The Uncertainty Conundrum in Climate Change Research and Applications

1
The Uncertainty Conundrum in Climate Change
Research and Applications
Linda O. Mearns Institute for the Study of
Society and Environment National Center for
Atmospheric Research Journalism Fellowship
Meeting June 2008
2
To know ones ignorance is the best part of
knowledge Lao Tzu
3
Doubt is not a pleasant condition, but certainty
is an absurd one.
-Voltaire
4
Any clear way, though it lead to death, is
preferable to the tangle of uncertainty.
- Charles Horton Cooley, U.S. sociologist
5
The quest for certainty blocks the search for
meaning.
- E. Fromm
6
Reasons for Research Explosion in Uncertainty

• Developments in climate modeling and computing
  more sophisticated ensembles and much greater
  computer power
• Heightened awareness of and involvement of
  stakeholders (i.e., decision makers)
• Ascendance of risk assessment approaches
• Enhanced interdisciplinarity (e.g., with
  statisticians and risk assessors)

7
Decision-making Taking Centre Stage
8
The Historical Perspective

• Williams, 1978 Carbon Dioxide, Climate and
  Society - IIASA Workshop
• EPRI and USEPA, 1992 Joint Climate Project to
  Address Decision Makers Uncertainties
• 21st Century Examples

9
H. L. Wiser, 1978
In
Scientific Needs of Policy and Decision Makers
Can the uncertainties in data and models be
quantified?
10
EPRI and USEPA, 1992

• Resolution of the uncertainties related to the
  policy-relevant questions is not a pre-requisite
  for action.

11
Early attempt at quantifying uncertainty
Everywhere Here 3 or 4 in Agreement
Kellogg and Zhao
12
The Probability Distribution Race
21st Century

• Probability distributions of what and for whom ?

13
Probabilities for Whom?

• Decision makers (Policy, Resource Management)
  which ones, what spatial scales, autonomous vs.
  planned adaptations, etc.
• Climate Science Researchers - climate modelers,
  impacts researchers (decisions about research)

14
Examples of Probabilistic Uncertainty Analysis

• Allen et al., 2000
• Jones, 2000 - regional risk assessment
  irrigation needs
• Webster and Sokolov, 2000
• Andronova and Schlesinger, 2001 - climate
  sensitivity
• Schneider, 2001
• Wigley and Raper, 2001 - future global
  temperature emissions x climate models
• Stott and Kettleborough, 2002
• Knutti et al., 2002
• Forest et al., 2002
• Palmer and Raisanen, 2002 probabilities of
  extreme precipitation
• Webster et al., 2003
• Giorgi and Mearns, 2003 - probabilities of
  regional climate change
• Mastrandrea and Schneider, 2004 - dangerous
  climate change
• Murphy et al., 2004 mulit-parameter ensembles
  weighting with CPI
• Tebaldi et al., 2004, 2005 Bayesian model of
  regional climate change multi-model ensembles
• Yates, et al., 2006 - distributions of water
  flows in California under future climate
• Tebaldi and Knutti, 2007

15
Global Scale Probabilities

• In the absence of climate mitigation policies,
  the 90 confidence interval for 1990 to 2100
  warming is 1.7 to 4.9 C

Wigley and Raper, Science, 2001
Similar calculation by Webster et al., (2001)
gives a 90 confidence interval of 1.1 to 4.5 C
16
Cumulative distributions of climate sensitivity
IPCC WG1, 2008 Chap 10
17
Approaching the Scale Where People Live

• Regional scale probabilities
• Multi-model ensembles (21 models)
• Bayesian approach statistical model with
  weighting based on bias and convergence Giorgi
  and Mearns, 2003 Tebaldi et al., 2005

18
Probabilistic Information on Climate Change -
Aspen
Tebaldi, 2006
19
Joint Distribution - Bayesian Model
C. Tebaldi
20
Regional Probabilities of Climate Change
21
PDFs are no better than their underlying
assumptionsStudies are providing methods
development, scenarios of PDFs, not actual
forecasts
22
What is the value of probabilistic information to
water resource managers?

• Climate change and water management in the Chino
  Basin, CA
• Characterizations of uncertainty used in
  workshops
• Traditional scenarios without probabilities
• Probability-weighted scenarios
• Scenarios constructed through robust decision
  making methods

Lempert et al.
23
Inland Empire Utilities Agency (IEUA), based in
Chino, CA Faces Significant Water Challenges

• IEUA currently serves 800,000 people
• May add 300,000 by 2025

• Current water sources include
• Groundwater 56
• Imports 32
• Recycled 1
• Surface 8
• Desalter 2

24
Very Preliminary Results

• Traditional scenarios appear to give participants
  much of the information they needed
• Emphasized importance of achieving goals of 20
  Year Plan to address climate change in addition
  to population growth
• But this was their first exposure to climate
  change information
• Probabilities raised potential of low likelihood,
  extremely large shortages
• IEUA has significant adaptive capacity to
  address historic natural variability of
  California climate
• Probabilistic information quickly prompted
  discussion of strengths and limits of adaptive
  capacity

25
Integrated Uncertainty Analysis Water Resources
in Northern California
Decisions on Adaptation Planning (e.g. new
storage infrastructure)
Probabilities of Climate Change
Probabilities of Hydrological Variables
Yates et al.
26
What to do with uncertainties difficult to
describe with probabilities?

• Incomplete knowledge of physical processes
• Model structure (including important feedbacks
  within the climate system)

27
Combining probabilistic and qualitative
information

• Careful process-based expert judgment of
  confidence in regional projections from
  multi-model ensembles
• Plus bias-weighted pdfs
• Method adumbrated in IPCC WG1 Chapter 11
• How to communicate this effectively in decision
  making contexts?

28
Characterizing and Communicating versus
Reducing Uncertainty

• Different parts of the climate change problem are
  in different states of characterization of the
  relevant uncertainties.
• e.g., Climate model sensitivity - 1.5 - 4.5 deg.
  - uncertainties well enough known to work on
  uncertainty reduction?
• Future Emissions of GHGs - still need to be more
  completely characterized. Unlikely to be reduced?
  (The reflexsivity problem)

29
Goal 3 of the CCSP Reduce uncertainty in
projections of how the Earths climate and
related systems may change in the future.
Do we understand what this means and is it an
appropriate first order goal?
30
IPCC AR4 Likelihood Scale

• Virtually certain gt 99 probability
• Very likely 90 to 99
• Likely 66 to 90
• About as likely as not 33 to 66
• Unlikely 10 to 33
• Very unlikely 1 to 10
• Exceptionally unlikely lt 1

31
Communicating the Odds of Temperature Change
M. Webster, 2002
32
Communicating the Role of Policy
Stringent Policy
No Policy
Webster
33
How have policy makers in the US (e.g. governor
of Arizona) reacted to some of the new regional
information about climate change, e.g., that it
is likely that annual mean precipitation will
decrease in the southwest US?
34
(No Transcript)
35
A Few Words on Extremes

• Recent release of U.S. Climate Change Science
  Program Product 3.3,
• Weather and Climate Extremes in a Changing
  Climate (June 2008)

36
Whats New?

• Much more integrated approach
• Why extremes matter
• Societal Vulnerability to Extremes
• Then physical science discussion
• Some new science
• Hurricane rainfall and wind speeds likely to
  increase
• More frequent strong storms outside tropics ,
  with stronger winds

37
Karl et al, 2008
38
Weather and Climate Extremes
Atmospheric Processes
Modeling of Extremes
Trends in Observations
Climate Change
Weather
Impacts and Vulnerabilities
Extreme Value Theory
39
Societal Impacts of and Vulnerability to Extremes

• Identification of extremes significant to society
• Modeling of impacts of extreme events
• Reducing societal vulnerability to extremes
• Understanding vulnerability requires knowledge of
  the behavior and interactions of all systems
  involved in an extreme event.
• E.g., town storm
  flood
• culture meteorology hydrology

40
Climate
Air Quality
Air Pollution Heat Waves
Adaptation Scenarios
Human Health
Vulnerable Populations
Adaptive Capacity
41
END