there is a large technical 
problem since they have proved that they are able to do it. So 
I think it is just access to the computing.
    And one more point.
    Senator Kerry. Do you agree with the figure that Dr. 
Sarachik gave us, $1 billion a year?
    Dr. MacDonald. I guess to understand that figure, I would 
have to know what all do we include, whether we include all of 
our satellite systems and so on. So I cannot comment on it.
    But I did want to make an additional point that part of the 
reason I think we can advance rapidly in climate modeling is 
that we did the same thing in weather models. When I started my 
career, they were really poor on precipitation and we got 
higher and higher resolution. They got better and better. I 
guess some would argue with that, but we think they are a lot 
better.
    Senator Kerry. I saw another hand. Yes, Dr. Walsh.
    Dr. Walsh. I would like to pick up on this analogy to the 
weather models. I realize weather and climate prediction are 
certainly different beasts, but I think that there is a lesson 
we can learn from the weather prediction community that may 
help us accelerate progress toward meeting the user needs. That 
is the statistical adjustment of model output. The adjustments 
are made for specific locations based on algorithms that have 
been developed using observational data. Now, that type of 
approach shortcuts to some extent the slow progress in model 
resolution and in model capabilities.
    But there are two requirements there. One is for an 
integrated observational system that is truly integrated with 
what the modeling community is doing, and the second is it 
requires some coordination, some organization in the national-
level program, whether it is through multi-agency or not. It is 
not going to happen piecemeal.
    Senator Kerry. Senator Stevens?
    Senator Stevens. Well, thank you very much. I woke up this 
morning and thought about this problem that just hit Myanmar, 
whatever we call it now.
    Senator Kerry. Burma.
    Senator Stevens. Do you remember, Dr. Walsh, we had a 
typhoon off of the northern coast of Alaska? We had cyclones, 
hurricanes. We seem to be unable to predict these things, and I 
think the total damage that comes from not being able to 
predict them are fairly obvious to everyone.
    Have we used space enough in terms of getting that data for 
you all to use in your computers? Goldwater used to believe we 
could get a lot of information by just observing what is going 
on on the globe as a whole rather than just one spot. Have we 
ever proceeded on any of this?
    Dr. Walsh. The satellite information seems to have been 
most useful in the weather prediction arena. In the sense that 
climate includes the statistics of weather over time, I think 
we may have more of a challenge in incorporating the satellite 
information into an enhancement of the climate models. I see 
the payoff more in the weather prediction arena.
    Senator Stevens. Dr. MacDonald, my staff thinks I ought to 
ask you the question of whether the budget this year will keep 
NOAA on track in obtaining supercomputing resources for 
improved climate modeling. Do you have enough money?
    Dr. MacDonald. Senator Stevens, we have been able to get 
the support that we need, and I think our climate modeling is 
going quite well. And what we are talking about here is kind of 
the next big jump up in the 5 to 10 year timeframe.
    Senator Stevens. This is not the place to get into it. I 
have been worried about the predictions that many people are 
relying on in terms of some of these, for instance, the IPCC 
because of what went into their computers. Do we have enough 
reliable information about the past to really feed the stuff 
into the computers as we are doing now?
    Dr. MacDonald. I think within NOAA and elsewhere, there are 
a lot of programs where we do try and look at not only the last 
100 years but the last 1,000 years. It is tough work. You are 
drilling into glaciers and trying to see what it was. And that 
has taught us a lot about our models. I think if we look 
objectively, our models have improved greatly from the 1990s 
and I think it is partly because we were able to look at the 
past and see what it was. So I am kind of an optimist to think 
we are making great progress, but the need for understanding 
climate is so great, that we are really looking for trying to 
get much better.
    Senator Kerry. Can I just extrapolate on that? What is the 
level of accuracy about the longer-term predictions of 
consequences of climate change? I think that is part of what 
Senator Stevens was asking. Has enough data gone in here that 
is good data to be able to say the sea level rise is accurate, 
that the vegetation migration is accurate, these expectations 
that we are now factoring in?
    Dr. Sarachik. I think you can say that on a large enough 
space scale. The problem is that that is very good for the 
IPCC, which is interested in mitigation. To simplify, 
mitigation is global, but adaptation is local. We do not have 
that sort of information on a local scale.
    Senator Kerry. You just do not know where that is going to 
happen, but you know it is going to happen.
    Dr. Sarachik. It is not clear to me it is going to happen 
because I think it depends on the health of the climate 
community. I work in the trenches, or I used to. I have had 10 
students getting Ph.D.'s, five of whom are no longer in the 
field because there were no opportunities for them in the 
field. This is a field which is simply not providing enough 
opportunities because there is not enough money. I do not know 
what it is like for government organizations, but the money is 
not working its way down to the universities where a lot of 
this work is done and a lot of people are trained.
    Senator Stevens. Dr. Sarachik, again a question. I am told 
the NOAA research program has been active for more than a 
decade. That program in Alaska was started in 2006. What are 
the challenges in translating scientific research and complex 
modeling, particularly the data, information that can be used 
by people, by just ordinary citizens?
    Dr. Sarachik. We deal with that pretty much every day in my 
center, and they need information that is translated into 
resource predictions. We, for example, use whatever climate 
information we can get from the models. We correct the models 
as best we can. It is not a very well defined procedure. Then 
we make hydrological predictions for things like stream flow, 
and from stream flow, we can make energy predictions, because 
most of our energy is hydropower, and salmon predictions and 
water availability predictions, irrigation, and agricultural 
predictions. That is the sort of thing we need to do. It is not 
being done well enough because the global models are not good 
enough, and it is not being done in enough places in the United 
States.
    Senator Stevens. Well, I guess I could not get too 
specific. And Dr. Walsh might correct me on this if I am wrong, 
but it is my understanding that few of the computers that are 
producing information that looks at the Arctic for the future 
have taken into account the vast amount of Atlantic water that 
has gone into the Arctic Ocean at a fairly deep level and that 
that water has been very warm and that the thawing in our area 
has been from the bottom up, not from the top down, not from 
warming on top, but from the warm water that is coming from the 
Atlantic Oscillation and bringing up more warm water than ever 
before. And theoretically, it may stop at any time. It may 
reverse and go back to its normal pattern.
    How do computers figure that in? The net result of the 
computers today say ice is going to disappear in our Arctic 
Ocean by 2020, 2030. Our people dispute that. As I understand 
it, we believe it is thinning and we are going to lose summer 
ice, but we are not going to be ice-free as these computers 
predict. How do we really get any balance in terms of the 
public information as to the results of something like these 
computers that are fed one basic group of statistics and other 
statistics that are not made available to them?
    Dr. Sarachik. I think all of the models currently being 
done for the IPCC process do include deep water coming up, 
thermohaline circulation, if you will. The fact that the models 
do not agree among themselves is an indication that we are a 
long way from making reasonable predictions.
    The policy issue is, how much do you need in order to make 
decisions? How much do you need to know about the future? And 
the future is always murky. The more we know about the future, 
the better those decisions will be.
    My attitude toward models is it tells us the range of 
things that could happen. It does not necessarily tell us what 
would happen. And that is the best we can do at the moment, but 
I think we can do better.
    Senator Stevens. I thank you.
    John, what coordination now exists between our 
International Arctic Research Center and those entities that 
are producing these global climate models that we are hearing 
so much about? Are they really feeding in some of the 
information that you have gathered through the International 
Arctic Research Center now for over 20 years?
    Dr. Walsh. Well, I think you touched on a good example with 
the inflow of Atlantic water into the Arctic Ocean. As Dr. 
Sarachik mentioned, there is a wide range among models and how 
they simulate that inflow. What we need is a good observational 
assessment to pin down which models are doing things right for 
the right reason. So I think what we are pointing to here is 
the need for, again, a coordination between the observations 
and the models. In this case, it is the model assessment side 
of the modeling enterprise.
    Senator Stevens. I do not want to offend my friend here. 
But around here if I criticize IPCC, I am criticizing 
motherhood. And yet, I think that their models are deficient in 
terms of the information base that has been made available to 
us in the Arctic. Am I wrong?
    Dr. Walsh. There was a polar chapter in the IPCC 
assessment. But you are right that it contained very little use 
of the model output and very little critical assessment of the 
models.
    Senator Stevens. And now we face the problem of having the 
polar bear declared endangered because its habitat may be 
affected, and that question of whether the habitat is going to 
be affected comes from this IPCC model that was deficient to 
start with, as far as I am concerned.
    Now, I do not want to put this on my friend from Alaska. 
But what do you do about this, Dr. Sarachik? How do we find 
some way where we can obtain models that the public as a whole 
can rely on without the hype that comes from something like 
IPCC? We do not have hype with the Alaska models, but they have 
been financed, by the way, by the Federal Government for 20 
years.
    Dr. Sarachik. We will never have perfect information about 
the future. There will always be an uncertainty in the policy 
decisions that need to be made.
    What we now know is that there is a possibility of large 
ice depletion in the Arctic. We do not know how much it is 
going to be. We have seen one example of complete melting of 
summer ice in the opening of the Northern Passage. Nobody would 
have expected that 25 years ago, but some of the models, in 
fact, have predicted that, but some of them have not.
    Senator Stevens. Would it surprise you to know that we know 
in history that it has been open before for substantial periods 
of time?
    Dr. Sarachik. If you go back in the geological record, yes, 
of course.
    Senator Stevens. I am talking about in recent history, the 
last 800 years.
    Dr. Sarachik. I do not believe that there is firm evidence 
that the Northern Passage has been open during that time.
    Senator Stevens. Well, it was open several times, as a 
matter of fact.
    Senator Kerry. Where is that documented?
    Dr. Sarachik. The ability to go across the Arctic Ocean in 
the summer which would shorten the distance between Asia and 
Europe considerably and the fact that it was open----
    Senator Stevens. The question is what is open and how long 
it has to be open in order to be classified as being open. But 
very clearly, there have been periods of time when people could 
go from the Atlantic to the Pacific across the top of this 
continent.
    And now it is being predicted that it will be open for a 
period of time, a substantial period of time. Many people 
believe that will be year-round. I am told it will not be year-
round. It might be open for a period of time in the summertime, 
but winter ice is not going to be gone. Would you disagree with 
that?
    Dr. Sarachik. I would plead ignorance because some models 
say that it will and some models say that it will not.
    I think one of the things we should recognize is that a lot 
of--I know this is on the one hand and on the other hand, but 
predictions are not made by models. Predictions are made by the 
emissions that go into the models. So the models simply give 
you the response to those predictions. If in fact we emit more 
greenhouse gases, more CO2, then the climate will be 
warmer and a lot of these things will happen more. The models 
simply describe the response to the emissions of the various 
greenhouse gases.
    Senator Stevens. Mr. Chairman, we mentioned this question 
of earmarks for the last 4 years. Through earmarks we have kept 
four vessels taking statistics on the Arctic Ocean for a period 
of time in the summertime. I do not know how long. We thought 
that would disappear because of the inability to get the 
earmark, but thankfully NOAA has agreed now to finance the same 
concept and keep it going so we get reliable predictions over a 
period of time of what the actual changes are in the Arctic 
Ocean. So I look forward to having accurate observational 
statistics coming at us now in this period ahead of us, and I 
hope we can do that in places where there are areas of real 
controversy like what is going to happen in the Arctic because 
those vessels, being from various nations, are collecting the 
same type of data throughout the Arctic Ocean, which is not 
just a little pond. It is an enormous place. If we can get that 
information and feed it into the computers, we are liable to 
start getting some accurate predictions, Doctor. So I agree 
with you. Stuff in and stuff out. So we want to put the right 
stuff in.
    Thank you.
    Senator Kerry. Can I say to my friend from Alaska--I just 
want to pick up on this--I think the key of what Dr. Sarachik 
just said is you have to look at what the input is to whatever 
the model is that you are looking at. And he said that if 
emissions continue to go up, it will get warmer.
    Now, on the current track that we are on, emissions are 
absolutely guaranteed to go up at an alarming rate. Is that 
correct? Does anybody disagree? Good. And if they go up----
    Senator Stevens. You are talking about CO2.
    Senator Kerry. I am talking about CO2. I am 
talking about all greenhouse gases. Greenhouse gases are going 
to go up. If we reach 600 to 900 parts per million, which the 
current rate of China's and India's and our own pulverized coal 
powerplant production levels are, the ice is going to continue 
to melt. And then the polar bear is going to be threatened.
    So it is a question of your input. You and others have to 
look at the input and make a public policy judgment as a person 
whether you find it accurate and concerning or not.
    I do not disagree. I have always said this, that there is a 
level of inexactitude in the modeling. We cannot tell you 
exactly what is going to happen in a lot of different places, 
but we get a big enough picture, do we not, gentlemen, that 
gives you pretty good indications of trend lines, which as a 
matter of public policy indicates you better take notice or not 
take notice?
    I mean, you have seen these transitions in Alaska. We are 
spending, what is it? $100 million and some to move a village. 
Your permafrost is melting, is it not?
    Senator Stevens. We would like to have that $100 million, 
though. We need $100 million.
    Senator Kerry. And you, sir, have about as much ability as 
anybody here in the Senate to make sure it will happen.
    [Laughter.]
    Senator Kerry. That is why I like sitting by you here in 
this Committee.
    Anyway, the point is made that I think we know we want to 
try to bring this down to a greater level of exactitude, and 
the question I am trying to get at is how rapidly can we do 
that, at what kind of expense. I think it is important that you 
have said we have to do this with much greater coordination. We 
have got to coordinate more effectively, and we have got to 
look in this committee at how you do that. We have to look at 
the question of whether or not you bring this under one roof. 
Correct? We need to get the National Academy perhaps involved 
in how we can best do this is what I am trying to glean out of 
this.
    What else do we need to do as a committee and as a Congress 
to try to get us on the right track here as fast as we can?
    Dr. Sarachik. I have served on a lot of Academy committees 
which have talked precisely about this problem, and this has 
been over a course of the last 15 years I would say.
    Senator Kerry. So are the studies already there?
    Dr. Sarachik. A lot of the studies are there, yes. In 
particular, there was a study called Pathways which was done in 
the late 1990s. I served on that committee. And it described 
the balanced approach to things, and it also objected to the 
way that research was currently being carried out in the United 
States by the USGCRP, the U.S. Global Change Research Program.
    I think climate science has made some tremendous advances. 
We now know that there only seem to be three major phenomena 
that we have to explain, El Nino, Pacific Decadal Oscillation, 
and North Atlantic Oscillation. If we can do that, we can get a 
large amount of the predictable part of climate in the future.
    I do not know of any programs in the United States which 
actually concentrate on that. I have been working on El Nino 
for 25 years, and at this moment, I do not know where I would 
apply for money in order to study that problem.
    Senator Stevens. You have come to the right place because I 
will sure help you if you could find some way to get a program 
that we could finance that would make some sense.
    Dr. Sarachik. I have made recommendations and the Academy 
has made recommendations. For example, I worked on a committee 
about decadal variability. The idea is that the basic problem 
in predicting El Nino was our inability to understand its 
decadal variations, and one of the big problems of global 
warming is the fact that it is being modulated by decadal 
variability. So we recommended only one thing, a program in 
decadal variability. When we presented this to the various 
agencies, they said we cannot do it. So they did not get 
together and form an initiative, which I would have hoped and 
expected that they would do.
    Senator Stevens. I am serious, Mr. Chairman. I wish you 
would really give Congress some recommendations along that 
line. In spite of the earmarks, I think it is high time we 
understood both the Atlantic Oscillation and the Pacific El 
Nino concepts and try to understand why they apparently are not 
there in the southern hemisphere. At least, I have not seen any 
sign of them having a reciprocal effect on, I think, the South 
Pole.
    Dr. Sarachik. Oh, you mean why the----
    Senator Stevens. Why was not the southern hemisphere 
affected the same way? If you go to the South Pole, you will 
find the ice has been piling up there for 40 years and not 
melting at all.
    Dr. Sarachik. --the basic reason for that is that there is 
a circumpolar current in the South Pole which goes all the way 
around, which allows water to come up from the deep. That water 
is extremely cold and will stay cold for a very, very long 
time.
    Senator Kerry. But I understand there was a very 
significant breach in the ice in the Antarctic just recently.
    Senator Stevens. That was because of the weight of the ice. 
It fell off.
    Dr. Sarachik. There is melting of the Antarctic continent, 
but in general, the predictions are that the Arctic will melt 
far more than the Antarctic basically because cold water will 
come up in the Antarctic which does not necessarily come up in 
the Arctic.
    Senator Stevens. Well, why do you not present us your 
recommendations? Maybe we can find some bipartisan way to get 
around this problem of no earmarks. I think that is the most 
significant thing that has come out of this. There really is 
not enough information to know about these oscillations and 
what it does to the North American continent.
    Dr. Sarachik. Correct.
    Senator Stevens. And I would like to join in demanding that 
the money be made available to do so.
    Senator Kerry. Would it have just fallen off if it stayed 
colder? That is OK.
    Dr. Hack, a quick question. Is your center over-subscribed?
    Dr. Hack. We are fully subscribed.
    Senator Kerry. Fully subscribed. And how do you prioritize 
and allocate the time for the computers?
    Dr. Hack. Right now, all of the time at the center is 
allocated through a program called INCITE. It is a program that 
is open to all comers.
    Senator Kerry. How much is allocated toward climate use?
    Dr. MacDonald. Fourteen percent of the total cycles are 
allocated to----
    Senator Kerry. If I could interrupt, let me just say, 
before Senator Stevens goes, if you could get the Committee in 
the next days your specific thoughts about how we address 
Senator Stevens' concern, we will go to work and see what we 
can do here and we will leave the record open. Fair enough?
    Senator Stevens. Thank you. I am sorry I have to leave. 
Thank you very much.
    Senator Kerry. I have to leave in about 5 minutes, folks. I 
have a foreign guest coming in. So I need to get up to that 
meeting.
    Dr. Hack. The center has had a very long history with the 
climate community back in the IPCC days when the AR4 
computations were being done.
    Senator Kerry. But it is competing. It is competing with 
these other interests. Right?
    What I am getting at is, do we need a climate-specific 
supercomputer center?
    Dr. Hack. It would be a tremendous asset to the climate 
community to have something like that I think.
    Senator Kerry. What would that cost?
    Dr. Hack. If you are talking something on the order of a 
petascale type of center, we are probably talking between $50 
million and $100 million.
    Senator Kerry. Where would be the preferred place of siting 
that other than Massachusetts?
    [Laughter.]
    Dr. Hack. Well, Oak Ridge would make a nice place.
    [Laughter.]
    Dr. Hack. But I think the main thing is to see--I think the 
harder thing is to try and coordinate this as an interagency 
question so that the agencies are all on board and one could 
tailor the needs of a center like that to meet all the 
disparate needs of the different agencies that would be running 
on the computer system.
    I just wanted to follow up on a couple of things that have 
come up. And that is that I think a lot of the questions, when 
we are asking about prediction and what is going to happen in 
the future, really come down to uncertainty in the modeling 
frameworks. How certain are the forecasts? And there is no one 
single thing you can put your finger on that is going to tell 
you why they are uncertain. Certainly resolution plays a very 
large role, as Dr. MacDonald showed. We have done our own 
experiments with resolution to illustrate the same sort of 
thing. You cannot capture, for example, orographic 
precipitation accurately with a very course model. That is a 
very simple thing.
    For things like ice, say, sea ice in the Arctic, the 
processes that are embodied in these models, the mathematical 
representations are approximations, and we improve those 
approximations through the observations. And as the 
observations get better, the approximations get better and the 
models get better and the uncertainties are reduced.
    So this is why I think putting one's finger just to say 
that this one magic pill that will solve all these things--I do 
not think that is the right way to look at it. I think that all 
these factors are interrelated. They all rely on one another. 
Computing is certainly as important as the investments in 
modeling and the investments in the observational systems to 
help improve the models.
    Senator Kerry. I guess with any of these models at some 
point you have to be willing to just draw a line and dismiss 
the imponderables, I assume, like the sunspot argument or dust 
storms or the Gulf Stream shuts down and all of a sudden that 
are unpredicted. There is a point, is there not, where you are 
able to take all of the potential variables that people can 
conjure up and adequately address them? Or is there just 
ultimately a level of imponderability here?
    Dr. Hack. I think there are gaps in what we understand 
about the climate system. I think it is encouraging that the 
models are as good as they are at their ability at least on 
global scales to reproduce the observed record.
    Senator Kerry. And the key is just really to look at what 
is going into it, is it not? You then decide, hey, what is the 
probability of this in a sense----
    Dr. Hack. As far as we are going to be able to do is to 
give a sense----
    Senator Kerry.--and take the data on that. We know there is 
going to be X amount of powerplants in China, X amount. We know 
there is going to be X amount of greenhouse gas. We are getting 
pretty good, I assume, at correlating the degrees, the 
Centigrade or Fahrenheit degrees of the warming level according 
to the greenhouse gases. As for forest migration or 
CO2 in the ocean, how do you bring all those 
together?
    Dr. Hack. I am optimistic that these kinds of problems 
can--the noise, let us say, and the uncertainty can be driven 
down with----
    Senator Kerry. How long will it take us to get there? 
Because time is ticking on us. We have got some skeptics around 
still.
    Dr. Hack. I think with a focused effort, goals are 
achievable within the decade.
    Senator Kerry. Within the decade.
    Dr. Hack. And the other aspect of this is the stakeholder 
community, the people I have interacted with. For example, a 
year ago, I was in a meeting with western water judges who were 
looking at a rule on water rights matters. The message was that 
they would much rather have data with uncertainty in it than no 
data at all. And the stakeholder community is a very 
sophisticated, intelligent community. They know how to use data 
that is not perfect, and as long as we make an attempt to try 
and establish what the error bounds are and start to be able to 
address some of the issues Dr. Sarachik talked about with 
regard to low frequency variability in the system, the answers 
that they are getting are going to be tighter and they will be 
of more use to their planning with regard to infrastructure and 
resources, resource management.
    Mr. Carlisle. Senator, I would like to echo that point, if 
I could. A decade is a long time for coastal communities who 
are faced with siting decisions every single day. So we are 
willing to accept a certain level of uncertainty. As long as we 
can frame it and base it back to sound science, we can at least 
start to have informed conversations. And even a little bit of 
information helps. So the sea level rise, with all the 
uncertainties around it--at least we can track that trajectory, 
and that is important. We can start to build in freeboard.
    So one of the things I will make a call for is while the 
modeling at the global scale and regionals is really important, 
we still can use things like high resolution topographical and 
bathymetry data and we can get a lot from that type of 
information. So these are very, very important, but we can also 
make progress while we are going through these long-term 
decadal research improvements to make some progress on the 
ground.
    Dr. Hack. I would like just to make one more statement 
about----
    Senator Kerry. You will have to do it quickly because I 
have got to wrap it up.
    Dr. Hack. That is that the IPCC showed very clearly that 
the models show predictive skill on subcontinental scales, 
certainly on continental scales. So the issue of resolution I 
think does provide an opportunity for a rather substantial 
improvement in predictive skill in the models if we can explore 
that part of parameter space. It is just too expensive and this 
is where the whole computational infrastructure issue comes 
into play. With dedicated facilities, these models can be 
configured to at least explore what the predictive skill of the 
models would be if you were able to run them at resolutions 
that are more typical of weather prediction models.
    Senator Kerry. Well, we will give you the chance in the 
next few days to get in to us what that best practice is going 
to be over the course of these next few years, as I will leave 
the record open. We really welcome that.
    Last question, Dr. MacDonald, just quickly. You talked 
about the three-legged stool, the increased computation 
measurement, et cetera. Are each of those legs equal today?
    Dr. MacDonald. I think that they are equal, and we are 
investing in them. We are trying to get the climate sensors 
onto NPOESS. We are trying to get the really big increases in 
computing, and we have expeditions up to the Arctic to try to 
understand what is happening with the ice. So they are equally 
important.
    Senator Kerry. And do we need to make an equal amount of 
advancement in each of them in order to get this level of 
predictability we want, or is there one more than the other 
that we ought to be focused on?
    Dr. MacDonald. No. I think of it as equal. That is why we 
like using the example of the stool. You cannot have one leg 
that is a lot longer. It is not a very good stool if it is.
    Senator Kerry. So we are back to our balance. Good enough.
    Folks, we could spend more time. I unfortunately cannot, 
not because I do not want to. Is that a vote we have on? Well, 
that is another reason we cannot.
    I am greatly appreciative. It has been very, very 
interesting certainly, and we will leave the record open for a 
week to allow any other colleague who wants to submit a 
question and to get your response back. And we thank you again. 
I know you have traveled a distance. Enjoy Washington for a day 
or so. And thank you all very, very much. I appreciate it.
    We stand adjourned.
    [Whereupon, at 4 p.m., the hearing was adjourned.]

                            A P P E N D I X

   Response to Written Questions Submitted by Hon. John F. Kerry to 
                           Bruce K. Carlisle

    Question 1. You emphasize the need for a national strategy on 
climate modeling to be coordinated with state, regional and local 
partners. How should states and cities inform the development of these 
models and the products generated based on the models?
    Answer. State coastal management programs are primarily using 
models and products from the Intergovernmental Panel on Climate Change, 
academia, and the National Oceanic and Atmospheric Administration. As 
these and other entities look to fine-tune and expand their models or 
begin working on next generation, they should engage with ``end-users'' 
at the regional, state, and local levels to assess their needs and 
identify opportunities for pilot applications. State governments are a 
very effective level to start at as many state programs have existing 
mechanisms for communicating, coordinating, and working directly with 
counties, cities, and towns. State coastal programs, in particular, 
have a demonstrated track record of working in close coordination with 
both Federal agencies and local communities to successfully provide 
high-quality products, services, and hands-on assistance to 
constituents in and beyond the coastal zone. The Coastal States 
Organization (CSO) serves as a central mechanism to coordinate input 
from and collaboration with state programs. In addition, ICLEI--an 
international association of local governments--represents another 
venue for communicating local needs. If modelers and product developers 
need to distribute and/or translate data into local tools and 
strategies, CSO can also work with the state programs and ICLEI to 
increase local awareness and implementation for real world change.

    Question 2. What information does the state need to advise cities, 
towns and citizens about the impacts of climate change and how they can 
prepare and adapt?
    Answer. For coastal communities, municipalities and citizens need 
to be aware of increased vulnerability to storms and sea-level rise. 
Therefore, information on the potential magnitude of impacts--including 
increased flooding, shoreline erosion, saltwater intrusion into fresh 
water aquifers, invasive species, harmful algal blooms, and the loss of 
coastal habitats such as beaches and marshes--within the next 20 to 50 
years is paramount for states to provide technical assistance to 
communities for effective climate change adaptation planning and 
implementation. Within these issue areas, high-resolution topographic 
and bathymetric elevation data are required, to be coupled with region-
specific tide data, sea level rise projections, and other key 
parameters in order to identify the areas and resources most vulnerable 
to accelerated sea level rise

    Question 3. How should that information be delivered to end-users?
    Answer. State coastal programs have the ability to work with the 
scientific community hand-in-hand to tailor high-quality products, 
services, and hands-on assistance to best suit the needs of both state 
and local decision-makers and resource managers. Massachusetts has 
found that concurrent, targeted outreach and technical assistance is 
essential to successful implementation. To that end, the Massachusetts 
Office of Coastal Zone Management has developed the StormSmart Coasts 
program, which is designed to give local decision-makers, and 
ultimately businesses and homeowners, information and tools on coastal 
resiliency through a user-friendly website, fact sheets, workshops, and 
direct technical assistance. We have received extensive positive 
feedback from municipalities, acclaim from national organizations, and 
interest from a multitude of state programs. A national version of 
StormSmart Coasts could be used to communicate current information on 
climate modeling.

    Question 4. In your written testimony, you discuss the state's role 
in developing high-resolution shoreline change data and coastal high-
hazard zone delineations. Are you capable of projecting that 
information into the future, given the likely impacts of global climate 
change?
    Answer. Shoreline change data and identified flood- and erosion-
hazard areas are extremely critical for coastal managers. However, 
identification of current and future risk zones is limited by the state 
of the science as well as our lack of resources to apply current 
scientific understanding. Erosion rates in Massachusetts and across 
much of the Nation have been increasing as a result of human 
alterations, changes in sediment supply, increasing frequency of 
storms, and sea-level rise, therefore, funding for updates of shoreline 
change data every five to 10 years is critical. More accurate and up-
to-date flood-hazard maps are also critical.

    Question 5. What additional information do you need in order to 
make those projections?
    Answer. Additional topographic and bathymetric data are needed by 
all coastal states. These data are often limited to sparse coverage 
over oceanfront shorelines and do not extend into bays or estuaries, 
where impacts will be experienced. Increased resolution of the 
following models is also essential:

   Sea-level rise--Coastal states will need more detailed and 
        complex models that incorporate local, embayment-scale changes 
        in coastal geomorphology, hydrological conditions, and human 
        alterations and responses (e.g., seawalls and beach 
        nourishment).

   Storm surge--Models that incorporate the unique 
        configurations of local embayments or coastline morphologies, 
        water depths, and physical features such as bridges and roads 
        are required.

   Sediment transport, wetland changes, and river hydrology--
        More information is needed to better understand erosion and 
        deposition cycles, improve our ability to predict changing 
        sediment transport, accretion and erosion regimes.

   Ground water and salt water--More information is required on 
        climate induced changes to local hydrologic cycles through 
        altered precipitation, evapotranspiration, and soil moisture 
        patterns.

    Atmospheric models would provide some input to the above, but they 
are at scales not directly useful to state coastal managers.
                                 ______
                                 
  Response to Written Questions Submitted by Hon. Daniel K. Inouye to 
                    Dr. Alexander (Sandy) MacDonald

    Question. Are regional climate models ready to be run today? If not 
what needs to be done to get them to the point where they can be used 
and deliver adequate information?
    Answer. Regional climate models with resolutions of 50 kilometer 
(km) and finer have been developed within NOAA and other U.S. 
Government agencies, and are ready to be used for regional projections. 
In the recent Intergovernmental Panel on Climate Change (IPCC) Fourth 
Assessment (AR4), NOAA and other U.S. Government agencies used a 
climate model with ocean resolution of 100 km and atmospheric 
resolution of 200 km. Since then, U.S. Government scientists have 
developed and validated models with much finer resolution (e.g., 50 km 
resolution in the atmosphere and 10-25 km resolution in the ocean). 
Implementing these new, finer resolution models to produce 
comprehensive climate projections for reports such as the IPCC Fifth 
Assessment Report (due out in 2013) would require a 100-fold increase 
in computer capacity, an estimate compatible with that reported in the 
2004 Federal Plan for High-End Computing: Report of the High-End 
Computing Revitalization Task Force (http://www.nitrd.gov/pubs/
2004_hecrtf/20040702_hecrtf
.pdf). NOAA is exploring a Memorandum of Agreement (MOA) with DOE to 
address HPC requirements collaboratively. This MOA would apply to NOAA 
use of DOE computing for prototyping models for climate research.
                                 ______
                                 
     Response to Written Questions Submitted by Hon. John Kerry to 
                    Dr. Alexander (Sandy) MacDonald

    Question 1. In your written testimony, you discuss the smaller-
resolution prototype models NOAA has developed. However, you note that 
NOAA's computer resources are inadequate to run comprehensive 
simulations of climate change using these models. What level of 
computing resources do you need? How much does that cost? Should these 
resources be centralized at NOAA, or is it appropriate to have several 
computing centers that can run these advanced models?
    Answer. The Nation's climate mission requires dedicated support for 
large scale High Performance Computing (HPC). The Administration's FY 
2009 Budget Request for NOAA provides a set of priorities to sustain 
core mission services and address some of our highest priority program 
needs. Roughly $19 million of the FY 2009 President's budget request 
for NOAA is for climate research HPC. Currently, these funds are 
allocated to the following high-priority activities:

   The development and application of the next generation of 
        climate change and Earth System Models, in preparation for the 
        IPCC Fifth Assessment Report (AR5; due out in 2013), including 
        new atmospheric and global ocean component models.

   Support for climate modeling requirements in developing the 
        Climate Change Science Program Synthesis and Assessment 
        products.

   Computational support for the World Meteorology 
        Organization/United Nations Environment Programme Stratospheric 
        Ozone Assessments.

   Computational support for developing a modeling capability 
        for monitoring and making predictions of Atlantic Meridional 
        Overturning Circulation changes.

   Continued limited integrations of high-resolution 
        atmospheric models to support the North American Climate Change 
        Assessment Program.

   Support for reanalysis and reforecast of 1979-2008 using the 
        coupled Climate Forecast System.

    NOAA's global climate models, as well as other U.S. models, are 
among the best in the world. Currently, the HPC available to the 
Nation's climate scientists allows global climate models to resolve 
climate research questions down to the scale of continents. Additional 
research HPC capacity for climate would be targeted toward using 
currently available higher resolution models to meet stakeholder demand 
for regional to local scale climate information. The additional HPC 
would also be used to produce more comprehensive climate outlooks with 
advanced models that improve treatments of processes critical to our 
understanding of climate change, such as aerosols and clouds. These 
advanced models would also include processes that are missing in 
today's models, such as ice sheet melting that is crucial to address 
sea-level rise. Another example of what advanced models would include 
are complex biogeochemical cycles that can be applied to answer 
questions about the carbon cycle and interaction of climate and 
ecosystems, such as the effects of ocean acidification.
    In July 2003, the Climate Change Science Program specifically 
identified two centers, NOAA's Geophysical Fluid Dynamics Laboratory 
and the National Centers for Atmospheric Research (NCAR), to produce 
sophisticated simulations, such as those required for assessment by the 
IPCC. Scientific uncertainty, numerical algorithm variations, non-
unique parameterizations of sub-grid size phenomena, and gaps in 
knowledge make it essential that multiple models be used to explore 
different approaches to improve understanding of the climate of the 
global integrated Earth system. At this time, NOAA is exploring 
partnerships with the Department of Energy and NCAR to identify the 
most cost-effective solution for facilities to house the Nation's 
climate computing. Should these activities be successful, leveraging 
these national partnerships and adopting a phased approach to 
implementing the required level of computing represents an executable 
strategy for meeting the Nation's growing climate information needs.

    Question 2. In your written testimony, you discuss NOAA's Modular 
Ocean Model. Are we capable of modeling the oceans with the same level 
of confidence that we model the atmosphere? Do we need more ocean 
observations to feed into those models?
    Answer. At the global scale, the ocean's role in climate change is 
governed by well understood scientific principles which are suitably 
represented by the present class of climate models. The NOAA 
Geophysical Fluid Dynamics Laboratory's Modular Ocean Model (MOM) is 
the world's most widely used numerical model for simulating the ocean 
circulation at the global scale and for understanding and predicting 
ocean climate phenomena. MOM is used for operational seasonal 
(including El Nino) forecasting at NOAA's National Weather Service, and 
was prominently used by several groups in the U.S. and worldwide in the 
recent IPCC Fourth Assessment Report.
    Uncertainty remains, however, when asking questions about regional 
spatial patterns and precise time scales of the ocean's response to 
climate change (e.g., how fast and how much will the Massachusetts 
coastal waters warm and the sea levels rise?). Such regional questions 
represent a grand challenge to be addressed by the next generation of 
global climate models.
    The geography of the world's ocean basins is extremely complex, 
with many relatively small scale features (e.g., continental shelves, 
narrow straits, and marginal seas) playing an important role affecting 
key features of large scale ocean properties (e.g., heat, salinity, 
nutrients). In addition, the spatial scales for ocean ``weather 
eddies'' is roughly 10 times smaller than the atmospheric weather 
eddies, thus making it roughly 10 x 10 times (factor of 10 for each of 
the two horizontal directions) more computationally expensive to 
represent ocean eddies in a numerical simulation. These two 
characteristics of the ocean underscore the benefits of model grid 
resolution finer than 10 km resolution, to address questions of 
regional climate impacts, including those most pertinent to the U.S. 
coastal zones.
    An ocean model is evaluated by confronting simulations with ocean 
observations. This evaluation in turn provides feedback to observing 
system design (i.e., do we need more observations, and if so, where?). 
The scientific reliability of global ocean climate simulations will 
match the level of atmospheric simulations through: the development of 
refined resolution global ocean climate simulations; targeted ocean 
field studies, observations, long-term monitoring; and theoretical 
studies, which enable a rigorous assessment of the models based on the 
real ocean system.

    Question 3. In your written testimony, you note that NOAA makes 
large amounts of your climate model output freely available. Is this 
information accessible only to advanced researchers, or are end-users 
able to access and utilize this data? Is the information available in a 
format that is useful for end-users?
    Answer. NOAA is committed to making our climate model output 
available to the public. With respect to access, the NOAA Operational 
Model Archive and Distribution System (NOMADS), provides open access to 
climate model output. NOAA's Geophysical Fluid Dynamics Laboratory 
modeling center provides climate data on the NOMADS publicly accessible 
data portal, and works directly with researchers to facilitate use of 
the data. Because information portals and access systems work best when 
we also invest in partnerships with decisionmakers, NOAA also works 
directly with end-users to help them interpret model projections in a 
manner useful to their needs. There are different types of users of our 
climate model output: climate researchers; researchers who study the 
impact of climate change on various sectors (e.g., agriculture, public 
health, air quality, water resources, migration, international 
security, travel, trade); and the engaged public (e.g., policymakers, 
urban planners, state and regional resource managers, or even curious 
students). Some examples of NOAA working successfully with different 
users include:

   Department of Energy's Program for Climate Model Diagnosis 
        and Intercomparison. Through this program, NOAA climate model 
        output from simulations of past, present and future climate was 
        used to prepare the IPCC's Fourth Assessment Report on Climate 
        Change.

   NOAA's Earth System Research Laboratory (ESRL), and the 
        Climate Program Office's Regional Integrated Sciences and 
        Assessments (RISA) Program generate regionally downscaled 
        projections of future climate change. Through sustained 
        interaction with stakeholders, ESRL and RISAs also provide 
        regionally tailored analyses that transform the global climate 
        projections into value-added decision-relevant information.

   The National Integrated Drought Information System Drought 
        Portal, an interagency effort coordinated by NOAA, provides 
        valuable information to stakeholders such as: early warning 
        about emerging and anticipated droughts; assimilated and 
        quality controlled data about droughts; model-based drought 
        outlooks and forecasts; information about risk and impact of 
        droughts to different agencies and stakeholders; information 
        about past droughts for comparison and to understand current 
        conditions; and explain how to plan for and manage the impacts 
        of droughts.

    NOAA, with its mission to act as a research and information service 
on environmental issues, is uniquely poised to serve the range of 
climate data needs, from researchers to end-users.

    Question 4. Given likely investment and innovation in computing 
infrastructure, when would data from the next generation of climate 
models be available to end-users and researchers?
    Answer. NOAA is committed to sharing climate model data with end-
users and researchers as rapidly as possible. However, before data can 
be shared, the data must be verified and validated for scientific 
credibility by peer review, and packaging and quality assurance tasks 
must be completed. As in the past, NOAA prioritizes computing 
infrastructure acquisitions for the following: simulations and analysis 
to understand and project climate change; archival storage of large 
volumes of data generated by these simulations; and for networking, to 
deliver the data to our partners and stakeholders. NOAA modeling 
centers have long-term experience in acquiring and maintaining a 
balanced infrastructure with available resources.
                                 ______
                                 
   Response to Written Questions Submitted by Hon. John F. Kerry to 
                            Edward Sarachik

    Question 1. The National Research Council (NRC) released reports in 
1998 and 2001 stating that the United States lagged behind other 
nations in our ability to model climate change. As chair of the 2001 
NRC Panel, do you believe that our capacity has improved since these 
reports were released? What improvements have been made? What 
challenges does the U.S. climate modeling community still face?
    Answer. Our capacity for the kind of climate modeling that the IPCC 
does has indeed increased--we now have higher resolution coupled 
climate-biogeochemistry models. But the kind of models the IPCC does is 
in support of the Framework Convention on Climate Change, and this 
involves getting global averages correct in order to avoid dangerous 
interference with the climate system. On the other hand, the kind of 
modeling that this hearing was about, namely climate modeling for the 
use of decisionmakers and end-users, involves getting the regional 
scales right and this has not improved. The IPCC itself recognizes that 
the results of its models are valid on space scales of 3,000 miles 
(continental scale) and this is not useful for any decisionmaking other 
than mitigation of greenhouse gases. The failure of our major large 
scale modeling institutions (we only have two--NCAR and GFDL) to 
address regional problems is the failure to get climate variability 
correct--annual cycle, El Nino, Pacific Decadal Oscillation and North 
Atlantic Oscillation. The failure to get climate variability correct is 
due to the general inadequacy of our (sub-critical) climate community 
to address new problems--a combination of lack of sustained 
observations, absence of the development of a model-based climate 
analysis (which would serve as the primary material for analysis of 
climate variability), and general inability of the CCSP to concentrate 
resources on research problems outside the direct interest of the 
participating government agencies.

    Question 2. In your written testimony, you emphasize that the 
university research community needs access to the supercomputers 
themselves, as well as access to the information generated by the 
models. What is the best way to facilitate that?
    Answer. It really wouldn't help to simply make supercomputer time 
available since the enormity of coupled climate models is generally 
beyond the capacity of individuals or small groups of individuals to 
deal with. The NCAR Community Climate Systems Model is an excellent 
template, one that has the broad community interacting with a core NCAR 
group--this is the best synergy between model builders at NCAR and 
model users in the distributed community and is far more than the sum 
of the parts. Supercomputer time needs to be made available to enable 
this synergy as well as funding. What the U.S. needs is many of these 
core model building institutions interacting with anyone that has 
something to contribute. At the present time, the Hadely Centre in 
England is funded at more than GFDL and NCAR CCSM combined, this in a 
country with 10 percent of the GDP of the U.S. By this standard, the 
U.S. should have ten or twenty modeling centers each with its own 
supercomputer, most interacting with the external community. Some of 
these centers should be regional centers concentrating on the climate 
problems of the regions in which they are sited.

    Question 3. Several witnesses have emphasized the need to integrate 
observational data into climate models. In your opinion, in addition to 
incorporating this data, do we need additional observational data?
    Answer. What we need is sustained and accurate data in the 
atmosphere, ocean, land and cryosphere adequate to define the long term 
climate patterns: annual cycle, El Nino-Southern Oscillation, Pacific 
Decadal Oscillation, and North Atlantic Oscillation, and physical 
climate impacts: soil moisture, stream flow, and glacier and ice sheet 
evolution. This is a finite (but expensive) aim. It would also be 
desirable to have data on ecological impacts, especially fisheries, 
forests, and disease vectors. We also need to be able to deal with 
opportunities as they arise--at the moment, the melting of the 
Greenland Ice sheet is so much on scientists' and the public's mind 
that one would expect our science agencies to undertake a large and 
concerted effort to measure the melting of Greenland--so far this 
hasn't happened.
                                 ______
                                 
   Response to Written Questions Submitted by Hon. John F. Kerry to 
                           Dr. Daniel A. Reed

    Question 1. In your written testimony, you discuss the need to 
integrate observational data into our climate models. What are the 
limitations in our ability to do that today?
    Answer. Data assimilation, the integration of observational data 
with computational models at each cycle, remains a technically 
challenging problem, both mathematically and technically. We have made 
progress, but much work remains. My climate modeling colleagues can 
best speak to the modeling issues, but I can comment on the 
computational and observational aspects. The diversity and scale of 
data--from current satellites to geological records--make the problem 
complex. Moreover, the spatial and temporal data gaps exacerbate the 
integration problems. Finally, there is simply the matter of scale. The 
volume of data we must manage and integrate grows daily.

    Question 2. How many different climate models do we need, here in 
the U.S. as well as internationally? Should we be collaborating with 
international partners on model development, or is there value in 
having separate models in separate countries?
    Answer. Again, this is a question best answered by my climate 
modeling colleagues. However, it is important to remember that any 
model is an approximation of the actual system. By necessity, each 
model includes simplifying abstractions that may fail to capture 
salient aspects of the real system. These simplifications make the 
model tractable and allow us to evaluate the models computationally in 
a reasonable time on available high-performance computing systems. 
Thus, the accuracy of any model depends on our current knowledge and 
understanding of climatic processes, the skill of the model builders 
and available computing resources. Advances in any of those areas can 
improve model accuracy. For this reason, we often evaluate ensembles of 
models, examining the common and differing behaviors to illuminate 
potential errors.
    Thus, I believe we need multiple models, with differing assumptions 
and approaches, enabled by a broad international collaboration. These 
models can and should be configured and specialized to understand 
regional climatic effects.

    Question 3. You describe the need for an integrated, interagency 
effort to address the range of research, software, data storage and 
computing challenge associated with climate modeling. How should that 
be structured? Should it be led by NSF, NOAA or another agency? Is the 
Global Change Research Program capable of managing such an effort?
    Answer. Many Federal agencies support climate change research, with 
differing scales and approaches. This has historically been the 
strength of the U.S. research funding environment. However, assessing 
the impact of climate change is an outcome-driven activity. I believe 
this is best managed by a single agency with the resources and the 
mandate to deliver detailed global and regional assessments, not a 
basic research agency.

    Question 4. In your written testimony, you say that ``climate 
change modeling is a deep and challenging scientific problem that 
requires computing infrastructure at the largest scale.'' The National 
Center for Atmospheric Research (NCAR) in Boulder is the only 
supercomputing facility focused strictly on climate change, but you 
indicate that NCAR cannot deliver timely and accurate predictions. 
Should we focus on establishing NCAR as the premier climate modeling 
center in the country and expanding our capabilities there? Or do we 
need a structure that supports advanced climate modeling at various 
institutions around the country?
    Answer. We need a balanced approach based on a pyramid model of 
computing. The pyramid apex is one or more premier high-performance 
computing systems for climate change modeling--substantially larger 
than anything available today. However, the apex must be supported by a 
diverse set of smaller systems spread across our universities and 
national laboratories.
    NCAR is one of the possible sites for an apex climate change 
modeling supercomputing center, but there are other viable sites as 
well (e.g., at one of our national laboratories). The site selection 
should be derived from a national analysis of available infrastructure 
(people and facilities), costs and community engagement.

    Question 5. Given the apparent limitations of using off-the-shelf 
parallel processors for the purpose of climate modeling, should we be 
building special-purpose supercomputers for this purpose? If so, is 
there any role for the private sector in developing these systems?
    Answer. Yes, for this and many other proposes of national 
importance. We need greater investment in purpose-built supercomputers 
that have been architected for critical national problems, just as we 
invest in purpose-built defense infrastructure. Climate modeling is but 
one critical example of such a national scientific problem; there are 
many others related to national security, biomedicine, energy research 
and other domains.
    I believe a coordinated program of research, development and 
production must involve government, academia and private industry. In 
the end, the systems will be built by industry if the government is a 
willing supporter and purchaser of the purpose-built systems.

                                  
